calibre/recipes/economist_search.recipe
2025-09-13 14:31:07 +05:30

323 lines
16 KiB
Python

#!/usr/bin/env python
import json
import time
from datetime import datetime, timedelta
from urllib.parse import quote, urlencode
from uuid import uuid4
from html5_parser import parse
from lxml import etree
from mechanize import Request
from calibre import browser
from calibre.ptempfile import PersistentTemporaryFile
from calibre.web.feeds.news import BasicNewsRecipe
def safe_dict(data, *names):
ans = data
for x in names:
ans = ans.get(x) or {}
return ans
def process_web_list(li_node):
li_html = ''
for li in li_node['items']:
if li.get('textHtml'):
li_html += f'<li>{li["textHtml"]}</li>'
elif li.get('textJson'):
li_html += f'<li>{parse_textjson(li["textJson"])}</li>'
else:
li_html += f'<li>{li.get("text", "")}</li>'
return '<ul>' + li_html + '</ul>'
def process_info_box(bx):
info = ''
for x in safe_dict(bx, 'components'):
info += f'<blockquote>{process_web_node(x)}</blockquote>'
return info
def parse_txt(ty):
typ = ty.get('type', '')
children = ty.get('children', [])
href = '#'
attributes = ty.get('attributes') or ()
for a in attributes:
if a.get('name') == 'href':
href = a.get('value', href)
break
tag_map = {
'text': lambda: [ty.get('value', '')],
'scaps': lambda: [
f'<span style="text-transform: uppercase; font-size: 0.85em; letter-spacing: 0.05em;">{"".join(parse_txt(c))}</span>'
for c in children
],
'bold': lambda: [f'<b>{"".join(parse_txt(c))}</b>' for c in children],
'drop_caps': lambda: [f'<b>{"".join(parse_txt(c))}</b>' for c in children],
'italic': lambda: [f'<i>{"".join(parse_txt(c))}</i>' for c in children],
'linebreak': lambda: ['<br>'],
'external_link': lambda: [
f'<a href="{href}">{"".join(parse_txt(c))}</a>' for c in children
],
'internal_link': lambda: [
f'<a href="{href}">{"".join(parse_txt(c))}</a>' for c in children
],
'ufinish': lambda: [text for c in children for text in parse_txt(c)],
'subscript': lambda: [f'<sub>{"".join(parse_txt(c))}</sub>' for c in children],
'superscript': lambda: [f'<sup>{"".join(parse_txt(c))}</sup>' for c in children],
}
if typ in tag_map:
yield from tag_map[typ]()
else:
print('** ', typ)
def parse_textjson(nt):
return ''.join(''.join(parse_txt(n)) for n in nt)
def process_web_node(node):
ntype = node.get('type', '')
if ntype == 'CROSSHEAD':
if node.get('textHtml'):
return f'<h4>{node.get("textHtml")}</h4>'
return f'<h4>{node.get("text", "")}</h4>'
elif ntype in ['PARAGRAPH', 'BOOK_INFO']:
if node.get('textHtml'):
return f'\n<p>{node.get("textHtml")}</p>'
if node.get('textJson'):
return f'\n<p>{parse_textjson(node["textJson"])}</p>'
return f'\n<p>{node.get("text", "")}</p>'
elif (ntype == 'IMAGE') or (node.get('__typename', '') == 'ImageComponent'):
alt = '' if node.get('altText') is None else node.get('altText')
cap = ''
if node.get('caption'):
if node['caption'].get('textHtml') is not None:
cap = node['caption']['textHtml']
elif node['caption'].get('textJson') is not None:
cap = parse_textjson(node['caption']['textJson'])
elif node['caption'].get('text') is not None:
cap = node['caption']['text']
return f'<div><img src="{node["url"]}" title="{alt}"></div><div style="text-align:center; font-size:small;">{cap}</div>'
elif ntype == 'PULL_QUOTE':
if node.get('textHtml'):
return f'<blockquote>{node.get("textHtml")}</blockquote>'
if node.get('textJson'):
return f'<blockquote>{parse_textjson(node["textJson"])}</blockquote>'
return f'<blockquote>{node.get("text", "")}</blockquote>'
elif ntype == 'BLOCK_QUOTE':
if node.get('textHtml'):
return f'<blockquote><i>{node.get("textHtml")}</i></blockquote>'
if node.get('textJson'):
return f'<blockquote><i>{parse_textjson(node["textJson"])}</i></blockquote>'
return f'<blockquote><i>{node.get("text", "")}</i></blockquote>'
elif ntype == 'DIVIDER':
return '<hr>'
elif ntype == 'INFOGRAPHIC':
if node.get('fallback'):
return process_web_node(node['fallback'])
elif ntype == 'INFOBOX':
return process_info_box(node)
elif ntype == 'UNORDERED_LIST':
if node.get('items'):
return process_web_list(node)
elif ntype:
print('** ', ntype)
return ''
def load_article_from_web_json(raw):
# open('/t/raw.json', 'w').write(raw)
body = ''
data = json.loads(raw)['data']['findArticleByUrl']
body += f'<div style="color: red; font-size:small; font-weight:bold;">{data.get("flyTitle", "")}</div>'
body += f'<h1>{data["headline"]}</h1>'
if data.get('rubric') and data.get('rubric') is not None:
body += f'<div style="font-style: italic; color:#202020;">{data.get("rubric", "")}</div>'
try:
date = data['dateModified']
except Exception:
date = data['datePublished']
