def source(request, source=''):
s = db.get_source(source)
if s is None:
return _404_response()
start, end, limit = _parse_params(request)
articles = db.list_articles_by_source(source, start=start, end=end, limit=limit)
s['articles'] = articles
return json_response(s)
def source(request):
s = _get(request, 's')
if not s:
return HttpResponseRedirect('/')
articles = db.list_articles_by_source(s)
if not articles:
return HttpResponseRedirect('/')
for article in articles:
article['url'] = urllib.unquote(article['url'])
source = db.get_source(s)
return render_to_response('source.html', dict(articles=articles,
source=source))
import time
def fix_0A(a):
if '0A' in a['url']:
print a['url']
a['url'] = a['url'].replace('0A', '0')
return True
return False
def fetch_text(a):
if a['cached']:
return False
time.sleep(0.2)
txt = lp.fetch_text(urllib.unquote(a['url']))
if txt:
print a['url_date'], a['title'], a['url']
a['cached'] = txt
return True
return False
source = u'自由時報'
articles = db.list_articles_by_source(source, addition_cols=['cached'])
fixed = 0
for a in articles:
if any((fix_0A(a), fetch_text(a))):
#print 'fixed', a
db.ensure_article_exists(a, overwrite=True)
fixed += 1
print 'total: %s, fixed: %s' % (len(articles), fixed)
