def get_metadata(stream, cover=True):
with TemporaryDirectory('_pdf_metadata_read') as pdfpath:
stream.seek(0)
with open(os.path.join(pdfpath, 'src.pdf'), 'wb') as f:
shutil.copyfileobj(stream, f)
info = read_info(pdfpath, bool(cover))
if info is None:
raise ValueError('Could not read info dict from PDF')
covpath = os.path.join(pdfpath, 'cover.jpg')
cdata = None
if cover and os.path.exists(covpath):
with open(covpath, 'rb') as f:
cdata = f.read()
title = info.get('Title', None) or 'Unknown'
au = info.get('Author', None)
if au is None:
au = ['Unknown']
else:
au = string_to_authors(au)
mi = MetaInformation(title, au)
creator = info.get('Creator', None)
if creator:
mi.book_producer = creator
keywords = info.get('Keywords', None)
mi.tags = []
if keywords:
mi.tags = [x.strip() for x in keywords.split(',')]
isbn = [check_isbn(x) for x in mi.tags if check_isbn(x)]
if isbn:
mi.isbn = isbn = isbn[0]
mi.tags = [x for x in mi.tags if check_isbn(x) != isbn]
subject = info.get('Subject', None)
if subject:
mi.tags.insert(0, subject)
if 'xmp_metadata' in info:
from ebook_converter.ebooks.metadata.xmp import consolidate_metadata
mi = consolidate_metadata(mi, info)
# Look for recognizable identifiers in the info dict, if they were not
# found in the XMP metadata
for scheme, check_func in {'doi': check_doi, 'isbn': check_isbn}.items():
if scheme not in mi.get_identifiers():
for k, v in info.items():
if k != 'xmp_metadata':
val = check_func(v)
if val:
mi.set_identifier(scheme, val)
break
if cdata:
mi.cover_data = ('jpeg', cdata)
return mi
def _parse_isbn(root, mi, ctx):
# some people try to put several isbn in this field, but it is not
# allowed. try to stick to the 1-st one in this case
isbn = ctx.XPath('normalize-space(//fb:publish-info/fb:isbn/text())')(root)
if isbn:
# some people try to put several isbn in this field, but it is not
# allowed. try to stick to the 1-st one in this case
if ',' in isbn:
isbn = isbn[:isbn.index(',')]
if check_isbn(isbn):
mi.isbn = isbn
def finalize(scheme, val):
if not scheme or not val:
return None, None
scheme = scheme.lower()
if scheme in ('http', 'https'):
return None, None
if scheme.startswith('isbn'):
scheme = 'isbn'
if scheme == 'isbn':
val = val.split(':')[-1]
val = check_isbn(val)
if val is None:
return None, None
return scheme, val
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 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