diff options
Diffstat (limited to 'module/plugins/internal')
| -rw-r--r-- | module/plugins/internal/CaptchaService.py | 16 | ||||
| -rw-r--r-- | module/plugins/internal/SimpleHoster.py | 68 | ||||
| -rw-r--r-- | module/plugins/internal/UnRar.py | 14 | ||||
| -rw-r--r-- | module/plugins/internal/XFSAccount.py | 14 | ||||
| -rw-r--r-- | module/plugins/internal/XFSHoster.py | 35 |
5 files changed, 84 insertions, 63 deletions
diff --git a/module/plugins/internal/CaptchaService.py b/module/plugins/internal/CaptchaService.py index 38775f519..7009e6986 100644 --- a/module/plugins/internal/CaptchaService.py +++ b/module/plugins/internal/CaptchaService.py @@ -7,7 +7,7 @@ from random import random class CaptchaService: __name__ = "CaptchaService" - __version__ = "0.14" + __version__ = "0.15" __description__ = """Base captcha service plugin""" __license__ = "GPLv3" @@ -52,7 +52,7 @@ class CaptchaService: class ReCaptcha(CaptchaService): __name__ = "ReCaptcha" - __version__ = "0.07" + __version__ = "0.08" __description__ = """ReCaptcha captcha service plugin""" __license__ = "GPLv3" @@ -92,7 +92,6 @@ class ReCaptcha(CaptchaService): raise TypeError(errmsg) js = self.plugin.req.load("http://www.google.com/recaptcha/api/challenge", get={'k': key}) - try: challenge = re.search("challenge : '(.+?)',", js).group(1) server = re.search("server : '(.+?)',", js).group(1) @@ -101,6 +100,8 @@ class ReCaptcha(CaptchaService): result = self.result(server, challenge) + self.plugin.logDebug("ReCaptcha result: %s" % result, "challenge: %s" % challenge) + return challenge, result @@ -111,7 +112,7 @@ class ReCaptcha(CaptchaService): class AdsCaptcha(CaptchaService): __name__ = "AdsCaptcha" - __version__ = "0.04" + __version__ = "0.05" __description__ = """AdsCaptcha captcha service plugin""" __license__ = "GPLv3" @@ -154,7 +155,6 @@ class AdsCaptcha(CaptchaService): CaptchaId, PublicKey = key js = self.plugin.req.load("http://api.adscaptcha.com/Get.aspx", get={'CaptchaId': CaptchaId, 'PublicKey': PublicKey}) - try: challenge = re.search("challenge: '(.+?)',", js).group(1) server = re.search("server: '(.+?)',", js).group(1) @@ -163,6 +163,8 @@ class AdsCaptcha(CaptchaService): result = self.result(server, challenge) + self.plugin.logDebug("AdsCaptcha result: %s" % result, "challenge: %s" % challenge) + return challenge, result @@ -173,7 +175,7 @@ class AdsCaptcha(CaptchaService): class SolveMedia(CaptchaService): __name__ = "SolveMedia" - __version__ = "0.05" + __version__ = "0.06" __description__ = """SolveMedia captcha service plugin""" __license__ = "GPLv3" @@ -202,6 +204,8 @@ class SolveMedia(CaptchaService): result = self.result(server, challenge) + self.plugin.logDebug("SolveMedia result: %s" % result, "challenge: %s" % challenge) + return challenge, result diff --git a/module/plugins/internal/SimpleHoster.py b/module/plugins/internal/SimpleHoster.py index 01702d423..24a2fa6b0 100644 --- a/module/plugins/internal/SimpleHoster.py +++ b/module/plugins/internal/SimpleHoster.py @@ -5,8 +5,6 @@ import re from time import time from urlparse import urlparse -from pycurl import FOLLOWLOCATION - from module.PyFile import statusMap as _statusMap from module.network.CookieJar import CookieJar from module.network.RequestFactory import getURL @@ -34,7 +32,11 @@ def _error(self, reason, type): #@TODO: Remove in 0.4.10 def _wait(self, seconds, reconnect): if seconds: - self.setWait(seconds, reconnect) + self.setWait(seconds) + + if reconnect is not None: + self.wantReconnect = reconnect + super(SimpleHoster, self).wait() @@ -57,12 +59,13 @@ def parseHtmlTagAttrValue(attr_name, tag): return m.group(2) if m else None -def parseHtmlForm(attr_str, html, input_names=None): - for form in re.finditer(r"(?P<tag><form[^>]*%s[^>]*>)(?P<content>.