diff options
Diffstat (limited to 'module/plugins/hoster/FilefactoryCom.py')
| -rw-r--r-- | module/plugins/hoster/FilefactoryCom.py | 90 | 
1 files changed, 0 insertions, 90 deletions
diff --git a/module/plugins/hoster/FilefactoryCom.py b/module/plugins/hoster/FilefactoryCom.py deleted file mode 100644 index ada498a51..000000000 --- a/module/plugins/hoster/FilefactoryCom.py +++ /dev/null @@ -1,90 +0,0 @@ -# -*- coding: utf-8 -*- - -import re - -from urlparse import urljoin - -from module.network.RequestFactory import getURL -from module.plugins.internal.SimpleHoster import SimpleHoster, parseFileInfo - - -def getInfo(urls): -    for url in urls: -        h = getURL(url, just_header=True) -        m = re.search(r'Location: (.+)\r\n', h) -        if m and not re.match(m.group(1), FilefactoryCom.__pattern__):  #: It's a direct link! Skipping -            yield (url, 0, 3, url) -        else:  #: It's a standard html page -            yield parseFileInfo(FilefactoryCom, url, getURL(url)) - - -class FilefactoryCom(SimpleHoster): -    __name__    = "FilefactoryCom" -    __type__    = "hoster" -    __version__ = "0.52" - -    __pattern__ = r'https?://(?:www\.)?filefactory\.com/(file|trafficshare/\w+)/\w+' - -    __description__ = """Filefactory.com hoster plugin""" -    __license__     = "GPLv3" -    __authors__     = [("stickell", "l.stickell@yahoo.it"), -                       ("Walter Purcaro", "vuolter@gmail.com")] - - -    INFO_PATTERN = r'<div id="file_name"[^>]*>\s*<h2>(?P<N>[^<]+)</h2>\s*<div id="file_info">\s*(?P<S>[\d.,]+) (?P<U>[\w^_]+) uploaded' -    OFFLINE_PATTERN = r'<h2>File Removed</h2>|This file is no longer available' - -    LINK_PATTERN = r'"([^"]+filefactory\.com/get.+?)"' - -    WAIT_PATTERN = r'<div id="countdown_clock" data-delay="(\d+)">' -    PREMIUM_ONLY_PATTERN = r'>Premium Account Required' - -    COOKIES = [("filefactory.com", "locale", "en_US.utf8")] - - -    def handleFree(self): -        if "Currently only Premium Members can download files larger than" in self.html: -            self.fail(_("File too large for free download")) -        elif "All free download slots on this server are currently in use" in self.html: -            self.retry(50, 15 * 60, _("All free slots are busy")) - -        m = re.search(self.LINK_PATTERN, self.html) -        if m is None: -            self.error(_("Free download link not found")) - -        dl_link = m.group(1) - -        m = re.search(self.WAIT_PATTERN, self.html) -        if m: -            self.wait(int(m.group(1))) - -        self.download(dl_link, disposition=True) - -        check = self.checkDownload({'multiple': "You are currently downloading too many files at once.", -                                    'error': '<div id="errorMessage">'}) - -        if check == "multiple": -            self.logDebug("Parallel downloads detected; waiting 15 minutes") -            self.retry(wait_time=15 * 60, reason=_("Parallel downloads")) -        elif check == "error": -            self.error(_("Unknown error")) - - -    def handlePremium(self): -        header = self.load(self.pyfile.url, just_header=True) - -        if 'location' in header: -            url = header['location'].strip() -            if not url.startswith("http://"): -                url = urljoin("http://www.filefactory.com", url) -        elif 'content-disposition' in header: -            url = self.pyfile.url -        else: -            html = self.load(self.pyfile.url) -            m = re.search(self.LINK_PATTERN, html) -            if m: -                url = m.group(1) -            else: -                self.error(_("Premium download link not found")) - -        self.download(url, disposition=True)  | 
