feat: show number of tokens remaining in UI (#1388)
When using the OpenAI Responses API, we now record the `usage` field for
a `"response.completed"` event, which includes metrics about the number
of tokens consumed. We also introduce `openai_model_info.rs`, which
includes current data about the most common OpenAI models available via
the API (specifically `context_window` and `max_output_tokens`). If
Codex does not recognize the model, you can set `model_context_window`
and `model_max_output_tokens` explicitly in `config.toml`.
When then introduce a new event type to `protocol.rs`, `TokenCount`,
which includes the `TokenUsage` for the most recent turn.
Finally, we update the TUI to record the running sum of tokens used so
the percentage of available context window remaining can be reported via
the placeholder text for the composer:

We could certainly get much fancier with this (such as reporting the
estimated cost of the conversation), but for now, we are just trying to
achieve feature parity with the TypeScript CLI.
Though arguably this improves upon the TypeScript CLI, as the TypeScript
CLI uses heuristics to estimate the number of tokens used rather than
using the `usage` information directly:
https://github.com/openai/codex/blob/296996d74e345b1b05d8c3451a06ace21c5ada96/codex-cli/src/utils/approximate-tokens-used.ts#L3-L16
Fixes https://github.com/openai/codex/issues/1242
2025-06-25 23:31:11 -07:00
|
|
|
|
use codex_core::protocol::TokenUsage;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use crossterm::event::KeyCode;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use crossterm::event::KeyEvent;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use crossterm::event::KeyModifiers;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::buffer::Buffer;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use ratatui::layout::Constraint;
|
|
|
|
|
|
use ratatui::layout::Layout;
|
|
|
|
|
|
use ratatui::layout::Margin;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::layout::Rect;
|
2025-07-31 00:43:21 -07:00
|
|
|
|
use ratatui::style::Color;
|
2025-08-07 00:45:47 -07:00
|
|
|
|
use ratatui::style::Modifier;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::style::Style;
|
2025-07-31 00:43:21 -07:00
|
|
|
|
use ratatui::style::Styled;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::style::Stylize;
|
|
|
|
|
|
use ratatui::text::Line;
|
2025-07-31 00:43:21 -07:00
|
|
|
|
use ratatui::text::Span;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use ratatui::widgets::Block;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::widgets::BorderType;
|
|
|
|
|
|
use ratatui::widgets::Borders;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use ratatui::widgets::StatefulWidgetRef;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
use ratatui::widgets::WidgetRef;
|
|
|
|
|
|
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
use super::chat_composer_history::ChatComposerHistory;
|
|
|
|
|
|
use super::command_popup::CommandPopup;
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
use super::file_search_popup::FileSearchPopup;
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
use crate::app_event::AppEvent;
|
2025-05-15 14:50:30 -07:00
|
|
|
|
use crate::app_event_sender::AppEventSender;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use crate::bottom_pane::textarea::TextArea;
|
|
|
|
|
|
use crate::bottom_pane::textarea::TextAreaState;
|
2025-06-28 15:04:23 -07:00
|
|
|
|
use codex_file_search::FileMatch;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use std::cell::RefCell;
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
2025-08-07 03:29:33 -07:00
|
|
|
|
const BASE_PLACEHOLDER_TEXT: &str = "Ask Codex to do anything";
|
2025-07-12 15:32:00 -07:00
|
|
|
|
/// If the pasted content exceeds this number of characters, replace it with a
|
|
|
|
|
|
/// placeholder in the UI.
|
|
|
|
|
|
const LARGE_PASTE_CHAR_THRESHOLD: usize = 1000;
|
feat: show number of tokens remaining in UI (#1388)
When using the OpenAI Responses API, we now record the `usage` field for
a `"response.completed"` event, which includes metrics about the number
of tokens consumed. We also introduce `openai_model_info.rs`, which
includes current data about the most common OpenAI models available via
the API (specifically `context_window` and `max_output_tokens`). If
Codex does not recognize the model, you can set `model_context_window`
and `model_max_output_tokens` explicitly in `config.toml`.
When then introduce a new event type to `protocol.rs`, `TokenCount`,
which includes the `TokenUsage` for the most recent turn.
Finally, we update the TUI to record the running sum of tokens used so
the percentage of available context window remaining can be reported via
the placeholder text for the composer:

We could certainly get much fancier with this (such as reporting the
estimated cost of the conversation), but for now, we are just trying to
achieve feature parity with the TypeScript CLI.
Though arguably this improves upon the TypeScript CLI, as the TypeScript
CLI uses heuristics to estimate the number of tokens used rather than
using the `usage` information directly:
https://github.com/openai/codex/blob/296996d74e345b1b05d8c3451a06ace21c5ada96/codex-cli/src/utils/approximate-tokens-used.ts#L3-L16
Fixes https://github.com/openai/codex/issues/1242
2025-06-25 23:31:11 -07:00
|
|
|
|
|
2025-05-14 10:13:29 -07:00
|
|
|
|
/// Result returned when the user interacts with the text area.
|
|
|
|
|
|
pub enum InputResult {
|
|
|
|
|
|
Submitted(String),
|
|
|
|
|
|
None,
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
struct TokenUsageInfo {
|
2025-08-07 05:17:18 -07:00
|
|
|
|
total_token_usage: TokenUsage,
|
|
|
|
|
|
last_token_usage: TokenUsage,
|
2025-08-03 11:31:35 -07:00
|
|
|
|
model_context_window: Option<u64>,
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
pub(crate) struct ChatComposer {
|
|
|
|
|
|
textarea: TextArea,
|
|
|
|
|
|
textarea_state: RefCell<TextAreaState>,
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
active_popup: ActivePopup,
|
2025-05-15 14:50:30 -07:00
|
|
|
|
app_event_tx: AppEventSender,
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
history: ChatComposerHistory,
|
2025-06-27 13:37:11 -04:00
|
|
|
|
ctrl_c_quit_hint: bool,
|
2025-07-31 17:30:44 -07:00
|
|
|
|
use_shift_enter_hint: bool,
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
dismissed_file_popup_token: Option<String>,
|
|
|
|
|
|
current_file_query: Option<String>,
|
2025-07-12 15:32:00 -07:00
|
|
|
|
pending_pastes: Vec<(String, String)>,
|
2025-08-03 11:31:35 -07:00
|
|
|
|
token_usage_info: Option<TokenUsageInfo>,
|
|
|
|
|
|
has_focus: bool,
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Popup state – at most one can be visible at any time.
|
|
|
|
|
|
enum ActivePopup {
|
|
|
|
|
|
None,
|
|
|
|
|
|
Command(CommandPopup),
|
|
|
|
|
|
File(FileSearchPopup),
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
impl ChatComposer {
|
2025-07-31 17:30:44 -07:00
|
|
|
|
pub fn new(
|
|
|
|
|
|
has_input_focus: bool,
|
|
|
|
|
|
app_event_tx: AppEventSender,
|
|
|
|
|
|
enhanced_keys_supported: bool,
|
|
|
|
|
|
) -> Self {
|
|
|
|
|
|
let use_shift_enter_hint = enhanced_keys_supported;
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
Self {
|
|
|
|
|
|
textarea: TextArea::new(),
|
|
|
|
|
|
textarea_state: RefCell::new(TextAreaState::default()),
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
active_popup: ActivePopup::None,
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
app_event_tx,
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
history: ChatComposerHistory::new(),
|
2025-06-27 13:37:11 -04:00
|
|
|
|
ctrl_c_quit_hint: false,
|
2025-07-31 17:30:44 -07:00
|
|
|
|
use_shift_enter_hint,
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
dismissed_file_popup_token: None,
|
|
|
|
|
|
current_file_query: None,
|
2025-07-12 15:32:00 -07:00
|
|
|
|
pending_pastes: Vec::new(),
|
2025-08-03 11:31:35 -07:00
|
|
|
|
token_usage_info: None,
|
|
|
|
|
|
has_focus: has_input_focus,
|
|
|
|
|
|
}
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
pub fn desired_height(&self, width: u16) -> u16 {
|
|
|
|
|
|
self.textarea.desired_height(width - 1)
|
2025-07-30 17:06:55 -07:00
|
|
|
|
+ match &self.active_popup {
|
2025-07-31 00:43:21 -07:00
|
|
|
|
ActivePopup::None => 1u16,
|
2025-07-30 17:06:55 -07:00
|
|
|
|
ActivePopup::Command(c) => c.calculate_required_height(),
|
|
|
|
|
|
ActivePopup::File(c) => c.calculate_required_height(),
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
pub fn cursor_pos(&self, area: Rect) -> Option<(u16, u16)> {
|
|
|
|
|
|
let popup_height = match &self.active_popup {
|
|
|
|
|
|
ActivePopup::Command(popup) => popup.calculate_required_height(),
|
|
|
|
|
|
ActivePopup::File(popup) => popup.calculate_required_height(),
|
|
|
|
|
|
ActivePopup::None => 1,
|
|
|
|
|
|
};
|
|
|
|
|
|
let [textarea_rect, _] =
|
|
|
|
|
|
Layout::vertical([Constraint::Min(0), Constraint::Max(popup_height)]).areas(area);
|
|
|
|
|
|
let mut textarea_rect = textarea_rect;
|
|
|
|
|
|
textarea_rect.width = textarea_rect.width.saturating_sub(1);
|
|
|
|
|
|
textarea_rect.x += 1;
|
|
|
|
|
|
let state = self.textarea_state.borrow();
|
|
|
|
|
|
self.textarea.cursor_pos_with_state(textarea_rect, &state)
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-07-16 08:59:26 -07:00
|
|
|
|
/// Returns true if the composer currently contains no user input.
|
|
|
|
|
|
pub(crate) fn is_empty(&self) -> bool {
|
|
|
|
|
|
self.textarea.is_empty()
|
|
|
|
|
|
}
|
|
|
|
|
|
|
feat: show number of tokens remaining in UI (#1388)
When using the OpenAI Responses API, we now record the `usage` field for
a `"response.completed"` event, which includes metrics about the number
of tokens consumed. We also introduce `openai_model_info.rs`, which
includes current data about the most common OpenAI models available via
the API (specifically `context_window` and `max_output_tokens`). If
Codex does not recognize the model, you can set `model_context_window`
and `model_max_output_tokens` explicitly in `config.toml`.
