Files
llmx/codex-rs/tui/src/bottom_pane/chat_composer.rs

361 lines
13 KiB
Rust
Raw Normal View History

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: ![Screenshot 2025-06-25 at 11 20 55 PM](https://github.com/user-attachments/assets/6fd6982f-7247-4f14-84b2-2e600cb1fd49) 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;
use crossterm::event::KeyEvent;
use ratatui::buffer::Buffer;
use ratatui::layout::Alignment;
use ratatui::layout::Rect;
use ratatui::style::Style;
use ratatui::style::Stylize;
use ratatui::text::Line;
use ratatui::widgets::BorderType;
use ratatui::widgets::Borders;
use ratatui::widgets::Widget;
use ratatui::widgets::WidgetRef;
use tui_textarea::Input;
use tui_textarea::Key;
use tui_textarea::TextArea;
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;
use crate::app_event::AppEvent;
use crate::app_event_sender::AppEventSender;
/// Minimum number of visible text rows inside the textarea.
const MIN_TEXTAREA_ROWS: usize = 1;
/// Rows consumed by the border.
const BORDER_LINES: u16 = 2;
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: ![Screenshot 2025-06-25 at 11 20 55 PM](https://github.com/user-attachments/assets/6fd6982f-7247-4f14-84b2-2e600cb1fd49) 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
const BASE_PLACEHOLDER_TEXT: &str = "send a message";
/// Result returned when the user interacts with the text area.
pub enum InputResult {
Submitted(String),
None,
}
pub(crate) struct ChatComposer<'a> {
textarea: TextArea<'a>,
command_popup: Option<CommandPopup>,
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,
}
impl ChatComposer<'_> {
pub fn new(has_input_focus: bool, app_event_tx: AppEventSender) -> Self {
let mut textarea = TextArea::default();
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: ![Screenshot 2025-06-25 at 11 20 55 PM](https://github.com/user-attachments/assets/6fd6982f-7247-4f14-84b2-2e600cb1fd49) 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
textarea.set_placeholder_text(BASE_PLACEHOLDER_TEXT);
textarea.set_cursor_line_style(ratatui::style::Style::default());
let mut this = Self {
textarea,
command_popup: None,
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(),
};
this.update_border(has_input_focus);
this
}
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: ![Screenshot 2025-06-25 at 11 20 55 PM](https://github.com/user-attachments/assets/6fd6982f-7247-4f14-84b2-2e600cb1fd49) 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,
token_usage: TokenUsage,
model_context_window: Option<u64>,
) {
let placeholder = match (token_usage.total_tokens, model_context_window) {
(total_tokens, Some(context_window)) => {
let percent_remaining: u8 = if context_window > 0 {
// Calculate the percentage of context left.
let percent = 100.0 - (total_tokens as f32 / context_window as f32 * 100.0);
percent.clamp(0.0, 100.0) as u8
} else {
// If we don't have a context window, we cannot compute the
// percentage.
100
};
if percent_remaining > 25 {
format!("{BASE_PLACEHOLDER_TEXT}{percent_remaining}% context left")
} else {
format!(
"{BASE_PLACEHOLDER_TEXT} — {percent_remaining}% context left (consider /compact)"
)
}
}
(total_tokens, None) => {
format!("{BASE_PLACEHOLDER_TEXT}{total_tokens} tokens used")
}
};
self.textarea.set_placeholder_text(placeholder);
}
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 {
self.history
.on_entry_response(log_id, offset, entry, &mut self.textarea)
}
pub fn set_input_focus(&mut self, has_focus: bool) {
self.update_border(has_focus);
}
/// Handle a key event coming from the main UI.
pub fn handle_key_event(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
let result = match self.command_popup {
Some(_) => self.handle_key_event_with_popup(key_event),
None => self.handle_key_event_without_popup(key_event),
};
// Update (or hide/show) popup after processing the key.
self.sync_command_popup();
result
}
/// Handle key event when the slash-command popup is visible.
fn handle_key_event_with_popup(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
let Some(popup) = self.command_popup.as_mut() else {
tracing::error!("handle_key_event_with_popup called without an active popup");
return (InputResult::None, false);
};
match key_event.into() {
Input { key: Key::Up, .. } => {
popup.move_up();
(InputResult::None, true)
}
Input { key: Key::Down, .. } => {
popup.move_down();
(InputResult::None, true)
}
Input { key: Key::Tab, .. } => {
if let Some(cmd) = popup.selected_command() {
let first_line = self
.textarea
.lines()
.first()
.map(|s| s.as_str())
.unwrap_or("");
let starts_with_cmd = first_line
.trim_start()
.starts_with(&format!("/{}", cmd.command()));
if !starts_with_cmd {
self.textarea.select_all();
self.textarea.cut();
let _ = self.textarea.insert_str(format!("/{} ", cmd.command()));
}
}
(InputResult::None, true)
}
Input {
key: Key::Enter,
shift: false,
alt: false,
ctrl: false,
} => {
if let Some(cmd) = popup.selected_command() {
// Send command to the app layer.
