[rendering-scripts] fix: add pretty-print format support to copilot log parser by github-actions[bot] · Pull Request #24070 · github/gh-aw

github-actions · 2026-04-02T11:27:43Z

Discovered via run §23896942706 (Issue Monster, copilot engine).

Problem

When the Copilot CLI is invoked with --log-dir (which routes debug JSON logs to a file), agent-stdio.log only captures the pretty-print stdout format that uses ✗/●/✓ markers for tool calls. The existing parser tried three formats — JSON array, [DEBUG] (redacted) debug blocks, and JSONL — and fell through to "Log format not recognized".

This affected every Copilot workflow run that uses the standard --log-dir flag.

Fix

Added parsePrettyPrintFormat() to parse_copilot_log.cjs as a fourth fallback. The function:

Detects the format by looking for ✗, ●, or ✓ markers
Filters Docker Compose lifecycle lines ( Container awf-...) and infra log lines ([INFO], [entrypoint], etc.)
Extracts failed tool calls (✗), successful tool calls (● or ✓), and their indented continuation output (└)
Collects agent text paragraphs
Parses token counts from the Breakdown by AI model: section
Parses the Turns: line from the statistics section for accurate turn count (falls back to tool entry count when absent)
Sets model_info.billing.is_premium = true when premium model usage is detected

Also updated additionalInfoCallback to read _premium_requests from the result entry (set by the new parser) instead of re-scanning the raw log content.

Testing

All 29 parse_copilot_log tests pass (23 original + 6 new)
New tests cover: ●/✗ marker parsing, ✓ marker parsing, continuation/indented output capture, token count extraction, premium flag detection, and Turns: count parsing
Manually verified against run §23896942706: summary grew from 147 bytes ("Log format not recognized") to 815 bytes with actual conversation content

Generated by Daily Rendering Scripts Verifier · ◷

expires on Apr 5, 2026, 11:27 AM UTC

When the Copilot CLI runs with --log-dir pointing to a directory, debug JSON logs go to that directory and stdout only contains the pretty-print format (✗/● markers for tool calls). The existing parser only handled JSON array, debug log (with [DEBUG] data: blocks), and JSONL formats, so all Copilot runs with a --log-dir produced "Log format not recognized" in the step summary instead of useful output. Adds parsePrettyPrintFormat() which detects and parses the ✗/● marker format by: - Filtering Docker Compose and infrastructure log lines - Extracting failed tool calls (✗), successful tool calls (●), and their indented continuation lines (└) - Collecting agent text paragraphs - Parsing token usage from the model breakdown section Also updates additionalInfoCallback to read _premium_requests from the result entry (set by parsePrettyPrintFormat) rather than re-scanning the raw log content. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Copilot

Pull request overview

Adds support for parsing the Copilot CLI “pretty-print” stdout log format (tool markers + indented continuations) so Copilot workflow runs using --log-dir produce a usable Agent Log Summary instead of “format not recognized”.

Changes:

Adds a new parsePrettyPrintFormat() fallback in parseCopilotLog() when JSON array / debug-block / JSONL parsing yields no entries.
Extracts tool calls (success/failure), agent text, token usage, and premium usage metadata from the pretty-print format.
Updates the Information section’s premium request count logic to prefer a parsed _premium_requests value when available.

Comments suppressed due to low confidence (1)

actions/setup/js/parse_copilot_log.cjs:159

SUCCESS_TOOL_RE currently matches only the ● marker. If the CLI emits ✓ for successful tool calls, those lines will be misclassified as agent text instead of tool calls. Consider expanding this regex to match both ✓ and ● (and keep the comment/docstring consistent with the accepted markers).

  const FAILED_TOOL_RE = /^✗\s+(\S+)/;
  const SUCCESS_TOOL_RE = /^●\s+(\S+)/;
  const CONTINUATION_RE = /^\s+[└│]/;

