X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/6ec15224f3482ea18d339df7e2acc160d799be43..7afa41f2cd2ae1c240f15e147ebacb9fb03bb6ca:/librarian/dcparser.py diff --git a/librarian/dcparser.py b/librarian/dcparser.py index cca9c6f..7a2a69e 100644 --- a/librarian/dcparser.py +++ b/librarian/dcparser.py @@ -1,155 +1,272 @@ # -*- coding: utf-8 -*- # # This file is part of Librarian, licensed under GNU Affero GPLv3 or later. -# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. +# Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information. # +from __future__ import unicode_literals + from xml.parsers.expat import ExpatError from datetime import date +from functools import total_ordering import time +import re +import six +from librarian.util import roman_to_int -from librarian import ValidationError, NoDublinCore, ParseError, DCNS, RDFNS +from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS, + XMLNS, WLURI, WLNS, PLMETNS) -import lxml.etree as etree # ElementTree API using libxml2 +import lxml.etree as etree # ElementTree API using libxml2 from lxml.etree import XMLSyntaxError +class TextPlus(six.text_type): + pass + + +class DatePlus(date): + pass + + # ============== # = Converters = # ============== +@six.python_2_unicode_compatible +@total_ordering class Person(object): """Single person with last name and a list of first names.""" def __init__(self, last_name, *first_names): self.last_name = last_name self.first_names = first_names - + @classmethod def from_text(cls, text): - parts = [ token.strip() for token in text.split(',') ] + parts = [token.strip() for token in text.split(',')] if len(parts) == 1: surname = parts[0] names = [] elif len(parts) != 2: - raise ValueError("Invalid person name. There should be at most one comma: \"%s\"." % text) + raise ValueError("Invalid person name. There should be at most one comma: \"%s\"." % text.encode('utf-8')) else: surname = parts[0] if len(parts[1]) == 0: # there is no non-whitespace data after the comma raise ValueError("Found a comma, but no names given: \"%s\" -> %r." % (text, parts)) - names = [ name for name in parts[1].split() if len(name) ] # all non-whitespace tokens + names = parts[1].split() return cls(surname, *names) - + + def readable(self): + return u" ".join(self.first_names + (self.last_name,)) + def __eq__(self, right): return self.last_name == right.last_name and self.first_names == right.first_names - - - def __unicode__(self): + + def __lt__(self, other): + return (self.last_name, self.first_names) < (other.last_name, other.first_names) + + def __hash__(self): + return hash((self.last_name, self.first_names)) + + def __str__(self): if len(self.first_names) > 0: return '%s, %s' % (self.last_name, ' '.join(self.first_names)) else: return self.last_name - + def __repr__(self): return 'Person(last_name=%r, first_names=*%r)' % (self.last_name, self.first_names) + def as_date(text): + """Dates for digitization of pictures. It seems we need the following: +ranges: '1350-1450', +centuries: "XVIII w.' +half centuries/decades: '2 poł. XVIII w.', 'XVII w., l. 20' +later-then: 'po 1450' +circa 'ok. 1813-1814', 'ok.1876-ok.1886 +turn: 1893/1894 +for now we will translate this to some single date losing information of course. + """ try: - try: - t = time.strptime(text, '%Y-%m-%d') - except ValueError: - t = time.strptime(text, '%Y') - return date(t[0], t[1], t[2]) - except ValueError, e: + # check out the "N. poł X w." syntax + if isinstance(text, six.binary_type): + text = text.decode("utf-8") + + century_format = u"(?:([12]) *poł[.]? +)?([MCDXVI]+) *w[.,]*(?: *l[.]? *([0-9]+))?" + vague_format = u"(?:po *|ok. *)?