def check_opf(container):
errors = []
opf_version = container.opf_version_parsed
if container.opf.tag != OPF('package'):
err = BaseError(_('The OPF does not have the correct root element'), container.opf_name, container.opf.sourceline)
err.HELP = xml(_(
'The opf must have the root element <package> in namespace {0}, like this: <package xmlns="{0}">')).format(OPF2_NS)
errors.append(err)
elif container.opf.get('version') is None and container.book_type == 'epub':
err = BaseError(_('The OPF does not have a version'), container.opf_name, container.opf.sourceline)
err.HELP = xml(_(
'The <package> tag in the OPF must have a version attribute. This is usually version="2.0" for EPUB2 and AZW3 and version="3.0" for EPUB3'))
errors.append(err)
for tag in ('metadata', 'manifest', 'spine'):
if not container.opf_xpath('/opf:package/opf:' + tag):
errors.append(MissingSection(container.opf_name, tag))
all_ids = set(container.opf_xpath('//*/@id'))
if '' in all_ids:
for empty_id_tag in container.opf_xpath('//*[@id=""]'):
errors.append(EmptyID(container.opf_name, empty_id_tag.sourceline))
all_ids.discard('')
for elem in container.opf_xpath('//*[@idref]'):
if elem.get('idref') not in all_ids:
errors.append(IncorrectIdref(container.opf_name, elem.get('idref'), elem.sourceline))
nl_items = [elem.sourceline for elem in container.opf_xpath('//opf:spine/opf:itemref[@linear="no"]')]
if nl_items:
errors.append(NonLinearItems(container.opf_name, nl_items))
seen, dups = {}, {}
for item in container.opf_xpath('/opf:package/opf:manifest/opf:item'):
href = item.get('href', None)
if href is None:
errors.append(NoHref(container.opf_name, item.get('id', None), item.sourceline))
else:
hname = container.href_to_name(href, container.opf_name)
if not hname or not container.exists(hname):
errors.append(MissingHref(container.opf_name, href, item.sourceline))
if href in seen:
if href not in dups:
dups[href] = [seen[href]]
dups[href].append(item.sourceline)
else:
seen[href] = item.sourceline
errors.extend(DuplicateHref(container.opf_name, eid, locs) for eid, locs in iteritems(dups))
seen, dups = {}, {}
for item in container.opf_xpath('/opf:package/opf:spine/opf:itemref[@idref]'):
ref = item.get('idref')
if ref in seen:
if ref not in dups:
dups[ref] = [seen[ref]]
dups[ref].append(item.sourceline)
else:
seen[ref] = item.sourceline
errors.extend(DuplicateHref(container.opf_name, eid, locs, for_spine=True) for eid, locs in iteritems(dups))
spine = container.opf_xpath('/opf:package/opf:spine[@toc]')
if spine:
spine = spine[0]
mitems = [x for x in container.opf_xpath('/opf:package/opf:manifest/opf:item[@id]') if x.get('id') == spine.get('toc')]
if mitems:
mitem = mitems[0]
if mitem.get('media-type', '') != guess_type('a.ncx'):
errors.append(IncorrectToc(container.opf_name, mitem.sourceline, bad_mimetype=mitem.get('media-type')))
else:
errors.append(IncorrectToc(container.opf_name, spine.sourceline, bad_idref=spine.get('toc')))
else:
spine = container.opf_xpath('/opf:package/opf:spine')
if spine:
spine = spine[0]
ncx = container.manifest_type_map.get(guess_type('a.ncx'))
if ncx:
ncx_name = ncx[0]
rmap = {v:k for k, v in iteritems(container.manifest_id_map)}
ncx_id = rmap.get(ncx_name)
if ncx_id:
errors.append(MissingNCXRef(container.opf_name, spine.sourceline, ncx_id))
if opf_version.major > 2:
existing_nav = find_existing_nav_toc(container)
if existing_nav is None:
errors.append(MissingNav(container.opf_name, 0))
else:
toc = parse_nav(container, existing_nav)
if len(toc) == 0:
errors.append(EmptyNav(existing_nav, 0))
covers = container.opf_xpath('/opf:package/opf:metadata/opf:meta[@name="cover"]')
if len(covers) > 0:
if len(covers) > 1:
errors.append(MultipleCovers(container.opf_name, [c.sourceline for c in covers]))
manifest_ids = set(container.opf_xpath('/opf:package/opf:manifest/opf:item/@id'))
for cover in covers:
if cover.get('content', None) not in manifest_ids:
errors.append(IncorrectCover(container.opf_name, cover.sourceline, cover.get('content', '')))
raw = etree.tostring(cover)
try:
n, c = raw.index(b'name="'), raw.index(b'content="')
except ValueError:
n = c = -1
if n > -1 and c > -1 and n > c:
errors.append(NookCover(container.opf_name, cover.sourceline))
uid = container.opf.get('unique-identifier', None)
if uid is None or not container.opf_xpath('/opf:package/opf:metadata/dc:identifier[@id=%r]' % uid):
errors.append(NoUID(container.opf_name))
for elem in container.opf_xpath('/opf:package/opf:metadata/dc:identifier'):
if not elem.text or not elem.text.strip():
errors.append(EmptyIdentifier(container.opf_name, elem.sourceline))
for item, name, linear in container.spine_iter:
mt = container.mime_map[name]
if mt != XHTML_MIME:
iid = item.get('idref', None)
lnum = None
if iid:
mitem = container.opf_xpath('/opf:package/opf:manifest/opf:item[@id=%r]' % iid)
if mitem:
lnum = mitem[0].sourceline
else:
iid = None
