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
|
2012-10-11 10:03:12 +00:00
|
|
|
from .util import fetchUrl, 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."""
|
2012-06-20 19:58:13 +00:00
|
|
|
url = (imageUrl, pageUrl)[usePageUrl]
|
|
|
|
return getQueryParams(url)[paramName][0]
|
|
|
|
return _namer
|
|
|
|
|
|
|
|
|
|
|
|
def regexNamer(regex):
|
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."""
|
2012-11-13 18:12:28 +00:00
|
|
|
mo = regex.search(imageUrl)
|
|
|
|
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."""
|
2013-02-05 18:51:46 +00:00
|
|
|
url1 = fetchUrl(url, cls.prevSearch, session=cls.session)
|
|
|
|
if not url1:
|
|
|
|
raise ValueError("could not find prevSearch pattern %r in %s" % (cls.prevSearch.pattern, url))
|
|
|
|
url2 = fetchUrl(url1, nextSearch, session=cls.session)
|
2012-12-12 16:41:29 +00:00
|
|
|
if not url2:
|
2013-02-05 18:51:46 +00:00
|
|
|
raise ValueError("could not find nextSearch pattern %r in %s" % (nextSearch.pattern, url1))
|
2012-12-12 16:41:29 +00:00
|
|
|
return url2
|
2012-06-20 19:58:13 +00:00
|
|
|
return _starter
|
|
|
|
|
|
|
|
|
|
|
|
def indirectStarter(baseUrl, 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."""
|
|
|
|
url = fetchUrl(baseUrl, latestSearch, session=cls.session)
|
2012-11-21 20:57:26 +00:00
|
|
|
if not url:
|
|
|
|
raise ValueError("could not find latestSearch pattern %r in %s" % (latestSearch.pattern, baseUrl))
|
|
|
|
return url
|
2012-06-20 19:58:13 +00:00
|
|
|
return _starter
|
|
|
|
|