Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

Refactor "Fokus.se" recipe

Browse Source
This commit is contained in:
Henrik Holm 2025-05-24 01:57:50 +02:00
parent 8abd9b706e
commit 9a75607748
No known key found for this signature in database
1 changed files with 101 additions and 193 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

294
recipes/fokus.recipe
View File

@ -1,9 +1,6 @@
#!/usr/bin/env python
# vim:fileencoding=utf-8
import time
from datetime import datetime, timedelta
from mechanize import Request
from datetime import datetime, timedelta, timezone
from calibre.ebooks.BeautifulSoup import BeautifulSoup
from calibre.web.feeds.news import BasicNewsRecipe
@ -16,7 +13,7 @@ class NoArticles(Exception):
class Fokus(BasicNewsRecipe):
title = 'Fokus'
main_url = 'https://www.fokus.se'
description = "The last 7 days of news and articles from the Swedish current-affairs magazine 'Fokus'"
description = "The current week's edition of Swedish current-affairs magazine 'Fokus'"
encoding = 'utf-8'
__author__ = 'Henrik Holm (https://github.com/h-holm)'
language = 'sv'
@ -25,8 +22,6 @@ class Fokus(BasicNewsRecipe):
no_stylesheets = True
compress_news_images = True
needs_subscription = 'optional'
oldest_article = 7 # days
max_articles_per_feed = 15
use_embedded_content = False
remove_empty_feeds = True
scale_news_images_to_device = True
@ -51,117 +46,122 @@ class Fokus(BasicNewsRecipe):
dict(name='p', class_='article-metadata'), # Dynamically created by the recipe.
dict(name='figure', class_='Single__thumbnail'), # Image of "Single" type articles.
dict(name='figure', class_='Longread__thumbnail'), # Image of "Longread" type articles.
dict(name='p', class_='Meta__author'), # Author of the article.
dict(name='time', class_='Meta__updated'), # Last updated date of the article.
dict(name='div', class_='sesamy-protected-content'), # Article body.
]
def get_cover_url(self) -> str:
# Create a `mechanize.Request` object.
req = Request(url=self.main_url, method='POST')
def extract_cover_url(self, a_tag) -> str:
'''Given the <a> tag of the current edition, extract the URL of the highest-resolution cover image.'''
figure_tag = a_tag.find('figure')
img_tag = figure_tag.find('img')
# Open the requested URL in the built-in browser of the `BasicNewsRecipe` parent class.
browser = self.get_browser()
response = browser.open(req)
# The `srcset` attribute contains a comma-separated list of URLs and their resolutions.
cover_urls = img_tag['srcset'].split(', ')
cover_urls = [src.split(' ') for src in cover_urls]
# Parse the response into a BeautifulSoup soup.
soup = BeautifulSoup(response.get_data(), 'html.parser')
# The second item of each tuple should be the resolution, e.g., '578w' or '821w'. Remove the 'w' suffix, cast
# to an integer and sort in descending order.
cover_urls = [(url, int(resolution[:-1])) for url, resolution in cover_urls]
cover_urls = sorted(cover_urls, key=lambda x: x[1], reverse=True)
# The cover image of the current edition is located in a <figure> tag with class 'Issue__thumbnail'.
# The first item of the sorted list is now the URL of the highest-resolution image.
self.cover_url = cover_urls[0][0]
self.log(f"Identified cover URL: '{self.cover_url}'")
return
def get_current_edition_url(self) -> str:
'''Return the URL of the current (weekly) edition of Fokus.'''
current_year = datetime.now().year
try:
figure_tag = soup.find('figure', class_='Issue__thumbnail')
img_tag = figure_tag.find('img')
# Set the `img_tag` to `None` if it is falsy. This way, we can force an `AttributeError` if no cover URL
# can be found.
img_tag = img_tag if img_tag else None
cover_url = img_tag['src']
except AttributeError:
self.log.error("Failed to identify the cover image URL. Does an 'Issue__thumbnail' figure still exist?")
return ''
soup = self.index_to_soup(f"{self.main_url}/vara-utgavor")
return cover_url
# Identify all <a> tags of class 'Issue' that have an href attribute containing the current year.
a_tags = soup.find_all('a', class_='Issue', href=True)
def get_browser(self):
br = BasicNewsRecipe.get_browser(self)
if self.username and self.password:
br.open('https://www.fokus.se/auth/logga-in')
br.select_form(name='loginForm')
br['j_username'] = self.username
br['j_password'] = self.password
br.submit()
return br
# Keep only the href, and subset to only those links that contain the current year.
edition_links = [a_tag['href'] for a_tag in a_tags if str(current_year) in a_tag['href']]
def get_web_sections(self, main_url: str) -> dict[str, str]:
'''Return a dict of (1) section URL and (2) section name key-value pairs found at `main_url`.
