Browse Source

chore(release): unwrap CHANGELOG paragraphs for GitHub Release notes

GitHub renders release-note Markdown with GFM hard breaks, so every
`\n` becomes `<br>`. The CHANGELOG is hard-wrapped at ~75 chars for
readable diffs, which renders as awkward visible line breaks on the
release page (see https://github.com/colbymchenry/codegraph/releases/tag/v0.7.10).

Add `scripts/extract-release-notes.mjs` to extract a version block
and join indented continuation lines into a single line per bullet.
Nested list items, headings, and link references are preserved.
`scripts/release.sh` now uses this helper instead of the inline awk
extractor — repo-level CHANGELOG.md viewing is unaffected because
CommonMark there treats newlines as spaces.

Also fix the 0.7.10 entry: "Two underlying fixes" -> "Three", "Rust
file-/level" broken hyphen, and move the closes/credit line above
the nested list so it doesn't strand as a top-level paragraph.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Colby McHenry 1 month ago
parent
commit
03f08ff899
3 changed files with 128 additions and 15 deletions
  1. 7 8
      CHANGELOG.md
  2. 116 0
      scripts/extract-release-notes.mjs
  3. 5 7
      scripts/release.sh

+ 7 - 8
CHANGELOG.md

@@ -42,24 +42,23 @@ and adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
   …) accept `module::symbol` (Rust / C++ / Ruby), `Module.symbol`
   …) accept `module::symbol` (Rust / C++ / Ruby), `Module.symbol`
   (TS / JS / Python), and `module/symbol` (path-style) — multi-level
   (TS / JS / Python), and `module/symbol` (path-style) — multi-level
   forms (`crate::configurator::stage_apply::run`) and Rust path
   forms (`crate::configurator::stage_apply::run`) and Rust path
-  prefixes (`crate`, `super`, `self`) are handled. Two underlying
-  fixes:
+  prefixes (`crate`, `super`, `self`) are handled. Closes
+  [#173](https://github.com/colbymchenry/codegraph/issues/173). Thanks
+  to [@joselhurtado](https://github.com/joselhurtado) for the detailed
+  reproduction. Three underlying fixes:
     - The FTS5 query builder now treats `::` as a token separator
     - The FTS5 query builder now treats `::` as a token separator
       instead of stripping it to nothing, so `stage_apply::run` no
       instead of stripping it to nothing, so `stage_apply::run` no
       longer collapses to the unsearchable `stage_applyrun`.
       longer collapses to the unsearchable `stage_applyrun`.
     - `matchesSymbol` falls back to a file-path containment check when
     - `matchesSymbol` falls back to a file-path containment check when
-      `qualifiedName` doesn't carry the module hierarchy (Rust file-
-      level functions, Python free functions in a package): a `run`
-      in `src/configurator/stage_apply.rs` now matches
+      `qualifiedName` doesn't carry the module hierarchy (Rust
+      file-level functions, Python free functions in a package): a
+      `run` in `src/configurator/stage_apply.rs` now matches
       `stage_apply::run` because `stage_apply` appears as a path
       `stage_apply::run` because `stage_apply` appears as a path
       segment.
       segment.
     - Qualified lookups that don't match the qualifier no longer fall
     - Qualified lookups that don't match the qualifier no longer fall
       through to fuzzy text matches — `stage_apply::nonexistent_fn`
       through to fuzzy text matches — `stage_apply::nonexistent_fn`
       returns `null` instead of resolving to an unrelated `rollback`
       returns `null` instead of resolving to an unrelated `rollback`
       in the same file.
       in the same file.
-  Closes [#173](https://github.com/colbymchenry/codegraph/issues/173).
-  Thanks to [@joselhurtado](https://github.com/joselhurtado) for the
-  detailed reproduction.
 
 
 [0.7.10]: https://github.com/colbymchenry/codegraph/releases/tag/v0.7.10
 [0.7.10]: https://github.com/colbymchenry/codegraph/releases/tag/v0.7.10
 
 

