strip HTML when calculating field checksum (users will need to checkDB)

anki/models.py

@@ -4,7 +4,7 @@
 
 import copy, re
 from anki.utils import intTime, hexifyID, joinFields, splitFields, ids2str, \
-    timestampID, fieldChecksum, json
+    timestampID, checksum, json
 from anki.lang import _
 from anki.consts import *
 from anki.hooks import runHook
@@ -456,7 +456,7 @@ select id from notes where mid = ?)""" % " ".join(map),
             s += t['name']
             s += t['qfmt']
             s += t['afmt']
-        return fieldChecksum(s)
+        return checksum(s)
 
     # Required field/text cache
     ##########################################################################

anki/notes.py

@@ -136,7 +136,7 @@ insert or replace into notes values (?,?,?,?,?,?,?,?,?,?,?)""",
         for flds in self.col.db.list(
             "select flds from notes where csum = ? and id != ? and mid = ?",
             csum, self.id or 0, self.mid):
-            if splitFields(flds)[0] == self.fields[0]:
+            if stripHTML(splitFields(flds)[0]) == stripHTML(self.fields[0]):
                 return 2
         return False
 

anki/utils.py

@@ -245,7 +245,7 @@ def checksum(data):
 
 def fieldChecksum(data):
     # 32 bit unsigned number from first 8 digits of sha1 hash
-    return int(checksum(data.encode("utf-8"))[:8], 16)
+    return int(checksum(stripHTML(data).encode("utf-8"))[:8], 16)
 
 # Temp files
 ##############################################################################