def sample_results(self):
m1 = Metadata('The Great Gatsby', ['Francis Scott Fitzgerald'])
m2 = Metadata('The Great Gatsby', ['F. Scott Fitzgerald'])
m1.has_cached_cover_url = True
m2.has_cached_cover_url = False
m1.comments = 'Some comments '*10
m1.tags = ['tag%d'%i for i in range(20)]
m1.rating = 4.4
m1.language = 'en'
m2.language = 'fr'
m1.pubdate = utcnow()
m2.pubdate = fromordinal(1000000)
m1.publisher = 'Publisher 1'
m2.publisher = 'Publisher 2'
return [m1, m2]
def sample_results(self):
m1 = Metadata('The Great Gatsby', ['Francis Scott Fitzgerald'])
m2 = Metadata('The Great Gatsby - An extra long title to test resizing', ['F. Scott Fitzgerald'])
m1.has_cached_cover_url = True
m2.has_cached_cover_url = False
m1.comments = 'Some comments '*10
m1.tags = ['tag%d'%i for i in range(20)]
m1.rating = 4.4
m1.language = 'en'
m2.language = 'fr'
m1.pubdate = utcnow()
m2.pubdate = fromordinal(1000000)
m1.publisher = 'Publisher 1'
m2.publisher = 'Publisher 2'
return [m1, m2]
def sample_results(self):
m1 = Metadata("The Great Gatsby", ["Francis Scott Fitzgerald"])
m2 = Metadata("The Great Gatsby - An extra long title to test resizing", ["F. Scott Fitzgerald"])
m1.has_cached_cover_url = True
m2.has_cached_cover_url = False
m1.comments = "Some comments " * 10
m1.tags = ["tag%d" % i for i in range(20)]
m1.rating = 4.4
m1.language = "en"
m2.language = "fr"
m1.pubdate = utcnow()
m2.pubdate = fromordinal(1000000)
m1.publisher = "Publisher 1"
m2.publisher = "Publisher 2"
return [m1, m2]
def merge(self, results, min_year, do_asr=True):
ans = Metadata(_('Unknown'))
# We assume the shortest title has the least cruft in it
ans.title = self.length_merge('title', results, null_value=ans.title)
# No harm in having extra authors, maybe something useful like an
# editor or translator
ans.authors = self.length_merge('authors', results,
null_value=ans.authors, shortest=False)
# We assume the shortest publisher has the least cruft in it
ans.publisher = self.length_merge('publisher', results,
null_value=ans.publisher)
# We assume the smallest set of tags has the least cruft in it
ans.tags = self.length_merge('tags', results,
null_value=ans.tags, shortest=msprefs['fewer_tags'])
# We assume the longest series has the most info in it
ans.series = self.length_merge('series', results,
null_value=ans.series, shortest=False)
for r in results:
if r.series and r.series == ans.series:
ans.series_index = r.series_index
break
# Average the rating over all sources
ratings = []
for r in results:
rating = r.rating
if rating and rating > 0 and rating <= 5:
ratings.append(rating)
if ratings:
ans.rating = int(round(sum(ratings)/len(ratings)))
# Smallest language is likely to be valid
ans.language = self.length_merge('language', results,
null_value=ans.language)
# Choose longest comments
ans.comments = self.length_merge('comments', results,
null_value=ans.comments, shortest=False)
# Published date
if min_year:
for r in results:
year = getattr(r.pubdate, 'year', None)
if year == min_year:
ans.pubdate = r.pubdate
break
if getattr(ans.pubdate, 'year', None) == min_year:
min_date = datetime(min_year, ans.pubdate.month, ans.pubdate.day,
tzinfo=utc_tz)
else:
min_date = datetime(min_year, 1, 2, tzinfo=utc_tz)
ans.pubdate = min_date
else:
min_date = datetime(3001, 1, 1, tzinfo=utc_tz)
for r in results:
if r.pubdate is not None:
candidate = as_utc(r.pubdate)
if candidate < min_date:
min_date = candidate
if min_date.year < 3000:
ans.pubdate = min_date
# Identifiers
for r in results:
ans.identifiers.update(r.identifiers)
# Cover URL
ans.has_cached_cover_url = bool([r for r in results if
getattr(r, 'has_cached_cover_url', False)])
# Merge any other fields with no special handling (random merge)
touched_fields = set()
for r in results:
if hasattr(r, 'identify_plugin'):
touched_fields |= r.identify_plugin.touched_fields
for f in touched_fields:
if f.startswith('identifier:') or not ans.is_null(f):
continue
setattr(ans, f, self.random_merge(f, results,
null_value=getattr(ans, f)))
if do_asr:
avg = [x.relevance_in_source for x in results]
avg = sum(avg)/len(avg)
ans.average_source_relevance = avg
return ans