From 4cd864cbc6efc3318feaac91aff457bf402f65c2 Mon Sep 17 00:00:00 2001
From: unkn0w7n <51942695+unkn0w7n@users.noreply.github.com>
Date: Tue, 22 Aug 2023 11:24:33 +0530
Subject: [PATCH] firstpost.recipe

---
 recipes/firstpost.recipe    |  82 ++++++++++++++++++++++++++++++++++++
 recipes/icons/firstpost.png | Bin 0 -> 1051 bytes
 2 files changed, 82 insertions(+)
 create mode 100644 recipes/firstpost.recipe
 create mode 100644 recipes/icons/firstpost.png

diff --git a/recipes/firstpost.recipe b/recipes/firstpost.recipe
new file mode 100644
index 0000000000..ef42c90e89
--- /dev/null
+++ b/recipes/firstpost.recipe
@@ -0,0 +1,82 @@
+from calibre.web.feeds.news import BasicNewsRecipe, classes
+from calibre.ptempfile import PersistentTemporaryFile
+
+# Firstpost feeds mix sections into other feeds, like explainers end up in opinion feed and opinions end up in India feed.
+# change google_feeds to True to fetch right sections.
+google_feeds = False
+
+class firstpost(BasicNewsRecipe):
+    title = 'Firstpost'
+    __author__ = 'unkn0wn'
+    description = (
+        'Firstpost.com will serve as a trusted guide to the crush of news and ideas around you.'
+        ' With thoughtful analysis and fearless views our team of editors and writers will track'
+        ' news in India and the world and provide a perspective that is reflective of a changing dynamic.'
+    )
+    no_stylesheets = True
+    use_embedded_content = False
+    encoding = 'utf-8'
+    language = 'en_IN'
+    remove_attributes = ['height', 'width', 'style']
+    masthead_url = 'https://images.firstpost.com/wp-content/uploads/2016/03/FP-Logo.png?impolicy=website&width=600&height=60'
+    max_articles_per_feed = 25
+    remove_empty_feeds = True
+    ignore_duplicate_articles = {'title', 'url'}
+    extra_css = '''
+        .category-name, .author-info { font-size:small; color:#202020; }
+        .wp-caption-text { font-size:small; text-align:center; }
+    '''
+
+    keep_only_tags = [
+        classes('article-sect')
+    ]
+
+    remove_tags = [
+        classes('art-rel-articles tags-wrap'),
+        dict(name='svg'),
+    ]
+
+    feeds = []
+
+    sections = [
+        'india', 'politics', 'opinion', 'explainers', 'business',
+        'world', 'web-stories', 'tech', 'artandculture', 'health', 'health-supplement',
+        # 'photos', 'entertainment', 'living', 'education', 'sports', 'firstcricket',
+    ]
+    if not google_feeds:
+        for sec in sections:
+            a = 'https://www.firstpost.com/rss/{}.xml'
+            feeds.append((sec.capitalize(), a.format(sec)))
+    else:
+        articles_are_obfuscated = True
+
+        def get_obfuscated_article(self, url):
+            br = self.get_browser()
+            soup = self.index_to_soup(url)
