X-Git-Url: https://git.phdru.name/?a=blobdiff_plain;f=parse_html%2Fbkmk_parse_html.py;h=6afb7f9df735f6a487b6b7537fc23f2c3daa71c2;hb=cb9c36b39ed72cd1fa272130d2bcf162a89c3013;hp=c14d055dfe9213556b48bf60fc4ee467d3888a39;hpb=b9c8d112b8d8d0f7c726ee7dd07a89b6569c90a1;p=bookmarks_db.git diff --git a/parse_html/bkmk_parse_html.py b/parse_html/bkmk_parse_html.py index c14d055..6afb7f9 100644 --- a/parse_html/bkmk_parse_html.py +++ b/parse_html/bkmk_parse_html.py @@ -186,12 +186,19 @@ def parse_html(html_text, charset=None, log=None): if log: log(" title : %s" % title) if parser.charset != universal_charset: try: - converted_title = unicode(title, parser.charset).encode(universal_charset) + converted_title = unicode(title, parser.charset).\ + encode(universal_charset) except UnicodeError: - if log: log(" incorrect conversion from %s, converting from %s" % (parser.charset, DEFAULT_CHARSET)) - converted_title = unicode(title, DEFAULT_CHARSET, "replace").encode(universal_charset, "replace") + if log: + log(" incorrect conversion from %s," + "converting from %s" + % (parser.charset, DEFAULT_CHARSET)) + converted_title = \ + unicode(title, DEFAULT_CHARSET, "replace").\ + encode(universal_charset, "replace") parser.charset = DEFAULT_CHARSET - if log and (converted_title != title): log(" converted title: %s" % converted_title) + if log and (converted_title != title): + log(" converted title: %s" % converted_title) except LookupError: if log: log(" unknown charset: '%s'" % parser.charset) else: @@ -201,7 +208,8 @@ def parse_html(html_text, charset=None, log=None): final_title = recode_entities(converted_title, universal_charset) parts = [s.strip() for s in final_title.replace('\r', '').split('\n')] final_title = ' '.join([s for s in parts if s]) - if log and (final_title != converted_title): log(" final title : %s" % final_title) + if log and (final_title != converted_title): + log(" final title : %s" % final_title) parser.title = final_title icon = parser.icon