When then introduce a new event type to `protocol.rs`, `TokenCount`,
which includes the `TokenUsage` for the most recent turn.
Finally, we update the TUI to record the running sum of tokens used so
the percentage of available context window remaining can be reported via
the placeholder text for the composer:

We could certainly get much fancier with this (such as reporting the
estimated cost of the conversation), but for now, we are just trying to
achieve feature parity with the TypeScript CLI.
Though arguably this improves upon the TypeScript CLI, as the TypeScript
CLI uses heuristics to estimate the number of tokens used rather than
using the `usage` information directly:
https://github.com/openai/codex/blob/296996d74e345b1b05d8c3451a06ace21c5ada96/codex-cli/src/utils/approximate-tokens-used.ts#L3-L16
Fixes https://github.com/openai/codex/issues/1242
2025-06-25 23:31:11 -07:00
|
|
|
|
/// Update the cached *context-left* percentage and refresh the placeholder
|
|
|
|
|
|
/// text. The UI relies on the placeholder to convey the remaining
|
|
|
|
|
|
/// context when the composer is empty.
|
|
|
|
|
|
pub(crate) fn set_token_usage(
|
|
|
|
|
|
&mut self,
|
2025-08-07 05:17:18 -07:00
|
|
|
|
total_token_usage: TokenUsage,
|
|
|
|
|
|
last_token_usage: TokenUsage,
|
feat: show number of tokens remaining in UI (#1388)
When using the OpenAI Responses API, we now record the `usage` field for
a `"response.completed"` event, which includes metrics about the number
of tokens consumed. We also introduce `openai_model_info.rs`, which
includes current data about the most common OpenAI models available via
the API (specifically `context_window` and `max_output_tokens`). If
Codex does not recognize the model, you can set `model_context_window`
and `model_max_output_tokens` explicitly in `config.toml`.
When then introduce a new event type to `protocol.rs`, `TokenCount`,
which includes the `TokenUsage` for the most recent turn.
Finally, we update the TUI to record the running sum of tokens used so
the percentage of available context window remaining can be reported via
the placeholder text for the composer:

We could certainly get much fancier with this (such as reporting the
estimated cost of the conversation), but for now, we are just trying to
achieve feature parity with the TypeScript CLI.
Though arguably this improves upon the TypeScript CLI, as the TypeScript
CLI uses heuristics to estimate the number of tokens used rather than
using the `usage` information directly:
https://github.com/openai/codex/blob/296996d74e345b1b05d8c3451a06ace21c5ada96/codex-cli/src/utils/approximate-tokens-used.ts#L3-L16
Fixes https://github.com/openai/codex/issues/1242
2025-06-25 23:31:11 -07:00
|
|
|
|
model_context_window: Option<u64>,
|
|
|
|
|
|
) {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.token_usage_info = Some(TokenUsageInfo {
|
2025-08-07 05:17:18 -07:00
|
|
|
|
total_token_usage,
|
|
|
|
|
|
last_token_usage,
|
2025-08-03 11:31:35 -07:00
|
|
|
|
model_context_window,
|
|
|
|
|
|
});
|
feat: show number of tokens remaining in UI (#1388)
When using the OpenAI Responses API, we now record the `usage` field for
a `"response.completed"` event, which includes metrics about the number
of tokens consumed. We also introduce `openai_model_info.rs`, which
includes current data about the most common OpenAI models available via
the API (specifically `context_window` and `max_output_tokens`). If
Codex does not recognize the model, you can set `model_context_window`
and `model_max_output_tokens` explicitly in `config.toml`.
When then introduce a new event type to `protocol.rs`, `TokenCount`,
which includes the `TokenUsage` for the most recent turn.
Finally, we update the TUI to record the running sum of tokens used so
the percentage of available context window remaining can be reported via
the placeholder text for the composer:

We could certainly get much fancier with this (such as reporting the
estimated cost of the conversation), but for now, we are just trying to
achieve feature parity with the TypeScript CLI.
Though arguably this improves upon the TypeScript CLI, as the TypeScript
CLI uses heuristics to estimate the number of tokens used rather than
using the `usage` information directly:
https://github.com/openai/codex/blob/296996d74e345b1b05d8c3451a06ace21c5ada96/codex-cli/src/utils/approximate-tokens-used.ts#L3-L16
Fixes https://github.com/openai/codex/issues/1242
2025-06-25 23:31:11 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
/// Record the history metadata advertised by `SessionConfiguredEvent` so
|
|
|
|
|
|
/// that the composer can navigate cross-session history.
|
|
|
|
|
|
pub(crate) fn set_history_metadata(&mut self, log_id: u64, entry_count: usize) {
|
|
|
|
|
|
self.history.set_metadata(log_id, entry_count);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Integrate an asynchronous response to an on-demand history lookup. If
|
|
|
|
|
|
/// the entry is present and the offset matches the current cursor we
|
|
|
|
|
|
/// immediately populate the textarea.
|
|
|
|
|
|
pub(crate) fn on_history_entry_response(
|
|
|
|
|
|
&mut self,
|
|
|
|
|
|
log_id: u64,
|
|
|
|
|
|
offset: usize,
|
|
|
|
|
|
entry: Option<String>,
|
|
|
|
|
|
) -> bool {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let Some(text) = self.history.on_entry_response(log_id, offset, entry) else {
|
|
|
|
|
|
return false;
|
|
|
|
|
|
};
|
|
|
|
|
|
self.textarea.set_text(&text);
|
|
|
|
|
|
self.textarea.set_cursor(0);
|
|
|
|
|
|
true
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
2025-07-12 15:32:00 -07:00
|
|
|
|
pub fn handle_paste(&mut self, pasted: String) -> bool {
|
|
|
|
|
|
let char_count = pasted.chars().count();
|
|
|
|
|
|
if char_count > LARGE_PASTE_CHAR_THRESHOLD {
|
|
|
|
|
|
let placeholder = format!("[Pasted Content {char_count} chars]");
|
2025-08-14 16:58:51 -04:00
|
|
|
|
self.textarea.insert_element(&placeholder);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
self.pending_pastes.push((placeholder, pasted));
|
|
|
|
|
|
} else {
|
|
|
|
|
|
self.textarea.insert_str(&pasted);
|
|
|
|
|
|
}
|
|
|
|
|
|
self.sync_command_popup();
|
|
|
|
|
|
self.sync_file_search_popup();
|
|
|
|
|
|
true
|
|
|
|
|
|
}
|
|
|
|
|
|
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
/// Integrate results from an asynchronous file search.
|
2025-06-28 15:04:23 -07:00
|
|
|
|
pub(crate) fn on_file_search_result(&mut self, query: String, matches: Vec<FileMatch>) {
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
// Only apply if user is still editing a token starting with `query`.
|
|
|
|
|
|
let current_opt = Self::current_at_token(&self.textarea);
|
|
|
|
|
|
let Some(current_token) = current_opt else {
|
|
|
|
|
|
return;
|
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
if !current_token.starts_with(&query) {
|
|
|
|
|
|
return;
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
if let ActivePopup::File(popup) = &mut self.active_popup {
|
|
|
|
|
|
popup.set_matches(&query, matches);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-06-27 13:37:11 -04:00
|
|
|
|
pub fn set_ctrl_c_quit_hint(&mut self, show: bool, has_focus: bool) {
|
|
|
|
|
|
self.ctrl_c_quit_hint = show;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.set_has_focus(has_focus);
|
2025-06-27 13:37:11 -04:00
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-11 14:15:41 -07:00
|
|
|
|
pub(crate) fn insert_str(&mut self, text: &str) {
|
|
|
|
|
|
self.textarea.insert_str(text);
|
|
|
|
|
|
self.sync_command_popup();
|
|
|
|
|
|
self.sync_file_search_popup();
|
|
|
|
|
|
}
|
|
|
|
|
|
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
/// Handle a key event coming from the main UI.
|
2025-05-14 10:13:29 -07:00
|
|
|
|
pub fn handle_key_event(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
let result = match &mut self.active_popup {
|
|
|
|
|
|
ActivePopup::Command(_) => self.handle_key_event_with_slash_popup(key_event),
|
|
|
|
|
|
ActivePopup::File(_) => self.handle_key_event_with_file_popup(key_event),
|
|
|
|
|
|
ActivePopup::None => self.handle_key_event_without_popup(key_event),
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
// Update (or hide/show) popup after processing the key.
|
|
|
|
|
|
self.sync_command_popup();
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
if matches!(self.active_popup, ActivePopup::Command(_)) {
|
|
|
|
|
|
self.dismissed_file_popup_token = None;
|
|
|
|
|
|
} else {
|
|
|
|
|
|
self.sync_file_search_popup();
|
|
|
|
|
|
}
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
|
|
|
|
|
result
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Handle key event when the slash-command popup is visible.