([0-9]{4})(-[0-9]{2}-[0-9]{2})?" + + m = re.match(century_format, text) + m2 = re.match(vague_format, text) + if m: + half = m.group(1) + decade = m.group(3) + century = roman_to_int(m.group(2)) + if half is not None: + if decade is not None: + raise ValueError("Bad date format. Cannot specify both half and decade of century") + half = int(half) + t = ((century*100 + (half-1)*50), 1, 1) + else: + decade = int(decade or 0) + t = ((century*100 + decade), 1, 1) + elif m2: + year = m2.group(1) + mon_day = m2.group(2) + if mon_day: + t = time.strptime(year + mon_day, "%Y-%m-%d") + else: + t = time.strptime(year, '%Y') + else: + raise ValueError + + return DatePlus(t[0], t[1], t[2]) + except ValueError as e: raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.") + def as_person(text): return Person.from_text(text) + def as_unicode(text): - if isinstance(text, unicode): + if isinstance(text, six.text_type): return text else: - return text.decode('utf-8') + return TextPlus(text.decode('utf-8')) + + +def as_wluri_strict(text): + return WLURI.strict(text) + class Field(object): - def __init__(self, uri, attr_name, type=as_unicode, multiple=False, salias=None, **kwargs): + def __init__(self, uri, attr_name, validator=as_unicode, strict=None, multiple=False, salias=None, **kwargs): self.uri = uri self.name = attr_name - self.validator = type + self.validator = validator + self.strict = strict self.multiple = multiple self.salias = salias - self.required = kwargs.get('required', True) and not kwargs.has_key('default') + self.required = kwargs.get('required', True) and 'default' not in kwargs self.default = kwargs.get('default', [] if multiple else [None]) - def validate_value(self, val): + def validate_value(self, val, strict=False): + if strict and self.strict is not None: + validator = self.strict + else: + validator = self.validator try: if self.multiple: - if self.validator is None: + if validator is None: return val - return [ self.validator(v) if v is not None else v for v in val ] + new_values = [] + for v in val: + nv = v + if v is not None: + nv = validator(v) + if hasattr(v, 'lang'): + setattr(nv, 'lang', v.lang) + new_values.append(nv) + return new_values elif len(val) > 1: - raise ValidationError("Mulitply values not allowed for field '%s'" % self.uri) + raise ValidationError("Multiple values not allowed for field '%s'" % self.uri) elif len(val) == 0: raise ValidationError("Field %s has no value to assign. Check your defaults." % self.uri) else: - if self.validator is None or val[0] is None: + if validator is None or val[0] is None: return val[0] - return self.validator(val[0]) - except ValueError, e: + nv = validator(val[0]) + if hasattr(val[0], 'lang'): + setattr(nv, 'lang', val[0].lang) + return nv + except ValueError as e: raise ValidationError("Field '%s' - invald value: %s" % (self.uri, e.message)) - def validate(self, fdict): - if not fdict.has_key(self.uri): + def validate(self, fdict, fallbacks=None, strict=False): + if fallbacks is None: + fallbacks = {} + if self.uri not in fdict: if not self.required: - f = self.default + # Accept single value for single fields and saliases. + if self.name in fallbacks: + if self.multiple: + f = fallbacks[self.name] + else: + f = [fallbacks[self.name]] + elif self.salias and self.salias in fallbacks: + f = [fallbacks[self.salias]] + else: + f = self.default else: raise ValidationError("Required field %s not found" % self.uri) else: f = fdict[self.uri] - return self.validate_value(f) + return self.validate_value(f, strict=strict) + + def __eq__(self, other): + if isinstance(other, Field) and other.name == self.name: + return True + return False + +class DCInfo(type): + def __new__(mcs, 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, mcs).__new__(mcs, classname, bases, class_dict) -class BookInfo(object): + +class WorkInfo(six.with_metaclass(DCInfo, object)): FIELDS = ( - Field( DCNS('creator'), 'author', as_person), - 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('date'), 'created_at', as_date), - Field( DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False), - 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('publisher'), 'publisher'), - Field( DCNS('source'), 'source_name', required=False), - Field( DCNS('source.URL'), 'source_url', required=False), - Field( DCNS('identifier.url'), 'url'), - Field( DCNS('relation.hasPart'), 'parts', multiple=True, required=False), - Field( DCNS('rights.license'), 'license', required=False), - Field( DCNS('rights'), 'license_description'), + Field(DCNS('creator'), 'authors', as_person, salias='author', multiple=True), + Field(DCNS('title'), 'title'), + Field(DCNS('type'), 