def __init__(self, pattern):
SGMLParser.__init__(self)
self.urls = []
self.is_done = False
self.target_ref = ""
self.pattern = pattern
self.dirs = {}
def __init__(self, email, password):
SGMLParser.__init__(self)
self.h3 = False
self.h3_is_ready = False
self.div = False
self.h3_and_div = False
self.a = False
self.depth = 0
self.names = ""
self.dic = {}
self.email = email
self.password = password
self.domain = 'renren.com'
try:
cookie = cookielib.CookieJar()
cookieProc = urllib2.HTTPCookieProcessor(cookie)
except:
raise
else:
proxy_handler = urllib2.ProxyHandler({'http':'http://10.17.75.253:3128'})
proxy_auth_handler = urllib2.ProxyBasicAuthHandler()
# opener = urllib2.build_opener(proxy_handler, proxy_auth_handler)
opener = urllib2.build_opener(cookieProc, proxy_handler, proxy_auth_handler)
urllib2.install_opener(opener)
def __init__(self):
SGMLParser.__init__(self)
self.is_tbl = False
self.is_tr = False
self.is_td = False
self.row_count = 0
self.out = []
def __init__(self): SGMLParser.__init__(self) self.is_div_day = False self.get_title_data = False self.level = 0 self.comments = [] self.postTitle = []
def __init__(self, markup='', parseOnlyThese=None, fromEncoding=None, markupMassage=True, smartQuotesTo=XML_ENTITIES, convertEntities=None, selfClosingTags=None, isHTML=False):
self.parseOnlyThese = parseOnlyThese
self.fromEncoding = fromEncoding
self.smartQuotesTo = smartQuotesTo
self.convertEntities = convertEntities
if self.convertEntities:
self.smartQuotesTo = None
if (convertEntities == self.HTML_ENTITIES):
self.convertXMLEntities = False
self.convertHTMLEntities = True
self.escapeUnrecognizedEntities = True
elif (convertEntities == self.XHTML_ENTITIES):
self.convertXMLEntities = True
self.convertHTMLEntities = True
self.escapeUnrecognizedEntities = False
elif (convertEntities == self.XML_ENTITIES):
self.convertXMLEntities = True
self.convertHTMLEntities = False
self.escapeUnrecognizedEntities = False
else:
self.convertXMLEntities = False
self.convertHTMLEntities = False
self.escapeUnrecognizedEntities = False
self.instanceSelfClosingTags = buildTagMap(None, selfClosingTags)
SGMLParser.__init__(self)
if hasattr(markup, 'read'):
markup = markup.read()
self.markup = markup
self.markupMassage = markupMassage
try:
self._feed(isHTML=isHTML)
except StopParsing:
pass
self.markup = None
def __init__(self):
SGMLParser.__init__(self)
self.flag = False
self.a_flag = False
self.b_flag = False
self.urls = []
self.titles = []
def __init__(self, context=None, request=None):
SGMLParser.__init__(self)
self.current_status = None
self.context = context
self.request = request
self.pieces = []
self.in_link = False
def __init__(self, term, verbose=False):
SGMLParser.__init__(self)
self.inside_pre = False
self.inside_a = False
self.term = term
self.verbose = verbose
self.major = Major.objects.all()[0]
def __init__(self):
SGMLParser.__init__(self)
self.channels_data = []
self.state = ["init"]
self.current_href = None
self.success = False
def __init__(self): SGMLParser.__init__(self) self.is_div = "" self.name = [] self.is_a = "" self.div_counter = 0 self.a_counter = 0
def __init__(self,url):
SGMLParser.__init__(self)
SGMLParser.reset(self)
self.vul_value = []
self.url_value = ""
self.is_table = None
self.is_date = None
self.is_tr = None
self.is_span = None
self.is_vul_name = None
self.is_td = None
self.is_tr = None
self.is_name = None
self.is_vul = None
self.is_name_desc = None
self.is_date_desc = None
self.is_cvss_desc = None
self.is_a = None
self.is_cvss = None
self.is_href_desc = None
self.is_url = None
self.is_vul_desc = None
self.table_count = 0
self.cvss_name = ""
self.vul_name = ""
self.href_name = ""
self.vul_type_name = ""
self.date_name = ""
self.td_count = 0
self.cve_name = ""
self.a_count = 0
self.desc_name = ""
self.url_name = url
def __init__(self, input_file, mapping): input_file.seek(0) self.input_file = input_file self.mapping = mapping self.current_offset = 0 self.count = 0 SGMLParser.__init__(self)
def __init__(self, httpResponse, normalizeMarkup=True, verbose=0):
abstractParser.__init__( self, httpResponse )
SGMLParser.__init__(self, verbose)
# Set some constants
self._tagsContainingURLs = ('go', 'a','img', 'link', 'script', 'iframe', 'object',
'embed', 'area', 'frame', 'applet', 'input', 'base',
'div', 'layer', 'ilayer', 'bgsound', 'form')
self._urlAttrs = ('href', 'src', 'data', 'action' )
# And some internal variables
self._tag_and_url = []
