include fractional portion of chunk that exceeds stdout/stderr limit (#497)

I saw cases where the first chunk of output from `ls -R` could be large
enough to exceed `MAX_OUTPUT_BYTES` or `MAX_OUTPUT_LINES`, in which case
the loop would exit early in `createTruncatingCollector()` such that
nothing was appended to the `chunks` array. As a result, the reported
`stdout` of `ls -R` would be empty.

I asked Codex to add logic to handle this edge case and write a unit
test. I used this as my test:

```
./codex-cli/dist/cli.js -q 'what is the output of `ls -R`'
```

now it appears to include a ton of stuff whereas before this change, I
saw:

```
{"type":"function_call_output","call_id":"call_a2QhVt7HRJYKjb3dIc8w1aBB","output":"{\"output\":\"\\n\\n[Output truncated: too many lines or bytes]\",\"metadata\":{\"exit_code\":0,\"duration_seconds\":0.5}}"}
```

codex-cli/src/utils/agent/sandbox/create-truncating-collector.ts

@@ -0,0 +1,78 @@
+// Maximum output cap: either MAX_OUTPUT_LINES lines or MAX_OUTPUT_BYTES bytes,
+// whichever limit is reached first.
+const MAX_OUTPUT_BYTES = 1024 * 10; // 10 KB
+const MAX_OUTPUT_LINES = 256;
+
+/**
+ * Creates a collector that accumulates data Buffers from a stream up to
+ * specified byte and line limits. After either limit is exceeded, further
+ * data is ignored.
+ */
+export function createTruncatingCollector(
+  stream: NodeJS.ReadableStream,
+  byteLimit: number = MAX_OUTPUT_BYTES,
+  lineLimit: number = MAX_OUTPUT_LINES,
+): {
+  getString: () => string;
+  hit: boolean;
+} {
+  const chunks: Array<Buffer> = [];
+  let totalBytes = 0;
+  let totalLines = 0;
+  let hitLimit = false;
+
+  stream?.on("data", (data: Buffer) => {
+    if (hitLimit) {
+      return;
+    }
+    const dataLength = data.length;
+    let newlineCount = 0;
+    for (let i = 0; i < dataLength; i++) {
+      if (data[i] === 0x0a) {
+        newlineCount++;
+      }
+    }
+    // If entire chunk fits within byte and line limits, take it whole
+    if (
+      totalBytes + dataLength <= byteLimit &&
+      totalLines + newlineCount <= lineLimit
+    ) {
+      chunks.push(data);
+      totalBytes += dataLength;
+      totalLines += newlineCount;
+    } else {
+      // Otherwise, take a partial slice up to the first limit breach
+      const allowedBytes = byteLimit - totalBytes;
+      const allowedLines = lineLimit - totalLines;
+      let bytesTaken = 0;
+      let linesSeen = 0;
+      for (let i = 0; i < dataLength; i++) {
+        // Stop if byte or line limit is reached
+        if (bytesTaken === allowedBytes || linesSeen === allowedLines) {
+          break;
+        }
+        const byte = data[i];
+        if (byte === 0x0a) {
+          linesSeen++;
+        }
+        bytesTaken++;
+      }
+      if (bytesTaken > 0) {
+        chunks.push(data.slice(0, bytesTaken));
+        totalBytes += bytesTaken;
+        totalLines += linesSeen;
+      }
+      hitLimit = true;
+    }
+  });
+
+  return {
+    getString() {
+      return Buffer.concat(chunks).toString("utf8");
+    },
+    /** True if either byte or line limit was exceeded */
+    get hit(): boolean {
+      return hitLimit;
+    },
+  };
+}

codex-cli/src/utils/agent/sandbox/raw-exec.ts

@@ -9,14 +9,10 @@ import type {
 
 import { log } from "../../logger/log.js";
 import { adaptCommandForPlatform } from "../platform-commands.js";
+import { createTruncatingCollector } from "./create-truncating-collector";
 import { spawn } from "child_process";
 import * as os from "os";
 
-// Maximum output cap: either MAX_OUTPUT_LINES lines or MAX_OUTPUT_BYTES bytes,
-// whichever limit is reached first.
-const MAX_OUTPUT_BYTES = 1024 * 10; // 10 KB
-const MAX_OUTPUT_LINES = 256;
-
 /**
  * This function should never return a rejected promise: errors should be
  * mapped to a non-zero exit code and the error message should be in stderr.
@@ -204,49 +200,6 @@ export function exec(
   });
 }
 
-/**
- * Creates a collector that accumulates data Buffers from a stream up to
- * specified byte and line limits. After either limit is exceeded, further
- * data is ignored.
- */
-function createTruncatingCollector(
-  stream: NodeJS.ReadableStream,
-  byteLimit: number = MAX_OUTPUT_BYTES,
-  lineLimit: number = MAX_OUTPUT_LINES,
-) {
-  const chunks: Array<Buffer> = [];
-  let totalBytes = 0;
-  let totalLines = 0;
-  let hitLimit = false;
-
-  stream?.on("data", (data: Buffer) => {
-    if (hitLimit) {
-      return;
-    }
-    totalBytes += data.length;
-    for (let i = 0; i < data.length; i++) {
-      if (data[i] === 0x0a) {
-        totalLines++;
-      }
-    }
-    if (totalBytes <= byteLimit && totalLines <= lineLimit) {
-      chunks.push(data);
-    } else {
-      hitLimit = true;
-    }
-  });
-
-  return {
-    getString() {
-      return Buffer.concat(chunks).toString("utf8");
-    },
-    /** True if either byte or line limit was exceeded */
-    get hit(): boolean {
-      return hitLimit;
-    },
-  };
-}
-
 /**
  * Adds a truncation warnings to stdout and stderr, if appropriate.
  */