release 2014.04.07.3

youtube_dl/YoutubeDL.py

@@ -286,6 +286,9 @@ class YoutubeDL(object):
         """Print message to stdout if not in quiet mode."""
         return self.to_stdout(message, skip_eol, check_quiet=True)
 
+    def _write_string(self, s, out=None):
+        write_string(s, out=out, encoding=self.params.get('encoding'))
+
     def to_stdout(self, message, skip_eol=False, check_quiet=False):
         """Print message to stdout if not in quiet mode."""
         if self.params.get('logger'):
@@ -295,7 +298,7 @@ class YoutubeDL(object):
             terminator = ['\n', ''][skip_eol]
             output = message + terminator
 
-            write_string(output, self._screen_file)
+            self._write_string(output, self._screen_file)
 
     def to_stderr(self, message):
         """Print message to stderr."""
@@ -305,7 +308,7 @@ class YoutubeDL(object):
         else:
             message = self._bidi_workaround(message)
             output = message + '\n'
-            write_string(output, self._err_file)
+            self._write_string(output, self._err_file)
 
     def to_console_title(self, message):
         if not self.params.get('consoletitle', False):
@@ -315,21 +318,21 @@ class YoutubeDL(object):
             # already of type unicode()
             ctypes.windll.kernel32.SetConsoleTitleW(ctypes.c_wchar_p(message))
         elif 'TERM' in os.environ:
-            write_string('\033]0;%s\007' % message, self._screen_file)
+            self._write_string('\033]0;%s\007' % message, self._screen_file)
 
     def save_console_title(self):
         if not self.params.get('consoletitle', False):
             return
         if 'TERM' in os.environ:
             # Save the title on stack
-            write_string('\033[22;0t', self._screen_file)
+            self._write_string('\033[22;0t', self._screen_file)
 
     def restore_console_title(self):
         if not self.params.get('consoletitle', False):
             return
         if 'TERM' in os.environ:
             # Restore the title from stack
-            write_string('\033[23;0t', self._screen_file)
+            self._write_string('\033[23;0t', self._screen_file)
 
     def __enter__(self):
         self.save_console_title()
@@ -1211,9 +1214,16 @@ class YoutubeDL(object):
         if not self.params.get('verbose'):
             return
 
-        write_string('[debug] Encodings: locale %s, fs %s, out %s, pref %s\n' %
-                 (locale.getpreferredencoding(), sys.getfilesystemencoding(), sys.stdout.encoding, self.get_encoding()))
-        write_string('[debug] youtube-dl version ' + __version__ + '\n')
+        write_string(
+            '[debug] Encodings: locale %s, fs %s, out %s, pref %s\n' % (
+                locale.getpreferredencoding(),
+                sys.getfilesystemencoding(),
+                sys.stdout.encoding,
+                self.get_encoding()),
+            encoding=None
+        )
+
+        self._write_string('[debug] youtube-dl version ' + __version__ + '\n')
         try:
             sp = subprocess.Popen(
                 ['git', 'rev-parse', '--short', 'HEAD'],
@@ -1222,20 +1232,20 @@ class YoutubeDL(object):
             out, err = sp.communicate()
             out = out.decode().strip()
             if re.match('[0-9a-f]+', out):
-                write_string('[debug] Git HEAD: ' + out + '\n')
+                self._write_string('[debug] Git HEAD: ' + out + '\n')
         except:
             try:
                 sys.exc_clear()
             except:
                 pass
-        write_string('[debug] Python version %s - %s' %
+        self._write_string('[debug] Python version %s - %s' %
                      (platform.python_version(), platform_name()) + '\n')
 
         proxy_map = {}
         for handler in self._opener.handlers:
             if hasattr(handler, 'proxies'):
                 proxy_map.update(handler.proxies)
-        write_string('[debug] Proxy map: ' + compat_str(proxy_map) + '\n')
+        self._write_string('[debug] Proxy map: ' + compat_str(proxy_map) + '\n')
 
     def _setup_opener(self):
         timeout_val = self.params.get('socket_timeout')

youtube_dl/extractor/__init__.py

@@ -208,6 +208,7 @@ from .rottentomatoes import RottenTomatoesIE
 from .roxwel import RoxwelIE
 from .rtlnow import RTLnowIE
 from .rts import RTSIE
+from .rtve import RTVEALaCartaIE
 from .rutube import (
     RutubeIE,
     RutubeChannelIE,

