From 9aa202d7bce0c0311b43795e70080f31e5837018 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 22:43:08 +0200 Subject: [PATCH 1/6] fix encoding for trojmiasto.pl recipe --- recipes/trojmiasto_pl.recipe | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/recipes/trojmiasto_pl.recipe b/recipes/trojmiasto_pl.recipe index 2977211191..8959e0e0b1 100644 --- a/recipes/trojmiasto_pl.recipe +++ b/recipes/trojmiasto_pl.recipe @@ -8,7 +8,7 @@ class Trojmiasto(BasicNewsRecipe): description = u'Wiadomości, imprezy, wydarzenia, spektakle.Gdańsk, Gdynia, Sopot - NOCLEGI, Katalog firm, repertuar kin, wydarzenia, przewodnik, mapa, kwatery, hotele. Portal regionalny trojmiasto.pl' # noqa category = '' language = 'pl' - encoding = 'iso-8859-2' + encoding = 'utf-8' extra_css = 'ul {list-style: none; padding:0; margin:0;}' cover_url = 'http://www.trojmiasto.pl/_img/toplong2/logo_trojmiasto.gif' use_embedded_content = False From ec5fea9c5e02445340a2b01c050839c5372d1a1a Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 23:20:56 +0200 Subject: [PATCH 2/6] fix fetching articles --- recipes/tanuki.recipe | 13 +------------ 1 file changed, 1 insertion(+), 12 deletions(-) diff --git a/recipes/tanuki.recipe b/recipes/tanuki.recipe index fbcad64d20..cf85fbdd83 100644 --- a/recipes/tanuki.recipe +++ b/recipes/tanuki.recipe @@ -11,6 +11,7 @@ class tanuki(BasicNewsRecipe): language = 'pl' max_articles_per_feed = 100 encoding = 'utf-8' + autocleanup = True extra_css = 'ul {list-style: none; padding: 0; margin: 0;} .kadr{float: left;} .dwazdania {float: right;}' preprocess_regexps = [(re.compile(u'

', re.DOTALL), lambda match: ''), (re.compile( unicode(r'
Zobacz jak ocenili
'), re.DOTALL), lambda match: '')] @@ -42,15 +43,3 @@ class tanuki(BasicNewsRecipe): appendtag.insert(pos, pagetext) for r in appendtag.findAll(attrs={'class': 'nextarrow'}): r.extract() - - def preprocess_html(self, soup): - self.append_page(soup, soup.body) - for a in soup('a'): - if a.has_key('href') and 'http://' not in a['href'] and 'https://' not in a['href']: # noqa - if 'tanuki-anime' in soup.title.string.lower(): - a['href'] = 'http://anime.tanuki.pl' + a['href'] - elif 'tanuki-manga' in soup.title.string.lower(): - a['href'] = 'http://manga.tanuki.pl' + a['href'] - elif 'tanuki-czytelnia' in soup.title.string.lower(): - a['href'] = 'http://czytelnia.tanuki.pl' + a['href'] - return soup From 988a6ffeceee41df8cff0bb16f7b8db0918d301e Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 23:31:15 +0200 Subject: [PATCH 3/6] fix article parsing --- recipes/tablety_pl.recipe | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/recipes/tablety_pl.recipe b/recipes/tablety_pl.recipe index 36bf58a8ef..30b7475239 100644 --- a/recipes/tablety_pl.recipe +++ b/recipes/tablety_pl.recipe @@ -16,7 +16,7 @@ class Tablety_pl(BasicNewsRecipe): max_articles_per_feed = 100 preprocess_regexps = [(re.compile(u'

Przeczytaj także.*?

', re.DOTALL), lambda match: ''), (re.compile(u'

Przeczytaj koniecznie.*?