dt = datetime.fromisoformat(date[:-1]) + timedelta(seconds=time.timezone)
dt = dt.strftime('%b %d, %Y %I:%M %p')
if data.get('dateline') is None:
body += f'<p style="color: gray; font-size: small;">{dt}</p>'
else:
body += f'<p style="color: gray; font-size: small;">{dt + " | " + (data["dateline"])}</p>'
main_image_url = safe_dict(data, 'leadComponent') or ''
if main_image_url:
body += process_web_node(data['leadComponent'])
if data.get('byline'):
if data['byline'] is not None:
body += f'<p style="color: gray; font-size: small;"><i>{"By " + data["byline"]}</i></p>'
for node in data.get('body'):
body += process_web_node(node)
return '<html><body><article>' + body + '</article></body></html>'
class NoArticles(Exception):
pass
def get_content(url_):
headers = {
'User-Agent': 'TheEconomist-Liskov-android',
'accept': 'multipart/mixed; deferSpec=20220824, application/json',
'accept-encoding': 'gzip',
'content-type': 'application/json',
'x-app-trace-id': str(uuid4()),
'x-economist-consumer': 'TheEconomist-Liskov-android',
'x-teg-client-name': 'Economist-Android',
'x-teg-client-os': 'Android',
'x-teg-client-version': '4.40.0',
}
br = browser()
req = Request(
url_,
headers=headers,
)
res = br.open(req)
return res.read()
def process_url(url):
if url.startswith('/'):
url = 'https://www.economist.com' + url
return url
class econ_search(BasicNewsRecipe):
title = 'The Economist - Search'
language = 'en_GB'
encoding = 'utf-8'
__author__ = 'unkn0wn'
description = (
'Use the Advanced section of the recipe to search.'
)
ignore_duplicate_articles = {'url'}
extra_css = '''
em { color:#202020; }
img {display:block; margin:0 auto;}
'''
# economist.com has started throttling after about 60% of the total has
# downloaded with connection reset by peer (104) errors.
delay = 1
browser_type = 'webengine'
recipe_specific_options = {
'q': {
'short': 'Text Search',
'default': 'schools brief'
},
's': {
'short': 'Sort by (date/relevance)',
'long': 'you can sort by date or relevance',
'default': 'relevance'
},
'p': {
'short': 'number of pages',
'long': 'number of pages of search results you want',
'default': '2'
},
'res': {
'short': 'For hi-res images, select a resolution from the\nfollowing options: 834, 960, 1096, 1280, 1424',
'long': 'This is useful for non e-ink devices, and for a lower file size\nthan the default, use from 480, 384, 360, 256.',
'default': '600',
},
}
def get_browser(self, *args, **kwargs):
kwargs['user_agent'] = (
'Mozilla/5.0 (Linux; Android 14) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/127.0.6533.103 Mobile Safari/537.36 Liskov'
)
br = BasicNewsRecipe.get_browser(self, *args, **kwargs)
return br
def parse_index(self):
url = 'https://www.economist.com/search?q={query}&sort={sort}&page={page}'
search = self.recipe_specific_options.get('q')
sort_type = self.recipe_specific_options.get('s')
page = self.recipe_specific_options.get('p')
self.title = 'The Economist - ' + search
ans = []
for num in range(1, int(page) + 1):
ans.extend(self.economist_parse_index(url.format(query=search.replace(' ', '+'), sort=sort_type, page=num)))
return [('Articles', ans)]
def economist_parse_index(self, url):
self.log('Page ', url.rsplit('=', 1)[-1])
soup = self.index_to_soup(url)
results = soup.find('ol', id='search-results')
if not results:
self.log('\tPage ', url.rsplit('=', 1)[-1], ' not found')
return
for a in results.findAll('a', attrs={'class': '_search-result'}):
url = a['href']
title = self.tag_to_string(a.find(attrs={'class': '_headline'}))
desc = self.tag_to_string(a.find(attrs={'class': '_snippet'}))
self.log('\t', title, '\n\t', desc, '\n\t\t', url)
yield {'title': title, 'url': url, 'description': desc}
def preprocess_html(self, soup):
width = '600'
w = self.recipe_specific_options.get('res')
if w and isinstance(w, str):
width = w
for img in soup.findAll('img', src=True):
qua = (
'economist.com/cdn-cgi/image/width=' + width + ',quality=80,format=auto/'
)
img['src'] = img['src'].replace('economist.com/', qua)
return soup
def preprocess_raw_html(self, raw, url):
# open('/t/raw.html', 'wb').write(raw.encode('utf-8'))
html = load_article_from_web_json(raw)
root = parse(html)
# the economist uses <small> for small caps with a custom font
for init in root.xpath('//span[@data-caps="initial"]'):
init.set('style', 'font-weight:bold;')
for x in root.xpath('//small'):
if x.text and len(x) == 0:
x.text = x.text.upper()
x.tag = 'span'
x.set(
'style',
'text-transform: uppercase; font-size: 0.85em; letter-spacing: 0.05em;',
)
for h2 in root.xpath('//h2'):
h2.tag = 'h4'
for x in root.xpath('//figcaption'):
x.set('style', 'text-align:center; font-size:small;')
