def __init__(self, http_entry): assert isinstance(http_entry, HTTPLogEntry) Node.__init__(self) self._dummy = 0 # help to process dummy nodes self.aem_pred = None # the classified type for preceeding entity by AEM model self.aem_last = None # the classified type for LAST entity by AEM model self.fake_link = True # if the link is fake derived from referrer self.pl = http_entry # http log entries self.tag = self.gen_tag() # formated tag to show by tree
def __init__(self, *args, document, **kwargs):
if 'tag' not in kwargs:
kwargs['tag'] = self.__class__.__name__
name = kwargs.get('name', kwargs['tag'])
self.document = document
# node consumes data
Node.__init__(self, *args, **kwargs)
State.__init__(
self,
name,
on_enter=kwargs.get('on_enter'),
on_exit=kwargs.get('on_exit'),
ignore_invalid_triggers=kwargs.get('ignore_invalid_triggers'))
self.data = self
def __init__( self, record, offset, tag=None, identifier=None, expanded=True ):
Node.__init__( self, tag=tag, identifier=identifier, expanded=expanded )
self.record = record
self.offset = offset
self.payload_offset = 0
self.xattrs = {}
self.xattrs[ "offset" ] = str( offset )
for k, v in record.headers:
self.xattrs[ k ] = v
if record.type == WarcRecord.RESPONSE and record.url.startswith( "http" ):
mime, data = record.content
if data.startswith( "HTTP" ):
match = re.search( "\r?\n(\r?\n)+", data, re.MULTILINE )
self.xattrs[ "http.headers" ] = data[ 0:match.end() ]
self.payload_offset = match.end()
