2012-06-20 20:41:04 +00:00
|
|
|
# -*- coding: iso-8859-1 -*-
|
|
|
|
# Copyright (C) 2004-2005 Tristan Seligmann and Jonathan Jacobs
|
2013-02-05 18:51:46 +00:00
|
|
|
# Copyright (C) 2012-2013 Bastian Kleineidam
|
2012-11-21 20:57:26 +00:00
|
|
|
|
2013-04-10 16:19:11 +00:00
|
|
|
from re import compile, escape
|
2012-06-20 19:58:13 +00:00
|
|
|
|
2012-10-11 10:03:12 +00:00
|
|
|
from ..scraper import _BasicScraper
|
|
|
|
from ..helpers import indirectStarter
|
2012-11-20 17:53:53 +00:00
|
|
|
from ..util import tagre
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
|
|
|
class Galaxion(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://galaxioncomics.com/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-02-04 20:00:26 +00:00
|
|
|
stripUrl = url + '%s/'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '1-comic/the-story-so-far/the-story-so-far'
|
2013-04-10 16:19:11 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(%scomics/[^"]+)' % rurl))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(%s[^"]+)' % rurl, after="prev"))
|
2012-11-21 20:57:26 +00:00
|
|
|
help = 'Index format: n-comic/book-n/chapter-n/title-nnn'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
|
|
|
class Garanos(_BasicScraper):
|
2013-04-10 16:19:11 +00:00
|
|
|
baseurl = 'http://garanos.alexheberling.com/'
|
|
|
|
rurl = escape(baseurl)
|
|
|
|
url = baseurl + 'pages/page-1/'
|
2013-02-04 20:00:26 +00:00
|
|
|
starter = indirectStarter(url,
|
2013-04-10 16:19:11 +00:00
|
|
|
compile(tagre("a", "href", r'(%spages/[^"]+)' % rurl, after="navi-last")))
|
|
|
|
stripUrl = baseurl + 'pages/page-%s'
|
|
|
|
imageSearch = compile(tagre("img", "src", r'(%scomics/[^"]+)' % rurl))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(%spages/[^"]+)' % rurl, after="prev"))
|
2013-04-10 21:57:09 +00:00
|
|
|
help = 'Index format: n (unpadded)'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
2013-03-06 19:21:10 +00:00
|
|
|
class GastroPhobia(_BasicScraper):
|
|
|
|
url = 'http://www.gastrophobia.com/'
|
|
|
|
stripUrl = url + 'index.php?date=%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '2008-07-30'
|
2013-03-06 19:21:10 +00:00
|
|
|
imageSearch = compile(r'<img src="(http://gastrophobia.com/comix/[^"]+)"[^>]*>(?!<br>)')
|
|
|
|
prevSearch = compile(r'<a href="(.+?)"><img src="pix/prev.gif" ')
|
|
|
|
help = 'Index format: yyyy-mm-dd'
|
|
|
|
|
|
|
|
|
|
|
|
class Geeks(_BasicScraper):
|
|
|
|
url = 'http://sevenfloorsdown.com/geeks/'
|
|
|
|
stripUrl = url + 'archives/%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '10'
|
2013-03-06 19:21:10 +00:00
|
|
|
imageSearch = compile(r'<img src=\'(http://sevenfloorsdown.com/geeks/comics/.+?)\'')
|
|
|
|
prevSearch = compile(r'<a href="(.+?)">« Previous')
|
|
|
|
help = 'Index format: nnn'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
2013-03-11 21:51:45 +00:00
|
|
|
class GeeksNextDoor(_BasicScraper):
|
|
|
|
url = 'http://www.geeksnextcomic.com/'
|
|
|
|
stripUrl = url + '%s.html'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '2010-10-04'
|
2013-03-11 21:51:45 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(images/GND\d+[^"]+)'))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(\d+-\d+-\d+\.html)') +
|
|
|
|
tagre("img", "src", r'images/nav_prev\.png'))
|
|
|
|
help = 'Index format: yyyy-mm-dd'
|
|
|
|
|
|
|
|
|
2012-06-20 19:58:13 +00:00
|
|
|
class GirlsWithSlingshots(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://www.girlswithslingshots.com/'
|
2013-04-10 21:57:09 +00:00
|
|
|
rurl = escape(url)
|
|
|
|
stripUrl = url + 'comic/gws%s/'
|
|
|
|
firstStripUrl = stripUrl % '1'
|
|
|
|
imageSearch = (
|
|
|
|
compile(tagre("img", "src", r'(%scomics/[^"]+)' % rurl)),
|
|
|
|
compile(tagre("img", "src", r'(http://cdn\.girlswithslingshots\.com/comics/[^"]+)')),
|
|
|
|
)
|
2013-04-11 16:27:43 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(%scomic/[^"]+)' % rurl, after="prev"))
|
2012-06-20 19:58:13 +00:00
|
|
|
help = 'Index format: nnn'
|
|
|
|
|
|
|
|
|
2013-03-06 19:21:10 +00:00
|
|
|
class GlassHalfEmpty(_BasicScraper):
|
|
|
|
url = 'http://www.defectivity.com/ghe/index.php'
|
|
|
|
stripUrl = url + '?strip_id=%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '0'
|
2013-03-06 19:21:10 +00:00
|
|
|
imageSearch = compile(r'src="(comics/.+?)"')
