X-Git-Url: http://git.bitcoin.ninja/index.cgi?p=youtube-dl;a=blobdiff_plain;f=youtube_dl%2Futils.py;h=2fe9cf585db817e1d86831c71ef28502b4a16ee5;hp=acc4f987bba1e5133d1d7eda011c527ea2fb2054;hb=c384d537f882efab10a78a56ce6dcb0a30f54b47;hpb=4ed2d7b7d1f67e499a46e507d957616e364565ca diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index acc4f987b..2fe9cf585 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -39,6 +39,7 @@ from .compat import ( compat_HTMLParser, compat_basestring, compat_chr, + compat_ctypes_WINFUNCTYPE, compat_etree_fromstring, compat_expanduser, compat_html_entities, @@ -159,6 +160,8 @@ DATE_FORMATS = ( '%Y-%m-%dT%H:%M', '%b %d %Y at %H:%M', '%b %d %Y at %H:%M:%S', + '%B %d %Y at %H:%M', + '%B %d %Y at %H:%M:%S', ) DATE_FORMATS_DAY_FIRST = list(DATE_FORMATS) @@ -1328,24 +1331,24 @@ def _windows_write_string(s, out): if fileno not in WIN_OUTPUT_IDS: return False - GetStdHandle = ctypes.WINFUNCTYPE( + GetStdHandle = compat_ctypes_WINFUNCTYPE( ctypes.wintypes.HANDLE, ctypes.wintypes.DWORD)( - (b'GetStdHandle', ctypes.windll.kernel32)) + ('GetStdHandle', ctypes.windll.kernel32)) h = GetStdHandle(WIN_OUTPUT_IDS[fileno]) - WriteConsoleW = ctypes.WINFUNCTYPE( + WriteConsoleW = compat_ctypes_WINFUNCTYPE( ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE, ctypes.wintypes.LPWSTR, ctypes.wintypes.DWORD, ctypes.POINTER(ctypes.wintypes.DWORD), - ctypes.wintypes.LPVOID)((b'WriteConsoleW', ctypes.windll.kernel32)) + ctypes.wintypes.LPVOID)(('WriteConsoleW', ctypes.windll.kernel32)) written = ctypes.wintypes.DWORD(0) - GetFileType = ctypes.WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)((b'GetFileType', ctypes.windll.kernel32)) + GetFileType = compat_ctypes_WINFUNCTYPE(ctypes.wintypes.DWORD, ctypes.wintypes.DWORD)(('GetFileType', ctypes.windll.kernel32)) FILE_TYPE_CHAR = 0x0002 FILE_TYPE_REMOTE = 0x8000 - GetConsoleMode = ctypes.WINFUNCTYPE( + GetConsoleMode = compat_ctypes_WINFUNCTYPE( ctypes.wintypes.BOOL, ctypes.wintypes.HANDLE, ctypes.POINTER(ctypes.wintypes.DWORD))( - (b'GetConsoleMode', ctypes.windll.kernel32)) + ('GetConsoleMode', ctypes.windll.kernel32)) INVALID_HANDLE_VALUE = ctypes.wintypes.DWORD(-1).value def not_a_console(handle): @@ -1835,10 +1838,20 @@ def parse_duration(s): days, hours, mins, secs, ms = m.groups() else: m = re.match( - r'''(?ix)(?:P?T)? + r'''(?ix)(?:P? + (?: + [0-9]+\s*y(?:ears?)?\s* + )? + (?: + [0-9]+\s*m(?:onths?)?\s* + )? + (?: + [0-9]+\s*w(?:eeks?)?\s* + )? (?: (?P[0-9]+)\s*d(?:ays?)?\s* )? + T)? (?: (?P[0-9]+)\s*h(?:ours?)?\s* )? @@ -1933,7 +1946,7 @@ class PagedList(object): class OnDemandPagedList(PagedList): - def __init__(self, pagefunc, pagesize, use_cache=False): + def __init__(self, pagefunc, pagesize, use_cache=True): self._pagefunc = pagefunc self._pagesize = pagesize self._use_cache = use_cache @@ -2254,7 +2267,7 @@ def js_to_json(code): "(?:[^"\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^"\\]*"| '(?:[^'\\]*(?:\\\\|\\['"nurtbfx/\n]))*[^'\\]*'| {comment}|,(?={skip}[\]}}])| - [a-zA-Z_][.a-zA-Z_0-9]*| + (?:(? ' + (t.file || t.sourceURL) + ': ' + t.line - + (t.function ? ' (in function ' + t.function +')' : '')); - }}); - }} - console.error(msgStack.join('\n')); - phantom.exit(1); - }}; - var page = require('webpage').create(); - var fs = require('fs'); - var read = {{ mode: 'r', charset: 'utf-8' }}; - var write = {{ mode: 'w', charset: 'utf-8' }}; - JSON.parse(fs.read("{cookies}", read)).forEach(function(x) {{ - phantom.addCookie(x); - }}); - page.settings.resourceTimeout = {timeout}; - page.settings.userAgent = "{ua}"; - page.onLoadStarted = function() {{ - page.evaluate(function() {{ - delete window._phantom; - delete window.callPhantom; - }}); - }}; - var saveAndExit = function() {{ - fs.write("{html}", page.content, write); - fs.write("{cookies}", JSON.stringify(phantom.cookies), write); - phantom.exit(); - }}; - page.onLoadFinished = function(status) {{ - if(page.url === "") {{ - page.setContent(fs.read("{html}", read), "{url}"); - }} - else {{ - {jscode} - }} - }}; - page.open(""); - ''' - - _TMP_FILE_NAMES = ['script', 'html', 'cookies'] - - @staticmethod - def _version(): - return get_exe_version('phantomjs', version_re=r'([0-9.]