From c68208089072653637df62205e6ce22a74b2a1f2 Mon Sep 17 00:00:00 2001
From: Damien Elmes <git@ichi2.net>
Date: Tue, 5 Apr 2011 20:28:21 +0900
Subject: [PATCH] make it easier to get media dir; remove tidyHTML()

---
 anki/deck.py              |  2 +-
 anki/importing/csvfile.py |  1 -
 anki/media.py             |  3 +++
 anki/utils.py             | 24 ------------------------
 4 files changed, 4 insertions(+), 26 deletions(-)

diff --git a/anki/deck.py b/anki/deck.py
index fb333f900..bef46d789 100644
--- a/anki/deck.py
+++ b/anki/deck.py
@@ -4,7 +4,7 @@
 
 import time, os, random, re, stat, simplejson, datetime, copy
 from anki.lang import _, ngettext
-from anki.utils import parseTags, tidyHTML, ids2str, hexifyID, \
+from anki.utils import parseTags, ids2str, hexifyID, \
      checksum, fieldChecksum, addTags, delTags, stripHTML, intTime, \
      splitFields
 from anki.hooks import runHook, runFilter
diff --git a/anki/importing/csvfile.py b/anki/importing/csvfile.py
index b75c50489..3d68b2e8c 100644
--- a/anki/importing/csvfile.py
+++ b/anki/importing/csvfile.py
@@ -6,7 +6,6 @@ import codecs, csv, re
 from anki.importing import Importer, ForeignCard
 from anki.lang import _
 from anki.errors import *
-from anki.utils import tidyHTML
 
 class TextImporter(Importer):
 
diff --git a/anki/media.py b/anki/media.py
index dc909177e..2ea090001 100644
--- a/anki/media.py
+++ b/anki/media.py
@@ -20,8 +20,11 @@ class MediaRegistry(object):
         self._updateDir()
 
     def dir(self, create=False):
+        "Call with create=None to retrieve dir without creating."
         if self._dir:
             return self._dir
+        elif create == None:
+            return self._updateDir(create)
         elif create:
             self._updateDir(True)
         return self._dir
diff --git a/anki/utils.py b/anki/utils.py
index 19b99d831..2d6722b15 100644
--- a/anki/utils.py
+++ b/anki/utils.py
@@ -149,30 +149,6 @@ def minimizeHTML(s):
                '<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
-    html = re.sub(u".*<body.*?>(.*)</body></html>",
-                  "\\1", html.replace("\n", u""))
-    # strip superfluous Qt formatting
-    html = re.sub(u"(?:-qt-table-type: root; )?"
-                  "margin-top:\d+px; margin-bottom:\d+px; margin-left:\d+px; "
-                  "margin-right:\d+px;(?: -qt-block-indent:0; "
-                  "text-indent:0px;)?", u"", html)
-    html = re.sub(u"-qt-paragraph-type:empty;", u"", html)
-    # strip leading space in style execute, and remove if no contents
-    html = re.sub(u'style=" ', u'style="', html)
-    html = re.sub(u' style=""', u"", html)
-    # convert P tags into SPAN and/or BR
-    html = re.sub(u'<p( style=.+?)>(.*?)</p>', u'<span\\1>\\2</span><br>', html)
-    html = re.sub(u'<p>(.*?)</p>', u'\\1<br>', html)
-    html = re.sub(u'<br>$', u'', html)
-    html = re.sub(u"^<table><tr><td style=\"border: none;\">(.*)<br></td></tr></table>$", 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):
     def fixup(m):
         text = m.group(0)