', re.DOTALL), lambda match: '')] - keep_only_tags = [dict(id='news_block')] + keep_only_tags = [dict(attrs={'class': ['featured-image', 'article-content clearfix']})] remove_tags = [dict(attrs={'class': ['comments_icon', 'wp-polls', 'entry-comments', 'wp-polls-loading', 'ts-fab-wrapper', 'entry-footer', 'social-custom']})] feeds = [(u'Najnowsze posty', u'http://www.tablety.pl/feed/')] From 244d428bbac9069f6ac7fba5bfa655522022e915 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 23:40:11 +0200 Subject: [PATCH 4/6] site revamped into tv station site --- recipes/icons/stopklatka.png | Bin 266 -> 0 bytes recipes/stopklatka.recipe | 71 ----------------------------------- 2 files changed, 71 deletions(-) delete mode 100644 recipes/icons/stopklatka.png delete mode 100644 recipes/stopklatka.recipe diff --git a/recipes/icons/stopklatka.png b/recipes/icons/stopklatka.png deleted file mode 100644 index 380737339ce3e775cd94640ef1d785b6eec79678..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 266 zcmV+l0rmcgP)U1~|G}mIvXK6vcK?xP{)gUl0CM3;#g|ACf1!0001$ zNklvbMSf_FC0viKUt*Yh}VCM8#8cQ@7(RNyG@iKJUYm_d3OpY03z@;T_-Zd QiU0rr07*qoM6N<$g0UHT$p8QV diff --git a/recipes/stopklatka.recipe b/recipes/stopklatka.recipe deleted file mode 100644 index 6507eacdde..0000000000 --- a/recipes/stopklatka.recipe +++ /dev/null @@ -1,71 +0,0 @@ -__license__ = 'GPL v3' -from calibre.web.feeds.news import BasicNewsRecipe -from calibre.ebooks.BeautifulSoup import Comment - - -class Stopklatka_pl(BasicNewsRecipe): - title = u'Stopklatka.pl' - __author__ = 'fenuks' - description = u'Stopklatka.pl to najdłużej działający polski portal filmowy. Baza filmów, seriali i aktorów, repertuar kin, program tv, wydarzenia ze świata filmu' # noqa - category = 'movies' - language = 'pl' - encoding = 'utf-8' - extra_css = 'img {display: block;} ul {list-style-type: none;}' - cover_url = 'http://static1.stopklatka.pl/images/20/19/11501.jpg' - use_embedded_content = False - oldest_article = 7 - BASEURL = 'http://stopklatka.pl' - max_articles_per_feed = 100 - no_stylesheets = True - remove_empty_feeds = True - remove_javascript = True - remove_attributes = ['style', 'font'] - ignore_duplicate_articles = {'title', 'url'} - - keep_only_tags = [dict( - attrs={'class': 'asset-full-content default-asset-publisher show-asset-title'})] - remove_tags = [dict(attrs={'class': ['metadata-entry metadata-tags', 'print-action', - 'asset-flag', 'asset-ratings', 'ad-nav']}), dict(id='contest')] - - feeds = [ - (u'Wiadomo\u015bci', u'http://stopklatka.pl/wiadomosci/-/asset_publisher/Hl7x4Ku4GpZj/rss?p_p_cacheability=cacheLevelPage'), - (u'Artyku\u0142y', u'http://stopklatka.pl/artykuly/-/asset_publisher/pKhn5s0IxqSc/rss?p_p_cacheability=cacheLevelPage'), - (u'Premiery i zapowiedzi', u'http://stopklatka.pl/premiery-i-zapowiedzi?p_p_id=eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL&p_p_lifecycle=2&p_p_state=normal&p_p_mode=view&p_p_cacheability=cacheLevelPage&p_p_col_id=column-1&p_p_col_pos=1&p_p_col_count=3&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13393201&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13760176&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=15238425&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13470227&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13913324&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=20234402&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13917041&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13905169&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=14253975&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=21586017&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13540662&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=12999052&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=45280408&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=14826890&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13459998&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13070805&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=20209965&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=21741457&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=35577381&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