# In order to successfully sort the links chronologically, first convert the data structure to a dict, wherein
# the key consists of only the date part of the URL and the value consists of the entire (unmodified) URL.
edition_links = {link.removesuffix('/').split('/')[-1]: link for link in edition_links}
For example, if the Fokus website currently includes an 'Aktuellt' section, the dict should include an entry on
the form: `{'https://www.fokus.se/aktuellt': 'Aktuellt'}`.
# Then, shorten the key further by keeping only the part after the first hyphen. This removes the year and
# typically results in only the calendar week number remaining, e.g., '1', '21' or '52'. Note however that
# editions can sometimes cover multiples weeks, e.g., '1-2', '01-03' or '50-51-52'. In order to sort correctly,
# it is therefore necessary to additionally keep only the first part of the week number(s) after the hyphen.
edition_links = {key.split('-', 1)[-1].split('-', 1)[0]: value for key, value in edition_links.items()}
Args:
main_url (str): The entrypoint URL of the Fokus website.
# Now, convert the resulting keys to integers
edition_links = {int(key): value for key, value in edition_links.items()}
Yields:
dict[str, str]: (1) URLs and (2) human-readable names of Fokus sections.
'''
self.log(f"Identifying all sections under '{main_url}'...")
soup = self.index_to_soup(main_url)
# Finally, sort in descending order, so that the most recent edition is first.
edition_links = dict(sorted(edition_links.items(), reverse=True))
current_edition_url = edition_links[list(edition_links.keys())[0]]
# Identify all unique <li> tags of class 'menu-item-type-taxonomy'. The class subsetting excludes sections that
# are not suited for reading, e.g., the "Podcast" and "Läs E-Tidningen" sections.
section_urls_and_names = {}
for li_tag in soup.find_all('li', class_='menu-item-type-taxonomy'):
# The <li> tag contains (should contain) an <a> anchor that in turn contains the URL and link name.
a_tag = li_tag.find('a')
url = a_tag.get('href').rstrip('/')
section_name = a_tag.text.strip()
self.log(f"Identified {len(edition_links)} editions, of which the most recent is '{current_edition_url}'.")
if url in section_urls_and_names:
# If this section URL has already been extracted from another <li> tag, it can be the case that the
# section name differs within this duplicate pair. In this case, use whichever section name is longer.
if len(section_name) >= len(section_urls_and_names[url]):
section_urls_and_names[url] = section_name
# Now that we know the URL of the current edition, we can use it to identify the cover image. The cover
# image URL exists in the `src` attribute of the <img> child tag of the <figure> child tag of the <a> tag
# of the current edition.
current_edition_a_tag = soup.find('a', class_='Issue', href=current_edition_url)
self.extract_cover_url(current_edition_a_tag)
except Exception as exc:
self.log.error(f"Failed to identify the current edition URL: {e}")
raise NoArticles(
f"Could not find the URL of the current edition. Either the '{self.main_url}' server is experiencing "
'issues, in which case you should try again later, or the website format has changed and the recipe '
'needs updating.'
) from exc
return current_edition_url
self.log(f"Identified section '{section_name}' at URL '{url}'.")
section_urls_and_names[url] = section_name
self.log(f'Identified a total of {len(section_urls_and_names)} unique sections.')
return section_urls_and_names
def parse_article_blurb(self, article_blurb) -> dict[str, str, str, str] | None:
def parse_article_blurb(self, article_blurb) -> dict[str, str, str, str, str] | None:
'''Given a <article> tag of class 'Blurb', parse it into a dict.
Args:
article_blurb (Tag): An <article> tag hosting metadata and the URL of an article.
Returns:
dict[str, str, str, str]: A dict on a `{'url': str, 'title': str, 'description': str, 'date': str}` format.
A dict on a `{'url': str, 'title': str, 'date': str, 'category': str, 'description': str}` format.
'''
if a_tag := article_blurb.find('a', href=True):
url = a_tag['href'].strip().rstrip('/')
if url.startswith('/'):
url = f'{self.main_url}{url}'
if title_tag := a_tag.find('h2', {'class': 'PostBlurb__title'}):
if title_tag := a_tag.find('h2', {'class': 'Blurb__title'}):
title = self.tag_to_string(title_tag).strip()
desc = ''
if desc_tag := a_tag.find('div', {'class': 'PostBlurb__excerpt'}):
if date_tag := a_tag.find('time', {'class': 'Blurb__date'}):
# Results in a Swedish date format, e.g., '23 MAJ 2025'.
date = self.tag_to_string(date_tag).strip()
