Python getTagPattern Examples

Example #1

File: NavigationResource.py Project: stuartjbrown/plugin.video.zdf_de_2016

from de.generia.kodi.plugin.backend.web.HtmlResource import HtmlResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile

leftNavPattern = getTagPattern('ul', 'left-nav')
dropdownLinksPattern = compile(
    '<a\s*class="[^"]*dropdown-link[^"]*"[^>]*href="([^"]*)"[^>]*data-title="([^"]*)"'
)


class Rubric(object):
    def __init__(self, title, url):
        self.title = title
        self.url = url

    def __str__(self):
        return "<Rubric '%s' url='%s'>" % (self.title, self.url)


class NavigationResource(HtmlResource):
    def __init__(self, url):
        super(NavigationResource, self).__init__(url)

    def parse(self):
        super(NavigationResource, self).parse()
        leftNavMatch = leftNavPattern.search(self.content)
        if leftNavMatch is None:
            self.warn(

Example #2

File: NavigationResource.py Project: gedisony/repo-plugins

from sets import Set
from de.generia.kodi.plugin.backend.web.HtmlResource import HtmlResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile

leftNavPattern = getTagPattern('ul', 'left-nav')
dropdownLinksPattern = compile('<a\s*class="[^"]*dropdown-link[^"]*"[^>]*href="([^"]*)"[^>]*data-title="([^"]*)"')

class Rubric(object):

    def __init__(self, title, url):
        self.title = title
        self.url = url
                        
    def __str__(self):
        return "<Rubric '%s' url='%s'>" % (self.title, self.url)
    
    
class NavigationResource(HtmlResource):

    def __init__(self, url):
        super(NavigationResource, self).__init__(url)

    def parse(self):
        super(NavigationResource, self).parse()
        leftNavMatch = leftNavPattern.search(self.content)
        if leftNavMatch is None:
            self.warn("can't find navigation in page '{}', no rubrics will be available ...", self.url)

Example #3

File: RubricResource.py Project: generia/plugin.video.zdf_de_2016

from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile
from de.generia.kodi.plugin.backend.zdf.Regex import stripTag

from de.generia.kodi.plugin.backend.zdf.Teaser import Teaser
from de.generia.kodi.plugin.backend.zdf.TeaserLazyload import TeaserLazyload

from urlparse import urlparse

fallbackTitlePattern = compile(
    '<li\s*class="item current"[^>]*>[^<]*<a[^>]*>([^<]*)</a>')
fallbackTitlePattern2 = compile(
    '<h\d\s*class="[^"]*stage-title[^"]*"[^>]*>([^<]*)</h\d>')

moduleItemPattern = getTagPattern('div', 'item-caption')
moduleItemTextPattern = compile(
    'class="item-description"[^>]*>([^<]*)</?[^>]*>')
moduleItemDatePattern = compile('<time[^>]*>([^<]*)</time>')
moduleItemImagePattern = compile(
    'class="item-img[^"]*"[^>]*data-srcset="([^"]*)"')
moduleItemVideoPattern = compile('&quot;1280x720&quot;:&quot;([^\?]*)\?cb')

postContentPattern = getTagPattern('div', 'details')

stageTeaserPattern = getTagPattern('div', 'title-table')
stageTeaserTextPattern = compile('class="teaser-text"[^>]*>([^<]*)</?[^>]*>')

listPattern = compile(
    'class="([^"]*b-cluster|[^"]*b-cluster\s[^"]*|[^"]*b-content-teaser-list[^"]*|[^"]*b-group-persons[^"]*|[^"]*b-post-content[^"]*|[^"]*b-(content|video)-module[^"]*|[^"]*stage-content[^"]|[^"]*(b-topics-module|b-newsstream)[^"]*)"[^>]*>'
)

Example #4

File: Teaser.py Project: gedisony/repo-plugins

from datetime import datetime

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import stripTag
from de.generia.kodi.plugin.backend.zdf.Regex import cleanTags
from de.generia.kodi.plugin.backend.zdf.Regex import compile

teaserPattern = getTagPattern('article', 'b-content-teaser-item')
sourcePattern = compile('class="m-16-9"[^>]*data-srcset="([^"]*)"')
labelPattern = getTagPattern('div', 'teaser-label')
iconPattern = compile('class="icon-[0-9]*_([^ ]*) icon">')
catPattern = compile('class="teaser-cat"[^>]*>([^<]*)</[^>]*>')
aPattern = compile('href="([^"]*)"[^>]*>')
titleIconPattern = compile('class="title-icon icon-[0-9]*_([^"]*)">')
textPattern = compile('class="teaser-text"[^>]*>([^<]*)</[^>]*>')
datePattern = compile('class="video-airing"[^>]*>([^<]*)</[^>]*>')

    
def compareTeasers(t1, t2):
    if t1 is None and t2 is None:
        return 0
    if t1 is None and t2 is not None:
        return 1
    if t1 is not None and t2 is None:
        return -1
    
