def metadata(self):
mi = Metadata(_('Unknown'))
dp_name, ap_name = self.get_document_properties_names()
if dp_name:
try:
raw = self.read(dp_name)
except KeyError:
pass
else:
read_doc_props(raw, mi, self.namespace.XPath)
if mi.is_null('language'):
try:
raw = self.read('word/styles.xml')
except KeyError:
pass
else:
read_default_style_language(raw, mi, self.namespace.XPath)
ap_name = self.relationships.get(self.namespace.names['APPPROPS'], None)
if ap_name:
try:
raw = self.read(ap_name)
except KeyError:
pass
else:
read_app_props(raw, mi)
return mi
def metadata(self):
mi = Metadata(_('Unknown'))
dp_name, ap_name = self.get_document_properties_names()
if dp_name:
try:
raw = self.read(dp_name)
except KeyError:
pass
else:
read_doc_props(raw, mi, self.namespace.XPath)
if mi.is_null('language'):
try:
raw = self.read('word/styles.xml')
except KeyError:
pass
else:
read_default_style_language(raw, mi, self.namespace.XPath)
ap_name = self.relationships.get(self.namespace.names['APPPROPS'], None)
if ap_name:
try:
raw = self.read(ap_name)
except KeyError:
pass
else:
read_app_props(raw, mi)
return mi
def get_metadata(book_id):
oldmi = db.get_metadata(book_id, index_is_id=True, get_cover=True, cover_as_data=True)
opf, cov = id_map[book_id]
if opf is None:
newmi = Metadata(oldmi.title, authors=tuple(oldmi.authors))
else:
with open(opf, 'rb') as f:
newmi = OPF(f, basedir=os.path.dirname(opf), populate_spine=False).to_book_metadata()
newmi.cover, newmi.cover_data = None, (None, None)
for x in ('title', 'authors'):
if newmi.is_null(x):
# Title and author are set to null if they are
# the same as the originals as an optimization,
# we undo that, as it is confusing.
newmi.set(x, copy.copy(oldmi.get(x)))
if cov:
with open(cov, 'rb') as f:
newmi.cover_data = ('jpg', f.read())
return oldmi, newmi
def metadata(self):
mi = Metadata(_('Unknown'))
name = self.relationships.get(DOCPROPS, None)
if name is None:
names = tuple(n for n in self.names
if n.lower() == 'docprops/core.xml')
if names:
name = names[0]
if name:
try:
raw = self.read(name)
except KeyError:
pass
else:
read_doc_props(raw, mi)
if mi.is_null('language'):
try:
raw = self.read('word/styles.xml')
except KeyError:
pass
else:
read_default_style_language(raw, mi)
name = self.relationships.get(APPPROPS, None)
if name is None:
names = tuple(n for n in self.names
if n.lower() == 'docprops/app.xml')
if names:
name = names[0]
if name:
try:
raw = self.read(name)
except KeyError:
pass
else:
read_app_props(raw, mi)
return mi
def metadata(self):
mi = Metadata(_('Unknown'))
name = self.relationships.get(DOCPROPS, None)
if name is None:
names = tuple(n for n in self.names if n.lower() == 'docprops/core.xml')
if names:
name = names[0]
if name:
try:
raw = self.read(name)
except KeyError:
pass
else:
read_doc_props(raw, mi)
if mi.is_null('language'):
try:
raw = self.read('word/styles.xml')
except KeyError:
pass
else:
read_default_style_language(raw, mi)
name = self.relationships.get(APPPROPS, None)
if name is None:
names = tuple(n for n in self.names if n.lower() == 'docprops/app.xml')
if names:
name = names[0]
if name:
try:
raw = self.read(name)
except KeyError:
pass
else:
read_app_props(raw, mi)
return mi
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
