2016-04-03 20:58:01 +00:00
|
|
|
# -*- coding: utf-8 -*-
|
2016-10-28 22:21:41 +00:00
|
|
|
# Copyright (C) 2004-2008 Tristan Seligmann and Jonathan Jacobs
|
2014-01-05 15:50:57 +00:00
|
|
|
# Copyright (C) 2012-2014 Bastian Kleineidam
|
2016-04-03 20:58:01 +00:00
|
|
|
# Copyright (C) 2015-2016 Tobias Gruetzmacher
|
|
|
|
|
|
|
|
from __future__ import absolute_import, division, print_function
|
2012-06-20 19:58:13 +00:00
|
|
|
|
2013-04-10 16:19:11 +00:00
|
|
|
from re import compile, escape
|
2016-04-03 20:58:01 +00:00
|
|
|
|
2015-05-07 11:57:10 +00:00
|
|
|
from ..scraper import _BasicScraper, _ParserScraper
|
2013-04-25 19:06:12 +00:00
|
|
|
from ..helpers import bounceStarter, indirectStarter
|
2012-11-21 20:57:26 +00:00
|
|
|
from ..util import tagre
|
2016-05-01 23:25:34 +00:00
|
|
|
from .common import (_ComicControlScraper, _WordPressScraper, _WPNaviIn,
|
|
|
|
WP_LATEST_SEARCH)
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
2013-04-25 19:06:12 +00:00
|
|
|
class Lackadaisy(_BasicScraper):
|
|
|
|
baseUrl = 'http://lackadaisy.foxprints.com/'
|
|
|
|
url = baseUrl + 'comic.php'
|
|
|
|
stripUrl = baseUrl + 'comic.php?comicid=%s'
|
|
|
|
firstStripUrl = stripUrl % '1'
|
|
|
|
imageSearch = compile(tagre("img", "src", r'(http://www\.lackadaisycats\.com/comic/[^"]*)'))
|
2016-04-03 20:58:01 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r"(/comic\.php\?comicid=[0-9]+)") +
|
|
|
|
"< Previous")
|
2016-04-12 21:11:39 +00:00
|
|
|
nextSearch = compile(tagre("a", "href", r"(/comic.php\?comicid=[0-9]+)") +
|
|
|
|
"Next")
|
2013-04-25 19:06:12 +00:00
|
|
|
help = 'Index format: n'
|
2016-04-13 18:01:51 +00:00
|
|
|
starter = bounceStarter
|
2013-04-25 19:06:12 +00:00
|
|
|
|
2016-04-21 06:20:49 +00:00
|
|
|
def namer(self, image_url, page_url):
|
2013-04-25 19:06:12 +00:00
|
|
|
"""Use comic id for filename."""
|
2016-04-21 06:20:49 +00:00
|
|
|
num = page_url.rsplit('=', 1)[-1]
|
|
|
|
ext = image_url.rsplit('.', 1)[-1]
|
2013-04-25 19:06:12 +00:00
|
|
|
return 'lackadaisy_%s.%s' % (num, ext)
|
|
|
|
|
|
|
|
|
2016-04-03 20:58:01 +00:00
|
|
|
class Laiyu(_WordPressScraper):
|
|
|
|
url = 'http://www.flowerlarkstudios.com/comic/preliminary-concepts/welcome/'
|
|
|
|
firstStripUrl = url
|
2016-04-12 21:11:39 +00:00
|
|
|
latestSearch = WP_LATEST_SEARCH
|
2016-04-13 18:01:51 +00:00
|
|
|
starter = indirectStarter
|
2016-04-03 20:58:01 +00:00
|
|
|
|
|
|
|
|
2012-06-20 19:58:13 +00:00
|
|
|
class LasLindas(_BasicScraper):
|
2013-02-04 20:00:26 +00:00
|
|
|
url = 'http://laslindas.katbox.net/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-02-19 19:58:04 +00:00
|
|
|
stripUrl = url + 'comic/%s/'
|
2013-04-10 16:19:11 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(%swp-content/uploads/[^"]+)' % rurl, after="attachment-full"))
|
2013-04-04 16:30:02 +00:00
|
|
|
multipleImagesPerStrip = True
|
2013-04-10 16:19:11 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(%scomic/[^"]+)' % rurl, after="previous"))
|
2012-11-21 20:57:26 +00:00
|
|
|
help = 'Index format: stripname'
|
2012-06-20 19:58:13 +00:00
|
|
|
|
|
|
|
|
2016-01-03 01:16:58 +00:00
|
|
|
class LastNerdsOnEarth(_ParserScraper):
|
2016-05-16 21:16:29 +00:00