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-02T11:32:49Z

actions/setup/js/parse_copilot_log.cjs

+ * This format uses ✗ for failed tool calls and ● for successful ones.
+ * @param {string} logContent - Raw log content as a string
+ * @returns {Array} Array of log entries in structured format, or empty array if not detected
+ */
+function parsePrettyPrintFormat(logContent) {
+  // Only attempt this format if the characteristic markers are present
+  if (!/^[✗●]/m.test(logContent)) {
+    return [];
+  }
+
+  const INFRA_LINE_RE = /^\[(INFO|WARN|SUCCESS|ERROR|entrypoint|health-check)\]|^ (?:Container|Network|Volume) |^Process exiting with code:/;
+  const FAILED_TOOL_RE = /^✗\s+(\S+)/;
+  const SUCCESS_TOOL_RE = /^●\s+(\S+)/;


The pretty-print format detection only looks for lines starting with ✗ or ●. Copilot CLI output (and this PR’s example) appears to use ✓ for successful tool calls, and runs with only successful tools would fail detection and fall through to “Log format not recognized”. Consider accepting ✓ as a success marker (and including it in the initial marker detection).

This issue also appears on line 157 of the same file.

Suggested change

* This format uses ✗ for failed tool calls and ● for successful ones.

* @param {string} logContent - Raw log content as a string

* @returns {Array} Array of log entries in structured format, or empty array if not detected

*/

function parsePrettyPrintFormat(logContent) {

// Only attempt this format if the characteristic markers are present

if (!/^[✗●]/m.test(logContent)) {

return [];

}

const INFRA_LINE_RE = /^\[(INFO|WARN|SUCCESS|ERROR|entrypoint|health-check)\]|^ (?:Container|Network|Volume) |^Process exiting with code:/;

const FAILED_TOOL_RE = /^✗\s+(\S+)/;

const SUCCESS_TOOL_RE = /^●\s+(\S+)/;

* This format uses ✗ for failed tool calls and ● or ✓ for successful ones.

* @param {string} logContent - Raw log content as a string

* @returns {Array} Array of log entries in structured format, or empty array if not detected

*/

function parsePrettyPrintFormat(logContent) {

// Only attempt this format if the characteristic markers are present

if (!/^[✗●✓]/m.test(logContent)) {

return [];

}

const INFRA_LINE_RE = /^\[(INFO|WARN|SUCCESS|ERROR|entrypoint|health-check)\]|^ (?:Container|Network|Volume) |^Process exiting with code:/;

const FAILED_TOOL_RE = /^✗\s+(\S+)/;

const SUCCESS_TOOL_RE = /^(?:●|✓)\s+(\S+)/;

Copilot · 2026-04-02T11:32:49Z

actions/setup/js/parse_copilot_log.cjs

+      // If still nothing, try the pretty-print stdout format (✗/● markers)
+      if (!logEntries || logEntries.length === 0) {
+        const prettyPrintEntries = parsePrettyPrintFormat(logContent);
+        if (prettyPrintEntries && prettyPrintEntries.length > 0) {
+          logEntries = prettyPrintEntries;


New pretty-print fallback parsing is not covered by tests. Since this adds a new supported log format (and regex-heavy parsing), add vitest cases in parse_copilot_log.test.cjs exercising: (1) success + failure tool markers, (2) continuation/indented output capture, and (3) token + premium request extraction.

Copilot · 2026-04-02T11:32:49Z

actions/setup/js/parse_copilot_log.cjs

+  entries.push({
+    type: "result",
+    num_turns: toolEntries.length,


num_turns is currently set to toolEntries.length, which is likely not the same as the CLI’s “Turns:” count (e.g., a single turn may include multiple tool calls, or zero tool calls). This will make the Information section’s Turns value inaccurate. Consider parsing the “Turns:” line from the Statistics/usage section (if present) and storing that value in the result entry instead.

Suggested change

entries.push({

type: "result",

num_turns: toolEntries.length,

// Derive the number of turns from the CLI's "Turns:" statistic if available.