for x in root.xpath('//cite'):
x.tag = 'blockquote'
x.set('style', 'color:#404040;')
raw = etree.tostring(root, encoding='unicode')
return raw
def get_article(self, url):
query = {
'operationName': 'ArticleDeeplinkQuery',
'variables': '{{"ref":"{}"}}'.format(url),
'query': 'query ArticleDeeplinkQuery($ref: String!, $includeRelatedArticles: Boolean = true ) { findArticleByUrl(url: $ref) { __typename ...ArticleDataFragment } } fragment ContentIdentityFragment on ContentIdentity { articleType forceAppWebView leadMediaType } fragment NarrationFragment on Narration { album bitrate duration filename id provider url isAiGenerated fileHash } fragment ImageTeaserFragment on ImageComponent { altText height imageType source url width } fragment PodcastAudioFragment on PodcastEpisode { id audio { url durationInSeconds } } fragment ArticleTeaserFragment on Article { id tegId url rubric headline flyTitle brand byline dateFirstPublished dateline dateModified datePublished dateRevised estimatedReadTime wordCount printHeadline contentIdentity { __typename ...ContentIdentityFragment } section { tegId name } teaserImage { __typename type ...ImageTeaserFragment } leadComponent { __typename type ...ImageTeaserFragment } narration(selectionMethod: PREFER_ACTOR_NARRATION) { __typename ...NarrationFragment } podcast { __typename ...PodcastAudioFragment } } fragment AnnotatedTextFragment on AnnotatedText { text textJson annotations { type length index attributes { name value } } } fragment ImageComponentFragment on ImageComponent { altText caption { __typename ...AnnotatedTextFragment } credit height imageType mode source url width } fragment BlockQuoteComponentFragment on BlockQuoteComponent { text textJson annotations { type length index attributes { name value } } } fragment BookInfoComponentFragment on BookInfoComponent { text textJson annotations { type length index attributes { name value } } } fragment ParagraphComponentFragment on ParagraphComponent { text textJson annotations { type length index attributes { name value } } } fragment PullQuoteComponentFragment on PullQuoteComponent { text textJson annotations { type length index attributes { name value } } } fragment CrossheadComponentFragment on CrossheadComponent { text } fragment OrderedListComponentFragment on OrderedListComponent { items { __typename ...AnnotatedTextFragment } } fragment UnorderedListComponentFragment on UnorderedListComponent { items { __typename ...AnnotatedTextFragment } } fragment VideoComponentFragment on VideoComponent { url title thumbnailImage } fragment InfoboxComponentFragment on InfoboxComponent { components { __typename type ...BlockQuoteComponentFragment ...BookInfoComponentFragment ...ParagraphComponentFragment ...PullQuoteComponentFragment ...CrossheadComponentFragment ...OrderedListComponentFragment ...UnorderedListComponentFragment ...VideoComponentFragment } } fragment InfographicComponentFragment on InfographicComponent { url title width fallback { __typename ...ImageComponentFragment } altText height width } fragment ArticleDataFragment on Article { id url brand byline rubric headline layout { headerStyle } contentIdentity { __typename ...ContentIdentityFragment } dateline dateFirstPublished dateModified datePublished dateRevised estimatedReadTime narration(selectionMethod: PREFER_ACTOR_NARRATION) { __typename ...NarrationFragment } printFlyTitle printHeadline printRubric flyTitle wordCount section { tegId name articles(pagingInfo: { pagingType: OFFSET pageSize: 6 pageNumber: 1 } ) @include(if: $includeRelatedArticles) { edges { node { __typename ...ArticleTeaserFragment } } } } teaserImage { __typename type ...ImageComponentFragment } tegId leadComponent { __typename type ...ImageComponentFragment } body { __typename type ...BlockQuoteComponentFragment ...BookInfoComponentFragment ...ParagraphComponentFragment ...PullQuoteComponentFragment ...CrossheadComponentFragment ...OrderedListComponentFragment ...UnorderedListComponentFragment ...InfoboxComponentFragment ...ImageComponentFragment ...VideoComponentFragment ...InfographicComponentFragment } footer { __typename type ...ParagraphComponentFragment } tags { name } ads { adData } podcast { __typename ...PodcastAudioFragment } }', # noqa: E501
}
deep_url = 'https://cp2-graphql-gateway.p.aws.economist.com/graphql?' + urlencode(
query, safe='()!', quote_via=quote
)
raw = get_content(deep_url)
return raw
def print_version(self, url):
art_cont = self.get_article(url)
pt = PersistentTemporaryFile('.html')
pt.write(art_cont)
pt.close()
return 'file:///' + pt.name