mirror of
https://github.com/kovidgoyal/calibre.git
synced 2025-07-07 10:14:46 -04:00
Update bloomberg-business-week.recipe
This commit is contained in:
unkn0w7n
parent
ed2b528568
commit
1c2a8d3ecd
@ -5,6 +5,44 @@ import json
|
|||||||
import random
|
import random
|
||||||
import time
|
import time
|
||||||
|
|
||||||
|
def get_contents(x):
|
||||||
|
otype = x.get('type', '')
|
||||||
|
if otype == 'text':
|
||||||
|
return x.get('value', '')
|
||||||
|
elif otype == 'paragraph':
|
||||||
|
return '<p>' + x.get('value', '') + ''.join(map(get_contents, x.get('content'))) + '</p>'
|
||||||
|
elif otype == 'heading':
|
||||||
|
return '<h3>' + x.get('value', '') + ''.join(map(get_contents, x.get('content'))) + '</h3>'
|
||||||
|
elif otype == 'list':
|
||||||
|
return '<ul>' + ''.join(map(get_contents, x.get('content'))) + '</ul>'
|
||||||
|
elif otype == 'listItem':
|
||||||
|
return '<li>' + x.get('value', '') + ''.join(map(get_contents, x.get('content'))) + '</li>'
|
||||||
|
elif otype == 'quote':
|
||||||
|
return '<blockquote>' + x.get('value', '') + ''.join(map(get_contents, x.get('content'))) + '</blockquote>'
|
||||||
|
elif otype == 'media':
|
||||||
|
if x['subType'] == 'photo':
|
||||||
|
return '<div><div><img src="{}"></div><div>{}</div></div>'.format(
|
||||||
|
x['data']['photo']['src'], x['data']['photo']['caption'])
|
||||||
|
elif x['subType'] == 'chart':
|
||||||
|
if x['data'] and x['data']['chart']:
|
||||||
|
return '<div><img src="{}"></div>'.format(x['data']['chart']['fallback'])
|
||||||
|
elif otype == 'link':
|
||||||
|
if x['data'] and x['content'] and x['content'][0] and x['content'][0]['value']:
|
||||||
|
if 'href' in x['data']:
|
||||||
|
return '<a href="' + x['data']['href'] + '">' + x['content'][0]['value'] + '</a>'
|
||||||
|
return '<i>' + x['content'][0]['value'] + '</i>'
|
||||||
|
elif otype == 'entity':
|
||||||
|
if x['content'] and x['content'][0] and x['content'][0]['value']:
|
||||||
|
if x['subType'] == 'story':
|
||||||
|
if x['data'] and x['data']['link'] and x['data']['link']['destination']:
|
||||||
|
if 'web' in x['data']['link']['destination']:
|
||||||
|
return '<a href="' + x['data']['link']['destination']['web'] + '">' + x['content'][0]['value'] + '</a>'
|
||||||
|
return '<i>' + x['content'][0]['value'] + '</i>'
|
||||||
|
elif x['subType'] in ('person', 'security'):
|
||||||
|
return '<i>' + x['content'][0]['value'] + '</i>'
|
||||||
|
|
||||||
|
return ''
|
||||||
|
|
||||||
class Bloomberg(BasicNewsRecipe):
|
class Bloomberg(BasicNewsRecipe):
|
||||||
title = u'Bloomberg Businessweek'
|
title = u'Bloomberg Businessweek'
|
||||||
language = 'en'
|
language = 'en'
|
||||||
@ -124,142 +162,11 @@ class Bloomberg(BasicNewsRecipe):
|
|||||||
else:
|
else:
|
||||||
body = ''
|
body = ''
|
||||||
body_data = data['body']['content']
|
body_data = data['body']['content']
|
||||||
|
for x in body_data:
|
||||||
for objects in body_data:
|
|
||||||
|
|
||||||
pause = random.choice((0.5, 1, 1.25))
|
pause = random.choice((0.5, 1, 1.25))
|
||||||
time.sleep(pause)
|
time.sleep(pause)
|
||||||
|
body += get_contents(x)
|
||||||
if objects['type'] == 'media' and objects['subType'] == 'photo':
|
return '<html><body>' + cat + title + subhead + auth + lede + caption + '<div>' + body + '</div></body></html>'
|
||||||
body += '<p id="img"><img src="{}">'.format(objects['data']['photo']['src'])
|
|
||||||
body += '<span>' + objects['data']['photo']['caption'] + '</span></p>'
|
|
||||||
if objects['type'] == 'media' and objects['subType'] == 'chart':
|
|
||||||
if objects['data'] and objects['data']['chart']:
|
|
||||||
body += '<p id="img"><img src="{}">'.format(objects['data']['chart']['fallback'])
|
|
||||||
|
|
||||||
if objects['type'] == 'paragraph':
|
|
||||||
body += '<p>'
|
|
||||||
if 'value' in objects:
|
|
||||||
body += objects['value']
|
|
||||||
|
|
||||||
if 'content' not in objects:
|
|
||||||
continue
|
|
||||||
|
|
||||||
for item in objects['content']:
|
|
||||||
|
|
||||||
if item['type'] == 'text' and item['value']:
|
|
||||||
body += item['value']
|
|
||||||
|
|
||||||
if item['type'] == 'link' and item['data']:
|
|
||||||
if item['content'] and item['content'][0] and item['content'][0]['value']:
|
|
||||||
if 'href' in item['data']:
|
|
||||||
body += '<a href="' + item['data']['href'] + '">' + item['content'][0]['value'] + '</a>'
|
|
||||||
else:
|
|
||||||
body += '<i>' + item['content'][0]['value'] + '</i>'
|
|
||||||
|
|
||||||
if item['type'] == 'entity':
|
|
||||||
if item['content'] and item['content'][0] and item['content'][0]['value']:
|
|
||||||