youtube_dl/extractor/rtve.py

@@ -0,0 +1,84 @@
+# encoding: utf-8
+from __future__ import unicode_literals
+
+import re
+import base64
+
+from .common import InfoExtractor
+from ..utils import (
+    struct_unpack,
+)
+
+
+class RTVEALaCartaIE(InfoExtractor):
+    IE_NAME = 'rtve.es:alacarta'
+    IE_DESC = 'RTVE a la carta'
+    _VALID_URL = r'http://www\.rtve\.es/alacarta/videos/[^/]+/[^/]+/(?P<id>\d+)'
+
+    _TEST = {
+        'url': 'http://www.rtve.es/alacarta/videos/balonmano/o-swiss-cup-masculina-final-espana-suecia/2491869/',
+        'md5': '18fcd45965bdd076efdb12cd7f6d7b9e',
+        'info_dict': {
+            'id': '2491869',
+            'ext': 'mp4',
+            'title': 'Balonmano - Swiss Cup masculina. Final: España-Suecia',
+        },
+    }
+
+    def _decrypt_url(self, png):
+        encrypted_data = base64.b64decode(png)
+        text_index = encrypted_data.find(b'tEXt')
+        text_chunk = encrypted_data[text_index-4:]
+        length = struct_unpack('!I', text_chunk[:4])[0]
+        # Use bytearray to get integers when iterating in both python 2.x and 3.x
+        data = bytearray(text_chunk[8:8+length])
+        data = [chr(b) for b in data if b != 0]
+        hash_index = data.index('#')
+        alphabet_data = data[:hash_index]
+        url_data = data[hash_index+1:]
+
+        alphabet = []
+        e = 0
+        d = 0
+        for l in alphabet_data:
+            if d == 0:
+                alphabet.append(l)
+                d = e = (e + 1) % 4
+            else:
+                d -= 1
+        url = ''
+        f = 0
+        e = 3
+        b = 1
+        for letter in url_data:
+            if f == 0:
+                l = int(letter)*10
+                f = 1
+            else:
+                if e == 0:
+                    l += int(letter)
+                    url += alphabet[l]
+                    e = (b + 3) % 4
+                    f = 0
+                    b += 1
+                else:
+                    e -= 1
+
+        return url
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+        info = self._download_json(
+            'http://www.rtve.es/api/videos/%s/config/alacarta_videos.json' % video_id,
+            video_id)['page']['items'][0]
+        png_url = 'http://www.rtve.es/ztnr/movil/thumbnail/default/videos/%s.png' % video_id
+        png = self._download_webpage(png_url, video_id, 'Downloading url information')
+        video_url = self._decrypt_url(png)
+
+        return {
+            'id': video_id,
+            'title': info['title'],
+            'url': video_url,
+            'thumbnail': info['image'],
+        }

youtube_dl/extractor/teamcoco.py

@@ -9,7 +9,7 @@ from ..utils import (
 
 
 class TeamcocoIE(InfoExtractor):
-    _VALID_URL = r'http://teamcoco\.com/video/(?P<video_id>[0-9]+)?/?(?P<url_title>.*)'
+    _VALID_URL = r'http://teamcoco\.com/video/(?P<video_id>[0-9]+)?/?(?P<display_id>.*)'
     _TESTS = [
     {
         'url': 'http://teamcoco.com/video/80187/conan-becomes-a-mary-kay-beauty-consultant',
@@ -33,21 +33,19 @@ class TeamcocoIE(InfoExtractor):
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        if mobj is None:
-            raise ExtractorError('Invalid URL: %s' % url)
-        url_title = mobj.group('url_title')
-        webpage = self._download_webpage(url, url_title)
+
+        display_id = mobj.group('display_id')
+        webpage = self._download_webpage(url, display_id)
         
         video_id = mobj.group("video_id")
-        if video_id == '':
+        if not video_id:
             video_id = self._html_search_regex(
                 r'<article class="video" data-id="(\d+?)"',
                 webpage, 'video id')
-        
-        self.report_extraction(video_id)
 
         data_url = 'http://teamcoco.com/cvp/2.0/%s.xml' % video_id
-        data = self._download_xml(data_url, video_id, 'Downloading data webpage')
+        data = self._download_xml(
+            data_url, display_id, 'Downloading data webpage')
 
