X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=Robots%2Fparse_html.py;h=47cbb190527d25c3a1bdb74bf0b694a0fe7df909;hb=73f38dccb4a11e857ae3497a61089a4ea89cd654;hp=e676504a9741765fd5f7bc98fb61cd9758b739a2;hpb=fd325219011aaed865f074275b2270112b5e9732;p=bookmarks_db.git diff --git a/Robots/parse_html.py b/Robots/parse_html.py index e676504..47cbb19 100755 --- a/Robots/parse_html.py +++ b/Robots/parse_html.py @@ -32,7 +32,7 @@ num_entity_re = re.compile("(&#[0-9]+;)") def recode_entities(title, charset): output = [] for part in entity_re.split(title): - if part not in ("&", "<", ">", ""e;") and \ + if part not in ("&", "<", ">", """) and \ entity_re.match(part): part = unichr(name2codepoint.get(part[1:-1], part)).encode(charset) output.append(part)