diff options
author | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2016-04-10 13:50:50 +0200 |
---|---|---|
committer | Cédric Bonhomme <cedric@cedricbonhomme.org> | 2016-04-10 13:50:50 +0200 |
commit | c7467cd9935a67418ac07cf8a0bec29bd74f9dc1 (patch) | |
tree | 15254c0905ef065cb7f03bfdb523a13a991923a7 /src | |
parent | Moved notifications module. (diff) | |
download | newspipe-c7467cd9935a67418ac07cf8a0bec29bd74f9dc1.tar.gz newspipe-c7467cd9935a67418ac07cf8a0bec29bd74f9dc1.tar.bz2 newspipe-c7467cd9935a67418ac07cf8a0bec29bd74f9dc1.zip |
Moved notifications module.
Diffstat (limited to 'src')
-rwxr-xr-x | src/web/utils.py | 281 |
1 files changed, 0 insertions, 281 deletions
diff --git a/src/web/utils.py b/src/web/utils.py deleted file mode 100755 index 92bdde8a..00000000 --- a/src/web/utils.py +++ /dev/null @@ -1,281 +0,0 @@ -#! /usr/bin/env python -#-*- coding: utf-8 -*- - -# JARR - A Web based news aggregator. -# Copyright (C) 2010-2016 Cédric Bonhomme - https://www.cedricbonhomme.org -# -# For more information : https://github.com/JARR-aggregator/JARR -# -# This program is free software: you can redistribute it and/or modify -# it under the terms of the GNU Affero General Public License as -# published by the Free Software Foundation, either version 3 of the -# License, or (at your option) any later version. -# -# This program is distributed in the hope that it will be useful, -# but WITHOUT ANY WARRANTY; without even the implied warranty of -# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the -# GNU Affero General Public License for more details. -# -# You should have received a copy of the GNU Affero General Public License -# along with this program. If not, see <http://www.gnu.org/licenses/>. - -__author__ = "Cedric Bonhomme" -__version__ = "$Revision: 1.8 $" -__date__ = "$Date: 2010/12/07 $" -__revision__ = "$Date: 2016/04/10 $" -__copyright__ = "Copyright (c) Cedric Bonhomme" -__license__ = "AGPLv3" - -# -# This file provides functions used for: -# - import from a JSON file; -# - generation of tags cloud; -# - HTML processing. -# - -import re -import sys -import glob -import opml -import json -import logging -import datetime -import operator -import urllib -import subprocess -import sqlalchemy -try: - from urlparse import urlparse, parse_qs, urlunparse -except: - from urllib.parse import urlparse, parse_qs, urlunparse, urljoin -from bs4 import BeautifulSoup -from collections import Counter -from contextlib import contextmanager -from flask import request - -import conf -from bootstrap import db -from web import controllers -from web.models import User, Feed, Article -from web.lib.utils import clear_string - -logger = logging.getLogger(__name__) - -ALLOWED_EXTENSIONS = set(['xml', 'opml', 'json']) - -def is_safe_url(target): - """ - Ensures that a redirect target will lead to the same server. - """ - ref_url = urlparse(request.host_url) - test_url = urlparse(urljoin(request.host_url, target)) - return test_url.scheme in ('http', 'https') and \ - ref_url.netloc == test_url.netloc - -def get_redirect_target(): - """ - Looks at various hints to find the redirect target. - """ - for target in request.args.get('next'), request.referrer: - if not target: - continue - if is_safe_url(target): - return target - -def allowed_file(filename): - """ - Check if the uploaded file is allowed. - """ - return '.' in filename and \ - filename.rsplit('.', 1)[1] in ALLOWED_EXTENSIONS - -@contextmanager -def opened_w_error(filename, mode="r"): - try: - f = open(filename, mode) - except IOError as err: - yield None, err - else: - try: - yield f, None - finally: - f.close() - -def fetch(id, feed_id=None): - """ - Fetch the feeds in a new processus. - The "asyncio" crawler is launched with the manager. - """ - cmd = [sys.executable, conf.BASE_DIR + '/manager.py', 'fetch_asyncio', - str(id), str(feed_id)] - return subprocess.Popen(cmd, stdout=subprocess.PIPE) - -def history(user_id, year=None, month=None): - """ - Sort articles by year and month. - """ - articles_counter = Counter() - articles = controllers.ArticleController(user_id).read() - if None != year: - articles = articles.filter(sqlalchemy.extract('year', Article.date) == year) - if None != month: - articles = articles.filter(sqlalchemy.extract('month', Article.date) == month) - for article in articles.all(): - if None != year: - articles_counter[article.date.month] += 1 - else: - articles_counter[article.date.year] += 1 - return articles_counter, articles - -def import_opml(email, opml_content): - """ - Import new feeds from an OPML file. - """ - user = User.query.filter(User.email == email).first() - try: - subscriptions = opml.from_string(opml_content) - except: - logger.exception("Parsing OPML file failed:") - raise - - def read(subsubscription, nb=0): - """ - Parse recursively through the categories and sub-categories. - """ - for subscription in subsubscription: - if len(subscription) != 0: - nb = read(subscription, nb) - else: - try: - title = subscription.text - except: - title = "" - try: - description = subscription.description - except: - description = "" - try: - link = subscription.xmlUrl - except: - continue - if None != Feed.query.filter(Feed.user_id == user.id, Feed.link == link).first(): - continue - try: - site_link = subscription.htmlUrl - except: - site_link = "" - new_feed = Feed(title=title, description=description, - link=link, site_link=site_link, - enabled=True) - user.feeds.append(new_feed) - nb += 1 - return nb - nb = read(subscriptions) - db.session.commit() - return nb - -def import_json(email, json_content): - """ - Import an account from a JSON file. - """ - user = User.query.filter(User.email == email).first() - json_account = json.loads(json_content) - nb_feeds, nb_articles = 0, 0 - # Create feeds: - for feed in json_account["result"]: - if None != Feed.query.filter(Feed.user_id == user.id, - Feed.link == feed["link"]).first(): - continue - new_feed = Feed(title=feed["title"], - description="", - link=feed["link"], - site_link=feed["site_link"], - created_date=datetime.datetime. - fromtimestamp(int(feed["created_date"])), - enabled=feed["enabled"]) - user.feeds.append(new_feed) - nb_feeds += 1 - db.session.commit() - # Create articles: - for feed in json_account["result"]: - user_feed = Feed.query.filter(Feed.user_id == user.id, - Feed.link == feed["link"]).first() - if None != user_feed: - for article in feed["articles"]: - if None == Article.query.filter(Article.user_id == user.id, - Article.feed_id == user_feed.id, - Article.link == article["link"]).first(): - new_article = Article(link=article["link"], - title=article["title"], - content=article["content"], - readed=article["readed"], - like=article["like"], - retrieved_date=datetime.datetime. - fromtimestamp(int(article["retrieved_date"])), - date=datetime.datetime. - fromtimestamp(int(article["date"])), - user_id=user.id, - feed_id=user_feed.id) - user_feed.articles.append(new_article) - nb_articles += 1 - db.session.commit() - return nb_feeds, nb_articles - -def clean_url(url): - """ - Remove utm_* parameters - """ - parsed_url = urlparse(url) - qd = parse_qs(parsed_url.query, keep_blank_values=True) - filtered = dict((k, v) for k, v in qd.items() - if not k.startswith('utm_')) - return urlunparse([ - parsed_url.scheme, - parsed_url.netloc, - urllib.parse.quote(urllib.parse.unquote(parsed_url.path)), - parsed_url.params, - urllib.parse.urlencode(filtered, doseq=True), - parsed_url.fragment - ]).rstrip('=') - -def load_stop_words(): - """ - Load the stop words and return them in a list. - """ - stop_words_lists = glob.glob('./JARR/var/stop_words/*.txt') - stop_words = [] - - for stop_wods_list in stop_words_lists: - with opened_w_error(stop_wods_list, "r") as (stop_wods_file, err): - if err: - stop_words = [] - else: - stop_words += stop_wods_file.read().split(";") - return stop_words - -def top_words(articles, n=10, size=5): - """ - Return the n most frequent words in a list. - """ - stop_words = load_stop_words() - words = Counter() - wordre = re.compile(r'\b\w{%s,}\b' % size, re.I) - for article in articles: - for word in [elem.lower() for elem in - wordre.findall(clear_string(article.content)) \ - if elem.lower() not in stop_words]: - words[word] += 1 - return words.most_common(n) - -def tag_cloud(tags): - """ - Generates a tags cloud. - """ - tags.sort(key=operator.itemgetter(0)) - return '\n'.join([('<font size=%d><a href="/search?query=%s" title="Count: %s">%s</a></font>' % \ - (min(1 + count * 7 / max([tag[1] for tag in tags]), 7), word, format(count, ',d'), word)) \ - for (word, count) in tags]) - -if __name__ == "__main__": - import_opml("root@jarr.localhost", "./var/feeds_test.opml") - #import_opml("root@jarr.localhost", "./var/JARR.opml") |