2012-06-20 20:41:04 +00:00
|
|
|
# -*- coding: iso-8859-1 -*-
|
|
|
|
# Copyright (C) 2004-2005 Tristan Seligmann and Jonathan Jacobs
|
2013-02-05 18:52:10 +00:00
|
|
|
# Copyright (C) 2012-2013 Bastian Kleineidam
|
2013-02-11 18:43:46 +00:00
|
|
|
from .util import fetchUrl, getPageContent, getQueryParams
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
def queryNamer(paramName, usePageUrl=False):
|
2012-09-26 14:47:39 +00:00
|
|
|
"""Get name from URL query part."""
|
2012-12-12 16:41:29 +00:00
|
|
|
@classmethod
|
|
|
|
def _namer(cls, imageUrl, pageUrl):
|
|
|
|
"""Get URL query part."""
|
2013-03-07 17:22:49 +00:00
|
|
|
url = pageUrl if usePageUrl else imageUrl
|
2012-06-20 19:58:13 +00:00
|
|
|
return getQueryParams(url)[paramName][0]
|
|
|
|
return _namer
|
|
|
|
|
|
|
|
|
2013-03-07 17:22:49 +00:00
|
|
|
def regexNamer(regex, usePageUrl=False):
|
2012-09-26 14:47:39 +00:00
|
|
|
"""Get name from regular expression."""
|
2012-12-12 16:41:29 +00:00
|
|
|
@classmethod
|
|
|
|
def _namer(cls, imageUrl, pageUrl):
|
|
|
|
"""Get first regular expression group."""
|
2013-03-07 17:22:49 +00:00
|
|
|
url = pageUrl if usePageUrl else imageUrl
|
|
|
|
mo = regex.search(url)
|
2012-11-13 18:12:28 +00:00
|
|
|
if mo:
|
|
|
|
return mo.group(1)
|
2012-06-20 19:58:13 +00:00
|
|
|
return _namer
|
|
|
|
|
|
|
|
|
2013-02-05 18:51:46 +00:00
|
|
|
def bounceStarter(url, nextSearch):
|
2012-09-26 14:47:39 +00:00
|
|
|
"""Get start URL by "bouncing" back and forth one time."""
|
2012-06-20 19:58:13 +00:00
|
|
|
@classmethod
|
|
|
|
def _starter(cls):
|
2012-12-12 16:41:29 +00:00
|
|
|
"""Get bounced start URL."""
|
2013-02-12 16:55:13 +00:00
|
|
|
data, baseUrl = getPageContent(url, cls.session)
|
2013-02-11 18:43:46 +00:00
|
|
|
url1 = fetchUrl(url, data, baseUrl, cls.prevSearch)
|
2013-02-12 16:55:13 +00:00
|
|
|
data, baseUrl = getPageContent(url1, cls.session)
|
2013-02-11 18:43:46 +00:00
|
|
|
return fetchUrl(url1, data, baseUrl, nextSearch)
|
2012-06-20 19:58:13 +00:00
|
|
|
return _starter
|
|
|
|
|
|
|
|
|
2013-02-11 18:43:46 +00:00
|
|
|
def indirectStarter(url, latestSearch):
|
2012-09-26 14:47:39 +00:00
|
|
|
"""Get start URL by indirection."""
|
2012-12-12 16:41:29 +00:00
|
|
|
@classmethod
|
|
|
|
def _starter(cls):
|
|
|
|
"""Get indirect start URL."""
|
2013-02-12 16:55:13 +00:00
|
|
|
data, baseUrl = getPageContent(url, cls.session)
|
2013-02-11 18:43:46 +00:00
|
|
|
return fetchUrl(url, data, baseUrl, latestSearch)
|
2012-06-20 19:58:13 +00:00
|
|
|
return _starter
|