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
fn handle_key_event_with_slash_popup(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
|
|
|
|
|
|
let ActivePopup::Command(popup) = &mut self.active_popup else {
|
|
|
|
|
|
unreachable!();
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
};
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
match key_event {
|
|
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Up, ..
|
|
|
|
|
|
} => {
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
popup.move_up();
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Down,
|
|
|
|
|
|
..
|
|
|
|
|
|
} => {
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
popup.move_down();
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Tab, ..
|
|
|
|
|
|
} => {
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
if let Some(cmd) = popup.selected_command() {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let first_line = self.textarea.text().lines().next().unwrap_or("");
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
|
|
|
|
|
let starts_with_cmd = first_line
|
|
|
|
|
|
.trim_start()
|
|
|
|
|
|
.starts_with(&format!("/{}", cmd.command()));
|
|
|
|
|
|
|
|
|
|
|
|
if !starts_with_cmd {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.textarea.set_text(&format!("/{} ", cmd.command()));
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Enter,
|
|
|
|
|
|
modifiers: KeyModifiers::NONE,
|
|
|
|
|
|
..
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
} => {
|
|
|
|
|
|
if let Some(cmd) = popup.selected_command() {
|
|
|
|
|
|
// Send command to the app layer.
|
2025-05-15 14:50:30 -07:00
|
|
|
|
self.app_event_tx.send(AppEvent::DispatchCommand(*cmd));
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
|
|
|
|
|
// Clear textarea so no residual text remains.
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.textarea.set_text("");
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
|
|
|
|
|
// Hide popup since the command has been dispatched.
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
self.active_popup = ActivePopup::None;
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
return (InputResult::None, true);
|
|
|
|
|
|
}
|
|
|
|
|
|
// Fallback to default newline handling if no command selected.
|
|
|
|
|
|
self.handle_key_event_without_popup(key_event)
|
|
|
|
|
|
}
|
|
|
|
|
|
input => self.handle_input_basic(input),
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
/// Handle key events when file search popup is visible.
|
|
|
|
|
|
fn handle_key_event_with_file_popup(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
|
|
|
|
|
|
let ActivePopup::File(popup) = &mut self.active_popup else {
|
|
|
|
|
|
unreachable!();
|
|
|
|
|
|
};
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
match key_event {
|
|
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Up, ..
|
|
|
|
|
|
} => {
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
popup.move_up();
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Down,
|
|
|
|
|
|
..
|
|
|
|
|
|
} => {
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
popup.move_down();
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Esc, ..
|
|
|
|
|
|
} => {
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
// Hide popup without modifying text, remember token to avoid immediate reopen.
|
|
|
|
|
|
if let Some(tok) = Self::current_at_token(&self.textarea) {
|
|
|
|
|
|
self.dismissed_file_popup_token = Some(tok.to_string());
|
|
|
|
|
|
}
|
|
|
|
|
|
self.active_popup = ActivePopup::None;
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Tab, ..
|
|
|
|
|
|
}
|
|
|
|
|
|
| KeyEvent {
|
|
|
|
|
|
code: KeyCode::Enter,
|
|
|
|
|
|
modifiers: KeyModifiers::NONE,
|
|
|
|
|
|
..
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
} => {
|
|
|
|
|
|
if let Some(sel) = popup.selected_match() {
|
|
|
|
|
|
let sel_path = sel.to_string();
|
|
|
|
|
|
// Drop popup borrow before using self mutably again.
|
|
|
|
|
|
self.insert_selected_path(&sel_path);
|
|
|
|
|
|
self.active_popup = ActivePopup::None;
|
|
|
|
|
|
return (InputResult::None, true);
|
|
|
|
|
|
}
|
|
|
|
|
|
(InputResult::None, false)
|
|
|
|
|
|
}
|
|
|
|
|
|
input => self.handle_input_basic(input),
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Extract the `@token` that the cursor is currently positioned on, if any.
|
|
|
|
|
|
///
|
|
|
|
|
|
/// The returned string **does not** include the leading `@`.
|
|
|
|
|
|
///
|
|
|
|
|
|
/// Behavior:
|
|
|
|
|
|
/// - The cursor may be anywhere *inside* the token (including on the
|
|
|
|
|
|
/// leading `@`). It does **not** need to be at the end of the line.
|
|
|
|
|
|
/// - A token is delimited by ASCII whitespace (space, tab, newline).
|
2025-08-07 00:41:48 -07:00
|
|
|
|
/// - If the token under the cursor starts with `@`, that token is
|
|
|
|
|
|
/// returned without the leading `@`. This includes the case where the
|
|
|
|
|
|
/// token is just "@" (empty query), which is used to trigger a UI hint
|
2025-08-03 11:31:35 -07:00
|
|
|
|
fn current_at_token(textarea: &TextArea) -> Option<String> {
|
|
|
|
|
|
let cursor_offset = textarea.cursor();
|
|
|
|
|
|
let text = textarea.text();
|
|
|
|
|
|
|
|
|
|
|
|
// Adjust the provided byte offset to the nearest valid char boundary at or before it.
|
|
|
|
|
|
let mut safe_cursor = cursor_offset.min(text.len());
|
|
|
|
|
|
// If we're not on a char boundary, move back to the start of the current char.
|
|
|
|
|
|
if safe_cursor < text.len() && !text.is_char_boundary(safe_cursor) {
|
|
|
|
|
|
// Find the last valid boundary <= cursor_offset.
|
|
|
|
|
|
safe_cursor = text
|
|
|
|
|
|
.char_indices()
|
|
|
|
|
|
.map(|(i, _)| i)
|
|
|
|
|
|
.take_while(|&i| i <= cursor_offset)
|
|
|
|
|
|
.last()
|
|
|
|
|
|
.unwrap_or(0);
|
|
|
|
|
|
}
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
// Split the line around the (now safe) cursor position.
|
|
|
|
|
|
let before_cursor = &text[..safe_cursor];
|
|
|
|
|
|
let after_cursor = &text[safe_cursor..];
|
|
|
|
|
|
|
|
|
|
|
|
// Detect whether we're on whitespace at the cursor boundary.
|
|
|
|
|
|
let at_whitespace = if safe_cursor < text.len() {
|
|
|
|
|
|
text[safe_cursor..]
|
|
|
|
|
|
.chars()
|
|
|
|
|
|
.next()
|
|
|
|
|
|
.map(|c| c.is_whitespace())
|
|
|
|
|
|
.unwrap_or(false)
|
|
|
|
|
|
} else {
|
|
|
|
|
|
false
|
|
|
|
|
|
};
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
// Left candidate: token containing the cursor position.
|
|
|
|
|
|
let start_left = before_cursor
|
2025-07-08 05:43:31 +09:00
|
|
|
|
.char_indices()
|
|
|
|
|
|
.rfind(|(_, c)| c.is_whitespace())
|
|
|
|
|
|
.map(|(idx, c)| idx + c.len_utf8())
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
.unwrap_or(0);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let end_left_rel = after_cursor
|
2025-07-08 05:43:31 +09:00
|
|
|
|
.char_indices()
|
|
|
|
|
|
.find(|(_, c)| c.is_whitespace())
|
|
|
|
|
|
.map(|(idx, _)| idx)
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
.unwrap_or(after_cursor.len());
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let end_left = safe_cursor + end_left_rel;
|
|
|
|
|
|
let token_left = if start_left < end_left {
|
|
|
|
|
|
Some(&text[start_left..end_left])
|
|
|
|
|
|
} else {
|
|
|
|
|
|
None
|
|
|
|
|
|
};
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
// Right candidate: token immediately after any whitespace from the cursor.
|
|
|
|
|
|
let ws_len_right: usize = after_cursor
|
|
|
|
|
|
.chars()
|
|
|
|
|
|
.take_while(|c| c.is_whitespace())
|
|
|
|
|
|
.map(|c| c.len_utf8())
|
|
|
|
|
|
.sum();
|
|
|
|
|
|
let start_right = safe_cursor + ws_len_right;
|
|
|
|
|
|
let end_right_rel = text[start_right..]