+ 116 - 0
scripts/extract-release-notes.mjs

@@ -0,0 +1,116 @@
+#!/usr/bin/env node
+/**
+ * Extract a release-notes block from CHANGELOG.md for a given version,
+ * then unwrap hard-wrapped paragraphs.
+ *
+ * Why: GitHub renders release-note Markdown with GFM hard breaks, so
+ * every `\n` becomes `<br>`. The CHANGELOG is hard-wrapped at ~75
+ * chars for readable diffs, which then renders as awkward visible
+ * line breaks on the release page. This script joins indented
+ * continuation lines into a single line per bullet so the GFM
+ * renderer produces clean paragraphs.
+ *
+ * Repo-level CHANGELOG.md viewing is unaffected (CommonMark treats
+ * newlines as spaces there).
+ *
+ * Usage: extract-release-notes.mjs <version>
+ *        e.g. extract-release-notes.mjs 0.7.10
+ */
+
+import { readFileSync } from 'fs';
+
+const version = process.argv[2];
+if (!version) {
+  console.error('usage: extract-release-notes.mjs <version>');
+  process.exit(1);
+}
+
+const escaped = version.replace(/\./g, '\\.');
+const headerRe = new RegExp(`^## \\[${escaped}\\]`);
+const anyHeaderRe = /^## \[/;
+
+const lines = readFileSync('CHANGELOG.md', 'utf8').split('\n');
+const start = lines.findIndex((l) => headerRe.test(l));
+if (start === -1) {
+  console.error(`no '## [${version}]' entry found in CHANGELOG.md`);
+  process.exit(1);
+}
+const after = lines.findIndex((l, i) => i > start && anyHeaderRe.test(l));
+const block = lines.slice(start, after === -1 ? lines.length : after);
+
+// Find the indent of the most recent list item; a continuation line
+// whose indent is GREATER than that belongs to that item, otherwise
+// it might belong to an ancestor item further up the stack.
+//
+// Track a stack of `{ indent: number }` frames so we can attach a
+// continuation to the right ancestor. This correctly handles the
+// post-nested-list continuation pattern:
+//
+//     - top-level
+//         - nested
+//       back to top-level  <- 2-space indent, joins the top-level bullet
+const out = [];
+let buf = '';                                // pending list-item text being built
+let stack = [];                              // [{ indent: number }] open list items
+
+function flushBuf() {
+  if (buf !== '') {
+    out.push(buf);
+    buf = '';
+  }
+}
+
+function leadingSpaces(s) {
+  const m = s.match(/^(\s*)/);
+  return m ? m[1].length : 0;
+}
+
+const listItemRe = /^(\s*)([-*+]|\d+\.)\s+/;
+
+for (const line of block) {
+  if (/^\s*$/.test(line)) {
+    flushBuf();
+    out.push('');
+    continue;
+  }
+  if (/^#/.test(line)) {
+    flushBuf();
+    stack = [];
+    out.push(line);
+    continue;
+  }
+  const itemMatch = line.match(listItemRe);
+  if (itemMatch) {
+    flushBuf();
+    const indent = itemMatch[1].length;
+    while (stack.length > 0 && stack[stack.length - 1].indent >= indent) {
+      stack.pop();
+    }
+    stack.push({ indent });
+    buf = line;
+    continue;
+  }
+  if (/^\s/.test(line)) {
+    // Continuation. Pop any list frames deeper than this indent — the
+    // continuation belongs to the nearest enclosing list item.
+    const indent = leadingSpaces(line);
+    while (stack.length > 1 && stack[stack.length - 1].indent >= indent) {
+      // Closes the deeper item — its buffered text is already in `buf`
+      // belonging to the most recent flush. We need to flush before
+      // re-buffering for the ancestor item.
+      flushBuf();
+      stack.pop();
+    }
+    const trimmed = line.replace(/^\s+/, '');
+    buf = buf === '' ? trimmed : `${buf} ${trimmed}`;
+    continue;
+  }
+  // Top-level non-list, non-heading (e.g. `[0.7.10]: https://...`)
+  flushBuf();
+  stack = [];
+  out.push(line);
+}
+flushBuf();
+
+process.stdout.write(out.join('\n'));
+if (!out[out.length - 1]?.endsWith('\n')) process.stdout.write('\n');

+ 5 - 7
scripts/release.sh

@@ -30,13 +30,11 @@ if ! grep -q "^## \[${VERSION}\]" CHANGELOG.md; then
   exit 1
   exit 1
 fi
 fi
 
 
-NOTES=$(awk -v v="${VERSION}" '
-  /^## \[/ {
-    if (p) exit
-    if ($0 ~ "^## \\[" v "\\]") p = 1
-  }
-  p
-' CHANGELOG.md)
+# Extract notes with paragraph unwrapping — GitHub Releases render with
+# GFM hard-breaks, so the CHANGELOG's hard-wrapped lines would show as
+# visible `<br>` breaks otherwise. The helper joins continuation lines
+# into a single line per bullet.
+NOTES=$(node scripts/extract-release-notes.mjs "${VERSION}")
 
 
 if [ -z "${NOTES}" ]; then
 if [ -z "${NOTES}" ]; then
   echo "error: failed to extract changelog notes for ${VERSION}" >&2
   echo "error: failed to extract changelog notes for ${VERSION}" >&2