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(\?strip_id=\d+)') + tagre("img", "src", r'\.\./images/arrowbuttons/onback\.jpg'))
|
|
|
|
help = 'Index format: nnn'
|
|
|
|
|
|
|
|
|
2012-11-20 17:53:53 +00:00
|
|
|
class GleefulNihilism(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://gleefulnihilism.com/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-02-04 20:00:26 +00:00
|
|
|
stripUrl = url + 'comics/%s/'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '2008/10/20/amoeba'
|
2013-04-10 16:19:11 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(%scomics/[^"]+)' % rurl))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(%scomics/[^"]+)' % rurl) + 'Previous')
|
2012-11-20 17:53:53 +00:00
|
|
|
help = 'Index format: yyyy/mm/dd/stripname'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
|
|
|
class Goats(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://www.goats.com/'
|
|
|
|
stripUrl = url + 'archive/%s.html'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '970401'
|
2012-06-20 19:58:13 +00:00
|
|
|
imageSearch = compile(r'<img.+?src="(/comix/.+?)"')
|
|
|
|
prevSearch = compile(r'<a href="(/archive/\d{6}.html)" class="button" title="go back">')
|
|
|
|
help = 'Index format: yymmdd'
|
|
|
|
|
|
|
|
|
2013-04-09 17:37:24 +00:00
|
|
|
class GoblinsComic(_BasicScraper):
|
|
|
|
url = 'http://www.goblinscomic.com/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-04-09 17:37:24 +00:00
|
|
|
stripUrl = url + '%s/'
|
|
|
|
firstStripUrl = stripUrl % '06252005'
|
2013-04-10 16:19:11 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(%s\d+/)' % rurl, after="prev"))
|
|
|
|
imageSearch = compile(tagre("img", "src", r'(%scomics/\d+\.[^"]+)' % rurl))
|
2013-04-09 17:37:24 +00:00
|
|
|
help = 'Index format: ddmmyyyy'
|
|
|
|
|
|
|
|
|
2012-06-20 19:58:13 +00:00
|
|
|
class GoneWithTheBlastwave(_BasicScraper):
|
2013-02-05 18:51:46 +00:00
|
|
|
url = 'http://www.blastwave-comic.com/index.php?p=comic&nro=1'
|
|
|
|
starter = indirectStarter(url,
|
2012-06-20 19:58:13 +00:00
|
|
|
compile(r'href="(index.php\?p=comic&nro=\d+)"><img src="images/page/default/latest'))
|
2013-02-05 18:51:46 +00:00
|
|
|
stripUrl = url[:-1] + '%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '1'
|
2012-06-20 19:58:13 +00:00
|
|
|
imageSearch = compile(r'<img.+src=".+(/comics/.+?)"')
|
|
|
|
prevSearch = compile(r'href="(index.php\?p=comic&nro=\d+)"><img src="images/page/default/previous')
|
|
|
|
help = 'Index format: n'
|
|
|
|
|
|
|
|
@classmethod
|
|
|
|
def namer(cls, imageUrl, pageUrl):
|
|
|
|
return '%02d' % int(compile(r'nro=(\d+)').search(pageUrl).group(1))
|
|
|
|
|
|
|
|
|
2013-01-29 20:42:10 +00:00
|
|
|
class GrrlPower(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://www.grrlpowercomic.com/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-02-04 20:00:26 +00:00
|
|
|
stripUrl = url + 'archives/%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '48'
|
2013-04-10 16:19:11 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(%scomics/[^"]+)' % rurl))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(%sarchives/\d+)' % rurl, after="navi-prev"))
|
2013-01-29 20:42:10 +00:00
|
|
|
help = 'Index format: number'
|
|
|
|
|
|
|
|
|
2012-06-20 19:58:13 +00:00
|
|
|
class GunnerkrigCourt(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://www.gunnerkrigg.com/'
|
|
|
|
stripUrl = url + '?p=%s'
|
2013-01-29 18:00:29 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(/comics/[^"]+)'))
|
|
|
|
prevSearch = compile(tagre("a", "href", r'(\?p=\d+)') + tagre("img", "src", "http://www\.gunnerkrigg\.com/images/prev_a\.jpg"))
|
|
|
|
help = 'Index format: number'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
|
|
|
class Gunshow(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://gunshowcomic.com/'
|
|
|
|
stripUrl = url + '%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '1'
|
2012-11-21 20:57:26 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(http://gunshowcomic\.com/comics/[^"]+)'))
|
2012-12-04 06:02:40 +00:00
|
|
|
multipleImagesPerStrip = True
|
|
|
|
prevSearch = compile(tagre("a", "href", r'([^"]+)') + tagre("img", "src", r'[^"]*menu/small/previous\.gif'))
|
2012-11-21 20:57:26 +00:00
|
|
|
help = 'Index format: n'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
2013-03-06 19:21:10 +00:00
|
|
|
class GUComics(_BasicScraper):
|
2013-03-07 22:08:17 +00:00
|
|
|
url = 'http://www.gucomics.com/'
|
|
|
|
stripUrl = url + '%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '20000710'
|
2013-03-06 19:21:10 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(/comics/\d{4}/gu_[^"]+)'))
|
2013-03-07 22:08:17 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(/\d+)') +
|
2013-03-06 19:21:10 +00:00
|
|
|
tagre("img", "src", r'/images/nav/prev\.png'))
|
|
|
|
help = 'Index format: yyyymmdd'
|