def test_input_meta_multi(self):
stream_meta = get_metadata(self.get_stream('meta_multi'))
canon_meta = Metadata('A Meta Tag & Title Ⓒ', ['George Washington', 'John Adams', 'Thomas Jefferson'])
canon_meta.publisher = 'Publisher A'
canon_meta.languages = ['English', 'Spanish']
canon_meta.pubdate = parse_date('2019-01-01')
canon_meta.timestamp = parse_date('2018-01-01')
canon_meta.series = 'Meta Series'
canon_meta.series_index = float(1)
canon_meta.rating = float(8)
canon_meta.comments = 'meta "comments" ♥ HTML &'
canon_meta.tags = ['tag a', 'tag b', 'tag c']
canon_meta.set_identifiers({'isbn': '1234567890', 'url': 'http://google.com/search?q=calibre'})
self.compare_metadata(stream_meta, canon_meta)
def test_input_comment_multi(self):
stream_meta = get_metadata(self.get_stream('comment_multi'))
canon_meta = Metadata('A Comment Tag & Title Ⓒ', ['James Madison', 'James Monroe', 'John Quincy Adams'])
canon_meta.publisher = 'Publisher C'
canon_meta.languages = ['French', 'Japanese']
canon_meta.pubdate = parse_date('2015-01-01')
canon_meta.timestamp = parse_date('2014-01-01')
canon_meta.series = 'Comment Series'
canon_meta.series_index = float(3)
canon_meta.rating = float(0)
canon_meta.comments = 'comment "comments" ♥ HTML -- too &'
canon_meta.tags = ['tag d', 'tag e', 'tag f']
canon_meta.set_identifiers({'isbn': '3456789012', 'url': 'http://google.com/search?q=calibre'})
self.compare_metadata(stream_meta, canon_meta)
def test_input_meta_single(self):
stream_meta = get_metadata(self.get_stream('meta_single'))
canon_meta = Metadata('A Meta Tag & Title Ⓒ', ['George Washington'])
canon_meta.publisher = 'Publisher A'
canon_meta.languages = ['English']
canon_meta.pubdate = parse_date('2019-01-01')
canon_meta.timestamp = parse_date('2018-01-01')
canon_meta.series = 'Meta Series'
canon_meta.series_index = float(1)
# canon_meta.rating = float(0)
# canon_meta.comments = ''
canon_meta.tags = ['tag a', 'tag b']
canon_meta.set_identifiers({'isbn': '1234567890'})
self.compare_metadata(stream_meta, canon_meta)
def consolidate_metadata(info_mi, info):
''' When both the PDF Info dict and XMP metadata are present, prefer the xmp
metadata unless the Info ModDate is never than the XMP MetadataDate. This
is the algorithm recommended by the PDF spec. '''
try:
raw = info['xmp_metadata'].rstrip()
if not raw:
return info_mi
xmp_mi = metadata_from_xmp_packet(raw)
except Exception:
import traceback
traceback.print_exc()
return info_mi
info_title = info_mi.title or 'Unknown'
info_authors = list(info_mi.authors or ())
info_tags = list(info_mi.tags or ())
info_mi.smart_update(xmp_mi, replace_metadata=True)
prefer_info = False
if 'ModDate' in info and hasattr(xmp_mi, 'metadata_date'):
try:
info_date = parse_date(info['ModDate'])
except Exception:
pass
else:
prefer_info = info_date > xmp_mi.metadata_date
if prefer_info:
info_mi.title = info_title
info_mi.authors = info_authors
info_mi.tags = info_tags
else:
# We'll use the xmp tags/authors but fallback to the info ones if the
# xmp does not have tags/authors. smart_update() should have taken care
# of the rest
info_mi.authors = (info_authors
if xmp_mi.is_null('authors') else xmp_mi.authors)
info_mi.tags = xmp_mi.tags or info_tags
return info_mi
def get_metadata_(src, encoding=None):
# Meta data definitions as in
# https://www.mobileread.com/forums/showpost.php?p=712544&postcount=9
if isinstance(src, bytes):
if not encoding:
src = xml_to_unicode(src)[0]
else:
src = src.decode(encoding, 'replace')
src = src[:150000] # Searching shouldn't take too long
comment_tags, meta_tags, meta_tag_ids, title_tag = parse_metadata(src)
def get_all(field):
ans = comment_tags.get(field, meta_tags.get(field, None))
if ans:
ans = [x.strip() for x in ans if x.strip()]
if not ans:
ans = None
return ans
def get(field):
ans = get_all(field)
if ans:
ans = ans[0]
return ans
# Title
title = get('title') or title_tag.strip() or 'Unknown'
# Author
authors = authors_to_string(get_all('authors')) or 'Unknown'
# Create MetaInformation with Title and Author
mi = Metadata(title, string_to_authors(authors))
# Single-value text fields
for field in ('publisher', 'isbn'):
val = get(field)
if val:
setattr(mi, field, val)
# Multi-value text fields
for field in ('languages',):
val = get_all(field)
if val:
setattr(mi, field, val)
# HTML fields
for field in ('comments',):
val = get(field)
if val:
