From 426a3f8ad53f09372d167c14f3bff8f917a9ea40 Mon Sep 17 00:00:00 2001 From: Oleg Broytman Date: Mon, 3 Mar 2008 21:03:35 +0000 Subject: [PATCH] Always log guessed charset even if it's utf-8. git-svn-id: file:///home/phd/archive/SVN/bookmarks_db/trunk@190 fdd5c36f-1aea-0310-aeeb-c58d7e2b6c23 --- Robots/parse_html.py | 21 ++++++++++++--------- 1 file changed, 12 insertions(+), 9 deletions(-) diff --git a/Robots/parse_html.py b/Robots/parse_html.py index ec93ea5..3e49491 100755 --- a/Robots/parse_html.py +++ b/Robots/parse_html.py @@ -86,23 +86,26 @@ def parse_html(filename, charset=None, log=None): if parser.charset: parser.charset = parser.charset.lower().replace("windows-", "cp") - if parser.charset and (parser.charset <> universal_charset): + if parser.charset and ( + (parser.charset <> universal_charset) or + ((not charset) or (charset <> parser.charset))): try: if parser.meta_charset: if log: log(" META charset : %s" % parser.charset) - elif (not charset) or charset <> parser.charset: + elif (not charset) or (charset <> parser.charset): if log: log(" guessed charset: %s" % parser.charset) if log: log(" current charset: %s" % universal_charset) if log: log(" title : %s" % title) - try: - 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") - parser.charset = DEFAULT_CHARSET + if parser.charset <> universal_charset: + try: + 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") + parser.charset = DEFAULT_CHARSET if log and (converted_title <> title): log(" converted title: %s" % converted_title) except LookupError: - if log: log(" unknown charset: `%s' or `%s'" % (parser.charset, universal_charset)) + if log: log(" unknown charset: '%s'" % parser.charset) else: if log: log(" title : %s" % title) -- 2.39.5