_assetEntryIds=13530138&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13392987%2Cmartwe-zlo&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13760162%2Cuklad-zamkniety&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F15238403%2Cwszyscy-w-naszej-rodzinie&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13470213%2Cdonoma&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13913310%2Ccristiada&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F20234381%2Craj-wiara&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13917027%2Cintruz&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13905155%2Cspring-breakers&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F14253957%2Ckrudowie&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F21586004%2Cswieta-czworca&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13540648%2Ckwartet&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F12999038%2Cimagine&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F45280404%2Cdom-na-kolkach&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F14826876%2Cg-i-joe-odwet&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13459984%2Cnieobliczalni&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13070591%2Csamotny-port-milosc&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F20209952%2Czanim-noc-nas-nie-rozdzieli&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F21741444%2Chemel&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35577377%2Czywie-bielarus-&_eventsearch_WAR_eventsearchportlet_INSTANCE_FLRWmpE7H8IL_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F13530124%2Cpanaceum'), # noqa - (u'Recenzje filmowe', u'http://stopklatka.pl/box-office/-/asset_publisher/3yxqotUEiqHJ/rss?p_p_cacheability=cacheLevelPage'), - (u'Recenzje', u'http://stopklatka.pl/recenzje/-/asset_publisher/5oZ3s2J3L0tG/rss?p_p_cacheability=cacheLevelPage'), - (u'Gwiazdy', u'http://stopklatka.pl/czerwony-dywan/-/asset_publisher/PqN7MDEGWGvh/rss?p_p_cacheability=cacheLevelPage'), - (u'Wywiady Stopklatki', u'http://stopklatka.pl/wywiady/-/asset_publisher/uVh3OrZCaLd7/rss?p_p_cacheability=cacheLevelPage'), - (u'Prosto z Hollywood', u'http://stopklatka.pl/wywiady-z-hollywood/-/asset_publisher/YsbU0JSoxb9G/rss?p_p_cacheability=cacheLevelPage'), - (u'Plotki', u'http://stopklatka.pl/czerwony-dywan/-/asset_publisher/XuF8EGAkVeTa/rss?p_p_cacheability=cacheLevelPage'), - (u'Box Office Polska', u'http://stopklatka.pl/box-office?p_p_id=eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ&p_p_lifecycle=2&p_p_state=normal&p_p_mode=view&p_p_cacheability=cacheLevelPage&p_p_col_id=column-1&p_p_col_pos=1&p_p_col_count=5&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_assetEntryIds=47982267&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_assetEntryIds=46685247&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_assetEntryIds=45280313&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47982263%2Cbox-office-weekendowy-polska-15-03-2013-17-03-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F46685243%2Cbox-office-weekendowy-polska-08-03-2013-10-03-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_Gqb98cI5dgSJ_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F45280309%2Cbox-office-weekendowy-polska-01-03-2013-03-03-2013'), # noqa - (u'Box Office USA', u'http://stopklatka.pl/box-office?p_p_id=eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd&p_p_lifecycle=2&p_p_state=normal&p_p_mode=view&p_p_cacheability=cacheLevelPage&p_p_col_id=column-1&p_p_col_pos=2&p_p_col_count=5&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_assetEntryIds=49047234&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_assetEntryIds=48879358&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_assetEntryIds=47605057&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_assetEntryIds=47809980&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_assetEntryIds=46505246&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F49047230%2Cbox-office-weekendowy-stany-zjednoczone-22-03-2013-24-03-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48879354%2Cbox-office-weekendowy-stany-zjednoczone-22-03-2013-24-03-2013-estymacja-&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47605053%2Cbox-office-weekendowy-stany-zjednoczone-15-03-2013-17-03-2013-estymacja-&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47809976%2Cbox-office-weekendowy-stany-zjednoczone-15-03-2013-17-03-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_24AXs0agMxJd_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F46505242%2Cbox-office-weekendowy-stany-zjednoczone-08-03-2013-10-03-2013'), # noqa - (u'Kalendarium imprez', - u'http://stopklatka.pl/kalendarium-imprez?p_p_id=eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh&p_p_lifecycle=2&p_p_state=normal&p_p_mode=view&p_p_cacheability=cacheLevelPage&p_p_col_id=column-1&p_p_col_pos=1&p_p_col_count=3&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47628974&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47627805&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=45317244&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=48884855&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47629292&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=48884742&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482058&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47627893&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482076&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47627838&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=48167620&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482067&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47811744&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482049&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47629615&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=45088670&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47628531&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481950&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481496&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482022&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=19323743&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47628034&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=47628064&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=45088819&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35482031&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481415&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481977&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=19323617&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481932&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_assetEntryIds=35481995&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47628970%2C4-festiwal-filmow-swiata-trzy-zywioly&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47627801%2Cwielka-podroz-krudow&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F45317240%2C6-przeglad-kina-rosyjskiego-nowe-kino-rosyjskie-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48884851%2C2-1-nowy-cykl-spotkan-literatury-z-filmem&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47629288%2C5-festiwal-polskich-filmow-krotkometrazowych-short-waves&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48884738%2Cmoico-enjoy-movies-przeglad-filmow-klasy-b-we-wroclawiu&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482054%2C1-ogolnopolski-festiwal-polskiej-animacji-o-pla-2013-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47627889%2Cviii-festiwal-filmow-afrykanskich-afrykamera-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482072%2C6-miedzynarodowy-festwial-kina-niezaleznego-off-plus-camera&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47627834%2C11-przeglad-filmow-studenckich-z-lodzkiej-filmowki-lodzia-po-wisle-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48167616%2Cweze-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482063%2Cxiv-festiwal-kina-amatorskiego-i-niezaleznego-kan&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47811740%2Cv-festiwal-muzyki-filmowej-krzysztofa-komedy&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482045%2Ckonkurs-scenariuszowy-script-pro-2013&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47629611%2C9-miedzynarodowy-festiwal-filmowy-%E2%80%9Ezydowskie-motywy%E2%80%9D&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F45088666%2C15-przeglad-filmowy-cieszyn-kino-na-granicy-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47628527%2Cdzien-filmowca-filmmaker-s-day&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481946%2C10-planete-doc-film-festival&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481492%2C66-miedzynarodowy-festiwal-filmowy-w-cannes&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482018%2C16-festiwal-filmow-kultowych&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F19323739%2C53-krakowski-festiwal-filmowy&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47628030%2C4-festiwal-filmow-mlodziezowych-18&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47628060%2Cvii-superorbitalny-festiwal-filmow-amatorskich-soffa&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F45088815%2Cxv-miedzynarodowy-festiwal-filmow-przyrodniczych-im-wlodzimierza-puchalskiego&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35482027%2C32-koszalinski-festiwal-debiutow-filmowych-mlodzi-i-film-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481411%2C6-miedzynarodowy-festiwal-filmow-animowanych-animator-&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481973%2C13-miedzynarodowy-festiwal-filmowy-sopot-film-festival&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F19323613%2C13-miedzynarodowy-festiwal-filmowy-t-mobile-nowe-horyzonty&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481928%2C7-festiwal-filmu-i-sztuki-dwa-brzegi-w-kazimierzu-dolnym-i-janowcu&_eventsearch_WAR_eventsearchportlet_INSTANCE_rHUXlm2Y2veh_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F35481991%2C3-miedzynarodowy-festiwal-filmu-i-muzyki-transatlantyk'), # noqa - (u'Konkursy', u'http://stopklatka.pl/konkursy?p_p_id=eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA&p_p_lifecycle=2&p_p_state=normal&p_p_mode=view&p_p_cacheability=cacheLevelPage&p_p_col_id=column-1&p_p_col_pos=3&p_p_col_count=5&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_assetEntryIds=47091950&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_assetEntryIds=48879762&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_assetEntryIds=48880109&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F47091941%2Cksiazki-dwie-kobiety-&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48879753%2Cdvd-rozmowy-noca-&_eventsearch_WAR_eventsearchportlet_INSTANCE_Yks1FKgVbrOA_urls=http%3A%2F%2Fstopklatka.pl%2F-%2F48880000%2Cdvd-milosc-'), # noqa - (u'Komiks Stopklatki', u'http://stopklatka.pl/komiks/-/asset_publisher/pKhn5s0IxqSc/rss?p_p_cacheability=cacheLevelPage')] - - def append_page(self, soup, appendtag): - tag = soup.find('a', attrs={'class': 'next'}) - if tag: - while tag: - url = tag['href'] - soup2 = self.index_to_soup(self.BASEURL + url) - tag = soup2.find('a', attrs={'class': 'next'}) - pagetext = soup2.find( - attrs={'class': 'journal-content-article'}) - comments = pagetext.findAll( - text=lambda text: isinstance(text, Comment)) - for comment in comments: - comment.extract() - pos = len(appendtag.contents) - appendtag.insert(pos, pagetext) - appendtag.find('a', attrs={'class': 'next'}).extract() - - def preprocess_html(self, soup): - self.append_page(soup, soup.body) - return soup - - def get_browser(self): - br = BasicNewsRecipe.get_browser(self) - br.open(self.BASEURL) - return br From 24df9008df028d49cb8f5db6bec5a443e12d73f2 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 23:42:51 +0200 Subject: [PATCH 5/6] remove the broken recipe - broken rss --- recipes/icons/telepolis_pl.png | Bin 171 -> 0 bytes recipes/telepolis_pl.recipe | 46 --------------------------------- 2 files changed, 46 deletions(-) delete mode 100644 recipes/icons/telepolis_pl.png delete mode 100644 recipes/telepolis_pl.recipe diff --git a/recipes/icons/telepolis_pl.png b/recipes/icons/telepolis_pl.png deleted file mode 100644 index 234df7ecd06739caaa48df02a2397d7ba7d79fd6..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 171 zcmeAS@N?(olHy`uVBq!ia0vp^0wB!61|;P_|4#%`MV>B>Ar-fh7iiB|y6_U;k*^9o zsgrXK)SLKBO3F#PF}Kaz&?