'type', required=False, multiple=True), + + Field(DCNS('contributor.editor'), 'editors', + as_person, salias='editor', multiple=True, required=False), + Field(DCNS('contributor.technical_editor'), 'technical_editors', + as_person, salias='technical_editor', multiple=True, required=False), + Field(DCNS('contributor.funding'), 'funders', salias='funder', multiple=True, required=False), + Field(DCNS('contributor.thanks'), 'thanks', required=False), + + Field(DCNS('date'), 'created_at'), + Field(DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False), + Field(DCNS('publisher'), 'publisher', multiple=True), + + 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, strict=as_wluri_strict), + Field(DCNS('rights.license'), 'license', required=False), + Field(DCNS('rights'), 'license_description'), + + Field(PLMETNS('digitisationSponsor'), 'sponsors', multiple=True, required=False), + Field(WLNS('digitisationSponsorNote'), 'sponsor_note', required=False), + Field(WLNS('developmentStage'), 'stage', required=False), ) - - @property - def slug(self): - return self.url.rsplit('/', 1)[1] @classmethod - def from_string(cls, xml): - from StringIO import StringIO - return cls.from_file(StringIO(xml)) - + def from_bytes(cls, xml, *args, **kwargs): + return cls.from_file(six.BytesIO(xml), *args, **kwargs) + @classmethod - def from_file(cls, xmlfile): - desc_tag = None + def from_file(cls, xmlfile, *args, **kwargs): + desc_tag = None try: - iter = etree.iterparse(xmlfile, ['start', 'end']) + iter = etree.iterparse(xmlfile, ['start', 'end']) for (event, element) in iter: if element.tag == RDFNS('RDF') and event == 'start': desc_tag = element @@ -165,43 +282,62 @@ class BookInfo(object): break # if there is no end, Expat should yell at us with an ExpatError - + # extract data from the element and make the info - return cls.from_element(desc_tag) - except XMLSyntaxError, e: + return cls.from_element(desc_tag, *args, **kwargs) + except XMLSyntaxError as e: raise ParseError(e) - except ExpatError, e: + except ExpatError as e: raise ParseError(e) @classmethod - def from_element(cls, rdf_tag): + def from_element(cls, rdf_tag, *args, **kwargs): # 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.") + lang = None + p = desc + while p is not None and lang is None: + lang = p.attrib.get(XMLNS('lang')) + p = p.getparent() + for e in desc.getchildren(): fv = field_dict.get(e.tag, []) - fv.append(e.text) + if e.text is not None: + text = e.text + if not isinstance(text, six.text_type): + text = text.decode('utf-8') + val = TextPlus(text) + val.lang = e.attrib.get(XMLNS('lang'), lang) + if e.tag == 'meta': + meta_id = e.attrib.get('id') + if meta_id and meta_id.endswith('-id'): + field_dict[meta_id] = [val.replace('ISBN-', 'ISBN ')] + else: + val = e.text + fv.append(val) field_dict[e.tag] = fv - - return cls( desc.attrib, field_dict ) - def __init__(self, rdf_attrs, dc_fields): + return cls(desc.attrib, field_dict, *args, **kwargs) + + def __init__(self, rdf_attrs, dc_fields, fallbacks=None, strict=False): """rdf_attrs should be a dictionary-like object with any attributes of the RDF:Description. - dc_fields - dictionary mapping DC fields (with namespace) to list of text values for the + dc_fields - dictionary mapping DC fields (with namespace) to list of text values for the given field. """ self.about = rdf_attrs.get(RDFNS('about')) self.fmap = {} for field in self.FIELDS: - value = field.validate( dc_fields ) + value = field.validate(dc_fields, fallbacks=fallbacks, strict=strict) setattr(self, 'prop_' + field.name, value) self.fmap[field.name] = field - if field.salias: self.fmap[field.salias] = field + if field.salias: + self.fmap[field.salias] = field def __getattribute__(self, name): try: @@ -209,11 +345,11 @@ class BookInfo(object): value = object.__getattribute__(self, 'prop_'+field.name) if field.name == name: return value - else: # singular alias + else: # singular alias if not field.multiple: raise "OUCH!! for field %s" % name - - return value[0] + + return value[0] if value else None except (KeyError, AttributeError): return object.__getattribute__(self, name) @@ -222,7 +358,7 @@ class BookInfo(object): field = object.