setattr(mi, field, val.replace('&', '&').replace('<', '<').replace('>', '>').replace('"', '"').replace("'", '''))
# Date fields
for field in ('pubdate', 'timestamp'):
try:
val = parse_date(get(field))
except:
pass
else:
if not is_date_undefined(val):
setattr(mi, field, val)
# SERIES
series = get('series')
if series:
pat = re.compile(r'\[([.0-9]+)\]$')
match = pat.search(series)
series_index = None
if match is not None:
try:
series_index = float(match.group(1))
except:
pass
series = series.replace(match.group(), '').strip()
mi.series = series
if series_index is None:
series_index = get('series_index')
try:
series_index = float(series_index)
except:
pass
if series_index is not None:
mi.series_index = series_index
# RATING
rating = get('rating')
if rating:
try:
mi.rating = float(rating)
if mi.rating < 0:
mi.rating = 0
if mi.rating > 10:
mi.rating = 0
except:
pass
# TAGS
tags = get_all('tags')
if tags:
tags = [x.strip() for s in tags for x in s.split(',') if x.strip()]
if tags:
mi.tags = tags
# IDENTIFIERS
for (k,v) in meta_tag_ids.items():
v = [x.strip() for x in v if x.strip()]
if v:
mi.set_identifier(k, v[0])
return mi
def process_metadata(self, idx, content, codec):
if idx == 100:
if self.mi.is_null('authors'):
self.mi.authors = []
au = clean_xml_chars(self.decode(content).strip())
# Author names in Amazon MOBI files are usually in LN, FN format,
# try to detect and auto-correct that.
m = re.match(r'([^,]+?)\s*,\s+([^,]+)$', au.strip())
if m is not None:
if tweaks['author_sort_copy_method'] != 'copy':
self.mi.authors.append(m.group(2) + ' ' + m.group(1))
else:
self.mi.authors.append(m.group())
if self.mi.is_null('author_sort'):
self.mi.author_sort = m.group()
else:
self.mi.authors.append(au)
elif idx == 101:
self.mi.publisher = clean_xml_chars(self.decode(content).strip())
if self.mi.publisher in {'Unknown', 'Unknown'}:
self.mi.publisher = None
elif idx == 103:
self.mi.comments = clean_xml_chars(self.decode(content).strip())
elif idx == 104:
raw = check_isbn(self.decode(content).strip().replace('-', ''))
if raw:
self.mi.isbn = raw
elif idx == 105:
if not self.mi.tags:
self.mi.tags = []
self.mi.tags.extend([
x.strip()
for x in clean_xml_chars(self.decode(content)).split(';')
])
self.mi.tags = uniq(self.mi.tags)
elif idx == 106:
try:
self.mi.pubdate = parse_date(self.decode(content),
as_utc=False)
except Exception:
pass
elif idx == 108:
self.mi.book_producer = clean_xml_chars(
self.decode(content).strip())
elif idx == 109:
self.mi.rights = clean_xml_chars(self.decode(content).strip())
elif idx == 112: # dc:source set in some EBSP amazon samples
try:
content = content.decode(codec).strip()
isig = 'urn:isbn:'
if content.lower().startswith(isig):
raw = check_isbn(content[len(isig):])
if raw and not self.mi.isbn:
self.mi.isbn = raw
elif content.startswith('calibre:'):
# calibre book uuid is stored here by recent calibre
# releases
cid = content[len('calibre:'):]
if cid:
self.mi.application_id = self.mi.uuid = cid
except:
pass
elif idx == 113: # ASIN or other id
try:
self.uuid = content.decode('ascii')
self.mi.set_identifier('mobi-asin', self.uuid)
except Exception:
self.uuid = None
elif idx == 116:
self.start_offset, = struct.unpack(b'>L', content)
elif idx == 121:
self.kf8_header, = struct.unpack(b'>L', content)
if self.kf8_header == NULL_INDEX:
self.kf8_header = None
def metadata_from_xmp_packet(raw_bytes):
root = parse_xmp_packet(raw_bytes)
mi = Metadata('Unknown')
title = first_alt('//dc:title', root)
if title:
if title.startswith(r'\376\377'):
# corrupted XMP packet generated by Nitro PDF. See
# https://bugs.launchpad.net/calibre/+bug/1541981
raise ValueError('Corrupted XMP metadata packet detected, '
'probably generated by Nitro PDF')
mi.title = title
authors = multiple_sequences('//dc:creator', root)
if authors:
mi.authors = authors
tags = multiple_sequences('//dc:subject', root) or multiple_sequences(
'//pdf:Keywords', root)
if tags:
mi.tags = tags
comments = first_alt('//dc:description', root)
if comments:
mi.comments = comments
publishers = multiple_sequences('//dc:publisher', root)
if publishers:
mi.publisher = publishers[0]
try:
pubdate = (parse_date(first_sequence('//dc:date', root)
or first_simple('//xmp:CreateDate', root),
assume_utc=False))
except Exception:
pass
else:
mi.pubdate = pubdate
bkp = first_simple('//xmp:CreatorTool', root)
if bkp:
mi.book_producer = bkp
md = safe_parse_date(first_simple('//xmp:MetadataDate', root))