errors.append(BadSpineMime(name, iid, mt, lnum, container.opf_name))
return errors
def check_opf(container):
errors = []
opf_version = container.opf_version_parsed
if container.opf.tag != OPF('package'):
err = BaseError(_('The OPF does not have the correct root element'), container.opf_name, container.opf.sourceline)
err.HELP = xml(_(
'The opf must have the root element <package> in namespace {0}, like this: <package xmlns="{0}">')).format(OPF2_NS)
errors.append(err)
elif container.opf.get('version') is None and container.book_type == 'epub':
err = BaseError(_('The OPF does not have a version'), container.opf_name, container.opf.sourceline)
err.HELP = xml(_(
'The <package> tag in the OPF must have a version attribute. This is usually version="2.0" for EPUB2 and AZW3 and version="3.0" for EPUB3'))
errors.append(err)
for tag in ('metadata', 'manifest', 'spine'):
if not container.opf_xpath('/opf:package/opf:' + tag):
errors.append(MissingSection(container.opf_name, tag))
all_ids = set(container.opf_xpath('//*/@id'))
for elem in container.opf_xpath('//*[@idref]'):
if elem.get('idref') not in all_ids:
errors.append(IncorrectIdref(container.opf_name, elem.get('idref'), elem.sourceline))
nl_items = [elem.sourceline for elem in container.opf_xpath('//opf:spine/opf:itemref[@linear="no"]')]
if nl_items:
errors.append(NonLinearItems(container.opf_name, nl_items))
seen, dups = {}, {}
for item in container.opf_xpath('/opf:package/opf:manifest/opf:item'):
href = item.get('href', None)
if href is None:
errors.append(NoHref(container.opf_name, item.get('id', None), item.sourceline))
else:
hname = container.href_to_name(href, container.opf_name)
if not hname or not container.exists(hname):
errors.append(MissingHref(container.opf_name, href, item.sourceline))
if href in seen:
if href not in dups:
dups[href] = [seen[href]]
dups[href].append(item.sourceline)
else:
seen[href] = item.sourceline
errors.extend(DuplicateHref(container.opf_name, eid, locs) for eid, locs in dups.iteritems())
seen, dups = {}, {}
for item in container.opf_xpath('/opf:package/opf:spine/opf:itemref[@idref]'):
ref = item.get('idref')
if ref in seen:
if ref not in dups:
dups[ref] = [seen[ref]]
dups[ref].append(item.sourceline)
else:
seen[ref] = item.sourceline
errors.extend(DuplicateHref(container.opf_name, eid, locs, for_spine=True) for eid, locs in dups.iteritems())
spine = container.opf_xpath('/opf:package/opf:spine[@toc]')
if spine:
spine = spine[0]
mitems = [x for x in container.opf_xpath('/opf:package/opf:manifest/opf:item[@id]') if x.get('id') == spine.get('toc')]
if mitems:
mitem = mitems[0]
if mitem.get('media-type', '') != guess_type('a.ncx'):
errors.append(IncorrectToc(container.opf_name, mitem.sourceline, bad_mimetype=mitem.get('media-type')))
else:
errors.append(IncorrectToc(container.opf_name, spine.sourceline, bad_idref=spine.get('toc')))
else:
spine = container.opf_xpath('/opf:package/opf:spine')
if spine:
spine = spine[0]
ncx = container.manifest_type_map.get(guess_type('a.ncx'))
if ncx:
ncx_name = ncx[0]
rmap = {v:k for k, v in container.manifest_id_map.iteritems()}
ncx_id = rmap.get(ncx_name)
if ncx_id:
errors.append(MissingNCXRef(container.opf_name, spine.sourceline, ncx_id))
if opf_version.major > 2:
existing_nav = find_existing_nav_toc(container)
if existing_nav is None:
errors.append(MissingNav(container.opf_name, 0))
else:
toc = parse_nav(container, existing_nav)
if len(toc) == 0:
errors.append(EmptyNav(existing_nav, 0))
covers = container.opf_xpath('/opf:package/opf:metadata/opf:meta[@name="cover"]')
if len(covers) > 0:
if len(covers) > 1:
errors.append(MultipleCovers(container.opf_name, [c.sourceline for c in covers]))
manifest_ids = set(container.opf_xpath('/opf:package/opf:manifest/opf:item/@id'))
for cover in covers:
if cover.get('content', None) not in manifest_ids:
errors.append(IncorrectCover(container.opf_name, cover.sourceline, cover.get('content', '')))
raw = etree.tostring(cover)
try:
n, c = raw.index('name="'), raw.index('content="')
except ValueError:
n = c = -1
if n > -1 and c > -1 and n > c:
errors.append(NookCover(container.opf_name, cover.sourceline))
uid = container.opf.get('unique-identifier', None)
if uid is None or not container.opf_xpath('/opf:package/opf:metadata/dc:identifier[@id=%r]' % uid):
errors.append(NoUID(container.opf_name))
for elem in container.opf_xpath('/opf:package/opf:metadata/dc:identifier'):
if not elem.text or not elem.text.strip():
errors.append(EmptyIdentifier(container.opf_name, elem.sourceline))
for item, name, linear in container.spine_iter:
mt = container.mime_map[name]
if mt != XHTML_MIME:
iid = item.get('idref', None)
lnum = None
if iid:
mitem = container.opf_xpath('/opf:package/opf:manifest/opf:item[@id=%r]' % iid)
if mitem:
lnum = mitem[0].sourceline
else:
iid = None
errors.append(BadSpineMime(name, iid, mt, lnum, container.opf_name))
return errors