|
|
|
|
|
|
.char_indices()
|
|
|
|
|
|
.find(|(_, c)| c.is_whitespace())
|
|
|
|
|
|
.map(|(idx, _)| idx)
|
|
|
|
|
|
.unwrap_or(text.len() - start_right);
|
|
|
|
|
|
let end_right = start_right + end_right_rel;
|
|
|
|
|
|
let token_right = if start_right < end_right {
|
|
|
|
|
|
Some(&text[start_right..end_right])
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
} else {
|
|
|
|
|
|
None
|
2025-08-03 11:31:35 -07:00
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
let left_at = token_left
|
2025-08-07 00:41:48 -07:00
|
|
|
|
.filter(|t| t.starts_with('@'))
|
2025-08-03 11:31:35 -07:00
|
|
|
|
.map(|t| t[1..].to_string());
|
|
|
|
|
|
let right_at = token_right
|
2025-08-07 00:41:48 -07:00
|
|
|
|
.filter(|t| t.starts_with('@'))
|
2025-08-03 11:31:35 -07:00
|
|
|
|
.map(|t| t[1..].to_string());
|
|
|
|
|
|
|
|
|
|
|
|
if at_whitespace {
|
2025-08-07 00:41:48 -07:00
|
|
|
|
if right_at.is_some() {
|
|
|
|
|
|
return right_at;
|
|
|
|
|
|
}
|
|
|
|
|
|
if token_left.is_some_and(|t| t == "@") {
|
|
|
|
|
|
return None;
|
|
|
|
|
|
}
|
|
|
|
|
|
return left_at;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
}
|
|
|
|
|
|
if after_cursor.starts_with('@') {
|
|
|
|
|
|
return right_at.or(left_at);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
left_at.or(right_at)
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Replace the active `@token` (the one under the cursor) with `path`.
|
|
|
|
|
|
///
|
|
|
|
|
|
/// The algorithm mirrors `current_at_token` so replacement works no matter
|
|
|
|
|
|
/// where the cursor is within the token and regardless of how many
|
|
|
|
|
|
/// `@tokens` exist in the line.
|
|
|
|
|
|
fn insert_selected_path(&mut self, path: &str) {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let cursor_offset = self.textarea.cursor();
|
|
|
|
|
|
let text = self.textarea.text();
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let before_cursor = &text[..cursor_offset];
|
|
|
|
|
|
let after_cursor = &text[cursor_offset..];
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
// Determine token boundaries.
|
|
|
|
|
|
let start_idx = before_cursor
|
|
|
|
|
|
.char_indices()
|
|
|
|
|
|
.rfind(|(_, c)| c.is_whitespace())
|
|
|
|
|
|
.map(|(idx, c)| idx + c.len_utf8())
|
|
|
|
|
|
.unwrap_or(0);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let end_rel_idx = after_cursor
|
|
|
|
|
|
.char_indices()
|
|
|
|
|
|
.find(|(_, c)| c.is_whitespace())
|
|
|
|
|
|
.map(|(idx, _)| idx)
|
|
|
|
|
|
.unwrap_or(after_cursor.len());
|
|
|
|
|
|
let end_idx = cursor_offset + end_rel_idx;
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
// Replace the slice `[start_idx, end_idx)` with the chosen path and a trailing space.
|
|
|
|
|
|
let mut new_text =
|
|
|
|
|
|
String::with_capacity(text.len() - (end_idx - start_idx) + path.len() + 1);
|
|
|
|
|
|
new_text.push_str(&text[..start_idx]);
|
|
|
|
|
|
new_text.push_str(path);
|
|
|
|
|
|
new_text.push(' ');
|
|
|
|
|
|
new_text.push_str(&text[end_idx..]);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.textarea.set_text(&new_text);
|
2025-08-07 00:58:06 +01:00
|
|
|
|
let new_cursor = start_idx.saturating_add(path.len()).saturating_add(1);
|
|
|
|
|
|
self.textarea.set_cursor(new_cursor);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
/// Handle key event when no popup is visible.
|
|
|
|
|
|
fn handle_key_event_without_popup(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
match key_event {
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
// -------------------------------------------------------------
|
|
|
|
|
|
// History navigation (Up / Down) – only when the composer is not
|
|
|
|
|
|
// empty or when the cursor is at the correct position, to avoid
|
|
|
|
|
|
// interfering with normal cursor movement.
|
|
|
|
|
|
// -------------------------------------------------------------
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Up | KeyCode::Down,
|
|
|
|
|
|
..
|
|
|
|
|
|
} => {
|
|
|
|
|
|
if self
|
|
|
|
|
|
.history
|
|
|
|
|
|
.should_handle_navigation(self.textarea.text(), self.textarea.cursor())
|
|
|
|
|
|
{
|
|
|
|
|
|
let replace_text = match key_event.code {
|
|
|
|
|
|
KeyCode::Up => self.history.navigate_up(&self.app_event_tx),
|
|
|
|
|
|
KeyCode::Down => self.history.navigate_down(&self.app_event_tx),
|
|
|
|
|
|
_ => unreachable!(),
|
|
|
|
|
|
};
|
|
|
|
|
|
if let Some(text) = replace_text {
|
|
|
|
|
|
self.textarea.set_text(&text);
|
|
|
|
|
|
self.textarea.set_cursor(0);
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
return (InputResult::None, true);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
self.handle_input_basic(key_event)
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
}
|
2025-08-03 11:31:35 -07:00
|
|
|
|
KeyEvent {
|
|
|
|
|
|
code: KeyCode::Enter,
|
|
|
|
|
|
modifiers: KeyModifiers::NONE,
|
|
|
|
|
|
..
|
2025-05-14 10:13:29 -07:00
|
|
|
|
} => {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let mut text = self.textarea.text().to_string();
|
|
|
|
|
|
self.textarea.set_text("");
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
|
2025-07-12 15:32:00 -07:00
|
|
|
|
// Replace all pending pastes in the text
|
|
|
|
|
|
for (placeholder, actual) in &self.pending_pastes {
|
|
|
|
|
|
if text.contains(placeholder) {
|
|
|
|
|
|
text = text.replace(placeholder, actual);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
self.pending_pastes.clear();
|
|
|
|
|
|
|
feat: record messages from user in ~/.codex/history.jsonl (#939)
This is a large change to support a "history" feature like you would
expect in a shell like Bash.
History events are recorded in `$CODEX_HOME/history.jsonl`. Because it
is a JSONL file, it is straightforward to append new entries (as opposed
to the TypeScript file that uses `$CODEX_HOME/history.json`, so to be
valid JSON, each new entry entails rewriting the entire file). Because
it is possible for there to be multiple instances of Codex CLI writing
to `history.jsonl` at once, we use advisory file locking when working
with `history.jsonl` in `codex-rs/core/src/message_history.rs`.
Because we believe history is a sufficiently useful feature, we enable
it by default. Though to provide some safety, we set the file
permissions of `history.jsonl` to be `o600` so that other users on the
system cannot read the user's history. We do not yet support a default
list of `SENSITIVE_PATTERNS` as the TypeScript CLI does:
https://github.com/openai/codex/blob/3fdf9df1335ac9501e3fb0e61715359145711e8b/codex-cli/src/utils/storage/command-history.ts#L10-L17
We are going to take a more conservative approach to this list in the
Rust CLI. For example, while `/\b[A-Za-z0-9-_]{20,}\b/` might exclude
sensitive information like API tokens, it would also exclude valuable
information such as references to Git commits.
As noted in the updated documentation, users can opt-out of history by
adding the following to `config.toml`:
```toml
[history]
persistence = "none"
```
Because `history.jsonl` could, in theory, be quite large, we take a[n
arguably overly pedantic] approach in reading history entries into
memory. Specifically, we start by telling the client the current number
of entries in the history file (`history_entry_count`) as well as the
inode (`history_log_id`) of `history.jsonl` (see the new fields on
`SessionConfiguredEvent`).
The client is responsible for keeping new entries in memory to create a
"local history," but if the user hits up enough times to go "past" the
end of local history, then the client should use the new
`GetHistoryEntryRequest` in the protocol to fetch older entries.
Specifically, it should pass the `history_log_id` it was given
originally and work backwards from `history_entry_count`. (It should
really fetch history in batches rather than one-at-a-time, but that is
something we can improve upon in subsequent PRs.)
The motivation behind this crazy scheme is that it is designed to defend
against:
* The `history.jsonl` being truncated during the session such that the
index into the history is no longer consistent with what had been read
up to that point. We do not yet have logic to enforce a `max_bytes` for
`history.jsonl`, but once we do, we will aspire to implement it in a way
that should result in a new inode for the file on most systems.
* New items from concurrent Codex CLI sessions amending to the history.
Because, in absence of truncation, `history.jsonl` is an append-only
log, so long as the client reads backwards from `history_entry_count`,
it should always get a consistent view of history. (That said, it will
not be able to read _new_ commands from concurrent sessions, but perhaps
we will introduce a `/` command to reload latest history or something
down the road.)
Admittedly, my testing of this feature thus far has been fairly light. I
expect we will find bugs and introduce enhancements/fixes going forward.