+uC=uF3PGau*g@)lp}{~tVjVn(A1v%{Lp-MnRzA`?r11t-jFaS?m!3YBwhtEej5gczH@Qs UoK^d`0c~gSboFyt=akR{06MQdasU7T diff --git a/recipes/telepolis_pl.recipe b/recipes/telepolis_pl.recipe deleted file mode 100644 index e1c325091f..0000000000 --- a/recipes/telepolis_pl.recipe +++ /dev/null @@ -1,46 +0,0 @@ -#!/usr/bin/env python2 - -__license__ = 'GPL v3' - -from calibre.web.feeds.news import BasicNewsRecipe - - -class telepolis(BasicNewsRecipe): - title = u'Telepolis.pl' - __author__ = 'Artur Stachecki , Tomasz Długosz ' - - language = 'pl' - description = u'Twój telekomunikacyjny serwis informacyjny.' - masthead_url = 'http://telepolis.pl/i/telepolis-logo2.gif' - no_stylesheets = True - use_embedded_content = False - - feeds = [ - (u'Wiadomości', u'http://www.telepolis.pl/rss,2,5,0.html') - ] - - keep_only_tags = [ - dict(name='div', attrs={'class': 'flol w510'}), - dict(name='div', attrs={'class': 'main_tresc'}), - dict(name='div', attrs={'class': 'main_tresc_news'}) - ] - - def append_page(self, soup, appendtag): - chpage = appendtag.find(attrs={'class': 'str'}) - if chpage: - for page in chpage.findAll('a'): - if page.renderContents() == 'Następna ›': - break - soup2 = self.index_to_soup(page['href']) - pagetext = soup2.find(attrs={'class': 'main_tresc'}) - pos = len(appendtag.contents) - appendtag.insert(pos, pagetext) - for r in appendtag.findAll(attrs={'class': 'str'}): - r.extract() - - def preprocess_html(self, soup): - self.append_page(soup, soup.body) - for image in soup.findAll('img'): - if 'm.jpg' in image['src']: - image['src'] = image['src'].replace('m.jpg', '.jpg') - return soup From 3c1e81f868d265ae7a121e93295c0f1445fc336b Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20D=C5=82ugosz?= Date: Sat, 6 Oct 2018 23:43:37 +0200 Subject: [PATCH 6/6] apparently no recent content --- recipes/icons/tawernarpg_pl.png | Bin 770 -> 0 bytes recipes/tawernarpg_pl.recipe | 36 -------------------------------- 2 files changed, 36 deletions(-) delete mode 100644 recipes/icons/tawernarpg_pl.png delete mode 100644 recipes/tawernarpg_pl.recipe diff --git a/recipes/icons/tawernarpg_pl.png b/recipes/icons/tawernarpg_pl.png deleted file mode 100644 index dac6d4bd2dfe8ba943368f523b4272099daca00a..0000000000000000000000000000000000000000 GIT binary patch literal 0 HcmV?d00001 literal 770 zcmV+d1O5DoP)ai*x)ZHtVk#fEKHP{ zp{;<8A&u0kq5y6SO%&S35siuM*gl_s+z;FLSb}6_fREsR)E|H6)Z6uyLe>S?;26a* z&C2&HahzODpO|yH-P*?9?$(N1D3&t084{Tk3ldW&xQ%ZfUtTcvU zkt7LzyGkx!z)%^aq0l-($h3!!E&=`yIMUhi!=2hPP!Ry<&*U#p=hR2GZ9Cm? z%-K^5xU&WFv&;PQ>jS29CMPnAcCAUXUZof53a|`dq^7L3=NGIib2HiS_E!5tp%ceW zdZkB?s%XPyVevGzMw?zJ={Nr(=M?Dm!?FkSCasU7T07*qoM6N<$f^}Pb AbpQYW diff --git a/recipes/tawernarpg_pl.recipe b/recipes/tawernarpg_pl.recipe deleted file mode 100644 index 0b0b3340ce..0000000000 --- a/recipes/tawernarpg_pl.recipe +++ /dev/null @@ -1,36 +0,0 @@ -__license__ = 'GPL v3' -import re -from calibre.web.feeds.news import BasicNewsRecipe - - -class TawernaRPG(BasicNewsRecipe): - title = u'Tawerna RPG' - __author__ = 'fenuks' - description = u'Tawerna RPG to ogólnopolski serwis zajmujący się fantastyką i grami fantastycznymi. Znajdziesz u nas zarówno gry fabularne, karciane, planszowe i komputerowe, a także recenzje, opowiadania i sporą dawkę humoru.' # noqa - category = 'fantasy, rpg, board games' - language = 'pl' - extra_css = '.slajd {list-style-type: none; padding-left: 0px; margin-left: 0px;} .lewanc {float: left; margin-right: 5px;} .srodek {display: block; margin-left: auto; margin-right: auto;}' # noqa - cover_url = 'http://www.tawerna.rpg.pl/img/logo.png' - preprocess_regexps = [(re.compile(u'

Dodaj komentarz

.*', - re.DOTALL | re.IGNORECASE), lambda match: '')] - use_embedded_content = False - oldest_article = 7 - max_articles_per_feed = 100 - no_stylesheets = True - remove_empty_feeds = True - remove_javascript = True - remove_attributes = ['style', 'font'] - ignore_duplicate_articles = {'title', 'url'} - - keep_only_tags = [dict(id='site')] - remove_tags = [dict(id=['player', 'komentarz'])] - remove_tags_after = dict(id='komentarz') - - feeds = [(u'Artykuły', 'http://www.tawerna.rpg.pl/css/rss.rss')] - - def preprocess_html(self, soup): - for r in soup.findAll(attrs={'class': 'powi'}): - r.parent.extract() - for c in soup.findAll(name=['li', 'ol', 'ul']): - c.name = 'div' - return soup