if item['subType'] == 'story':
|
|
||||||
if item['data'] and item['data']['link'] and item['data']['link']['destination']:
|
|
||||||
if 'web' in item['data']['link']['destination']:
|
|
||||||
body += '<a href="' + item['data']['link']['destination']['web'] + '">' + item['content'][0]['value'] + '</a>'
|
|
||||||
else:
|
|
||||||
body += '<i>' + item['content'][0]['value'] + '</i>'
|
|
||||||
|
|
||||||
elif item['subType'] == 'person' or 'security':
|
|
||||||
body += item['content'][0]['value']
|
|
||||||
|
|
||||||
if objects['type'] == 'heading':
|
|
||||||
body += '<p id="bold">'
|
|
||||||
if 'value' in objects:
|
|
||||||
body += objects['value']
|
|
||||||
|
|
||||||
if 'content' not in objects:
|
|
||||||
continue
|
|
||||||
|
|
||||||
for item in objects['content']:
|
|
||||||
|
|
||||||
if item['type'] == 'text' and item['value']:
|
|
||||||
body += item['value']
|
|
||||||
|
|
||||||
if item['type'] == 'link' and item['data']:
|
|
||||||
if item['content'] and item['content'][0] and item['content'][0]['value']:
|
|
||||||
if 'href' in item['data']:
|
|
||||||
body += '<a href="' + item['data']['href'] + '">' + item['content'][0]['value'] + '</a>'
|
|
||||||
else:
|
|
||||||
body += '<i>' + item['content'][0]['value'] + '</i>'
|
|
||||||
|
|
||||||
if item['type'] == 'entity':
|
|
||||||
if item['content'] and item['content'][0] and item['content'][0]['value']:
|
|
||||||
if item['subType'] == 'story':
|
|
||||||
if item['data'] and item['data']['link'] and item['data']['link']['destination']:
|
|
||||||
if 'web' in item['data']['link']['destination']:
|
|
||||||
body += '<a href="' + item['data']['link']['destination']['web'] + '">' + item['content'][0]['value'] + '</a>'
|
|
||||||
else:
|
|
||||||
body += '<i>' + item['content'][0]['value'] + '</i>'
|
|
||||||
|
|
||||||
elif item['subType'] == 'person' or 'security':
|
|
||||||
body += item['content'][0]['value']
|
|
||||||
|
|
||||||
if objects['type'] == 'quote':
|
|
||||||
if 'value' in objects:
|
|
||||||
body +='<blockquote id="subhead">' + objects['value'] + '</blockquote>'
|
|
||||||
if 'content' not in objects:
|
|
||||||
continue
|
|
||||||
for item in objects['content']:
|
|
||||||
if item['type'] == 'paragraph' and item['content'] and item['content'][0]:
|
|
||||||
if 'value' not in item['content'][0]:
|
|
||||||
continue
|
|
||||||
body += '<blockquote id="subhead">' + item['content'][0]['value'] + '</blockquote>'
|
|
||||||
|
|
||||||
if objects['type'] == 'list':
|
|
||||||
if 'content' not in objects:
|
|
||||||
continue
|
|
||||||
body += '<ul>'
|
|
||||||
for content in objects['content']:
|
|
||||||
if content['type'] == 'listItem':
|
|
||||||
body += '<li>'
|
|
||||||
if 'value' in content:
|
|
||||||
body += content['value']
|
|
||||||
elif 'content' in content:
|
|
||||||
for val_cont in content['content']:
|
|
||||||
if 'value' in val_cont:
|
|
||||||
body += val_cont['value']
|
|
||||||
elif 'content' in val_cont:
|
|
||||||
for cont2 in val_cont['content']:
|
|
||||||
if 'value' in cont2:
|
|
||||||
body += cont2['value']
|
|
||||||
body += '</ul>'
|
|
||||||
|
|
||||||
skip = ['ad', 'inline-newsletter', 'inline-recirc', 'tabularData', 'list', 'quote', 'heading', 'paragraph', 'media']
|
|
||||||
if not any(x in objects['type'] for x in skip):
|
|
||||||
body += '<p>'
|
|
||||||
if 'value' in objects:
|
|
||||||
body += objects['value']
|
|
||||||
if not 'content' in objects:
|
|
||||||
continue
|
|
||||||
for content in objects['content']:
|
|
||||||
if 'value' in content:
|
|
||||||
body += content['value']
|
|
||||||
elif 'content' in content:
|
|
||||||
for cont1 in content['content']:
|
|
||||||
if 'value' in cont1:
|
|
||||||
body += cont1['value']
|
|
||||||
elif 'content' in val_cont:
|
|
||||||
for cont2 in val_cont['content']:
|
|
||||||
if 'value' in cont2:
|
|
||||||
body += cont2['value']
|
|
||||||
elif 'content' in cont2:
|
|
||||||
for cont3 in cont2['content']:
|
|
||||||
if 'value' in cont3:
|
|
||||||
body += cont3['value']
|
|
||||||
elif 'content' in cont3:
|
|
||||||
for cont4 in cont3['content']:
|
|
||||||
if 'value' in cont4:
|
|
||||||
body += cont4['value']
|
|
||||||
|
|
||||||
html = '<html><body>' + cat + title + subhead + auth + lede + caption + '<div>' + body
|
|
||||||
return html
|
|
||||||
|
|
||||||
def preprocess_html(self, soup):
|
def preprocess_html(self, soup):
|
||||||
for icon in soup.findAll('img', attrs={'class':'video-player__play-icon'}):
|
for icon in soup.findAll('img', attrs={'class':'video-player__play-icon'}):
|
||||||
|
|||||||
Loading…
x
Reference in New Issue
Block a user