    '''
    if t1.date is not None and t2.date is not None:
        d1 = datetime.strptime(t1.date, "%d.%m.%Y")

Example #5

File: LiveTvResource.py Project: tischmi/plugin.video.zdf_de_2016

from de.generia.kodi.plugin.backend.zdf.AbstractPageResource import AbstractPageResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile
from de.generia.kodi.plugin.backend.zdf.Teaser import Teaser

livetvCellPattern = getTagPattern('div', 'js-livetv-scroller-cell')
titlePattern = compile('<h2[^>]*>([^<]*)</h2>')
contentNamePattern = compile('"embed_content"\s*:\s*"([^"]*)"')
imagePattern = compile('data-src="([^"]*)"')


class LiveTvResource(AbstractPageResource):
    def __init__(self, url):
        super(LiveTvResource, self).__init__(url)

    def parse(self):
        super(LiveTvResource, self).parse()
        livetvCellMatch = livetvCellPattern.search(self.content)
        if livetvCellMatch is None:
            #self.warn("can't find live-tv cells in page '{}', no channels will be available ...", self.url)
            return

        self.teasers = []
        while livetvCellMatch is not None:
            pos = livetvCellMatch.end(0)
            teaser = Teaser()
            pos = self._parseTitle(pos, teaser)
            pos = self._parseContentName(pos, teaser)

Example #6

from de.generia.kodi.plugin.backend.web.HtmlResource import HtmlResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile

from de.generia.kodi.plugin.backend.zdf.Teaser import Teaser

fallbackTitlePattern = compile(
    '<li\s*class="item current"[^>]*>[^<]*<a[^>]*>([^<]*)</a>')

moduleItemPattern = getTagPattern('div', 'item-caption')
moduleItemTextPattern = compile('class="item-description"[^>]*>([^<]*)<span')
moduleItemDatePattern = compile('<time[^>]*>([^<]*)</time>')

listPattern = compile(
    'class="([^"]*b-cluster|[^"]*b-cluster [^"]*|[^"]*b-content-teaser-list[^"]*|[^"]*b-content-module[^"]*)"[^>]*>'
)

sectionTitlePattern = compile(
    '<h2\s*class="[^"]*title[^"]*"[^>]*>([^<]*)</h2>')
sectionItemPattern = getTagPattern('article', 'b-content-teaser-item')

clusterTitlePattern = compile(
    '<h2\s*class="[^"]*cluster-title[^"]*"[^>]*>([^<]*)</h2>')
clusterItemPattern = getTagPattern('article', 'b-cluster-teaser')


class Cluster(object):
    def __init__(self, title, listType, listStart, listEnd=-1):

Example #7

from datetime import datetime

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import stripTag
from de.generia.kodi.plugin.backend.zdf.Regex import cleanTags
from de.generia.kodi.plugin.backend.zdf.Regex import compile

teaserPattern = getTagPattern('article', 'b-content-teaser-item')
sourcePattern = compile('class="m-16-9"[^>]*data-srcset="([^"]*)"')
labelPattern = getTagPattern('div', 'teaser-label')
iconPattern = compile('class="icon-[0-9]*_([^ ]*) icon">')
catPattern = compile('class="teaser-cat\s*[^"]*"[^>]*>')
catCategoryPattern = compile(
    'class="teaser-cat-category\s*[^"]*"[^>]*>([^<]*)</[^>]*>')
catBrandPattern = compile(
    'class="teaser-cat-brand\s*[^"]*"[^>]*>([^<]*)</[^>]*>')
aPattern = compile('href="([^"]*)"[^>]*>')
textPattern = compile('class="teaser-text"[^>]*>([^<]*)</[^>]*>')
footPattern = compile('class="teaser-foot\s*[^"]*"[^>]*>')
footIconPattern = compile('class="[^"]*icon-[0-9]*_(play)[^"]*">')
teaserInfoPattern = getTagPattern('dd', 'teaser-info')
teaserInfoIsTiviPattern = getTagPattern('span', 'is-tivi')
teaserInfoDurationPattern = compile('(.*) min')
teaserInfoEpisodePattern = compile('S(..) F(..)')
apiTokenPattern = compile('"apiToken"\s*:\s*"([^"]*)"')


def compareTeasers(t1, t2):
    if t1 is None and t2 is None:

