X-Git-Url: https://projects.mako.cc/source/iron-blogger/blobdiff_plain/24183af125853dcca9ea51fab8ece103ed4dc5ac..a1b80d8bf647891f36787c19c0dfff41584a7946:/scan-feeds.py diff --git a/scan-feeds.py b/scan-feeds.py index bc82608..bf60aef 100755 --- a/scan-feeds.py +++ b/scan-feeds.py @@ -2,13 +2,18 @@ import yaml import feedparser import datetime +import sys from dateutil.parser import parse import dateutil.tz as tz with open('bloggers.yml') as f: users = yaml.safe_load(f.read()) -log = {} +try: + with open('out/report.yml') as f: + log = yaml.safe_load(f.read()) +except IOError: + log = {} START = datetime.datetime(2009, 12, 21, 6) @@ -16,15 +21,17 @@ def parse_published(pub): return parse(pub).astimezone(tz.tzlocal()).replace(tzinfo=None) def get_date(post): - if 'published' in post: - return post.published - return post.updated + for k in ('published', 'created', 'updated'): + if k in post: + return post[k] def get_link(post): return post.link def parse_feeds(weeks, uri): feed = feedparser.parse(uri) + if not feed.entries: + print >>sys.stderr, "WARN: no entries for ", uri for post in feed.entries: date = parse_published(get_date(post)) @@ -34,19 +41,17 @@ def parse_feeds(weeks, uri): while len(weeks) <= wn: weeks.append([]) - weeks[wn].append(dict( - date=date, - title=post.title, - url=get_link(post))) + + post = dict(date=date, + title=post.title, + url=get_link(post)) + if post['url'] not in [p['url'] for p in weeks[wn]]: + weeks[wn].append(post) for (username, u) in users.items(): - weeks = [] - print "[%s]" % (username) + weeks = log.setdefault(username, []) for l in u['links']: parse_feeds(weeks, l[2]) - log[username] = weeks - for (i, w) in enumerate(weeks): - print " [%d]: %s" % (i, w) with open('out/report.yml', 'w') as f: yaml.safe_dump(log, f)