# Add a newline before the date to make it more readable.
date = f'\n{date}'
# Assign the article to its first listed category as inferred from the first <li> tag of class
# 'Blurb__category'. Default to 'Fokus' if no such tag is found.
category = 'Fokus'
if category_tag := a_tag.find('li', {'class': 'Blurb__category'}):
category = self.tag_to_string(category_tag).strip()
desc = ''
if desc_tag := a_tag.find('div', {'class': 'Blurb__summary'}):
desc = self.tag_to_string(desc_tag).strip()
return {'url': url, 'title': title, 'description': desc}
return {'url': url, 'title': title, 'date': date, 'category': category, 'description': desc}
return
def _get_article_blurbs(self, soup) -> dict[str, dict[str, str, str, str]]:
def get_article_blurbs(self, soup) -> dict[str, dict[str, str, str, str, str]]:
'''Given a Fokus webpage `soup`, return a dict of unique article entries found on the page.
The key of a given entry in the output dictionary is the article URL. The corresponding value is a dictionary
on a `{'url': str, 'title': str, 'description': str, 'date': str}` format.
on a `{'url': str, 'title': str, 'date': str, 'category': str, 'description': str}` format.
Args:
soup (BeautifulSoup): The `bs4.BeautifulSoup` soup of a Fokus webpage.
Returns:
dict[str, dict[str, str, str, str]]: A dict with article URLs as keys and 'article dicts' as values.
dict[str, dict[str, str, str, str, str]]: A dict with article URLs as keys and 'article dicts' as values.
'''
def _log(article) -> None:
@ -171,8 +171,10 @@ class Fokus(BasicNewsRecipe):
log_message += f" : {article['description']}"
self.log.debug(log_message)
# Identify all <article> tags of class 'Blurb' that have an href attribute.
self.log(f'Identifying all articles...')
try:
article_blurbs = soup.find_all('article', {'class': 'PostBlurb'})
article_blurbs = soup.find_all('article', {'class': 'Blurb'})
except AttributeError:
article_blurbs = []
@ -184,94 +186,49 @@ class Fokus(BasicNewsRecipe):
for article_blurb in article_blurbs:
if article := self.parse_article_blurb(article_blurb):
_log(article)
# If an entry with the same URL already exists, keep whichever entry has the longer description.
if article['url'] in article_blurbs:
if len(article['description']) <= len(parsed_blurbs[article['url']]['description']):
continue
parsed_blurbs[article['url']] = article
return parsed_blurbs
def get_article_blurbs(self, sections: dict[str, str]) -> dict[str, dict[str, str, str, str]]:
'''Create and return a dict of all unique article blurbs found in all `sections`.
The key of a given entry in the output dictionary is the article URL. The corresponding value is a dictionary
on a `{'url': str, 'title': str, 'description': str, 'date': str}` format.
Args:
sections (dict[str, str]): A dict on a `{section_url: section_name}` format.
Returns:
dict[str, dict[str, str, str, str]]: A dict with article URLs as keys and 'article dicts' as values.
'''
self.log(f'Identifying all articles under all {len(sections)} sections...')
article_blurbs = {}
for section_url, section_title in sections.items():
try:
section_soup = self.index_to_soup(section_url)
except Exception:
self.log.error(f"Failed to download section '{section_title}' via URL '{section_url}'")
continue
self.log(f"Identifying all articles under '{section_url}'...")
for article_url, article_blurb in self._get_article_blurbs(section_soup).items():
# If the article URL has already been encountered, keep only the article blurb with the longer
# description string.
if article_url not in article_blurbs:
article_blurbs[article_url] = article_blurb
elif len(article_blurb['description']) > len(article_blurbs[article_url]['description']):
article_blurbs[article_url] = article_blurb
self.log(f'A total of {len(article_blurbs)} articles were identified in the {len(sections)} sections.')
return article_blurbs
def assign_articles_to_sections(
def convert_to_section_lists(
self,
sections: dict[str, str],
articles: dict[str, dict[str, str, str, str]],
articles: dict[str, dict[str, str, str, str, str]],
) -> dict[str, list[dict[str, str, str, str]]]:
'''Assign each article in `articles` to a section in `sections`.
'''Convert the `articles` dict of dicts to a dict of lists; each list holds the articles of a given section.
Args:
sections (dict[str, str]): A dict of section URLs as keys and section titles as values.
articles (dict[str, dict[str, str, str, str]]): A dict of article URLs as keys and article dicts as values.
articles (dict[str, dict[str, str, str, str, str]]): A dict of article URLs and article dicts.
Returns:
dict[str, list[dict[str, str, str, str]]]: A dict on a `{section_title: list[article_dict]}` format.
