You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@solr.apache.org by ab...@apache.org on 2026/09/28 08:27:11 UTC
(solr) branch branch_10x updated: changelog changes2html.py: fix author extraction swallowing trailing title text, and multi-issue items not all linking
This is an automated email from the ASF dual-hosted git repository.
sigram pushed a commit to branch branch_10x
in repository https://gitbox.apache.org/repos/asf/solr.git
The following commit(s) were added to refs/heads/branch_10x by this push:
new edd8fd92f65 changelog changes2html.py: fix author extraction swallowing trailing title text, and multi-issue items not all linking
edd8fd92f65 is described below
commit edd8fd92f65801b4b3496eb68c35cb9f699f600b
Author: David Smiley <ds...@apache.org>
AuthorDate: Sun Sep 27 14:33:53 2026 -0400
changelog changes2html.py: fix author extraction swallowing trailing title text, and multi-issue items not all linking
extract_issue_from_text() only removed the first markdown-formatted
issue link it found via re.search, leaving any additional issue
references (e.g. two related JIRA issues on one change) as raw,
unlinked markdown text in the rendered item.
Separately, extract_authors() ran on the text left after removing the
issue link, walking backward through every adjacent trailing '(...)'
group. Since removing the issue link closed the gap between it and
the description, a parenthetical at the end of the description itself
(an acronym expansion, a method's '()', a quoted aside) became
adjacent to the real author group and got swallowed into it, e.g.
'(RRF)' or '(field:* / [* TO *])' showing up as a bogus author.
Fix both by extracting every issue link (not just the first) and
using the position of the first one as a hard boundary: everything
before it is the description and is never touched by author
extraction, which is now scanned only from that point on.
Verified by rendering the actual 9.11.0 CHANGELOG.md before and after:
308 entries changed, all of them fixes of these two kinds (spot-checked
broadly, including old entries referencing many issues); identical
<li> count and no dropped/garbled entries.
Co-Authored-By: Claude Sonnet 5 <no...@anthropic.com>
(cherry picked from commit c9ad9c450b3876f87db067c086b81cdbc8d5eef3)
---
.../documentation/changes-to-html/changes2html.py | 77 +++++++++++++++-------
1 file changed, 55 insertions(+), 22 deletions(-)
diff --git a/gradle/documentation/changes-to-html/changes2html.py b/gradle/documentation/changes-to-html/changes2html.py
index 1d52618834e..287c3205145 100755
--- a/gradle/documentation/changes-to-html/changes2html.py
+++ b/gradle/documentation/changes-to-html/changes2html.py
@@ -142,21 +142,22 @@ class HTMLGenerator:
"""Format a single issue reference as an HTML anchor tag"""
return f'<a href="{url_prefix}{issue_id}">{label}</a>'
- def _extract_markdown_issue(self, text):
+ def _find_markdown_issue_matches(self, text):
"""
- Extract markdown-formatted JIRA/GitHub issues like [SOLR-123](url) or [PR#123](url).
- Returns (issue_link_html, text_without_issue) or (None, text) if not found.
+ Find every markdown-formatted JIRA/GitHub issue reference in text,
+ e.g. [SOLR-123](url) or [PR#123](url) -- an item can reference more
+ than one issue. Returns a list of (start, end, issue_link_html)
+ tuples in the order they appear.
"""
+ matches = []
for pattern, url_prefix, label_fmt in self.issue_patterns:
- match = re.search(pattern, text)
- if match:
+ for match in re.finditer(pattern, text):
issue_id = match.group(1)
label = label_fmt.format(issue_id)
issue_html = self._format_issue_link(url_prefix, issue_id, label)
- text_without = (text[:match.start()] + text[match.end():]).strip()
- return issue_html, text_without
-
- return None, text
+ matches.append((match.start(), match.end(), issue_html))
+ matches.sort(key=lambda m: m[0])
+ return matches
def _extract_plain_pr_references(self, text):
"""
@@ -187,17 +188,39 @@ class HTMLGenerator:
def extract_issue_from_text(self, text):
"""
- Extract the first issue reference from text.
- Tries in order: markdown JIRA/GitHub issues, plain GitHub PR references.
- Returns (issue_link_html, text_without_issue) or (None, text) if not found.
+ Extract issue reference(s) from text.
+ Tries markdown-formatted JIRA/GitHub issues first -- there may be
+ more than one, e.g. two related JIRA issues on the same change --
+ falling back to a plain GitHub PR reference like "#123" if none
+ are found.
+
+ Returns (issue_link_html_or_None, description_or_None, tail).
+ When markdown issues were found, `description` is the text before
+ the first issue reference and `tail` is everything from there on
+ with the issue reference(s) removed (normally just the trailing
+ author "(...)" groups). Author extraction must be bounded to
+ `tail`, never `description`, so a parenthetical at the end of the
+ description (e.g. an acronym expansion) is never mistaken for an
+ author group. When `description` is None, no markdown issue was
+ found and `tail` is just `text` with any plain PR reference
+ removed, if present.
"""
- # Try markdown-formatted issues first
- issue_html, text_without = self._extract_markdown_issue(text)
- if issue_html:
- return issue_html, text_without
+ matches = self._find_markdown_issue_matches(text)
+ if matches:
+ first_start = matches[0][0]
+ parts = []
+ pieces = []
+ last_end = first_start
+ for start, end, issue_html in matches:
+ pieces.append(text[last_end:start])
+ parts.append(issue_html)
+ last_end = end
+ pieces.append(text[last_end:])
+ return ' '.join(parts), text[:first_start], ''.join(pieces)
# Fall back to plain GitHub PR references
- return self._extract_plain_pr_references(text)
+ issue_html, text_without = self._extract_plain_pr_references(text)
+ return issue_html, None, text_without
def _format_single_author(self, author_text):
"""
@@ -372,11 +395,21 @@ class HTMLGenerator:
Format: [ISSUE](url) description (author1) (author2)
Output: <a href>ISSUE</a>: description<br><span class="attrib">(authors)</span>
"""
- # Extract the issue
- issue_html, text_after_issue = self.extract_issue_from_text(item_text)
-
- # Always try to extract authors, whether or not we found an issue
- authors_list, description = self.extract_authors(text_after_issue if issue_html else item_text)
+ # Extract the issue(s)
+ issue_html, description_head, tail = self.extract_issue_from_text(item_text)
+
+ if description_head is not None:
+ # Markdown issue(s) found: bound author extraction to the tail
+ # so it can't reach back into the description.
+ authors_list, tail_remainder = self.extract_authors(tail)
+ description = description_head.strip()
+ tail_remainder = tail_remainder.strip()
+ if tail_remainder:
+ description = f'{description} {tail_remainder}'.strip()
+ else:
+ # No markdown issue (plain PR reference, or no issue at all):
+ # extract authors from the whole remaining text, as before.
+ authors_list, description = self.extract_authors(tail if issue_html else item_text)
if issue_html:
# We have an issue link