def get_statistics_from_last_update_for_all_sources(db_root):
last_metainfo_by_day = dict()
# fetch the summary of the last day for every source
for source_name in get_all_provider_names(db_root):
p = Provider(db_root, source_name)
all_days = p.get_all_days()
if all_days:
last_day = utils.get_latest_day(all_days)
last_metainfo_by_day[source_name] = (last_day, p.get_cached_metainfos_for_day(last_day))
# not every source has data for the real last day, search for th
last_days = set([v[0] for k, v in last_metainfo_by_day.items()])
real_last_day = utils.get_latest_day(last_days)
# build the overall metainfos using only the source which have data for the real last day
overall_metainfo = defaultdict(int)
provider_count = 0
for name, data in last_metainfo_by_day.items():
day, metainfos = data
if day == real_last_day:
provider_count += 1
for k, v in metainfos.items():
overall_metainfo[k] += v
overall_metainfo.update(dict(provider_count=provider_count))
return overall_metainfo
def get_first_and_last_date(db_root):
source_names = get_all_provider_names(db_root)
p = Provider(db_root, source_names[0])
all_days = p.get_all_days()
return utils.make_date_from_string(all_days[0]), utils.make_date_from_string(all_days[-1])
def get_summary_from_last_update_for_all_sources(db_root):
source_names = get_all_provider_names(db_root)
last_update = list()
for name in source_names:
p = Provider(db_root, name)
all_days = p.get_all_days()
if all_days:
last_day = utils.get_latest_day(all_days)
summary = p.get_cached_metainfos_for_day(last_day)
last_update.append((name, utils.make_date_from_string(last_day), summary))
return last_update