update import script to handle superseded patch revisions

1 files changed, 34 insertions, 0 deletions

diff --git a/import_emails.py b/import_emails.py
index 079492f..7c9467e 100755
--- a/import_emails.py
+++ b/import_emails.py
@@ -3,18 +3,52 @@
 from bin import django_setup, get_logger
 django_setup()  # must be called to get sys.path and django settings in place
 
+import difflib
 import email
+import re
 import imaplib
 
 from patchwork.bin import parsemail
+from patchwork.models import Patch, State
 
 log = get_logger('import_emails')
 
 
+def find_old_revisions(patch):
+    log.debug('looking for old versions of patch %d', patch.id)
+    proj = patch.project
+    submitter = patch.submitter
+    potential = Patch.objects.filter(project=proj, submitter=submitter)
+    exclude_states = State.objects.filter(name__in=('Accepted', 'Superseded'))
+
+    for p in potential.exclude(state__in=exclude_states):
+        if p.id == patch.id:
+            continue  # skip ourself
+        matcher = difflib.SequenceMatcher(None, p.name, patch.name)
+        if matcher.ratio() > 0.9:
+            # title's seem to match, check commit
+            matcher = difflib.SequenceMatcher(None, p.content, patch.content)
+            if matcher.ratio() > 0.9:
+                # Some commits may be just reverting a previous one and in
+                # these cases the title and content ratios will be high and
+                # pass the two tests above, so we do one last check here to
+                # avoid false-positives in these cases.
+                pot = re.sub(r'\[[^\]]*\]', '', p.name).strip()
+                cur = re.sub(r'\[[^\]]*\]', '', patch.name).strip()
+                if (pot.startswith('Revert') and not cur.startswith('Revert')):
+                    continue
+            yield p
+
+
 def process_message(msg):
     msg = email.message_from_string(msg)
     # TODO how could we better check for errors in parse_mail
     parsemail.parse_mail(msg)
+    for patch in Patch.objects.filter(msgid=msg.get('Message-Id').strip()):
+        for old in find_old_revisions(patch):
+            log.info('marking patch %d as superseded by %d', patch.id, old.id)
+            old.state = State.objects.get(name='Superseded')
+            old.save()
 
 
 def move_message(mail, uid, folder):