[1/2] ty: Try subject matching for every commit before using trackers

Commit Message

Will Deacon March 1, 2024, 10:24 a.m. UTC

From: Will Deacon <[e-mail]>

Nicolin reports that 'b4 ty' produced a "thank you" letter with
incorrect commit IDs:

  https://lore.kernel.org/r/ZeDtfKMUXnOnJOic@Asurada-Nvidia

This appears to be because of the creative use of Message-Ids in the
series at:

  https://lore.kernel.org/r/[e-mail]

For example, comparing the Message-Id of patches 1 and 11 of that series,
we can see that one is a substring of the other:

  Message-Id: <[e-mail]>
  Message-Id: <[e-mail]>

This confuses auto_locate_series(), as it will end up matching the
Message-Id of patch 1 with the Link: tag in patch 11 and therefore get
the wrong patch/commit association.

Fix this problem by trying to match each commit based on the Subject
before falling back to the fuzzier tracker-based matching.

Reported-by: Nicolin Chen <[e-mail]>
Signed-off-by: Will Deacon <[e-mail]>
---
 src/b4/ty.py | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

Patch

diff --git a/src/b4/ty.py b/src/b4/ty.py
index a8bdabf5bcbc..950c357a89c5 100644
--- a/src/b4/ty.py
+++ b/src/b4/ty.py
@@ -217,7 +217,13 @@  def auto_locate_series(gitdir: Optional[str], jsondata: dict, branch: str,
                     success = True
                     matches += 1
                     break
-                elif len(patch) > 2 and len(patch[2]) and len(commit[2]):
+
+            if success:
+                continue
+
+            # try to locate by tracker
+            for pwhash, commit in commits.items():
+                if len(patch) > 2 and len(patch[2]) and len(commit[2]):
                     for tracker in commit[2]:
                         if tracker.find(patch[2]) >= 0:
                             logger.debug('Matched using recorded message-id')