X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/07fdba2c7fe8e11b6867712d47bdd608e88c29fb..2363bf73e1728ae729c13ec9db43298425badeb6:/librarian/dcparser.py diff --git a/librarian/dcparser.py b/librarian/dcparser.py index 557509c..a7215a1 100644 --- a/librarian/dcparser.py +++ b/librarian/dcparser.py @@ -1,16 +1,26 @@ # -*- 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. +# from xml.parsers.expat import ExpatError from datetime import date import time +import re +from librarian.util import roman_to_int -# Import ElementTree from anywhere -try: - import xml.etree.ElementTree as etree # Python >= 2.5 -except ImportError: - try: - import elementtree.ElementTree as etree # effbot's pure Python module - except ImportError: - import lxml.etree as etree # ElementTree API using libxml2 +from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS, + XMLNS, WLURI) + +import lxml.etree as etree # ElementTree API using libxml2 +from lxml.etree import XMLSyntaxError + + +class TextPlus(unicode): + pass + +class DatePlus(date): + pass # ============== @@ -21,177 +31,425 @@ class Person(object): 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(',') ] + 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) + 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 + 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 __cmp__(self, other): + return cmp((self.last_name, self.first_names), (other.last_name, other.first_names)) + + def __hash__(self): + return hash((self.last_name, self.first_names)) + def __unicode__(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: + # check out the "N. poł X w." syntax + if isinstance(text, str): text = text.decode("utf-8") -def str_to_unicode(value, previous): - return unicode(value) + 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(str(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 -def str_to_unicode_list(value, previous): - if previous is None: - previous = [] - previous.append(str_to_unicode(value, None)) - return previous + return DatePlus(t[0], t[1], t[2]) + except ValueError, e: + raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.") +def as_person(text): + return Person.from_text(text) -def str_to_person(value, previous): - comma_count = value.count(',') - - if comma_count == 0: - last_name, first_names = value, [] - elif comma_count == 1: - last_name, first_names = value.split(',') - first_names = [name for name in first_names.split(' ') if len(name)] +def as_unicode(text): + if isinstance(text, unicode): + return text else: - raise ValueError("value contains more than one comma: %r" % value) - - return Person(last_name.strip(), *first_names) + return TextPlus(text.decode('utf-8')) +def as_wluri_strict(text): + return WLURI.strict(text) -def str_to_date(value, previous): - try: - t = time.strptime(value, '%Y-%m-%d') - except ValueError: - t = time.strptime(value, '%Y') - return date(t[0], t[1], t[2]) +class Field(object): + 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 = validator + self.strict = strict + self.multiple = multiple + self.salias = salias + + self.required = kwargs.get('required', True) and not kwargs.has_key('default') + self.default = kwargs.get('default', [] if multiple else [None]) + + 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 validator is None: + return 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("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 validator is None or val[0] is None: + return val[0] + nv = validator(val[0]) + if hasattr(val[0], 'lang'): + setattr(nv, 'lang', val[0].lang) + return nv + except ValueError, e: + raise ValidationError("Field '%s' - invald value: %s" % (self.uri, e.message)) + def validate(self, fdict, fallbacks=None, strict=False): + if fallbacks is None: + fallbacks = {} + if not fdict.has_key(self.uri): + if not self.required: + # 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] -# ========== -# = Parser = -# ========== -class ParseError(Exception): - def __init__(self, message): - super(ParseError, self).__init__(message) + 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 XMLNamespace(object): - '''Represents XML namespace.''' - - def __init__(self, uri): - self.uri = uri - def __call__(self, tag): - return '{%s}%s' % (self.uri, tag) +class DCInfo(type): + def __new__(meta, classname, bases, class_dict): + fields = list(class_dict['FIELDS']) - def __contains__(self, tag): - return tag.startswith(str(self)) + 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) - def __repr__(self): - return 'XMLNamespace(%r)' % self.uri - - def __str__(self): - return '%s' % self.uri - - -class BookInfo(object): - RDF = XMLNamespace('http://www.w3.org/1999/02/22-rdf-syntax-ns#') - DC = XMLNamespace('http://purl.org/dc/elements/1.1/') - - mapping = { - DC('creator') : ('author', str_to_person), - DC('title') : ('title', str_to_unicode), - DC('subject.period') : ('epoch', str_to_unicode), - DC('subject.type') : ('kind', str_to_unicode), - DC('subject.genre') : ('genre', str_to_unicode), - DC('date') : ('created_at', str_to_date), - DC('date.pd') : ('released_to_public_domain_at', str_to_date), - DC('contributor.translator') : ('translator', str_to_person), - DC('contributor.technical_editor') : ('technical_editor', str_to_person), - DC('publisher') : ('publisher', str_to_unicode), - DC('source') : ('source_name', str_to_unicode), - DC('source.URL') : ('source_url', str_to_unicode), - DC('identifier.url') : ('url', str_to_unicode), - DC('relation.hasPart') : ('parts', str_to_unicode_list), - DC('rights.license') : ('license', str_to_unicode), - DC('rights') : ('license_description', str_to_unicode), - } + class_dict['FIELDS'] = tuple(fields) + return super(DCInfo, meta).