From 80863f5a9784c3fb610ac7e91651d9f9a71b8877 Mon Sep 17 00:00:00 2001
From: Kovid Goyal <kovid@kovidgoyal.net>
Date: Sun, 3 Apr 2022 08:31:50 +0530
Subject: [PATCH] The Diplomat by unkn0wn

---
 recipes/the_diplomat.recipe | 71 +++++++++++++++++++++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 recipes/the_diplomat.recipe

diff --git a/recipes/the_diplomat.recipe b/recipes/the_diplomat.recipe
new file mode 100644
index 0000000000..fde4a0f1a7
--- /dev/null
+++ b/recipes/the_diplomat.recipe
@@ -0,0 +1,71 @@
+#!/usr/bin/env  python
+
+from calibre.web.feeds.news import BasicNewsRecipe, classes
+
+
+class Diplomat(BasicNewsRecipe):
+    title = u'The Diplomat'
+    description = 'The Diplomat is the premier international current-affairs magazine for the Asia-Pacific region. READ THE DIPLOMAT, KNOW THE ASIA-PACIFIC'
+    language = 'en'
+    __author__ = 'unkn0wn'
+    oldest_article = 14  # days
+    max_articles_per_feed = 100
+    encoding = 'utf-8'
+    use_embedded_content = False
+    no_stylesheets = True
+    remove_attributes = ['style', 'height', 'width']
+    masthead_url = 'https://thediplomat.com/wp-content/themes/td_theme_v3/assets/logo/diplomat_logo_black.svg'
+    ignore_duplicate_articles = {'url'}
+    extra_css = '''
+        #td-meta { font-size: x-small }
+        #td-story-authors { font-size: small}
+        #td-story-image{font-size: small; font-style: italic;}
+        [id^="caption-attachment"] {font-size: small; font-style: italic;}
+    '''
+
+    def get_cover_url(self):
+        soup = self.index_to_soup('https://thediplomat.com')
+        tag = soup.find(attrs={'class': 'td-nav-mag'})
+        if tag:
+            url = tag.find('img')['src'].split("/")[-1]
+            self.cover_url = ('https://magazine.thediplomat.com/media/1080/' +
+                              url)
+            # /ads/magazine/cover/td-mag-s-1/issue_89_cover.jpg
+            # https://magazine.thediplomat.com/media/1080/issue_89_cover.jpg
+        return super().get_cover_url()
+
+    keep_only_tags = [
+        dict(name='header', attrs={'id': 'td-story-head'}),
+        classes('td-media--story td-prose td-story-magazine__link td-authors'),
+    ]
+
+    remove_tags = [
+        dict(name='aside', attrs={'id': 'td-box-newsletter'}),
+        classes('td-share td-ad td-ad-inline-txt'),
+    ]
+
+    feeds = [
+        ('Features', 'https://thediplomat.com/category/features/feed'),
+        ('Interviews', 'https://thediplomat.com/category/interviews/feed'),
+        ('Magazine Preview', 'https://thediplomat.com/category/magazine/feed'),
+        # REGIONS
+        ('Central Asia', 'https://thediplomat.com/regions/central-asia/feed'),
+        ('East Asia', 'https://thediplomat.com/regions/east-asia/feed'),
+        ('Oceania', 'https://thediplomat.com/regions/oceania-region/feed'),
+        ('South Asia', 'https://thediplomat.com/regions/south-asia/feed'),
+        ('South East Asia',
+         'https://thediplomat.com/regions/southeast-asia/feed'),
+        # TOPICS
+        ('Diplomacy', 'https://thediplomat.com/topics/diplomacy/feed'),
+        ('Economy', 'https://thediplomat.com/topics/economy/feed'),
+        ('Environment', 'https://thediplomat.com/topics/environment/feed'),
+        ('Opinion', 'https://thediplomat.com/topics/opinion/feed'),
+        ('Politics', 'https://thediplomat.com/topics/politics/feed'),
+        ('Security', 'https://thediplomat.com/topics/security/feed'),
+        ('Society', 'https://thediplomat.com/topics/society/feed'),
+    ]
+
+    def preprocess_html(self, soup):
+        for img in soup.findAll('img', attrs={'src': True}):
+            img['src'] = img['src'].replace("td-story-s-1", "td-story-s-2")
+        return soup