+            link = soup.find('a', href=True)
+            skip_sections =[ # add sections you want to skip
+                '/video/', '/videos/', '/media/', '/vantage/'
+            ]
+            if any(x in link['href'] for x in skip_sections):
+                self.log('Aborting Article ', link['href'])
+                self.abort_article('skipping video links')
+            self.log('Downloading ', link['href'])
+            html = br.open(link['href']).read()
+            pt = PersistentTemporaryFile('.html')
+            pt.write(html)
+            pt.close()
+            return pt.name
+
+        for sec in sections:
+            a = 'https://news.google.com/rss/search?q=when:27h+allinurl:firstpost.com{}&hl=en-IN&gl=IN&ceid=IN:en'
+            feeds.append((sec.capitalize(), a.format('%2F' + sec + '%2F')))
+        # feeds.append(('Others', a.format('')))
+
+    def preprocess_html(self, soup):
+        if h2 := soup.find('h2', attrs={'class':'category-name'}):
+            h2.name = 'p'
+        if h := soup.find('h2', attrs={'class':'inner-copy'}):
+            h.name = 'p'
+        for img in soup.findAll('img', attrs={'data-src':True}):
+            img['src'] = img['data-src']
+        return soup
diff --git a/recipes/icons/firstpost.png b/recipes/icons/firstpost.png
new file mode 100644
index 0000000000000000000000000000000000000000..693d6365d15604bfed512e088dbc6ac72ccc62df
GIT binary patch
literal 1051
zcmeAS@N?(olHy`uVBq!ia0vp^0wB!73?$#)eFPFvNcITwWnidMV_;}#VPN<L6l!?E
zz))(y!0;-8fx&791A}<MoLzx7KndvppAgso|NjG-5b)m59EgByAPE%jY|h>V<gk?l
z`2_>DNrxVIzuCVXD8iWJ?e4<R!7A$k<gk}``ns||=MZBxw8&!Kf1iPY$*U?Pq9iD>
zT%n*SKP@vSRiUJ^AXT9vw}64cqIYU&;3g$Qjz7;us#|yjJo=1hoxC5m_k8)ET#2tQ
zSDcSpJIk=;!UdCw`~KHmod2W3c+zH{{J0#$jgv#e^^S#|U3*)txpMaM%b7-p@A)1M
zie1I0X|-p?f$fp%x%;g6!ajy>U)5&L)b&7Mv(L>PCvMJO({R*LnlX)gdc^IUX47vh
zc5pDuGX6B7^!|>W>1y_g3D!qB_s)F~G`laVLwV=R%t@zaOYlG0+~E}wW0li=fW`ax
zD;{m{#<d@hSkLR`_~@}+G>_F^ET!Pbr=>P-VjtG5SyiUdt7*P?&$BN5riF7}{(N}$
z;nC={H=TEkSD$G4$a7;Ovx2g?dwr8&z6LkP#{V-|gB(_0;Mg|Lr__L@|MfmymP5;$
zZ=A?d|F%!uaRc9$Uz*QZ?$q6KiD<}?)>y8%T!^(HZjWF=%<eMrn3(JrZyY%GcKx%f
z(6|^VQ<K(Ge6s3Bi^9<!`8|8I-V_H~>+C(Ww{6b{;Shg;Sm$LI#oo@E^UjWg<<rHw
z73|d)qNj4*l$6qmXy)2deQ>k4v#X{@Yrs{LFhgJ7YDEEt%gjY59Gp1BUS3mwGB3HB
zC6CePe|Jq{-qi~stg2E)YrR&~-TIj+z`d?*rr)IfUT2n^$~|Z0#(sF#bpKN`%fjE8
z>)KC>2|mr1qw~SqCGg#rEw@_vew}GQA{$`(<Hf1Koo+?!RxiTs-c7XkP+0a~bkiQ^
zI(8qK>yDcafBqh6U-kC;rK?`&w}0IC^^rcK>f7o2KiAE^Y+tiE-#gIfWdJat`FOfG
zhHzX@_F!gaWn<gCMMYIrz*$$<Ga!KXlt4mKgOP*+4;ve+Mzf|T!<m4fW`hO>Ulq>)
zj$;BLEd~-2vCa#6J_M<-o#1V9&}7h@C0Y2wvd152u4;*EL`h0wNvc(HQ7VvPFfuSQ
z)-|xuH8Kh@FtswVv@$i*HZZU<Fc9V1JPk!dZhlH;S|x4`DdF5&Kn)sj8%i>BQ;SOy
zfjI%ECB{x#Ux9igL3)Do(@M${i&7aJQ}UBi6+Ckj(^G>|6H_V+Po{!$FoUP7pUXO@
GgeCwE{Fh_^

literal 0
HcmV?d00001