+)') - - def __init__(self, extractor, required_version=None, timeout=10000): - self.exe = check_executable('phantomjs', ['-v']) - if not self.exe: - raise ExtractorError('PhantomJS executable not found in PATH, ' - 'download it from http://phantomjs.org', - expected=True) - - self.extractor = extractor - - if required_version: - version = self._version() - if is_outdated_version(version, required_version): - self.extractor._downloader.report_warning( - 'Your copy of PhantomJS is outdated, update it to version ' - '%s or newer if you encounter any errors.' % required_version) - - self.options = { - 'timeout': timeout, - } - self._TMP_FILES = {} - for name in self._TMP_FILE_NAMES: - tmp = tempfile.NamedTemporaryFile(delete=False) - tmp.close() - self._TMP_FILES[name] = tmp - - def __del__(self): - for name in self._TMP_FILE_NAMES: - try: - os.remove(self._TMP_FILES[name].name) - except: - pass - - def _save_cookies(self, url): - cookies = cookie_jar_to_list(self.extractor._downloader.cookiejar) - for cookie in cookies: - if 'path' not in cookie: - cookie['path'] = '/' - if 'domain' not in cookie: - cookie['domain'] = compat_urlparse.urlparse(url).netloc - with open(self._TMP_FILES['cookies'].name, 'wb') as f: - f.write(json.dumps(cookies).encode('utf-8')) - - def _load_cookies(self): - with open(self._TMP_FILES['cookies'].name, 'rb') as f: - cookies = json.loads(f.read().decode('utf-8')) - for cookie in cookies: - if cookie['httponly'] is True: - cookie['rest'] = {'httpOnly': None} - if 'expiry' in cookie: - cookie['expire_time'] = cookie['expiry'] - self.extractor._set_cookie(**cookie) - - def get(self, url, html=None, video_id=None, note=None, note2='Executing JS on webpage', headers={}, jscode='saveAndExit();'): - """ - Downloads webpage (if needed) and executes JS - - Params: - url: website url - html: optional, html code of website - video_id: video id - note: optional, displayed when downloading webpage - note2: optional, displayed when executing JS - headers: custom http headers - jscode: code to be executed when page is loaded - - Returns tuple with: - * downloaded website (after JS execution) - * anything you print with `console.log` (but not inside `page.execute`!) - - In most cases you don't need to add any `jscode`. - It is executed in `page.onLoadFinished`. - `saveAndExit();` is mandatory, use it instead of `phantom.exit()` - It is possible to wait for some element on the webpage, for example: - var check = function() { - var elementFound = page.evaluate(function() { - return document.querySelector('#b.done') !== null; - }); - if(elementFound) - saveAndExit(); - else - window.setTimeout(check, 500); - } - - page.evaluate(function(){ - document.querySelector('#a').click(); - }); - check(); - """ - if 'saveAndExit();' not in jscode: - raise ExtractorError('`saveAndExit();` not found in `jscode`') - if not html: - html = self.extractor._download_webpage(url, video_id, note=note, headers=headers) - with open(self._TMP_FILES['html'].name, 'wb') as f: - f.write(html.encode('utf-8')) - - self._save_cookies(url) - - replaces = self.options - replaces['url'] = url - user_agent = headers.get('User-Agent') or std_headers['User-Agent'] - replaces['ua'] = user_agent.replace('"', '\\"') - replaces['jscode'] = jscode - - for x in self._TMP_FILE_NAMES: - replaces[x] = self._TMP_FILES[x].name.replace('\\', '\\\\').replace('"', '\\"') - - with open(self._TMP_FILES['script'].name, 'wb') as f: - f.write(self._TEMPLATE.format(**replaces).encode('utf-8')) - - if video_id is None: - self.extractor.to_screen('%s' % (note2,)) - else: - self.extractor.to_screen('%s: %s' % (video_id, note2)) - - p = subprocess.Popen([ - self.exe, '--ssl-protocol=any', - self._TMP_FILES['script'].name - ], stdout=subprocess.PIPE, stderr=subprocess.PIPE) - out, err = p.communicate() - if p.returncode != 0: - raise ExtractorError( - 'Executing JS failed\n:' + encodeArgument(err)) - with open(self._TMP_FILES['html'].name, 'rb') as f: - html = f.read().decode('utf-8') - - self._load_cookies() - - return (html, encodeArgument(out)) - - def random_birthday(year_field, month_field, day_field): return { year_field: str(random.randint(1950, 1995)),