*?)</?(form|body|html)[^>]*>" % attr_str, +def parseHtmlForm(attr_str, html, input_names={}): + for form in re.finditer(r"(?P<TAG><form[^>]*%s[^>]*>)(?P<CONTENT>.*?)</?(form|body|html)[^>]*>" % attr_str, html, re.S | re.I): inputs = {} - action = parseHtmlTagAttrValue("action", form.group('tag')) - for inputtag in re.finditer(r'(<(input|textarea)[^>]*>)([^<]*(?=</\2)|)', form.group('content'), re.S | re.I): + action = parseHtmlTagAttrValue("action", form.group('TAG')) + + for inputtag in re.finditer(r'(<(input|textarea)[^>]*>)([^<]*(?=</\2)|)', form.group('CONTENT'), re.S | re.I): name = parseHtmlTagAttrValue("name", inputtag.group(1)) if name: value = parseHtmlTagAttrValue("value", inputtag.group(1)) @@ -71,7 +74,7 @@ def parseHtmlForm(attr_str, html, input_names=None): else: inputs[name] = value - if isinstance(input_names, dict): + if input_names: # check input attributes for key, val in input_names.iteritems(): if key in inputs: @@ -111,23 +114,24 @@ def timestamp(): #@TODO: Move to hoster class in 0.4.10 def _getDirectLink(self, url): - self.req.http.c.setopt(FOLLOWLOCATION, 0) + header = self.load(url, ref=True, just_header=True, decode=True) - html = self.load(url, ref=True, decode=True) + if not 'code' in header or header['code'] != 302: + return "" - self.req.http.c.setopt(FOLLOWLOCATION, 1) + if not 'location' in header or not header['location']: + return "" - if self.getInfo(url, html)['status'] is not 2: - try: - return re.search(r'Location\s*:\s*(.+)', self.req.http.header, re.I).group(1).rstrip() #@TODO: Remove .rstrip() in 0.4.10 - except: - pass + # if 'content-type' in header and "text/plain" not in header['content-type']: + # return "" + + return header['location'] class SimpleHoster(Hoster): __name__ = "SimpleHoster" __type__ = "hoster" - __version__ = "0.61" + __version__ = "0.67" __pattern__ = r'^unmatchable$' @@ -198,7 +202,7 @@ class SimpleHoster(Hoster): @classmethod def getInfo(cls, url="", html=""): - info = {'name': urlparse(url).path.split('/')[-1] or _("Unknown"), 'size': 0, 'status': 3, 'url': url or ""} + info = {'name': urlparse(url).path.split('/')[-1] or _("Unknown"), 'size': 0, 'status': 3 if url else 1, 'url': url or ""} if not html: if url: @@ -225,13 +229,19 @@ class SimpleHoster(Hoster): except: pass - for pattern in ("INFO_PATTERN", "NAME_PATTERN", "SIZE_PATTERN", - "FILE_INFO_PATTERN", "FILE_NAME_PATTERN", "FILE_SIZE_PATTERN"): #@TODO: Remove in 0.4.10 + for pattern in ("FILE_INFO_PATTERN", "INFO_PATTERN", + "FILE_NAME_PATTERN", "NAME_PATTERN", + "FILE_SIZE_PATTERN", "SIZE_PATTERN"): #@TODO: Remove old patterns starting with "FILE_" in 0.4.10 try: attr = getattr(cls, pattern) - info.update(re.search(attr, html).groupdict()) + dict = re.search(attr, html).groupdict() + + if all(True for k in dict if k not in info): + info.update(dict) + except AttributeError: continue + else: online = True @@ -316,8 +326,7 @@ class SimpleHoster(Hoster): premium_only = 'error' in self.info and self.info['error'] == "premium-only" - info = self.getInfo(pyfile.url, self.html) - self._updateInfo(info) + self._updateInfo(self.getInfo(pyfile.url, self.html)) self.checkNameSize() @@ -409,7 +418,9 @@ class SimpleHoster(Hoster): #: Deprecated def getFileInfo(self): - return self.checkInfo() + self.info = {} + self.checkInfo() + return self.info def _updateInfo(self, info): @@ -419,14 +430,15 @@ class SimpleHoster(Hoster): def handleDirect(self): - self.link = _getDirectLink(self, self.pyfile.url) + link = _getDirectLink(self, self.pyfile.url) - if self.link: + if link: self.logInfo(_("Direct download link detected")) + self.link = link + self._updateInfo(self.getInfo(self.pyfile.url)) self.checkNameSize() - else: self.logDebug(_("Direct download link not found")) @@ -480,7 +492,7 @@ class SimpleHoster(Hoster): self.retry(max_tries=max_tries, reason=_("Download limit reached")) - def parseHtmlForm(self, attr_str='', input_names=None): + def parseHtmlForm(self, attr_str="", input_names={}): return parseHtmlForm(attr_str, self.html, input_names) diff --git a/module/plugins/internal/UnRar.py b/module/plugins/internal/UnRar.py index 12bd84d20..c15a4c96e 100644 --- a/module/plugins/internal/UnRar.py +++ b/module/plugins/internal/UnRar.py @@ -22,7 +22,7 @@ def renice(pid, value): class UnRar(AbtractExtractor): __name__ = "UnRar" - __version__ = "0.18" + __version__ = "0.19" __description__ = """Rar extractor plugin""" __license__ = "GPLv3" @@ -32,12 +32,12 @@ class UnRar(AbtractExtractor): CMD = "unrar" # there are some more uncovered rar formats - re_version = re.compile(r"(UNRAR 5[\d.]+(.*?)freeware)") - re_splitfile = re.compile(r"(.*)\.part(\d+)\.rar$", re.I) - re_partfiles = re.compile(r".*\.(rar|r\d+)", re.I) - re_filelist = re.compile(r"(.+)\s+(\d+)\s+(\d+)\s+") - re_filelist5 = re.compile(r"(.+)\s+(\d+)\s+\d\d-\d\d-\d\d\s+\d\d:\d\d\s+(.+)") - re_wrongpwd = re.compile("(Corrupt file or wrong password|password incorrect)", re.I) + re_version = re.compile(r'UNRAR ([\w .]+?) freeware') + re_splitfile = re.compile(r'(.*)\.part(\d+)\.rar$', re.I) + re_partfiles = re.compile(r'.*\.(rar|r\d+)', re.I) + re_filelist = re.compile(r'(.+)\s+(\d+)\s+(\d+)\s+') + re_filelist5 = re.compile(r'(.+)\s+(\d+)\s+\d\d-\d\d-\d\d\s+\d\d:\d\d\s+(.+)') + re_wrongpwd = re.compile(r'(Corrupt file or wrong password|password incorrect)', re.I) @staticmethod diff --git a/module/plugins/internal/XFSAccount.py b/module/plugins/internal/XFSAccount.py index f2a4e8a15..5a265c08a 100644 --- a/module/plugins/internal/XFSAccount.py +++ b/module/plugins/internal/XFSAccount.py @@ -12,7 +12,7 @@ from module.plugins.internal.SimpleHoster import parseHtmlForm, set_cookies class XFSAccount(Account): __name__ = "XFSAccount" __type__ = "account" - __version__ = "0.30" + __version__ = "0.31" __description__ = """XFileSharing account plugin""" __license__ = "GPLv3" @@ -109,12 +109,12 @@ class XFSAccount(Account): else: self.logDebug("TRAFFIC_LEFT_PATTERN not found") - m = re.finditer(self.LEECH_TRAFFIC_PATTERN, html) - if m: + leech = [m.groupdict() for m in re.finditer(self.LEECH_TRAFFIC_PATTERN, html)] + if leech: leechtraffic = 0 try: - for leech in m: - size = leech['S'] + for traffic in leech: + size = traffic['S'] if "nlimited" in size: leechtraffic = -1 @@ -122,8 +122,8 @@ class XFSAccount(Account): validuntil = -1 break else: - if 'U' in leech: - unit = leech['U'] + if 'U' in traffic: + unit = traffic['U'] elif isinstance(self.LEECH_TRAFFIC_UNIT, basestring): unit = self.LEECH_TRAFFIC_UNIT else: diff --git a/module/plugins/internal/XFSHoster.py b/module/plugins/internal/XFSHoster.py index 30864ef02..2aaf18b1a 100644 --- a/module/plugins/internal/XFSHoster.py +++ b/module/plugins/internal/XFSHoster.py @@ -16,7 +16,7 @@ from module.utils import html_unescape class XFSHoster(SimpleHoster): __name__ = "XFSHoster" __type__ = "hoster" - __version__ = "0.24" + __version__ = "0.26" __pattern__ = r'^unmatchable$' @@ -33,10 +33,10 @@ class XFSHoster(SimpleHoster): TEXT_ENCODING = False COOKIES = [(HOSTER_DOMAIN, "lang", "english")] CHECK_DIRECT_LINK = None - MULTI_HOSTER = False + MULTI_HOSTER = True #@NOTE: Should be default to False for safe, but I'm lazy... INFO_PATTERN = r'<tr><td align=right><b>Filename:</b></td><td nowrap>(?P<N>[^<]+)</td></tr>\s*.*?<small>\((?P<S>[^<]+)\)</small>' - NAME_PATTERN = r'(>Filename:</b></td><td nowrap>|name="fname" value="|<span class="name">|<[Tt]itle>.*?Download )(?P<N>.+?)(\s*<|")' + NAME_PATTERN = r'(>Filename:</b></td><td nowrap>|name="fname" value="|<span class="name">)(?P<N>.