         qualities = ['500k', '480p', '1000k', '720p', '1080p']
         formats = []
@@ -82,6 +80,7 @@ class TeamcocoIE(InfoExtractor):
 
         return {
             'id': video_id,
+            'display_id': display_id,
             'formats': formats,
             'title': self._og_search_title(webpage),
             'thumbnail': self._og_search_thumbnail(webpage),

youtube_dl/utils.py

@@ -910,25 +910,81 @@ def platform_name():
     return res
 
 
-def write_string(s, out=None):
+def _windows_write_string(s, out):
+    """ Returns True if the string was written using special methods,
+    False if it has yet to be written out."""
+    # Adapted from http://stackoverflow.com/a/3259271/35070
+
+    import ctypes
+    import ctypes.wintypes
+
+    WIN_OUTPUT_IDS = {
+        1: -11,
+        2: -12,
+    }
+
+    fileno = out.fileno()
+    if fileno not in WIN_OUTPUT_IDS:
+        return False
+
+    GetStdHandle = ctypes.WINFUNCTYPE(
+        ctypes.wintypes.HANDLE, ctypes.wintypes.DWORD)(
+        ("GetStdHandle", ctypes.windll.kernel32))
+    h = GetStdHandle(WIN_OUTPUT_IDS[fileno])
+
+    WriteConsoleW = ctypes.WINFUNCTYPE(
+        ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE, ctypes.wintypes.LPWSTR,
+        ctypes.wintypes.DWORD, ctypes.POINTER(ctypes.wintypes.DWORD),
+        ctypes.wintypes.LPVOID)(("WriteConsoleW", ctypes.windll.kernel32))
+    written = ctypes.wintypes.DWORD(0)
+
+    GetFileType = ctypes.WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)(("GetFileType", ctypes.windll.kernel32))
+    FILE_TYPE_CHAR = 0x0002
+    FILE_TYPE_REMOTE = 0x8000
+    GetConsoleMode = ctypes.WINFUNCTYPE(
+        ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE,
+        ctypes.POINTER(ctypes.wintypes.DWORD))(
+        ("GetConsoleMode", ctypes.windll.kernel32))
+    INVALID_HANDLE_VALUE = ctypes.wintypes.DWORD(-1).value
+
+    def not_a_console(handle):
+        if handle == INVALID_HANDLE_VALUE or handle is None:
+            return True
+        return ((GetFileType(handle) & ~FILE_TYPE_REMOTE) != FILE_TYPE_CHAR
+                or GetConsoleMode(handle, ctypes.byref(ctypes.wintypes.DWORD())) == 0)
+
+    if not_a_console(h):
+        return False
+
+    remaining = len(s)
+    while remaining > 0:
+        ret = WriteConsoleW(
+            h, s, min(len(s), 1024), ctypes.byref(written), None)
+        if ret == 0:
+            raise OSError('Failed to write string')
+        remaining -= written.value
+    return True
+
+
+def write_string(s, out=None, encoding=None):
     if out is None:
         out = sys.stderr
     assert type(s) == compat_str
 
+    if sys.platform == 'win32' and encoding is None and hasattr(out, 'fileno'):
+        if _windows_write_string(s, out):
+            return
+
     if ('b' in getattr(out, 'mode', '') or
             sys.version_info[0] < 3):  # Python 2 lies about mode of sys.stderr
-        s = s.encode(preferredencoding(), 'ignore')
-    try:
+        byt = s.encode(encoding or preferredencoding(), 'ignore')
+        out.write(byt)
+    elif hasattr(out, 'buffer'):
+        enc = encoding or getattr(out, 'encoding', None) or preferredencoding()
+        byt = s.encode(enc, 'ignore')
+        out.buffer.write(byt)
+    else:
         out.write(s)
-    except UnicodeEncodeError:
-        # In Windows shells, this can fail even when the codec is just charmap!?
-        # See https://wiki.python.org/moin/PrintFails#Issue
-        if sys.platform == 'win32' and hasattr(out, 'encoding'):
-            s = s.encode(out.encoding, 'ignore').decode(out.encoding)
-            out.write(s)
-        else:
-            raise
-
     out.flush()
 
 

youtube_dl/version.py

@@ -1,2 +1,2 @@
 
-__version__ = '2014.04.07'
+__version__ = '2014.04.07.3'