2025-05-15 16:26:23 -07:00
|
|
|
|
if text.is_empty() {
|
|
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
} else {
|
|
|
|
|
|
self.history.record_local_submission(&text);
|
|
|
|
|
|
(InputResult::Submitted(text), true)
|
|
|
|
|
|
}
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
input => self.handle_input_basic(input),
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
/// Handle generic Input events that modify the textarea content.
|
2025-08-03 11:31:35 -07:00
|
|
|
|
fn handle_input_basic(&mut self, input: KeyEvent) -> (InputResult, bool) {
|
2025-07-12 15:32:00 -07:00
|
|
|
|
// Normal input handling
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
self.textarea.input(input);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let text_after = self.textarea.text();
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
// Check if any placeholders were removed and remove their corresponding pending pastes
|
|
|
|
|
|
self.pending_pastes
|
|
|
|
|
|
.retain(|(placeholder, _)| text_after.contains(placeholder));
|
|
|
|
|
|
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
(InputResult::None, true)
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Synchronize `self.command_popup` with the current text in the
|
|
|
|
|
|
/// textarea. This must be called after every modification that can change
|
|
|
|
|
|
/// the text so the popup is shown/updated/hidden as appropriate.
|
|
|
|
|
|
fn sync_command_popup(&mut self) {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let first_line = self.textarea.text().lines().next().unwrap_or("");
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
let input_starts_with_slash = first_line.starts_with('/');
|
|
|
|
|
|
match &mut self.active_popup {
|
|
|
|
|
|
ActivePopup::Command(popup) => {
|
|
|
|
|
|
if input_starts_with_slash {
|
|
|
|
|
|
popup.on_composer_text_change(first_line.to_string());
|
|
|
|
|
|
} else {
|
|
|
|
|
|
self.active_popup = ActivePopup::None;
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
_ => {
|
|
|
|
|
|
if input_starts_with_slash {
|
|
|
|
|
|
let mut command_popup = CommandPopup::new();
|
|
|
|
|
|
command_popup.on_composer_text_change(first_line.to_string());
|
|
|
|
|
|
self.active_popup = ActivePopup::Command(command_popup);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/// Synchronize `self.file_search_popup` with the current text in the textarea.
|
|
|
|
|
|
/// Note this is only called when self.active_popup is NOT Command.
|
|
|
|
|
|
fn sync_file_search_popup(&mut self) {
|
|
|
|
|
|
// Determine if there is an @token underneath the cursor.
|
|
|
|
|
|
let query = match Self::current_at_token(&self.textarea) {
|
|
|
|
|
|
Some(token) => token,
|
|
|
|
|
|
None => {
|
|
|
|
|
|
self.active_popup = ActivePopup::None;
|
|
|
|
|
|
self.dismissed_file_popup_token = None;
|
|
|
|
|
|
return;
|
|
|
|
|
|
}
|
|
|
|
|
|
};
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
// If user dismissed popup for this exact query, don't reopen until text changes.
|
|
|
|
|
|
if self.dismissed_file_popup_token.as_ref() == Some(&query) {
|
|
|
|
|
|
return;
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
}
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
2025-08-07 00:41:48 -07:00
|
|
|
|
if !query.is_empty() {
|
|
|
|
|
|
self.app_event_tx
|
|
|
|
|
|
.send(AppEvent::StartFileSearch(query.clone()));
|
|
|
|
|
|
}
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
|
|
|
|
|
|
match &mut self.active_popup {
|
|
|
|
|
|
ActivePopup::File(popup) => {
|
2025-08-07 00:41:48 -07:00
|
|
|
|
if query.is_empty() {
|
|
|
|
|
|
popup.set_empty_prompt();
|
|
|
|
|
|
} else {
|
|
|
|
|
|
popup.set_query(&query);
|
|
|
|
|
|
}
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
_ => {
|
|
|
|
|
|
let mut popup = FileSearchPopup::new();
|
2025-08-07 00:41:48 -07:00
|
|
|
|
if query.is_empty() {
|
|
|
|
|
|
popup.set_empty_prompt();
|
|
|
|
|
|
} else {
|
|
|
|
|
|
popup.set_query(&query);
|
|
|
|
|
|
}
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
self.active_popup = ActivePopup::File(popup);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
self.current_file_query = Some(query);
|
|
|
|
|
|
self.dismissed_file_popup_token = None;
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
fn set_has_focus(&mut self, has_focus: bool) {
|
|
|
|
|
|
self.has_focus = has_focus;
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
impl WidgetRef for &ChatComposer {
|
2025-05-14 10:13:29 -07:00
|
|
|
|
fn render_ref(&self, area: Rect, buf: &mut Buffer) {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let popup_height = match &self.active_popup {
|
|
|
|
|
|
ActivePopup::Command(popup) => popup.calculate_required_height(),
|
|
|
|
|
|
ActivePopup::File(popup) => popup.calculate_required_height(),
|
|
|
|
|
|
ActivePopup::None => 1,
|
|
|
|
|
|
};
|
|
|
|
|
|
let [textarea_rect, popup_rect] =
|
|
|
|
|
|
Layout::vertical([Constraint::Min(0), Constraint::Max(popup_height)]).areas(area);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
match &self.active_popup {
|
|
|
|
|
|
ActivePopup::Command(popup) => {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
popup.render_ref(popup_rect, buf);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
ActivePopup::File(popup) => {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
popup.render_ref(popup_rect, buf);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
|
|
|
|
|
ActivePopup::None => {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let bottom_line_rect = popup_rect;
|
2025-07-31 00:43:21 -07:00
|
|
|
|
let key_hint_style = Style::default().fg(Color::Cyan);
|
2025-08-07 00:45:47 -07:00
|
|
|
|
let mut hint = if self.ctrl_c_quit_hint {
|
2025-07-31 00:43:21 -07:00
|
|
|
|
vec![
|
|
|
|
|
|
Span::from(" "),
|
|
|
|
|
|
"Ctrl+C again".set_style(key_hint_style),
|
|
|
|
|
|
Span::from(" to quit"),
|
|
|
|
|
|
]
|
|
|
|
|
|
} else {
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let newline_hint_key = if self.use_shift_enter_hint {
|
|
|
|
|
|
"Shift+⏎"
|
|
|
|
|
|
} else {
|
|
|
|
|
|
"Ctrl+J"
|
|
|
|
|
|
};
|
2025-07-31 00:43:21 -07:00
|
|
|
|
vec![
|
|
|
|
|
|
Span::from(" "),
|
|
|
|
|
|
"⏎".set_style(key_hint_style),
|
|
|
|
|
|
Span::from(" send "),
|
2025-07-31 17:30:44 -07:00
|
|
|
|
newline_hint_key.set_style(key_hint_style),
|
2025-07-31 00:43:21 -07:00
|
|
|
|
Span::from(" newline "),
|
|
|
|
|
|
"Ctrl+C".set_style(key_hint_style),
|
|
|
|
|
|
Span::from(" quit"),
|
|
|
|
|
|
]
|
|
|
|
|
|
};
|
2025-08-07 00:45:47 -07:00
|
|
|
|
|
|
|
|
|
|
// Append token/context usage info to the footer hints when available.
|
|
|
|
|
|
if let Some(token_usage_info) = &self.token_usage_info {
|
2025-08-07 05:17:18 -07:00
|
|
|
|
let token_usage = &token_usage_info.total_token_usage;
|
2025-08-07 00:45:47 -07:00
|
|
|
|
hint.push(Span::from(" "));
|
|
|
|
|
|
hint.push(
|
2025-08-11 07:19:15 -07:00
|
|
|
|
Span::from(format!("{} tokens used", token_usage.blended_total()))
|
2025-08-07 00:45:47 -07:00
|
|
|
|
.style(Style::default().add_modifier(Modifier::DIM)),
|
|
|
|
|
|
);
|
2025-08-07 05:17:18 -07:00
|
|
|
|
let last_token_usage = &token_usage_info.last_token_usage;
|
2025-08-07 00:45:47 -07:00
|
|
|
|
if let Some(context_window) = token_usage_info.model_context_window {
|
|
|
|
|
|
let percent_remaining: u8 = if context_window > 0 {
|
|
|
|
|
|
let percent = 100.0
|
2025-08-11 07:19:15 -07:00
|
|
|
|
- (last_token_usage.tokens_in_context_window() as f32
|
|
|
|
|
|
/ context_window as f32
|
2025-08-07 05:17:18 -07:00
|
|
|
|
* 100.0);
|
2025-08-07 00:45:47 -07:00
|
|
|
|
percent.clamp(0.0, 100.0) as u8
|
|
|
|
|
|
} else {
|
|
|
|
|
|
100
|
|
|
|
|
|
};
|
|
|
|
|
|
hint.push(Span::from(" "));
|
|
|
|
|
|
hint.push(
|
|
|
|
|
|
Span::from(format!("{percent_remaining}% context left"))
|
|
|
|
|
|
.style(Style::default().add_modifier(Modifier::DIM)),
|
|
|
|
|
|
);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-07-31 00:43:21 -07:00
|
|
|
|
Line::from(hint)
|
|
|
|
|
|
.style(Style::default().dim())
|
|
|
|
|
|
.render_ref(bottom_line_rect, buf);
|
feat: add support for @ to do file search (#1401)
Introduces support for `@` to trigger a fuzzy-filename search in the
composer. Under the hood, this leverages
https://crates.io/crates/nucleo-matcher to do the fuzzy matching and
https://crates.io/crates/ignore to build up the list of file candidates
(so that it respects `.gitignore`).
For simplicity (at least for now), we do not do any caching between
searches like VS Code does for its file search:
https://github.com/microsoft/vscode/blob/1d89ed699b2e924d418c856318a3e12bca67ff3a/src/vs/workbench/services/search/node/rawSearchService.ts#L212-L218
Because we do not do any caching, I saw queries take up to three seconds
on large repositories with hundreds of thousands of files. To that end,
we do not perform searches synchronously on each keystroke, but instead
dispatch an event to do the search on a background thread that
asynchronously reports back to the UI when the results are available.
This is largely handled by the `FileSearchManager` introduced in this
PR, which also has logic for debouncing requests so there is at most one
search in flight at a time.
While we could potentially polish and tune this feature further, it may
already be overengineered for how it will be used, in practice, so we
can improve things going forward if it turns out that this is not "good
enough" in the wild.