'''
self.log(f'Assigning each of the {len(articles)} articles to either of the {len(sections)} sections...')
self.log(f'Assigning each of the {len(articles)} articles to a section...')
section_to_articles = {}
for article_url, article_dict in articles.items():
last_url = article_url
while article_url not in sections and len(article_url) > len(self.main_url):
article_url = article_url.rsplit('/', 1)[0]
# Prevent an infinite loop.
if article_url == last_url:
break
last_url = article_url
# If no section corresponding to the URL exists, default to the 'Home Page' section.
section_title = sections[article_url] if article_url in sections else sections[self.main_url]
section_title = article_dict['category']
if section_title not in section_to_articles:
section_to_articles[section_title] = []
# Remove the 'category' key from the article dict, as it is not needed in the final output.
article_dict.pop('category')
section_to_articles[section_title].append(article_dict)
# Log how many sections contained no articles younger than `self.oldest_article`.
if diff := len(sections) - len(section_to_articles):
self.log(f'{diff} sections contained no articles younger than {self.oldest_article} days.')
return section_to_articles
def parse_index(self):
# Identify all sections in the web version of Fokus.
sections = self.get_web_sections(self.main_url)
current_edition_url = self.get_current_edition_url()
if not current_edition_url:
raise NoArticles(
f"Could not find the URL of the current edition. Either the '{self.main_url}' server is experiencing "
'issues, in which case you should try again later, or the website format has changed and the recipe '
'needs updating.'
)
self.log(f'Current edition URL: {current_edition_url}')
# Add an entry for the start page.
sections[self.main_url] = 'Home Page'
# Identify all sections in the web version of Fokus.
edition_soup = self.index_to_soup(current_edition_url)
# From the section URLs and the main URL, identify all unique articles.
articles = self.get_article_blurbs(sections)
articles = self.get_article_blurbs(edition_soup)
if not articles:
raise NoArticles(
f"Could not find any articles. Either the '{self.main_url}' server is experiencing issues, in which "
@ -279,56 +236,7 @@ class Fokus(BasicNewsRecipe):
)
# Assign each identified article to a section based on its URL.
section_to_articles = self.assign_articles_to_sections(sections, articles)
section_to_articles = self.convert_to_section_lists(articles)
# Convert to the expected `list[tuple[str, dict[str, str, str, str]]]` format.
feeds = list(section_to_articles.items())
num_articles = sum(len(article_dicts) for article_dicts in section_to_articles.values())
self.log(f'A total of {num_articles} articles belonging to {len(section_to_articles)} sections were kept.')
return feeds
def populate_article_metadata(self, article, soup, _):
# The article description/summary is found in the <p> tag of class 'Single__lead' or 'Longread__lead'.
lead_tag = soup.find('p', {'class': ['Single__lead', 'Longread__lead']})
article.summary = article.text_summary = lead_tag.get_text(strip=True)
# Extract the article timestamp from the `datetime` attribute of the first <time> tag of class 'Meta__updated'.
# The timestamp is on the ISO format. After the timestamp has been extracted, remove all such <time> tags from
# the soup (the article can contain several).
if time_tag := soup.find('time', {'class': 'Meta__updated'}):
dt = time_tag['datetime']
dt = datetime.fromisoformat(dt) + timedelta(seconds=time.timezone)
article.date = dt.strftime('%Y-%m-%d %H:%M')
for time_tag in soup.find_all('time', {'class': 'Meta__updated'}):
time_tag.decompose()
# Extract the author name from the first <p> tag of class 'Meta__author'. After the author name has been
# extracted, remove all such <p> tags from the soup (the article can contain several).
if author_tag := soup.find('p', {'class': 'Meta__author'}):
# If the tag contains an <a> child tag, extract the author name from it.
if a_tag := author_tag.find('a'):
author_info = a_tag.get_text(strip=True)
# To ensure a clean output, remove the <a> child tag.
a_tag.decompose()
else:
# If the tag does not contain an <a> child tag, extract the author name from the text of the <p> tag.
author_info = author_tag.get_text(strip=True)
# Remove the 'Text: ' prefix from the author name (if any).
if author_info.startswith('Text: '):
author_info = author_info[6:]
for author_tag in soup.find_all('p', {'class': 'Meta__author'}):
author_tag.decompose()
else:
# If the author name is empty, set it to 'Fokus'.
if not author_info:
author_info = 'Fokus'
# Concatenate the author name and the article date.
article_metadata = f"{author_info} | {article.date}"
# Finally, add a new <p> tag with the article metadata to the soup. Place it directly after the lead text.
new_tag = soup.new_tag('p')
new_tag['class'] = 'article-metadata'
new_tag.string = article_metadata
lead_tag.insert_after(new_tag)
# Convert to tuples.
return list(section_to_articles.items())