mod = safe_parse_date(first_simple('//xmp:ModifyDate', root))
fd = more_recent(md, mod)
if fd is not None:
mi.metadata_date = fd
rating = first_simple('//calibre:rating', root)
if rating is not None:
try:
rating = float(rating)
if 0 <= rating <= 10:
mi.rating = rating
except (ValueError, TypeError):
pass
series, series_index = read_series(root)
if series:
mi.series, mi.series_index = series, series_index
for x in ('title_sort', 'author_sort'):
for elem in XPath('//calibre:' + x)(root):
val = read_simple_property(elem)
if val:
setattr(mi, x, val)
break
for x in ('author_link_map', 'user_categories'):
val = first_simple('//calibre:' + x, root)
if val:
try:
setattr(mi, x, json.loads(val))
except Exception:
pass
languages = multiple_sequences('//dc:language', root)
if languages:
languages = list(filter(None, map(canonicalize_lang, languages)))
if languages:
mi.languages = languages
identifiers = {}
for xmpid in XPath('//xmp:Identifier')(root):
for scheme, value in read_xmp_identifers(xmpid):
if scheme and value:
identifiers[scheme.lower()] = value
for namespace in ('prism', 'pdfx'):
for scheme in KNOWN_ID_SCHEMES:
if scheme not in identifiers:
val = first_simple('//%s:%s' % (namespace, scheme), root)
scheme = scheme.lower()
if scheme == 'isbn':
val = check_isbn(val)
elif scheme == 'doi':
val = check_doi(val)
if val:
identifiers[scheme] = val
# Check Dublin Core for recognizable identifier types
for scheme, check_func in {'doi': check_doi, 'isbn': check_isbn}.items():
if scheme not in identifiers:
val = check_func(first_simple('//dc:identifier', root))
if val:
identifiers['doi'] = val
if identifiers:
mi.set_identifiers(identifiers)
read_user_metadata(mi, root)
return mi
def safe_parse_date(raw):
if raw:
try:
return parse_date(raw)
except Exception:
pass
def get_metadata(stream, extract_cover=True):
whitespace = re.compile(r'\s+')
def normalize(s):
return whitespace.sub(' ', s).strip()
with ZipFile(stream) as zf:
meta = zf.read('meta.xml')
root = fromstring(meta)
def find(field):
ns, tag = fields[field]
ans = root.xpath('//ns0:{}'.format(tag), namespaces={'ns0': ns})
if ans:
return normalize(
tostring(ans[0],
method='text',
encoding='unicode',
with_tail=False)).strip()
mi = MetaInformation(None, [])
title = find('title')
if title:
mi.title = title
creator = find('initial-creator') or find('creator')
if creator:
mi.authors = string_to_authors(creator)
desc = find('description')
if desc:
mi.comments = desc
lang = find('language')
if lang and canonicalize_lang(lang):
mi.languages = [canonicalize_lang(lang)]
kw = find('keyword') or find('keywords')
if kw:
mi.tags = [x.strip() for x in kw.split(',') if x.strip()]
data = {}
for tag in root.xpath('//ns0:user-defined',
namespaces={'ns0': fields['user-defined'][0]}):
name = (tag.get('{%s}name' % METANS) or '').lower()
vtype = tag.get('{%s}value-type' % METANS) or 'string'
val = tag.text
if name and val:
if vtype == 'boolean':
val = val == 'true'
data[name] = val
opfmeta = False # we need this later for the cover
opfnocover = False
if data.get('opf.metadata'):
# custom metadata contains OPF information
opfmeta = True
if data.get('opf.titlesort', ''):
mi.title_sort = data['opf.titlesort']
if data.get('opf.authors', ''):
mi.authors = string_to_authors(data['opf.authors'])
if data.get('opf.authorsort', ''):
mi.author_sort = data['opf.authorsort']
if data.get('opf.isbn', ''):
isbn = check_isbn(data['opf.isbn'])
if isbn is not None:
mi.isbn = isbn
if data.get('opf.publisher', ''):
mi.publisher = data['opf.publisher']
if data.get('opf.pubdate', ''):
mi.pubdate = parse_date(data['opf.pubdate'], assume_utc=True)
if data.get('opf.identifiers'):
try:
mi.identifiers = json.loads(data['opf.identifiers'])
except Exception:
pass
if data.get('opf.rating'):
try:
mi.rating = max(0, min(float(data['opf.rating']), 10))
except Exception:
pass
if data.get('opf.series', ''):
mi.series = data['opf.series']
if data.get('opf.seriesindex', ''):
try:
mi.series_index = float(data['opf.seriesindex'])
except Exception:
mi.series_index = 1.0
if data.get('opf.language', ''):
cl = canonicalize_lang(data['opf.language'])
if cl:
mi.languages = [cl]
opfnocover = data.get('opf.nocover', False)
if not opfnocover:
try:
read_cover(stream, zf, mi, opfmeta, extract_cover)
except Exception:
pass # Do not let an error reading the cover prevent reading other data
return mi