__new__(meta, classname, bases, class_dict) + + +class WorkInfo(object): + __metaclass__ = DCInfo + + FIELDS = ( + 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, default=[]), + Field( DCNS('contributor.technical_editor'), 'technical_editors', + as_person, salias='technical_editor', multiple=True, default=[]), + Field( DCNS('contributor.funding'), 'funders', + salias='funder', multiple=True, default=[]), + Field( DCNS('contributor.thanks'), 'thanks', required=False), + + 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, strict=as_wluri_strict), + Field( DCNS('rights.license'), 'license', required=False), + Field( DCNS('rights'), 'license_description'), + ) @classmethod - def from_string(cls, xml): + def from_string(cls, xml, *args, **kwargs): from StringIO import StringIO - return cls.from_file(StringIO(xml)) - + return cls.from_file(StringIO(xml), *args, **kwargs) + @classmethod - def from_file(cls, xml_file): - book_info = cls() - + def from_file(cls, xmlfile, *args, **kwargs): + desc_tag = None try: - tree = etree.parse(xml_file) + iter = etree.iterparse(xmlfile, ['start', 'end']) + for (event, element) in iter: + if element.tag == RDFNS('RDF') and event == 'start': + desc_tag = element + break + + if desc_tag is None: + raise NoDublinCore("DublinCore section not found. \ + Check if there are rdf:RDF and rdf:Description tags.") + + # continue 'till the end of RDF section + for (event, element) in iter: + if element.tag == RDFNS('RDF') and event == 'end': + 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, *args, **kwargs) + except XMLSyntaxError, e: + raise ParseError(e) except ExpatError, e: raise ParseError(e) - description = tree.find('//' + book_info.RDF('Description')) - book_info.wiki_url = description.get(cls.RDF('about'), None) - - if description is None: - raise ParseError('no Description tag found in document') - - for element in description.findall('*'): - book_info.parse_element(element) - - return book_info - - def parse_element(self, element): + @classmethod + 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')) + + 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, []) + if e.text is not None: + text = e.text + if not isinstance(text, unicode): + text = text.decode('utf-8') + val = TextPlus(text) + val.lang = e.attrib.get(XMLNS('lang'), lang) + else: + val = e.text + fv.append(val) + field_dict[e.tag] = fv + + 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 + given field. """ + + self.about = rdf_attrs.get(RDFNS('about')) + self.fmap = {} + + for field in self.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 + + def __getattribute__(self, name): + try: + field = object.__getattribute__(self, 'fmap')[name] + value = object.__getattribute__(self, 'prop_'+field.name) + if field.name == name: + return value + else: # singular alias + if not field.multiple: + raise "OUCH!! for field %s" % name + + return value[0] if value else None + except (KeyError, AttributeError): + return object.__getattribute__(self, name) + + def __setattr__(self, name, newvalue): try: - attribute, converter = self.mapping[element.tag] - setattr(self, attribute, converter(element.text, getattr(self, attribute, None))) - except KeyError: - pass + field = object.__getattribute__(self, 'fmap')[name] + if field.name == name: + object.__setattr__(self, 'prop_'+field.name, newvalue) + else: # singular alias + if not field.multiple: + raise "OUCH! while setting field %s" % name + + object.__setattr__(self, 'prop_'+field.name, [newvalue]) + except (KeyError, AttributeError): + return object.__setattr__(self, name, newvalue) - def to_xml(self): + def update(self, field_dict): + """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): + setattr(self, field.name, field_dict[field.name]) + + 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' - - root = etree.Element(self.RDF('RDF')) - description = etree.SubElement(root, self.RDF('Description')) - - if self.wiki_url: - description.set(self.RDF('about'), self.wiki_url) - - for tag, (attribute, converter) in self.mapping.iteritems(): - if hasattr(self, attribute): - e = etree.Element(tag) - e.text = unicode(getattr(self, attribute)) - description.append(e) - - return unicode(etree.tostring(root, 'utf-8'), 'utf-8') + #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 + for x in v: + e = etree.Element(field.uri) + if x is not None: + e.text = unicode(x) + description.append(e) + else: + e = etree.Element(field.uri) + e.text = unicode(v) + description.append(e) + + return root + + def serialize(self): + rdf = {} + 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 x is not None ] + else: + v = unicode(v) + + dc[field.name] = {'uri': field.uri, 'value': v} + rdf['fields'] = dc + return rdf def to_dict(self): - etree._namespace_map[str(self.RDF)] = 'rdf' - etree._namespace_map[str(self.DC)] = 'dc' - - result = {'about': self.wiki_url} - for tag, (attribute, converter) in self.mapping.iteritems(): - if hasattr(self, attribute): - result[attribute] = unicode(getattr(self, attribute)) - + result = {'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 + v = [ unicode(x) for x in v if x is not None ] + else: + v = unicode(v) + result[field.name] = v + + if field.salias: + v = getattr(self, field.salias) + if v is not None: result[field.salias] = unicode(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('contributor.translator'), 'translators', \ + as_person, salias='translator', multiple=True, default=[]), + 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), + ) -if __name__ == '__main__': - import sys - - info = parse(sys.argv[1]) - for attribute, _ in BookInfo.mapping.values(): - print '%s: %r' % (attribute, getattr(info, attribute, None)) +def parse(file_name, cls=BookInfo): + return cls.from_file(file_name)