|
|
|
baseUrl = 'http://www.lastnerdsonearth.com/'
|
|
|
|
url = baseUrl + 'latest/'
|
|
|
|
firstStripUrl = baseUrl + 'ch1p1'
|
2016-01-03 01:16:58 +00:00
|
|
|
imageSearch = '//div[@id="content"]/a/img'
|
2016-05-16 21:16:29 +00:00
|
|
|
prevSearch = '//div[@id="comicnav"]/a[img[contains(@src, "nav-prev")]]'
|
2016-01-03 01:16:58 +00:00
|
|
|
|
|
|
|
|
2013-03-06 19:21:10 +00:00
|
|
|
class LeastICouldDo(_BasicScraper):
|
|
|
|
url = 'http://www.leasticoulddo.com/'
|
2013-04-10 16:19:11 +00:00
|
|
|
rurl = escape(url)
|
2013-03-06 19:21:10 +00:00
|
|
|
stripUrl = url + 'comic/%s'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '20130109'
|
2013-11-12 17:33:14 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(%swp-content/uploads/\d+/\d+/\d{8,9}\.\w{1,4})' % rurl))
|
2016-04-03 20:58:01 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(%scomic/\d+/)' % rurl,
|
|
|
|
after="Previous"))
|
2016-04-12 21:11:39 +00:00
|
|
|
latestSearch = compile(tagre("a", "href", r'(%scomic/\d+/)' % rurl,
|
|
|
|
after="feature-comic"))
|
2016-04-13 18:01:51 +00:00
|
|
|
starter = indirectStarter
|
2013-03-06 19:21:10 +00:00
|
|
|
help = 'Index format: yyyymmdd'
|
|
|
|
|
|
|
|
|
2016-04-03 22:12:53 +00:00
|
|
|
class LetsSpeakEnglish(_ComicControlScraper):
|
|
|
|
url = 'http://www.marycagle.com'
|
|
|
|
|
|
|
|
|
2013-03-06 19:21:10 +00:00
|
|
|
class LittleGamers(_BasicScraper):
|
|
|
|
url = 'http://www.little-gamers.com/'
|
|
|
|
stripUrl = url + '%s/'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '2000/12/01/99'
|
2013-03-06 19:21:10 +00:00
|
|
|
imageSearch = compile(tagre("img", "src", r'(http://little-gamers\.com/comics/[^"]+)'))
|
2013-03-26 16:36:06 +00:00
|
|
|
prevSearch = compile(tagre("a", "href", r'(http://www\.little-gamers\.com/[^"]+)', before="comic-nav-prev-link"))
|
2013-03-06 19:21:10 +00:00
|
|
|
help = 'Index format: yyyy/mm/dd/name'
|
|
|
|
|
|
|
|
|
2015-05-31 23:33:50 +00:00
|
|
|
class LoadingArtist(_ParserScraper):
|
2016-08-16 09:20:35 +00:00
|
|
|
url = 'http://www.loadingartist.com/latest'
|
2015-05-31 23:33:50 +00:00
|
|
|
imageSearch = '//div[@class="comic"]//img'
|
|
|
|
prevSearch = "//a[contains(concat(' ', @class, ' '), ' prev ')]"
|
2013-03-06 19:00:30 +00:00
|
|
|
|
2016-04-03 20:58:01 +00:00
|
|
|
|
2016-05-01 23:25:34 +00:00
|
|
|
class LoFiJinks(_WPNaviIn):
|
|
|
|
url = 'http://hijinksensue.com/comic/learning-to-love-again/'
|
|
|
|
firstStripUrl = 'http://hijinksensue.com/comic/lo-fijinks-everything-i-know-anout-james-camerons-avatar-movie/'
|
|
|
|
endOfLife = True
|
|
|
|
|
|
|
|
|
2015-05-07 11:57:10 +00:00
|
|
|
class LookingForGroup(_ParserScraper):
|
2016-05-16 21:16:29 +00:00
|
|
|
url = 'http://www.lfg.co/'
|
2013-02-12 16:55:13 +00:00
|
|
|
stripUrl = url + 'page/%s/'
|
2013-04-10 21:57:09 +00:00
|
|
|
firstStripUrl = stripUrl % '1'
|
2015-05-07 11:57:10 +00:00
|
|
|
css = True
|
|
|
|
imageSearch = '#comic img'
|
|
|
|
prevSearch = '#comic-left > a'
|
2016-04-12 21:11:39 +00:00
|
|
|
latestSearch = '#header-dropdown-comic-lfg > a:nth-of-type(2)'
|
2016-04-13 18:01:51 +00:00
|
|
|
starter = indirectStarter
|
2012-06-20 19:58:13 +00:00
|
|
|
help = 'Index format: nnn'
|