self._parsed_URLs = []
self._re_URLs = []
self._encoding = httpResponse.getCharset()
self._forms = []
self._insideForm = False
self._insideSelect = False
self._insideTextarea = False
self._insideScript = False
self._commentsInDocument = []
self._scriptsInDocument = []
# Meta tags
self._metaRedirs = []
self._metaTags = []
self._normalizeMarkup = normalizeMarkup
# Fill self._re_URLs list with url objects
self._regex_url_parse( httpResponse )
# Now we are ready to work
self._preParse( httpResponse )
def __init__(self):
BaseModule.__init__(self)
SGMLParser.__init__(self)
self.is_b = 0
self.is_dl = 0
self.is_p = 0
self.is_br = 0
self.is_tr = 0
self.is_td = 0
self.is_a = 0
self.is_span = 0
self.is_font = 0
self.is_li = 0
self.is_div = 0
self.is_dt = 0
self.is_ul = 0
self.is_dd = 0
self.is_table = 0
self.is_em = 0
self.is_tbody = 0
self.is_img = 0
self.is_h1 = 0
self.is_h3 = 0
self.is_i = 0
self.req = None
def __init__(self, email, password):
SGMLParser.__init__(self)
self.h3 = False
self.h3_is_ready = False
self.div = False
self.h3_and_div = False
self.a = False
self.depth = 0
self.names = ""
self.dic = {}
self.email = email
self.password = password
self.domain = 'renren.com'
self.file = None
self.friend_file = None
self.ghost = Ghost()
self.cookie = None
self.group_url = "http://friend.renren.com/groupsdata"
self.group_home = "http://friend.renren.com/managefriends"
self.file_url = "renren_cookie.txt"
self.mongodb = MongoClient("127.0.0.1", 27017)
try:
self.cookie = cookielib.LWPCookieJar(self.file_url)
cookieProc = urllib2.HTTPCookieProcessor(self.cookie)
except:
raise
else:
opener = urllib2.build_opener(cookieProc)
urllib2.install_opener(opener)
print "init finished successfully!!"
def __init__(self, baseuri):
SGMLParser.__init__(self)
self.links = []
self.baseuri = baseuri
self.startlink = None
self.homelink = None
self.prearchiveslink = None
def __init__(self):
self.url = None
self.links = []
self.linkpos = {}
self.images = []
# Keywords
self.keywords = []
# Description of page
self.description = ''
# Title of page
self.title = ''
self.title_flag = True
# Fix for <base href="..."> links
self.base_href = False
# Base url for above
self.base = None
# anchor links flag
self._anchors = True
# For META robots tag
self.can_index = True
self.can_follow = True
# Current tag
self._tag = ''
SGMLParser.__init__(self)
# Type
self.typ = 0
def __init__(self):
SGMLParser.__init__(self)
self.majors = {}
self.inside_row = False
self.col = 0
self.capture = False
self.data = ''
def __init__(self,username,password):
SGMLParser.__init__(self)
self.username = username
self.password = password
cj = cookielib.CookieJar()
opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cj))
urllib2.install_opener(opener)
def __init__(self, site):
SGMLParser.__init__(self)
self.site=site
self.city=[]
self.name=[]
self.url=[]
self.state=StateBase(self)
def __init__(self):
self.result = []
self.data = ""
self.link = ""
self.TAG_BEG = False
self.TAG_END = False
SGMLParser.__init__(self, 0)
def __init__(self, base, username, password):
SGMLParser.__init__(self)
self.h3 = False
self.h3_is_ready = False
self.div = False
self.h3_and_div = False
self.a = False
self.depth = 0
self.names = ""
self.dic = {}
self.base = base
self.fastloginfield = "username"
self.username = username
self.password = password
self.quickforward = "yes"
self.handlekey = "ls"
try:
cookie = cookielib.CookieJar()
cookieProc = urllib2.HTTPCookieProcessor(cookie)
except:
raise
else:
opener = urllib2.build_opener(cookieProc)
urllib2.install_opener(opener)
def __init__( self, verbose=0 ):
SGMLParser.__init__( self, verbose )
self.savedata = None
self.title = ''
self.metatags = {}
self.body = ''
def __init__(self):
self.data = ""
self.links = []
self.title = ""
self.TAG_BEG = False
self.TAG_END = False
SGMLParser.__init__(self, 0)
def __init__(self):
SGMLParser.__init__(self)
self.infoFlag = ""
self.appurl=[]
self.appname=[]
self.descriptionFlag=''
self.descriptionPFlag=''
self.description=[]
self.appnameFlag=''
self.info=[]
self.iphonescreenshots=[]
self.ipadscreenshots=[]
self.whatisnewFlag=''
self.whatisnew=[]
self.allstartFlag=''
self.allstart=[]
self.customerstartFlag=''
self.customerstart=[]
self.customerFlag=''
self.customer=[]
self.iphonescreenshotsFlag=''
self.ipadscreenshotsFlag=''
self.nameFlag=''
self.name=''
self.icon=''
def __init__(self, wordname):
self._word = word.Word()
self._word.set_name(wordname)
