diff options
| -rw-r--r-- | pyaggr3g470r/lib/article_utils.py | 16 | ||||
| -rw-r--r-- | pyaggr3g470r/models/article.py | 8 |
2 files changed, 10 insertions, 14 deletions
diff --git a/pyaggr3g470r/lib/article_utils.py b/pyaggr3g470r/lib/article_utils.py index 115b6058..dbe29659 100644 --- a/pyaggr3g470r/lib/article_utils.py +++ b/pyaggr3g470r/lib/article_utils.py @@ -2,7 +2,6 @@ import logging import requests import dateutil.parser from datetime import datetime -from bs4 import BeautifulSoup import conf from pyaggr3g470r.lib.utils import to_hash @@ -10,9 +9,8 @@ from pyaggr3g470r.lib.utils import to_hash logger = logging.getLogger(__name__) -def extract_id(entry, keys=[('link', 'link'), - ('published', 'retrieved_date'), - ('updated', 'retrieved_date')], force_id=False): +def extract_id(entry, keys=[('link', 'link'), ('published', 'date'), + ('updated', 'date')], force_id=False): """For a given entry will return a dict that allows to identify it. The dict will be constructed on the uid of the entry. if that identifier is absent, the dict will be constructed upon the values of "keys". @@ -21,7 +19,7 @@ def extract_id(entry, keys=[('link', 'link'), if entry_id: return {'entry_id': entry_id} if not entry_id and force_id: - entry_id = to_hash("".join(entry[entry_key] for _, entry_key in keys + return to_hash("".join(entry[entry_key] for _, entry_key in keys if entry_key in entry).encode('utf8')) else: ids = {} @@ -38,8 +36,8 @@ def construct_article(entry, feed): if hasattr(feed, 'dump'): # this way can be a sqlalchemy obj or a dict feed = feed.dump() "Safe method to transorm a feedparser entry into an article" - date = datetime.now() - + now = datetime.now() + date = None for date_key in ('published', 'updated'): if entry.get(date_key): try: @@ -72,5 +70,5 @@ def construct_article(entry, feed): 'title': entry.get('title', 'No title'), 'readed': False, 'like': False, 'content': content, - 'retrieved_date': date.isoformat(), - 'date': date.isoformat()} + 'retrieved_date': now.isoformat(), + 'date': (date or now).isoformat()} diff --git a/pyaggr3g470r/models/article.py b/pyaggr3g470r/models/article.py index 16f15b9e..c98164d0 100644 --- a/pyaggr3g470r/models/article.py +++ b/pyaggr3g470r/models/article.py @@ -26,7 +26,6 @@ __revision__ = "$Date: 2014/04/12 $" __copyright__ = "Copyright (c) Cedric Bonhomme" __license__ = "GPLv3" -import json from bootstrap import db from datetime import datetime from sqlalchemy import asc, desc @@ -66,10 +65,9 @@ class Article(db.Model): .order_by(asc("Article.date")).first() def __repr__(self): - return json.dumps({"title": self.title, - "link": self.link, - "content": self.content - }) + return "<Article(id=%d, entry_id=%s, title=%r, " \ + "date=%r, retrieved_date=%r)>" % (self.id, self.entry_id, + self.title, self.date, self.retrieved_date) def dump(self): return {"id": self.id, |