X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Futils.py;h=2853ba50f228bb1f87900aa78a256b1637fcee2b;hb=54041793389bc72def59a625dc59a4672cdb4825;hp=0f903c64a0d8c71517378143ecad1caf07cb5893;hpb=9beb5af82ecfbb35ee534692e73fb38f4399698a;p=youtube-dl.git diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 0f903c64a..2853ba50f 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -11,16 +11,12 @@ import sys import zlib import urllib2 import email.utils +import json try: import cStringIO as StringIO except ImportError: import StringIO - -try: - import json -except ImportError: # Python <2.6, use trivialjson (https://github.com/phihag/trivialjson): - import trivialjson as json std_headers = { 'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:5.0.1) Gecko/20100101 Firefox/5.0.1', @@ -87,7 +83,7 @@ class IDParser(HTMLParser.HTMLParser): HTMLParser.HTMLParser.__init__(self) def error(self, message): - print self.getpos() + print >> sys.stderr, self.getpos() if self.error_count > 10 or self.started: raise HTMLParser.HTMLParseError(message, self.getpos()) self.rawdata = '\n'.join(self.html.split('\n')[self.getpos()[0]:]) # skip one line @@ -160,12 +156,6 @@ def clean_html(html): return html -def sanitize_title(utitle): - """Sanitizes a video title so it could be used as part of a filename.""" - utitle = unescapeHTML(utitle) - return utitle.replace(unicode(os.sep), u'%') - - def sanitize_open(filename, open_mode): """Try to open the given filename, and slightly tweak it if this fails. @@ -200,10 +190,14 @@ def timeconvert(timestr): if timetuple is not None: timestamp = email.utils.mktime_tz(timetuple) return timestamp - -def simplify_title(title): - expr = re.compile(ur'[^\w\d_\-]+', flags=re.UNICODE) - return expr.sub(u'_', title).strip(u'_') + +def sanitize_filename(s): + """Sanitizes a string so it could be used as part of a filename.""" + def replace_insane(char): + if char in u' .\\/|?*<>:"' or ord(char) < 32: + return '_' + return char + return u''.join(map(replace_insane, s)).strip('_') def orderedSet(iterable): """ Remove all duplicates from the input iterable """ @@ -294,6 +288,13 @@ class ContentTooShortError(Exception): self.expected = expected +class Trouble(Exception): + """Trouble helper exception + + This is an exception to be handled with + FileDownloader.trouble + """ + class YoutubeDLHandler(urllib2.HTTPHandler): """Handler for HTTP requests and responses.