simplify bold/italic/underline tags from qt in upgrade

anki/storage.py

@@ -420,6 +420,13 @@ def _postSchemaUpgrade(deck):
               "revCardsDue", "revCardsRandom", "acqCardsRandom",
               "acqCardsOld", "acqCardsNew"):
         deck.db.execute("drop view if exists %s" % v)
+    # minimize qt's bold/italics/underline cruft. we made need to use lxml to
+    # do this properly
+    from anki.utils import minimizeHTML
+    r = [(minimizeHTML(x[2]), x[0], x[1]) for x in deck.db.execute(
+        "select fid, fmid, val from fdata")]
+    deck.db.executemany("update fdata set val = ? where fid = ? and fmid = ?",
+                        r)
     # ensure all templates use the new style field format, and update cach
     for m in deck.allModels():
         for t in m.templates:

anki/utils.py

@@ -139,6 +139,16 @@ def stripHTMLMedia(s):
     s = re.sub("<img src=[\"']?([^\"'>]+)[\"']? ?/?>", " \\1 ", s)
     return stripHTML(s)
 
+def minimizeHTML(s):
+    "Correct Qt's verbose bold/underline/etc."
+    s = re.sub('<span style="font-weight:600;">(.*?)</span>', '<b>\\1</b>',
+               s)
+    s = re.sub('<span style="font-style:italic;">(.*?)</span>', '<i>\\1</i>',
+               s)
+    s = re.sub('<span style="text-decoration: underline;">(.*?)</span>',
+               '<u>\\1</u>', s)
+    return s
+
 def tidyHTML(html):
     "Remove cruft like body tags and return just the important part."
     # contents of body - no head or html tags
@@ -160,6 +170,7 @@ def tidyHTML(html):
     html = re.sub(u"^<table><tr><td style=\"border: none;\">(.*)<br></td></tr></table>$", u"\\1", html)
     # this is being added by qt's html editor, and leads to unwanted spaces
     html = re.sub(u"^<p dir='rtl'>(.*?)</p>$", u'\\1', html)
+    html = minimizeHTML(html)
     return html
 
 def entsToTxt(html):