X-Git-Url: https://projects.mako.cc/source/iron-blogger/blobdiff_plain/8c882ab61f0be6f76df016d98a20db04323c6be1..ab6966baf120f059c56d1f3185e65988e2adbb0b:/scan-feeds.py diff --git a/scan-feeds.py b/scan-feeds.py index 6373316..7e27b8c 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) @@ -21,20 +26,12 @@ def get_date(post): return post.updated def get_link(post): - if 'links' in post: - links = dict((l.rel, l) for l in post.links if 'html' in l.type) - if 'self' in links: - return links['self'].href - elif 'alternate' in links: - return links['alternate'].href - if 'href' in post: - return post.href - if 'link' in post: - return post.link - return None + 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)) @@ -44,19 +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)