jenkins-bot merged this change.

View Change

Fix regression: archivebot should preserve h1 headers

Follows-up: I644f17dc6b1c775de134c5e795c456423cc40147

Bug: T221445
Change-Id: Ibcad74be6faf9c8137f081847374283acc6e2f2f
---
M scripts/archivebot.py
M tests/archivebot_tests.py
2 files changed, 32 insertions(+), 2 deletions(-)

diff --git a/scripts/archivebot.py b/scripts/archivebot.py
index a44398e..b6f9187 100755
--- a/scripts/archivebot.py
+++ b/scripts/archivebot.py
@@ -441,10 +441,11 @@
         self.archives = {}
         self.archived_threads = 0
 
-        # Exclude non-thread headings
+        # Exclude unsupported headings (h1, h3, etc):
+        # adding the marker will make them ignored by extract_sections()
         text = self.get()
         marker = findmarker(text)
-        text = re.sub(r'^===', marker + r'===', text, flags=re.M)
+        text = re.sub(r'^((=|={3,})[^=])', marker + r'\1', text, flags=re.M)
 
         # Find threads, avoid archiving categories or interwiki
         header, threads, footer = extract_sections(text, self.site)
diff --git a/tests/archivebot_tests.py b/tests/archivebot_tests.py
index 66243e6..6896711 100644
--- a/tests/archivebot_tests.py
+++ b/tests/archivebot_tests.py
@@ -271,6 +271,35 @@
         page.load_page()
         self.assertEqual([x.title for x in page.threads], ['A', 'B'])
 
+    def testThreadsWithSubsections(self):
+        """Test recognizing threads with subsections.
+
+        Talk:For-pywikibot-archivebot/subsections must have::
+
+         {{User:MiszaBot/config
+         |archive = Talk:Main_Page/archive
+         |algo = old(30d)
+         }}
+         = Front matter =
+         placeholder
+         == A ==
+         foo bar
+         === A1 ===
+         foo bar bar
+         ==== A11 ====
+         foo
+         == B ==
+         foo bar bar bar
+        """
+        site = self.get_site()
+        page = pywikibot.Page(site, 'Talk:For-pywikibot-archivebot/testcase2')
+        tmpl = pywikibot.Page(site, 'User:MiszaBot/config')
+        archiver = archivebot.PageArchiver(
+            page=page, template=tmpl, salt='', force=False)
+        page = archivebot.DiscussionPage(page, archiver)
+        page.load_page()
+        self.assertEqual([x.title for x in page.threads], ['A', 'B'])
+
 
 if __name__ == '__main__':  # pragma: no cover
     try:

To view, visit change 505365. To unsubscribe, or for help writing mail filters, visit settings.