// Fallback to the number of tool entries to preserve existing behavior when absent.

let numTurns = toolEntries.length;

const turnsMatch = logContent && logContent.match(/Turns:\s*(\d+)/i);

if (turnsMatch && turnsMatch[1]) {

const parsedTurns = parseInt(turnsMatch[1], 10);

if (!Number.isNaN(parsedTurns) && parsedTurns > 0) {

numTurns = parsedTurns;

}

}

entries.push({

type: "result",

num_turns: numTurns,

pelikhan · 2026-04-02T11:34:48Z

@copilot review comments

- Add ✓ as success marker alongside ● in detection regex and SUCCESS_TOOL_RE - Parse Turns: line from log for num_turns with fallback to toolEntries.length - Add 6 tests covering: success/failure markers, ✓ marker, continuation output, token extraction, premium flag, and Turns: count parsing Agent-Logs-Url: https://github.com/github/gh-aw/sessions/74f36bfb-a0dd-4357-b3d4-74c9f6569bec Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

- Remove unnecessary `logContent &&` guard (logContent is already used without null checks) - Use optional chaining `resultEntry?.num_turns` in test to prevent potential null reference Agent-Logs-Url: https://github.com/github/gh-aw/sessions/74f36bfb-a0dd-4357-b3d4-74c9f6569bec Co-authored-by: pelikhan <4175913+pelikhan@users.noreply.github.com>

Copilot · 2026-04-02T11:43:31Z

@copilot review comments

Addressed all three review comments in commit 6cfc772:

✓ marker support — Updated the detection regex from /^[✗●]/m to /^[✗●✓]/m and changed SUCCESS_TOOL_RE from /^●\s+(\S+)/ to /^(?:●|✓)\s+(\S+)/ so runs with only ✓-marked tool calls are correctly detected.
num_turns accuracy — Added parsing of the Turns: line from the statistics section, with fallback to toolEntries.length when absent.
Tests — Added 6 new test cases covering: ●/✗ markers, ✓ marker, continuation/indented output capture, token count extraction from the breakdown section, premium flag detection, and Turns: count parsing. All 29 tests pass.

github-actions bot added automated-fix javascript Pull requests that update javascript code rendering labels Apr 2, 2026

pelikhan marked this pull request as ready for review April 2, 2026 11:29

Copilot AI review requested due to automatic review settings April 2, 2026 11:29

Copilot started reviewing on behalf of pelikhan April 2, 2026 11:30 View session

Copilot AI reviewed Apr 2, 2026

View reviewed changes

Copilot started work on behalf of pelikhan April 2, 2026 11:35 View session

Copilot AI and others added 2 commits April 2, 2026 11:42

Copilot finished work on behalf of pelikhan April 2, 2026 11:44

Copilot AI requested a review from pelikhan April 2, 2026 11:44

pelikhan merged commit b56a7f0 into main Apr 2, 2026
80 of 83 checks passed

pelikhan deleted the fix/copilot-pretty-print-parser-format-c26bdd4d87f202ae branch April 2, 2026 11:47

github-actions bot mentioned this pull request Apr 2, 2026

Smoke Test: Copilot - 23899414681 #24092

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[rendering-scripts] fix: add pretty-print format support to copilot log parser#24070

[rendering-scripts] fix: add pretty-print format support to copilot log parser#24070
pelikhan merged 3 commits intomainfrom
fix/copilot-pretty-print-parser-format-c26bdd4d87f202ae

github-actions bot commented Apr 2, 2026 •

edited by Copilot AI

Loading

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Apr 2, 2026

Uh oh!

Copilot AI Apr 2, 2026

Uh oh!

Copilot AI Apr 2, 2026

Uh oh!

pelikhan commented Apr 2, 2026

Uh oh!

Copilot AI commented Apr 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

-  entries.push({
-    type: "result",
-    num_turns: toolEntries.length,
+  // Derive the number of turns from the CLI's "Turns:" statistic if available.
+  // Fallback to the number of tool entries to preserve existing behavior when absent.
+  let numTurns = toolEntries.length;
+  const turnsMatch = logContent && logContent.match(/Turns:\s*(\d+)/i);
+  if (turnsMatch && turnsMatch[1]) {
+    const parsedTurns = parseInt(turnsMatch[1], 10);
+    if (!Number.isNaN(parsedTurns) && parsedTurns > 0) {
+      numTurns = parsedTurns;
+    }
+  }
+  entries.push({
+    type: "result",
+    num_turns: numTurns,

Conversation

github-actions bot commented Apr 2, 2026 • edited by Copilot AI Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Problem

Fix

Testing

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI Apr 2, 2026

Choose a reason for hiding this comment

Uh oh!

pelikhan commented Apr 2, 2026

Uh oh!

Copilot AI commented Apr 2, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

github-actions bot commented Apr 2, 2026 •

edited by Copilot AI

Loading