Note this feature does not work like `@` in the TypeScript CLI, which
was more like directory-based tab completion. In the Rust CLI, `@`
triggers a full-repo fuzzy-filename search.
Fixes https://github.com/openai/codex/issues/1261.
2025-06-28 13:47:42 -07:00
|
|
|
|
}
|
feat: add support for commands in the Rust TUI (#935)
Introduces support for slash commands like in the TypeScript CLI. We do
not support the full set of commands yet, but the core abstraction is
there now.
In particular, we have a `SlashCommand` enum and due to thoughtful use
of the [strum](https://crates.io/crates/strum) crate, it requires
minimal boilerplate to add a new command to the list.
The key new piece of UI is `CommandPopup`, though the keyboard events
are still handled by `ChatComposer`. The behavior is roughly as follows:
* if the first character in the composer is `/`, the command popup is
displayed (if you really want to send a message to Codex that starts
with a `/`, simply put a space before the `/`)
* while the popup is displayed, up/down can be used to change the
selection of the popup
* if there is a selection, hitting tab completes the command, but does
not send it
* if there is a selection, hitting enter sends the command
* if the prefix of the composer matches a command, the command will be
visible in the popup so the user can see the description (commands could
take arguments, so additional text may appear after the command name
itself)
https://github.com/user-attachments/assets/39c3e6ee-eeb7-4ef7-a911-466d8184975f
Incidentally, Codex wrote almost all the code for this PR!
2025-05-14 12:55:49 -07:00
|
|
|
|
}
|
2025-08-13 15:50:50 -07:00
|
|
|
|
let border_style = if self.has_focus {
|
|
|
|
|
|
Style::default().fg(Color::Cyan)
|
|
|
|
|
|
} else {
|
|
|
|
|
|
Style::default().add_modifier(Modifier::DIM)
|
|
|
|
|
|
};
|
2025-08-03 11:31:35 -07:00
|
|
|
|
Block::default()
|
|
|
|
|
|
.borders(Borders::LEFT)
|
|
|
|
|
|
.border_type(BorderType::QuadrantOutside)
|
2025-08-13 15:50:50 -07:00
|
|
|
|
.border_style(border_style)
|
2025-08-03 11:31:35 -07:00
|
|
|
|
.render_ref(
|
|
|
|
|
|
Rect::new(textarea_rect.x, textarea_rect.y, 1, textarea_rect.height),
|
|
|
|
|
|
buf,
|
|
|
|
|
|
);
|
|
|
|
|
|
let mut textarea_rect = textarea_rect;
|
|
|
|
|
|
textarea_rect.width = textarea_rect.width.saturating_sub(1);
|
|
|
|
|
|
textarea_rect.x += 1;
|
2025-08-07 00:46:45 -07:00
|
|
|
|
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let mut state = self.textarea_state.borrow_mut();
|
|
|
|
|
|
StatefulWidgetRef::render_ref(&(&self.textarea), textarea_rect, buf, &mut state);
|
|
|
|
|
|
if self.textarea.text().is_empty() {
|
2025-08-07 00:45:47 -07:00
|
|
|
|
Line::from(BASE_PLACEHOLDER_TEXT)
|
2025-08-03 11:31:35 -07:00
|
|
|
|
.style(Style::default().dim())
|
|
|
|
|
|
.render_ref(textarea_rect.inner(Margin::new(1, 0)), buf);
|
|
|
|
|
|
}
|
2025-05-14 10:13:29 -07:00
|
|
|
|
}
|
|
|
|
|
|
}
|
2025-07-08 05:43:31 +09:00
|
|
|
|
|
|
|
|
|
|
#[cfg(test)]
|
|
|
|
|
|
mod tests {
|
2025-08-06 09:10:23 -07:00
|
|
|
|
use crate::app_event::AppEvent;
|
2025-07-12 15:32:00 -07:00
|
|
|
|
use crate::bottom_pane::AppEventSender;
|
2025-07-08 05:43:31 +09:00
|
|
|
|
use crate::bottom_pane::ChatComposer;
|
2025-07-12 15:32:00 -07:00
|
|
|
|
use crate::bottom_pane::InputResult;
|
|
|
|
|
|
use crate::bottom_pane::chat_composer::LARGE_PASTE_CHAR_THRESHOLD;
|
2025-08-03 11:31:35 -07:00
|
|
|
|
use crate::bottom_pane::textarea::TextArea;
|
2025-07-08 05:43:31 +09:00
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_current_at_token_basic_cases() {
|
|
|
|
|
|
let test_cases = vec![
|
|
|
|
|
|
// Valid @ tokens
|
|
|
|
|
|
("@hello", 3, Some("hello".to_string()), "Basic ASCII token"),
|
|
|
|
|
|
(
|
|
|
|
|
|
"@file.txt",
|
|
|
|
|
|
4,
|
|
|
|
|
|
Some("file.txt".to_string()),
|
|
|
|
|
|
"ASCII with extension",
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
"hello @world test",
|
|
|
|
|
|
8,
|
|
|
|
|
|
Some("world".to_string()),
|
|
|
|
|
|
"ASCII token in middle",
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
"@test123",
|
|
|
|
|
|
5,
|
|
|
|
|
|
Some("test123".to_string()),
|
|
|
|
|
|
"ASCII with numbers",
|
|
|
|
|
|
),
|
|
|
|
|
|
// Unicode examples
|
|
|
|
|
|
("@İstanbul", 3, Some("İstanbul".to_string()), "Turkish text"),
|
|
|
|
|
|
(
|
|
|
|
|
|
"@testЙЦУ.rs",
|
|
|
|
|
|
8,
|
|
|
|
|
|
Some("testЙЦУ.rs".to_string()),
|
|
|
|
|
|
"Mixed ASCII and Cyrillic",
|
|
|
|
|
|
),
|
|
|
|
|
|
("@诶", 2, Some("诶".to_string()), "Chinese character"),
|
|
|
|
|
|
("@👍", 2, Some("👍".to_string()), "Emoji token"),
|
|
|
|
|
|
// Invalid cases (should return None)
|
|
|
|
|
|
("hello", 2, None, "No @ symbol"),
|
2025-08-07 00:41:48 -07:00
|
|
|
|
(
|
|
|
|
|
|
"@",
|
|
|
|
|
|
1,
|
|
|
|
|
|
Some("".to_string()),
|
|
|
|
|
|
"Only @ symbol triggers empty query",
|
|
|
|
|
|
),
|
2025-07-08 05:43:31 +09:00
|
|
|
|
("@ hello", 2, None, "@ followed by space"),
|
|
|
|
|
|
("test @ world", 6, None, "@ with spaces around"),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
for (input, cursor_pos, expected, description) in test_cases {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let mut textarea = TextArea::new();
|
2025-07-08 05:43:31 +09:00
|
|
|
|
textarea.insert_str(input);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
textarea.set_cursor(cursor_pos);
|
2025-07-08 05:43:31 +09:00
|
|
|
|
|
|
|
|
|
|
let result = ChatComposer::current_at_token(&textarea);
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
result, expected,
|
2025-07-10 20:08:16 +02:00
|
|
|
|
"Failed for case: {description} - input: '{input}', cursor: {cursor_pos}"
|
2025-07-08 05:43:31 +09:00
|
|
|
|
);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_current_at_token_cursor_positions() {
|
|
|
|
|
|
let test_cases = vec![
|
|
|
|
|
|
// Different cursor positions within a token
|
|
|
|
|
|
("@test", 0, Some("test".to_string()), "Cursor at @"),
|
|
|
|
|
|
("@test", 1, Some("test".to_string()), "Cursor after @"),
|
|
|
|
|
|
("@test", 5, Some("test".to_string()), "Cursor at end"),
|
|
|
|
|
|
// Multiple tokens - cursor determines which token
|
|
|
|
|
|
("@file1 @file2", 0, Some("file1".to_string()), "First token"),
|
|
|
|
|
|
(
|
|
|
|
|
|
"@file1 @file2",
|
|
|
|
|
|
8,
|
|
|
|
|
|
Some("file2".to_string()),
|
|
|
|
|
|
"Second token",
|
|
|
|
|
|
),
|
|
|
|
|
|
// Edge cases
|
2025-08-07 00:41:48 -07:00
|
|
|
|
("@", 0, Some("".to_string()), "Only @ symbol"),
|
2025-07-08 05:43:31 +09:00
|
|
|
|
("@a", 2, Some("a".to_string()), "Single character after @"),
|
|
|
|
|
|
("", 0, None, "Empty input"),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
for (input, cursor_pos, expected, description) in test_cases {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let mut textarea = TextArea::new();
|
2025-07-08 05:43:31 +09:00
|
|
|
|
textarea.insert_str(input);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
textarea.set_cursor(cursor_pos);
|
2025-07-08 05:43:31 +09:00
|
|
|
|
|
|
|
|
|
|
let result = ChatComposer::current_at_token(&textarea);
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
result, expected,
|
|
|
|
|
|
"Failed for cursor position case: {description} - input: '{input}', cursor: {cursor_pos}",
|
|
|
|
|
|
);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_current_at_token_whitespace_boundaries() {
|
|
|
|
|
|
let test_cases = vec![
|
|
|
|
|
|
// Space boundaries
|
|
|
|
|
|
(
|
|
|
|
|
|
"aaa@aaa",
|
|
|
|
|
|
4,
|
|
|
|
|
|
None,
|
|
|
|
|
|
"Connected @ token - no completion by design",
