X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/1f9c252b0da609f7dff02ab74b60f39f57dd2e3d..ac9ba215113d29655e142f23ab398c5c2765df1e:/src/librarian/dcparser.py diff --git a/src/librarian/dcparser.py b/src/librarian/dcparser.py index f402c04..910f5e1 100644 --- a/src/librarian/dcparser.py +++ b/src/librarian/dcparser.py @@ -1,11 +1,11 @@ # This file is part of Librarian, licensed under GNU Affero GPLv3 or later. -# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. +# Copyright © Fundacja Wolne Lektury. See NOTICE for more information. # from xml.parsers.expat import ExpatError from datetime import date +import io import time import re -import six from librarian.util import roman_to_int from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS, @@ -15,14 +15,13 @@ import lxml.etree as etree from lxml.etree import XMLSyntaxError from librarian.meta.types.bool import BoolValue -from librarian.meta.types.date import DateValue from librarian.meta.types.person import Person from librarian.meta.types.wluri import WLURI -from librarian.meta.types.text import TextValue +from librarian.meta.types import text -class Field(object): - def __init__(self, uri, attr_name, value_type=TextValue, +class Field: + def __init__(self, uri, attr_name, value_type=text.TextValue, multiple=False, salias=None, **kwargs): self.uri = uri self.name = attr_name @@ -104,7 +103,7 @@ class DCInfo(type): return super(DCInfo, mcs).__new__(mcs, classname, bases, class_dict) -class WorkInfo(six.with_metaclass(DCInfo, object)): +class WorkInfo(metaclass=DCInfo): FIELDS = ( Field(DCNS('creator'), 'authors', Person, salias='author', multiple=True), @@ -121,7 +120,7 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): Field(DCNS('contributor.thanks'), 'thanks', required=False), Field(DCNS('date'), 'created_at'), - Field(DCNS('date.pd'), 'released_to_public_domain_at', DateValue, + Field(DCNS('date.pd'), 'released_to_public_domain_at', required=False), Field(DCNS('publisher'), 'publisher', multiple=True), @@ -151,7 +150,7 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): @classmethod def from_bytes(cls, xml, *args, **kwargs): - return cls.from_file(six.BytesIO(xml), *args, **kwargs) + return cls.from_file(io.BytesIO(xml), *args, **kwargs) @classmethod def from_file(cls, xmlfile, *args, **kwargs): @@ -237,7 +236,6 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): for field in self.FIELDS: value = field.validate(dc_fields, fallbacks=fallbacks, strict=strict, validate_required=validate_required) - print(field.name, value) setattr(self, 'prop_' + field.name, value) self.fmap[field.name] = field if field.salias: @@ -303,11 +301,11 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): for x in v: e = etree.Element(field.uri) if x is not None: - e.text = six.text_type(x) + e.text = str(x) description.append(e) else: e = etree.Element(field.uri) - e.text = six.text_type(v) + e.text = str(v) description.append(e) return root @@ -322,9 +320,9 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): if field.multiple: if len(v) == 0: continue - v = [six.text_type(x) for x in v if x is not None] + v = [str(x) for x in v if x is not None] else: - v = six.text_type(v) + v = str(v) dc[field.name] = {'uri': field.uri, 'value': v} rdf['fields'] = dc @@ -339,32 +337,33 @@ class WorkInfo(six.with_metaclass(DCInfo, object)): if field.multiple: if len(v) == 0: continue - v = [six.text_type(x) for x in v if x is not None] + v = [str(x) for x in v if x is not None] else: - v = six.text_type(v) + v = str(v) result[field.name] = v if field.salias: v = getattr(self, field.salias) if v is not None: - result[field.salias] = six.text_type(v) + result[field.salias] = str(v) return result class BookInfo(WorkInfo): FIELDS = ( - Field(DCNS('audience'), 'audiences', salias='audience', multiple=True, + Field(DCNS('audience'), 'audiences', text.Audience, salias='audience', multiple=True, required=False), - Field(DCNS('subject.period'), 'epochs', salias='epoch', multiple=True, + Field(DCNS('subject.period'), 'epochs', text.Epoch, salias='epoch', multiple=True, required=False), - Field(DCNS('subject.type'), 'kinds', salias='kind', multiple=True, + Field(DCNS('subject.type'), 'kinds', text.Kind, salias='kind', multiple=True, required=False), - Field(DCNS('subject.genre'), 'genres', salias='genre', multiple=True, + Field(DCNS('subject.genre'), 'genres', text.Genre, salias='genre', multiple=True, required=False), - Field(WLNS('category.legimi'), 'legimi', required=False), - + Field('category.legimi', 'legimi', text.LegimiCategory, required=False), + Field('category.thema.main', 'thema_main', text.MainThemaCategory, required=False), + Field('category.thema', 'thema', text.ThemaCategory, required=False, multiple=True), Field(DCNS('subject.location'), 'location', required=False), Field(DCNS('contributor.translator'), 'translators',