__getattribute__(self, 'fmap')[name] if field.name == name: object.__setattr__(self, 'prop_'+field.name, newvalue) - else: # singular alias + else: # singular alias if not field.multiple: raise "OUCH! while setting field %s" % name @@ -231,58 +367,59 @@ class BookInfo(object): return object.__setattr__(self, name, newvalue) def update(self, field_dict): - """Update using field_dict. Verify correctness, but don't check if all + """Update using field_dict. Verify correctness, but don't check if all required fields are present.""" for field in self.FIELDS: - if field_dict.has_key(field.name): + if field.name in field_dict: setattr(self, field.name, field_dict[field.name]) - def to_etree(self, parent = None): + def to_etree(self, parent=None): """XML representation of this object.""" - #etree._namespace_map[str(self.RDF)] = 'rdf' - #etree._namespace_map[str(self.DC)] = 'dc' - + # etree._namespace_map[str(self.RDF)] = 'rdf' + # etree._namespace_map[str(self.DC)] = 'dc' + if parent is None: root = etree.Element(RDFNS('RDF')) else: root = parent.makeelement(RDFNS('RDF')) description = etree.SubElement(root, RDFNS('Description')) - + if self.about: description.set(RDFNS('about'), self.about) - + for field in self.FIELDS: v = getattr(self, field.name, None) if v is not None: if field.multiple: - if len(v) == 0: continue + if len(v) == 0: + continue for x in v: e = etree.Element(field.uri) - e.text = unicode(x) + if x is not None: + e.text = six.text_type(x) description.append(e) else: e = etree.Element(field.uri) - e.text = unicode(v) + e.text = six.text_type(v) description.append(e) - - return root + return root def serialize(self): - rdf = {} - rdf['about'] = { 'uri': RDFNS('about'), 'value': self.about } + rdf = {'about': {'uri': RDFNS('about'), 'value': self.about}} dc = {} for field in self.FIELDS: v = getattr(self, field.name, None) if v is not None: if field.multiple: - if len(v) == 0: continue - v = [ unicode(x) for x in v if v is not None ] + if len(v) == 0: + continue + v = [six.text_type(x) for x in v if x is not None] else: - v = unicode(v) - + v = six.text_type(v) + dc[field.name] = {'uri': field.uri, 'value': v} rdf['fields'] = dc return rdf @@ -294,17 +431,52 @@ class BookInfo(object): if v is not None: if field.multiple: - if len(v) == 0: continue - v = [ unicode(x) for x in v if v is not None ] + if len(v) == 0: + continue + v = [six.text_type(x) for x in v if x is not None] else: - v = unicode(v) + v = six.text_type(v) result[field.name] = v if field.salias: v = getattr(self, field.salias) - if v is not None: result[field.salias] = unicode(v) - + if v is not None: + result[field.salias] = six.text_type(v) + 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, required=False), + Field(DCNS('subject.type'), 'kinds', salias='kind', multiple=True, required=False), + Field(DCNS('subject.genre'), 'genres', salias='genre', multiple=True, required=False), + + Field(DCNS('subject.location'), 'location', required=False), + + Field(DCNS('contributor.translator'), 'translators', + as_person, salias='translator', multiple=True, required=False), + Field(DCNS('relation.hasPart'), 'parts', WLURI, strict=as_wluri_strict, multiple=True, required=False), + Field(DCNS('relation.isVariantOf'), 'variant_of', WLURI, strict=as_wluri_strict, required=False), + + Field(DCNS('relation.coverImage.url'), 'cover_url', required=False), + Field(DCNS('relation.coverImage.attribution'), 'cover_by', required=False), + Field(DCNS('relation.coverImage.source'), 'cover_source', required=False), + # WLCover-specific. + Field(WLNS('coverBarColor'), 'cover_bar_color', required=False), + Field(WLNS('coverBoxPosition'), 'cover_box_position', required=False), + Field(WLNS('coverClass'), 'cover_class', default=['default']), + Field(WLNS('coverLogoUrl'), 'cover_logo_urls', multiple=True, required=False), + + Field('pdf-id', 'isbn_pdf', required=False), + Field('epub-id', 'isbn_epub', required=False), + Field('mobi-id', 'isbn_mobi', required=False), + Field('txt-id', 'isbn_txt', required=False), + Field('html-id', 'isbn_html', required=False), + ) + + +def parse(file_name, cls=BookInfo): + return cls.from_file(file_name)