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
"aaa @aaa",
|
|
|
|
|
|
5,
|
|
|
|
|
|
Some("aaa".to_string()),
|
|
|
|
|
|
"@ token after space",
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
"test @file.txt",
|
|
|
|
|
|
7,
|
|
|
|
|
|
Some("file.txt".to_string()),
|
|
|
|
|
|
"@ token after space",
|
|
|
|
|
|
),
|
|
|
|
|
|
// Full-width space boundaries
|
|
|
|
|
|
(
|
|
|
|
|
|
"test @İstanbul",
|
2025-08-03 11:31:35 -07:00
|
|
|
|
8,
|
2025-07-08 05:43:31 +09:00
|
|
|
|
Some("İstanbul".to_string()),
|
|
|
|
|
|
"@ token after full-width space",
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
"@ЙЦУ @诶",
|
2025-08-03 11:31:35 -07:00
|
|
|
|
10,
|
2025-07-08 05:43:31 +09:00
|
|
|
|
Some("诶".to_string()),
|
|
|
|
|
|
"Full-width space between Unicode tokens",
|
|
|
|
|
|
),
|
|
|
|
|
|
// Tab and newline boundaries
|
|
|
|
|
|
(
|
|
|
|
|
|
"test\t@file",
|
|
|
|
|
|
6,
|
|
|
|
|
|
Some("file".to_string()),
|
|
|
|
|
|
"@ token after tab",
|
|
|
|
|
|
),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
for (input, cursor_pos, expected, description) in test_cases {
|
2025-08-03 11:31:35 -07:00
|
|
|
|
let mut textarea = TextArea::new();
|
2025-07-08 05:43:31 +09:00
|
|
|
|
textarea.insert_str(input);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
textarea.set_cursor(cursor_pos);
|
2025-07-08 05:43:31 +09:00
|
|
|
|
|
|
|
|
|
|
let result = ChatComposer::current_at_token(&textarea);
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
result, expected,
|
|
|
|
|
|
"Failed for whitespace boundary case: {description} - input: '{input}', cursor: {cursor_pos}",
|
|
|
|
|
|
);
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn handle_paste_small_inserts_text() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
let needs_redraw = composer.handle_paste("hello".to_string());
|
|
|
|
|
|
assert!(needs_redraw);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
assert_eq!(composer.textarea.text(), "hello");
|
2025-07-12 15:32:00 -07:00
|
|
|
|
assert!(composer.pending_pastes.is_empty());
|
|
|
|
|
|
|
|
|
|
|
|
let (result, _) =
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
|
|
|
|
|
|
match result {
|
|
|
|
|
|
InputResult::Submitted(text) => assert_eq!(text, "hello"),
|
|
|
|
|
|
_ => panic!("expected Submitted"),
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn handle_paste_large_uses_placeholder_and_replaces_on_submit() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
let large = "x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 10);
|
|
|
|
|
|
let needs_redraw = composer.handle_paste(large.clone());
|
|
|
|
|
|
assert!(needs_redraw);
|
|
|
|
|
|
let placeholder = format!("[Pasted Content {} chars]", large.chars().count());
|
2025-08-03 11:31:35 -07:00
|
|
|
|
assert_eq!(composer.textarea.text(), placeholder);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
assert_eq!(composer.pending_pastes.len(), 1);
|
|
|
|
|
|
assert_eq!(composer.pending_pastes[0].0, placeholder);
|
|
|
|
|
|
assert_eq!(composer.pending_pastes[0].1, large);
|
|
|
|
|
|
|
|
|
|
|
|
let (result, _) =
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
|
|
|
|
|
|
match result {
|
|
|
|
|
|
InputResult::Submitted(text) => assert_eq!(text, large),
|
|
|
|
|
|
_ => panic!("expected Submitted"),
|
|
|
|
|
|
}
|
|
|
|
|
|
assert!(composer.pending_pastes.is_empty());
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn edit_clears_pending_paste() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let large = "y".repeat(LARGE_PASTE_CHAR_THRESHOLD + 1);
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
composer.handle_paste(large);
|
|
|
|
|
|
assert_eq!(composer.pending_pastes.len(), 1);
|
|
|
|
|
|
|
|
|
|
|
|
// Any edit that removes the placeholder should clear pending_paste
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
|
|
|
|
|
|
assert!(composer.pending_pastes.is_empty());
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn ui_snapshots() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
use insta::assert_snapshot;
|
|
|
|
|
|
use ratatui::Terminal;
|
|
|
|
|
|
use ratatui::backend::TestBackend;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
|
|
|
|
|
let mut terminal = match Terminal::new(TestBackend::new(100, 10)) {
|
|
|
|
|
|
Ok(t) => t,
|
|
|
|
|
|
Err(e) => panic!("Failed to create terminal: {e}"),
|
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
let test_cases = vec![
|
|
|
|
|
|
("empty", None),
|
|
|
|
|
|
("small", Some("short".to_string())),
|
|
|
|
|
|
("large", Some("z".repeat(LARGE_PASTE_CHAR_THRESHOLD + 5))),
|
|
|
|
|
|
("multiple_pastes", None),
|
|
|
|
|
|
("backspace_after_pastes", None),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
for (name, input) in test_cases {
|
|
|
|
|
|
// Create a fresh composer for each test case
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender.clone(), false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
if let Some(text) = input {
|
|
|
|
|
|
composer.handle_paste(text);
|
|
|
|
|
|
} else if name == "multiple_pastes" {
|
|
|
|
|
|
// First large paste
|
|
|
|
|
|
composer.handle_paste("x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 3));
|
|
|
|
|
|
// Second large paste
|
|
|
|
|
|
composer.handle_paste("y".repeat(LARGE_PASTE_CHAR_THRESHOLD + 7));
|
|
|
|
|
|
// Small paste
|
|
|
|
|
|
composer.handle_paste(" another short paste".to_string());
|
|
|
|
|
|
} else if name == "backspace_after_pastes" {
|
|
|
|
|
|
// Three large pastes
|
|
|
|
|
|
composer.handle_paste("a".repeat(LARGE_PASTE_CHAR_THRESHOLD + 2));
|
|
|
|
|
|
composer.handle_paste("b".repeat(LARGE_PASTE_CHAR_THRESHOLD + 4));
|
|
|
|
|
|
composer.handle_paste("c".repeat(LARGE_PASTE_CHAR_THRESHOLD + 6));
|
|
|
|
|
|
// Move cursor to end and press backspace
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.set_cursor(composer.textarea.text().len());
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
terminal
|
|
|
|
|
|
.draw(|f| f.render_widget_ref(&composer, f.area()))
|
|
|
|
|
|
.unwrap_or_else(|e| panic!("Failed to draw {name} composer: {e}"));
|
|
|
|
|
|
|
|
|
|
|
|
assert_snapshot!(name, terminal.backend());
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-06 09:10:23 -07:00
|
|
|
|
#[test]
|
|
|
|
|
|
fn slash_init_dispatches_command_and_does_not_submit_literal_text() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
use std::sync::mpsc::TryRecvError;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
|
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
|
|
|
|
|
|
|
|
|
|
|
// Type the slash command.
|
|
|
|
|
|
for ch in [
|
|
|
|
|
|
'/', 'i', 'n', 'i', 't', // "/init"
|
|
|
|
|
|
] {
|
|
|
|
|
|
let _ = composer.handle_key_event(KeyEvent::new(KeyCode::Char(ch), KeyModifiers::NONE));
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
// Press Enter to dispatch the selected command.
|
|
|
|
|
|
let (result, _needs_redraw) =
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
|
|
|
|
|
|
|
|
|
|
|
|
// When a slash command is dispatched, the composer should not submit
|
|
|
|
|
|
// literal text and should clear its textarea.
|
|
|
|
|
|
match result {
|
|
|
|
|
|
InputResult::None => {}
|
|
|
|
|
|
InputResult::Submitted(text) => {
|
|
|
|
|
|
panic!("expected command dispatch, but composer submitted literal text: {text}")
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
assert!(composer.textarea.is_empty(), "composer should be cleared");
|
|
|
|
|
|
|
|
|
|
|
|
// Verify a DispatchCommand event for the "init" command was sent.
