X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/918a554570309fdddb8158b963fef3b3ae3c2c2b..5b407667ca47cf4d9752821fd49e5611737146d2:/librarian/dcparser.py diff --git a/librarian/dcparser.py b/librarian/dcparser.py index ae8466c..d99aaf0 100644 --- a/librarian/dcparser.py +++ b/librarian/dcparser.py @@ -118,35 +118,53 @@ class Field(object): return self.validate_value(f) + def __eq__(self, other): + if isinstance(other, Field) and other.name == self.name: + return True + return False +class DCInfo(type): + def __new__(meta, classname, bases, class_dict): + fields = list(class_dict['FIELDS']) + + for base in bases[::-1]: + if hasattr(base, 'FIELDS'): + for field in base.FIELDS[::-1]: + try: + fields.index(field) + except ValueError: + fields.insert(0, field) + + class_dict['FIELDS'] = tuple(fields) + return super(DCInfo, meta).__new__(meta, classname, bases, class_dict) + + +class WorkInfo(object): + __metaclass__ = DCInfo -class BookInfo(object): FIELDS = ( - Field( DCNS('creator'), 'author', as_person), + Field( DCNS('creator'), 'authors', as_person, salias='author', multiple=True), Field( DCNS('title'), 'title'), - Field( DCNS('subject.period'), 'epochs', salias='epoch', multiple=True), - Field( DCNS('subject.type'), 'kinds', salias='kind', multiple=True), - Field( DCNS('subject.genre'), 'genres', salias='genre', multiple=True), - Field( DCNS('audience'), 'audiences', salias='audience', multiple=True, - required=False), - Field( DCNS('date'), 'created_at', as_date), - Field( DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False), + Field( DCNS('type'), 'type', required=False, multiple=True), + Field( DCNS('contributor.editor'), 'editors', \ as_person, salias='editor', multiple=True, default=[]), - Field( DCNS('contributor.translator'), 'translators', \ - as_person, salias='translator', multiple=True, default=[]), Field( DCNS('contributor.technical_editor'), 'technical_editors', as_person, salias='technical_editor', multiple=True, default=[]), + + Field( DCNS('date'), 'created_at', as_date), + Field( DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False), Field( DCNS('publisher'), 'publisher'), + + Field( DCNS('language'), 'language'), + Field( DCNS('description'), 'description', required=False), + Field( DCNS('source'), 'source_name', required=False), Field( DCNS('source.URL'), 'source_url', required=False), Field( DCNS('identifier.url'), 'url', WLURI), - Field( DCNS('relation.hasPart'), 'parts', multiple=True, required=False), Field( DCNS('rights.license'), 'license', required=False), Field( DCNS('rights'), 'license_description'), - Field( DCNS('language'), 'language'), - Field( DCNS('description'), 'description', required=False), ) @classmethod @@ -186,7 +204,7 @@ class BookInfo(object): def from_element(cls, rdf_tag): # the tree is already parsed, so we don't need to worry about Expat errors field_dict = {} - desc = rdf_tag.find(".//" + RDFNS('Description') ) + desc = rdf_tag.find(".//" + RDFNS('Description')) if desc is None: raise NoDublinCore("No DublinCore section found.") @@ -196,7 +214,7 @@ class BookInfo(object): fv.append(e.text) field_dict[e.tag] = fv - return cls( desc.attrib, field_dict ) + return cls(desc.attrib, field_dict) def __init__(self, rdf_attrs, dc_fields): """rdf_attrs should be a dictionary-like object with any attributes of the RDF:Description. @@ -283,7 +301,6 @@ class BookInfo(object): return root - def serialize(self): rdf = {} rdf['about'] = { 'uri': RDFNS('about'), 'value': self.about } @@ -321,5 +338,25 @@ class BookInfo(object): return result -def parse(file_name): - return BookInfo.from_file(file_name) + +class BookInfo(WorkInfo): + FIELDS = ( + Field( DCNS('audience'), 'audiences', salias='audience', multiple=True, + required=False), + + Field( DCNS('subject.period'), 'epochs', salias='epoch', multiple=True), + Field( DCNS('subject.type'), 'kinds', salias='kind', multiple=True), + Field( DCNS('subject.genre'), 'genres', salias='genre', multiple=True), + + Field( DCNS('contributor.translator'), 'translators', \ + as_person, salias='translator', multiple=True, default=[]), + Field( DCNS('relation.hasPart'), 'parts', WLURI, multiple=True, required=False), + + Field( DCNS('relation.cover_image.url'), 'cover_url', required=False), + Field( DCNS('relation.cover_image.attribution'), 'cover_by', required=False), + Field( DCNS('relation.cover_image.source'), 'cover_source', required=False), + ) + + +def parse(file_name, cls=BookInfo): + return cls.from_file(file_name)