SGMLParser.__init__(self)
self._phonetic_found = 0
self._etcTrans_found = 0
self._etcTrans_li_found = 0
self._p_additional_found = 0
self._synonyms_found = 0
self._synonyms_a_found = 0
self._tPowerTrans_found = 0
self._tPowerTrans_ul_sense_ex_found = 0
self._tPowerTrans_ul_sense_ex_p_found = 0
self._examplesToggle_found = 0
self._examplesToggle_bilingual_found = 0
self._examplesToggle_bilingual_p_found = 0
self._examplesToggle_bilingual_p_noattrs_found = 0 #用来解决“报错”这个<p>
self._tmp_sentence_list = []
self._tmp_sentence = ''
def __init__(self):
SGMLParser.__init__(self)
self.data = ''
self.capture = False
self.inside_select = False
self.current_id = ''
self.terms = {}
def __init__(self):
SGMLParser.__init__(self)
self.data = ""
#instantiate an array of hyperlinks
self.hyperlinks = []
self.descriptions = []
self.inside_a_element = 0
def __init__(self, verbose=0): "Initialise an object, passing 'verbose' to the superclass." SGMLParser.__init__(self, verbose) self.hyperlinks = [] self.descriptions = [] self.inside_a_element = 0 self.starting_description = 0
def __init__(self, source, type):
self.source = source
self.type = type
self.is_title = 0
self.title_text = ""
self.title_list = []
self.href_list = []
SGMLParser.__init__(self)
def __init__(self, verbose=0):
SGMLParser.__init__(self, verbose)
self.links = []
self.iframes = []
self.starting_description = False
def __init__(self, content_start_comment='', content_end_comment=''):
SGMLParser.__init__(self)
self.buf = ""
self.output = ""
self.entitydefs.update({'auml':'ä','ouml':'ö','aring':'å','Auml':'Ä','Ouml':'Ö','Aring':'Å','nbsp':' ',})
def __init__(self):
SGMLParser.__init__(self)
self.data = None
self.currenttable = None
self.lastcaption = None
def __init__(self):
SGMLParser.__init__(self)
self.hrefs = []
self.srcs = []
#保存视频数组
self.hrefs2 = []
def __init__(self, verbose=0):
self._init()
# Fall-back defaults.
self._modFunct = None
self._as = 'http'
SGMLParser.__init__(self, verbose)
def __init__(self, valid_tags=None, nasty_tags=None):
SGMLParser.__init__(self)
self.result = ""
self.valid_tags = valid_tags or VALID_TAGS
self.nasty_tags = nasty_tags or NASTY_TAGS
def __init__(self, baseUrl): SGMLParser.__init__(self) self.baseUrl = baseUrl self.leaves = [] self.directories = []
def __init__(self, verbose=0):
SGMLParser.__init__(self, verbose)
self.title = self.data = None
self._reset()
def __init__(self):
SGMLParser.__init__(self)
def __init__(self) :
SGMLParser.__init__(self)
phpbb_prompt.PromptParser.__init__(self)
self._parsing_form = False
def __init__(self):
SGMLParser.__init__(self)
self.img_locations = []
def __init__(self):
SGMLParser.__init__(self)
self.scripts = []
"""HTML 2.0 parser.
def __init__(self): self.text = [] SGMLParser.__init__(self)
def __init__(self):
SGMLParser.__init__(self)
self.inside_table = False
self.capture = False
self.data = ''
self.section_ids = []
def __init__(self):
SGMLParser.__init__(self)
self.result = ""
def __init__(self, site="163"):
SGMLParser.__init__(self)
self.site = site
def __init__(self): SGMLParser.__init__(self) self.is_h4 = "" self.name = []
def __init__(self, verbose=0):
SGMLParser.__init__(self, verbose)
self.savedata = None
self.title = ''
self.metatags = {}
self.body = ''
def __init__(self):
SGMLParser.__init__(self)
self.docs = []
def __init__(self):
SGMLParser.__init__(self)
self.list = []
def __init__(self):
SGMLParser.__init__(self)
self.data = ''
self.capture = False
self.inside_select = False
self.majors = {}
def __init__(self,out): HTMLParser.__init__(self) self.out = out
def __init__(self):
"""
"""
SGMLParser.__init__(self)
def __init__(self):
SGMLParser.__init__(self)
self.is_package = ""
self.packages = []
self.count = 0
def __init__(self, url):
SGMLParser.__init__(self)
self._is_weather, self._is_td = False, False
self._page_datas, self._weather = None, [] # 网页数据，天气数据
self._get_page_datas(url)
self.feed(self._page_datas)
def __init__(self):
SGMLParser.__init__(self)
self.hasil = []
self.baris = []
self.data = []
self.last_error = None
def __init__(self):
SGMLParser.__init__(self)
self.majors = {}
self.data = ''
self.capture = False
self.major_list = False
def __init__(self):
SGMLParser.__init__(self)
self.is_a = ""
self.link = []
self.Title = ""
def __init__(self):
SGMLParser.__init__(self)
phpbb_prompt.PromptParser.__init__(self)
self.__parsing_search_result = False
self.__parsing_post_body = False
self.search_result = set()