|
|
|
|
|
|
match rx.try_recv() {
|
|
|
|
|
|
Ok(AppEvent::DispatchCommand(cmd)) => {
|
|
|
|
|
|
assert_eq!(cmd.command(), "init");
|
|
|
|
|
|
}
|
|
|
|
|
|
Ok(_other) => panic!("unexpected app event"),
|
|
|
|
|
|
Err(TryRecvError::Empty) => panic!("expected a DispatchCommand event for '/init'"),
|
|
|
|
|
|
Err(TryRecvError::Disconnected) => panic!("app event channel disconnected"),
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-08-11 14:15:41 -07:00
|
|
|
|
#[test]
|
|
|
|
|
|
fn slash_mention_dispatches_command_and_inserts_at() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
use std::sync::mpsc::TryRecvError;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
|
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
|
|
|
|
|
|
|
|
|
|
|
for ch in ['/', 'm', 'e', 'n', 't', 'i', 'o', 'n'] {
|
|
|
|
|
|
let _ = composer.handle_key_event(KeyEvent::new(KeyCode::Char(ch), KeyModifiers::NONE));
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
let (result, _needs_redraw) =
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
|
|
|
|
|
|
|
|
|
|
|
|
match result {
|
|
|
|
|
|
InputResult::None => {}
|
|
|
|
|
|
InputResult::Submitted(text) => {
|
|
|
|
|
|
panic!("expected command dispatch, but composer submitted literal text: {text}")
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
assert!(composer.textarea.is_empty(), "composer should be cleared");
|
|
|
|
|
|
|
|
|
|
|
|
match rx.try_recv() {
|
|
|
|
|
|
Ok(AppEvent::DispatchCommand(cmd)) => {
|
|
|
|
|
|
assert_eq!(cmd.command(), "mention");
|
|
|
|
|
|
composer.insert_str("@");
|
|
|
|
|
|
}
|
|
|
|
|
|
Ok(_other) => panic!("unexpected app event"),
|
|
|
|
|
|
Err(TryRecvError::Empty) => panic!("expected a DispatchCommand event for '/mention'"),
|
|
|
|
|
|
Err(TryRecvError::Disconnected) => {
|
|
|
|
|
|
panic!("app event channel disconnected")
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
assert_eq!(composer.textarea.text(), "@");
|
|
|
|
|
|
}
|
|
|
|
|
|
|
2025-07-12 15:32:00 -07:00
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_multiple_pastes_submission() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
// Define test cases: (paste content, is_large)
|
|
|
|
|
|
let test_cases = [
|
|
|
|
|
|
("x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 3), true),
|
|
|
|
|
|
(" and ".to_string(), false),
|
|
|
|
|
|
("y".repeat(LARGE_PASTE_CHAR_THRESHOLD + 7), true),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
// Expected states after each paste
|
|
|
|
|
|
let mut expected_text = String::new();
|
|
|
|
|
|
let mut expected_pending_count = 0;
|
|
|
|
|
|
|
|
|
|
|
|
// Apply all pastes and build expected state
|
|
|
|
|
|
let states: Vec<_> = test_cases
|
|
|
|
|
|
.iter()
|
|
|
|
|
|
.map(|(content, is_large)| {
|
|
|
|
|
|
composer.handle_paste(content.clone());
|
|
|
|
|
|
if *is_large {
|
|
|
|
|
|
let placeholder = format!("[Pasted Content {} chars]", content.chars().count());
|
|
|
|
|
|
expected_text.push_str(&placeholder);
|
|
|
|
|
|
expected_pending_count += 1;
|
|
|
|
|
|
} else {
|
|
|
|
|
|
expected_text.push_str(content);
|
|
|
|
|
|
}
|
|
|
|
|
|
(expected_text.clone(), expected_pending_count)
|
|
|
|
|
|
})
|
|
|
|
|
|
.collect();
|
|
|
|
|
|
|
|
|
|
|
|
// Verify all intermediate states were correct
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
states,
|
|
|
|
|
|
vec![
|
|
|
|
|
|
(
|
|
|
|
|
|
format!("[Pasted Content {} chars]", test_cases[0].0.chars().count()),
|
|
|
|
|
|
1
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
format!(
|
|
|
|
|
|
"[Pasted Content {} chars] and ",
|
|
|
|
|
|
test_cases[0].0.chars().count()
|
|
|
|
|
|
),
|
|
|
|
|
|
1
|
|
|
|
|
|
),
|
|
|
|
|
|
(
|
|
|
|
|
|
format!(
|
|
|
|
|
|
"[Pasted Content {} chars] and [Pasted Content {} chars]",
|
|
|
|
|
|
test_cases[0].0.chars().count(),
|
|
|
|
|
|
test_cases[2].0.chars().count()
|
|
|
|
|
|
),
|
|
|
|
|
|
2
|
|
|
|
|
|
),
|
|
|
|
|
|
]
|
|
|
|
|
|
);
|
|
|
|
|
|
|
|
|
|
|
|
// Submit and verify final expansion
|
|
|
|
|
|
let (result, _) =
|
|
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Enter, KeyModifiers::NONE));
|
|
|
|
|
|
if let InputResult::Submitted(text) = result {
|
|
|
|
|
|
assert_eq!(text, format!("{} and {}", test_cases[0].0, test_cases[2].0));
|
|
|
|
|
|
} else {
|
|
|
|
|
|
panic!("expected Submitted");
|
|
|
|
|
|
}
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_placeholder_deletion() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
// Define test cases: (content, is_large)
|
|
|
|
|
|
let test_cases = [
|
|
|
|
|
|
("a".repeat(LARGE_PASTE_CHAR_THRESHOLD + 5), true),
|
|
|
|
|
|
(" and ".to_string(), false),
|
|
|
|
|
|
("b".repeat(LARGE_PASTE_CHAR_THRESHOLD + 6), true),
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
// Apply all pastes
|
|
|
|
|
|
let mut current_pos = 0;
|
|
|
|
|
|
let states: Vec<_> = test_cases
|
|
|
|
|
|
.iter()
|
|
|
|
|
|
.map(|(content, is_large)| {
|
|
|
|
|
|
composer.handle_paste(content.clone());
|
|
|
|
|
|
if *is_large {
|
|
|
|
|
|
let placeholder = format!("[Pasted Content {} chars]", content.chars().count());
|
|
|
|
|
|
current_pos += placeholder.len();
|
|
|
|
|
|
} else {
|
|
|
|
|
|
current_pos += content.len();
|
|
|
|
|
|
}
|
|
|
|
|
|
(
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.text().to_string(),
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.pending_pastes.len(),
|
|
|
|
|
|
current_pos,
|
|
|
|
|
|
)
|
|
|
|
|
|
})
|
|
|
|
|
|
.collect();
|
|
|
|
|
|
|
|
|
|
|
|
// Delete placeholders one by one and collect states
|
|
|
|
|
|
let mut deletion_states = vec![];
|
|
|
|
|
|
|
|
|
|
|
|
// First deletion
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.set_cursor(states[0].2);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
|
|
|
|
|
|
deletion_states.push((
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.text().to_string(),
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.pending_pastes.len(),
|
|
|
|
|
|
));
|
|
|
|
|
|
|
|
|
|
|
|
// Second deletion
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.set_cursor(composer.textarea.text().len());
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
|
|
|
|
|
|
deletion_states.push((
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.text().to_string(),
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.pending_pastes.len(),
|
|
|
|
|
|
));
|
|
|
|
|
|
|
|
|
|
|
|
// Verify all states
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
deletion_states,
|
|
|
|
|
|
vec![
|
|
|
|
|
|
(" and [Pasted Content 1006 chars]".to_string(), 1),
|
|
|
|
|
|
(" and ".to_string(), 0),
|
|
|
|
|
|
]
|
|
|
|
|
|
);
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
#[test]
|
|
|
|
|
|
fn test_partial_placeholder_deletion() {
|
|
|
|
|
|
use crossterm::event::KeyCode;
|
|
|
|
|
|
use crossterm::event::KeyEvent;
|
|
|
|
|
|
use crossterm::event::KeyModifiers;
|
|
|
|
|
|
|
|
|
|
|
|
let (tx, _rx) = std::sync::mpsc::channel();
|
|
|
|
|
|
let sender = AppEventSender::new(tx);
|
2025-07-31 17:30:44 -07:00
|
|
|
|
let mut composer = ChatComposer::new(true, sender, false);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
|
|
|
|
|
|
// Define test cases: (cursor_position_from_end, expected_pending_count)
|
|
|
|
|
|
let test_cases = [
|
|
|
|
|
|
5, // Delete from middle - should clear tracking
|
|
|
|
|
|
0, // Delete from end - should clear tracking
|
|
|
|
|
|
];
|
|
|
|
|
|
|
|
|
|
|
|
let paste = "x".repeat(LARGE_PASTE_CHAR_THRESHOLD + 4);
|
|
|
|
|
|
let placeholder = format!("[Pasted Content {} chars]", paste.chars().count());
|
|
|
|
|
|
|
|
|
|
|
|
let states: Vec<_> = test_cases
|
|
|
|
|
|
.into_iter()
|
|
|
|
|
|
.map(|pos_from_end| {
|
|
|
|
|
|
composer.handle_paste(paste.clone());
|
|
|
|
|
|
composer
|
|
|
|
|
|
.textarea
|
2025-08-03 11:31:35 -07:00
|
|
|
|
.set_cursor((placeholder.len() - pos_from_end) as usize);
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.handle_key_event(KeyEvent::new(KeyCode::Backspace, KeyModifiers::NONE));
|
|
|
|
|
|
let result = (
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.text().contains(&placeholder),
|
2025-07-12 15:32:00 -07:00
|
|
|
|
composer.pending_pastes.len(),
|
|
|
|
|
|
);
|
2025-08-03 11:31:35 -07:00
|
|
|
|
composer.textarea.set_text("");
|
2025-07-12 15:32:00 -07:00
|
|
|
|
result
|
|
|
|
|
|
})
|
|
|
|
|
|
.collect();
|
|
|
|
|
|
|
|
|
|
|
|
assert_eq!(
|
|
|
|
|
|
states,
|
|
|
|
|
|
vec![
|
|
|
|
|
|
(false, 0), // After deleting from middle
|
|
|
|
|
|
(false, 0), // After deleting from end
|
|
|
|
|
|
]
|
|
|
|
|
|
);
|
|
|
|
|
|
}
|
2025-07-08 05:43:31 +09:00
|
|
|
|
}
|