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, httpResponse):
abstractParser.__init__(self , httpResponse)
self._parsed_URLs = []
self._re_URLs = []
# work !
self._preParse( httpResponse.getBody() )
def __init__(self, httpResponse):
abstractParser.__init__(self , httpResponse)
# Two lists with url objects
self._parsed_URLs = []
self._re_URLs = []
# Work !
self._preParse( httpResponse.getBody() )
def __init__(self, httpResponse):
abstractParser.__init__(self , httpResponse)
# To store results
self._parsed_URLs = []
self._re_URLs = []
# work !
swf = httpResponse.getBody()
if self._is_compressed( swf ):
try:
swf = self._inflate( swf )
except Exception, e:
# If the inflate fails... there is nothing else to do.
return