self.app_event_tx.send(AppEvent::DispatchCommand(*cmd));
// Clear textarea so no residual text remains.
self.textarea.select_all();
self.textarea.cut();
// Hide popup since the command has been dispatched.
self.command_popup = None;
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),
}
}
/// Handle key event when no popup is visible.
fn handle_key_event_without_popup(&mut self, key_event: KeyEvent) -> (InputResult, bool) {
let input: Input = key_event.into();
match input {
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.
// -------------------------------------------------------------
Input { key: Key::Up, .. } => {
if self.history.should_handle_navigation(&self.textarea) {
let consumed = self
.history
.navigate_up(&mut self.textarea, &self.app_event_tx);
if consumed {
return (InputResult::None, true);
}
}
self.handle_input_basic(input)
}
Input { key: Key::Down, .. } => {
if self.history.should_handle_navigation(&self.textarea) {
let consumed = self
.history
.navigate_down(&mut self.textarea, &self.app_event_tx);
if consumed {
return (InputResult::None, true);
}
}
self.handle_input_basic(input)
}
Input {
key: Key::Enter,
shift: false,
alt: false,
ctrl: false,
} => {
let text = self.textarea.lines().join("\n");
self.textarea.select_all();
self.textarea.cut();
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)
}
}
Input {
key: Key::Enter, ..
}
| Input {
key: Key::Char('j'),
ctrl: true,
alt: false,
shift: false,
} => {
self.textarea.insert_newline();
(InputResult::None, true)
}
input => self.handle_input_basic(input),
}
}
/// Handle generic Input events that modify the textarea content.
fn handle_input_basic(&mut self, input: Input) -> (InputResult, bool) {
self.textarea.input(input);
(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) {
// Inspect only the first line to decide whether to show the popup. In
// the common case (no leading slash) we avoid copying the entire
// textarea contents.
let first_line = self
.textarea
.lines()
.first()
.map(|s| s.as_str())
.unwrap_or("");
if first_line.starts_with('/') {
// Create popup lazily when the user starts a slash command.
let popup = self.command_popup.get_or_insert_with(CommandPopup::new);
// Forward *only* the first line since `CommandPopup` only needs
// the command token.
popup.on_composer_text_change(first_line.to_string());
} else if self.command_popup.is_some() {
// Remove popup when '/' is no longer the first character.
self.command_popup = None;
}
}
pub fn calculate_required_height(&self, area: &Rect) -> u16 {
let rows = self.textarea.lines().len().max(MIN_TEXTAREA_ROWS);
let num_popup_rows = if let Some(popup) = &self.command_popup {
popup.calculate_required_height(area)
} else {
0
};
rows as u16 + BORDER_LINES + num_popup_rows
}
fn update_border(&mut self, has_focus: bool) {
struct BlockState {
right_title: Line<'static>,
border_style: Style,
}
let bs = if has_focus {
BlockState {
right_title: Line::from("Enter to send | Ctrl+D to quit | Ctrl+J for newline")
.alignment(Alignment::Right),
border_style: Style::default(),
}
} else {
BlockState {
right_title: Line::from(""),
border_style: Style::default().dim(),
}
};
self.textarea.set_block(
ratatui::widgets::Block::default()
.title_bottom(bs.right_title)
.borders(Borders::ALL)
.border_type(BorderType::Rounded)
.border_style(bs.border_style),
);
}
pub(crate) fn is_command_popup_visible(&self) -> bool {
self.command_popup.is_some()
}
}
impl WidgetRef for &ChatComposer<'_> {
fn render_ref(&self, area: Rect, buf: &mut Buffer) {
if let Some(popup) = &self.command_popup {
let popup_height = popup.calculate_required_height(&area);
// Split the provided rect so that the popup is rendered at the
// *top* and the textarea occupies the remaining space below.
let popup_rect = Rect {
x: area.x,
y: area.y,
width: area.width,
height: popup_height.min(area.height),
};
let textarea_rect = Rect {
x: area.x,
y: area.y + popup_rect.height,
width: area.width,
height: area.height.saturating_sub(popup_rect.height),
};
popup.render(popup_rect, buf);
self.textarea.render(textarea_rect, buf);
} else {
self.textarea.render(area, buf);
}
}
}