+?)(\s*<|")' SIZE_PATTERN = r'(>Size:</b></td><td>|>File:.*>|<span class="size">)(?P<S>[\d.,]+)\s*(?P<U>[\w^_]+)' OFFLINE_PATTERN = r'>\s*\w+ (Not Found|file (was|has been) removed)' @@ -46,14 +46,17 @@ class XFSHoster(SimpleHoster): PREMIUM_ONLY_PATTERN = r'>This file is available for Premium Users only' ERROR_PATTERN = r'(?:class=["\']err["\'].*?>|<[Cc]enter><b>|>Error</td>|>\(ERROR:)(?:\s*<.+?>\s*)*(.+?)(?:["\']|<|\))' - OVR_LINK_PATTERN = r'<h2>Download Link</h2>\s*<textarea[^>]*>([^<]+)' - LINK_PATTERN = None #: final download url pattern + LEECH_LINK_PATTERN = r'<h2>Download Link</h2>\s*<textarea[^>]*>([^<]+)' + LINK_PATTERN = None #: final download url pattern CAPTCHA_PATTERN = r'(https?://[^"\']+?/captchas?/[^"\']+)' CAPTCHA_DIV_PATTERN = r'>Enter code.*?<div.*?>(.+?)</div>' RECAPTCHA_PATTERN = None SOLVEMEDIA_PATTERN = None + FORM_PATTERN = None + FORM_INPUTS_MAP = None #: dict passed as input_names to parseHtmlForm + def setup(self): self.chunkLimit = 1 @@ -143,7 +146,7 @@ class XFSHoster(SimpleHoster): #only tested with easybytez.com self.html = self.load("http://www.%s/" % self.HOSTER_DOMAIN) - action, inputs = self.parseHtmlForm('') + action, inputs = self.parseHtmlForm() upload_id = "%012d" % int(random() * 10 ** 12) action += upload_id + "&js_on=1&utype=prem&upload_type=url" @@ -184,11 +187,11 @@ class XFSHoster(SimpleHoster): self.fail(stmsg) #get easybytez.com link for uploaded file - m = re.search(self.OVR_LINK_PATTERN, self.html) + m = re.search(self.LEECH_LINK_PATTERN, self.html) if m is None: - self.error(_("OVR_LINK_PATTERN not found")) + self.error(_("LEECH_LINK_PATTERN not found")) - header = self.load(m.group(1).strip(), just_header=True, decode=True) #@TODO: Remove .strip() in 0.4.10 + header = self.load(m.group(1), just_header=True, decode=True) if 'location' in header: #: Direct download link self.link = header['location'] @@ -257,10 +260,10 @@ class XFSHoster(SimpleHoster): def getPostParameters(self): - if hasattr(self, "FORM_PATTERN"): - action, inputs = self.parseHtmlForm(self.FORM_PATTERN) + if self.FORM_PATTERN or self.FORM_INPUTS_MAP: + action, inputs = self.parseHtmlForm(self.FORM_PATTERN or "", self.FORM_INPUTS_MAP or {}) else: - action, inputs = self.parseHtmlForm(input_names={"op": re.compile("^download")}) + action, inputs = self.parseHtmlForm(input_names={'op': re.compile(r'^download')}) if not inputs: action, inputs = self.parseHtmlForm('F1') @@ -311,8 +314,8 @@ class XFSHoster(SimpleHoster): m = re.search(self.CAPTCHA_DIV_PATTERN, self.html, re.S) if m: captcha_div = m.group(1) + numerals = re.findall(r'<span.*?padding-left\s*:\s*(\d+).*?>(\d)</span>', html_unescape(captcha_div)) self.logDebug(captcha_div) - numerals = re.findall(r'<span.*?padding-left\s*:\s*(\d+).*?>(\d)</span>', html_unescape(captcha_div)) inputs['code'] = "".join([a[1] for a in sorted(numerals, key=lambda num: int(num[0]))]) self.logDebug("Captcha code: %s" % inputs['code'], numerals) return 2 @@ -322,9 +325,10 @@ class XFSHoster(SimpleHoster): captcha_key = re.search(self.RECAPTCHA_PATTERN, self.html).group(1) except: captcha_key = recaptcha.detect_key() + else: + self.logDebug("ReCaptcha key: %s" % captcha_key) if captcha_key: - self.logDebug("ReCaptcha key: %s" % captcha_key) inputs['recaptcha_challenge_field'], inputs['recaptcha_response_field'] = recaptcha.challenge(captcha_key) return 3 @@ -333,9 +337,10 @@ class XFSHoster(SimpleHoster): captcha_key = re.search(self.SOLVEMEDIA_PATTERN, self.html).group(1) except: captcha_key = solvemedia.detect_key() + else: + self.logDebug("SolveMedia key: %s" % captcha_key) if captcha_key: - self.logDebug("SolveMedia key: %s" % captcha_key) inputs['adcopy_challenge'], inputs['adcopy_response'] = solvemedia.challenge(captcha_key) return 4 |