Example #8

File: Teaser.py Project: tischmi/plugin.video.zdf_de_2016

from datetime import datetime

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import stripTag
from de.generia.kodi.plugin.backend.zdf.Regex import cleanTags
from de.generia.kodi.plugin.backend.zdf.Regex import compile

teaserPattern = getTagPattern('article', 'b-content-teaser-item')
sourcePattern = compile('class="m-16-9"[^>]*data-srcset="([^"]*)"')
labelPattern = getTagPattern('div', 'teaser-label')
iconPattern = compile('class="icon-[0-9]*_([^ ]*) icon">')
catPattern = compile('class="teaser-cat\s*[^"]*"[^>]*>')
catCategoryPattern = compile(
    'class="teaser-cat-category\s*[^"]*"[^>]*>([^<]*)</[^>]*>')
catBrandPattern = compile(
    'class="teaser-cat-brand\s*[^"]*"[^>]*>([^<]*)</[^>]*>')
aPattern = compile('href="([^"]*)"[^>]*>')
textPattern = compile('class="teaser-text"[^>]*>([^<]*)</[^>]*>')
footPattern = compile('class="teaser-foot"[^>]*>')
footIconPattern = compile('class="[^"]*icon-[0-9]*_(play)[^"]*">')
datePattern = compile('class="teaser-info"[^>]*>([^<]*)</[^>]*>')
apiTokenPattern = compile('"apiToken"\s*:\s*"([^"]*)"')


def compareTeasers(t1, t2):
    if t1 is None and t2 is None:
        return 0
    if t1 is None and t2 is not None:
        return 1

Example #9

File: LiveTvResource.py Project: gedisony/repo-plugins

from de.generia.kodi.plugin.backend.web.HtmlResource import HtmlResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile
from de.generia.kodi.plugin.backend.zdf.Teaser import Teaser

livetvCellPattern = getTagPattern('div', 'js-livetv-scroller-cell')
titlePattern = compile('<h2[^>]*>([^<]*)</h2>')
contentNamePattern = compile('data-zdfplayer-id="([^"]*)"')
imagePattern = compile('data-src="([^"]*)"')


class LiveTvResource(HtmlResource):

    def __init__(self, url):
        super(LiveTvResource, self).__init__(url)

    def parse(self):
        super(LiveTvResource, self).parse()
        livetvCellMatch = livetvCellPattern.search(self.content)
        if livetvCellMatch is None:
            #self.warn("can't find live-tv cells in page '{}', no channels will be available ...", self.url)
            return
        
        self.teasers = []
        while livetvCellMatch is not None:
            pos = livetvCellMatch.end(0)
            teaser = Teaser()
            pos = self._parseTitle(pos, teaser)

Example #10

File: RubricResource.py Project: gedisony/repo-plugins

from de.generia.kodi.plugin.backend.web.HtmlResource import HtmlResource

from de.generia.kodi.plugin.backend.zdf import stripHtml
from de.generia.kodi.plugin.backend.zdf.Regex import getTagPattern
from de.generia.kodi.plugin.backend.zdf.Regex import getTag
from de.generia.kodi.plugin.backend.zdf.Regex import compile

from de.generia.kodi.plugin.backend.zdf.Teaser import Teaser

fallbackTitlePattern = compile('<li\s*class="item current"[^>]*>[^<]*<a[^>]*>([^<]*)</a>')
fallbackTitlePattern2 = compile('<h\d\s*class="[^"]*stage-title[^"]*"[^>]*>([^<]*)</h\d>')

moduleItemPattern = getTagPattern('div', 'item-caption')
moduleItemTextPattern = compile('class="item-description"[^>]*>([^<]*)<span')
moduleItemDatePattern = compile('<time[^>]*>([^<]*)</time>')

listPattern = compile('class="([^"]*b-cluster|[^"]*b-cluster [^"]*|[^"]*b-content-teaser-list[^"]*|[^"]*b-content-module[^"]*)"[^>]*>')

sectionTitlePattern = compile('<h2\s*class="[^"]*title[^"]*"[^>]*>([^<]*)</h2>')
sectionItemPattern = getTagPattern('article', 'b-content-teaser-item')

clusterTitlePattern = compile('<h2\s*class="[^"]*cluster-title[^"]*"[^>]*>([^<]*)</h2>')
clusterItemPattern = getTagPattern('article', 'b-cluster-teaser')

class Cluster(object):

    def __init__(self, title, listType, listStart, listEnd=-1):
        self.title = title
        self.listType = listType
        self.listStart = listStart
        self.listEnd = listEnd