librarian/dcparser.py

   1 # -*- coding: utf-8 -*-
   2 #
   3 # This file is part of Librarian, licensed under GNU Affero GPLv3 or later.
   4 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   5 #
   6 from xml.parsers.expat import ExpatError
   7 from datetime import date
   8 import time
   9
  10 from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS,
  11                        WLURI)
  12
  13 import lxml.etree as etree  # ElementTree API using libxml2
  14 from lxml.etree import XMLSyntaxError
  15
  16
  17 # ==============
  18 # = Converters =
  19 # ==============
  20 class Person(object):
  21     """Single person with last name and a list of first names."""
  22     def __init__(self, last_name, *first_names):
  23         self.last_name = last_name
  24         self.first_names = first_names
  25
  26     @classmethod
  27     def from_text(cls, text):
  28         parts = [token.strip() for token in text.split(',')]
  29         if len(parts) == 1:
  30             surname = parts[0]
  31             names = []
  32         elif len(parts) != 2:
  33             raise ValueError("Invalid person name. There should be at most one comma: \"%s\"." % text)
  34         else:
  35             surname = parts[0]
  36             if len(parts[1]) == 0:
  37                 # there is no non-whitespace data after the comma
  38                 raise ValueError("Found a comma, but no names given: \"%s\" -> %r." % (text, parts))
  39             names = [name for name in parts[1].split() if len(name)]  # all non-whitespace tokens
  40         return cls(surname, *names)
  41
  42     def readable(self):
  43         return u" ".join(self.first_names + (self.last_name,))
  44
  45     def __eq__(self, right):
  46         return self.last_name == right.last_name and self.first_names == right.first_names
  47
  48     def __cmp__(self, other):
  49         return cmp((self.last_name, self.first_names), (other.last_name, other.first_names))
  50
  51     def __hash__(self):
  52         return hash((self.last_name, self.first_names))
  53
  54     def __unicode__(self):
  55         if len(self.first_names) > 0:
  56             return '%s, %s' % (self.last_name, ' '.join(self.first_names))
  57         else:
  58             return self.last_name
  59
  60     def __repr__(self):
  61         return 'Person(last_name=%r, first_names=*%r)' % (self.last_name, self.first_names)
  62
  63
  64 def as_date(text):
  65     try:
  66         try:
  67             t = time.strptime(text, '%Y-%m-%d')
  68         except ValueError:
  69             t = time.strptime(text, '%Y')
  70         return date(t[0], t[1], t[2])
  71     except ValueError, e:
  72         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
  73
  74
  75 def as_person(text):
  76     return Person.from_text(text)
  77
  78
  79 def as_unicode(text):
  80     if isinstance(text, unicode):
  81         return text
  82     else:
  83         return text.decode('utf-8')
  84
  85
  86 def as_wluri_strict(text):
  87     return WLURI.strict(text)
  88
  89
  90 class Field(object):
  91     def __init__(self, uri, attr_name, validator=as_unicode, strict=None, multiple=False, salias=None, **kwargs):
  92         self.uri = uri
  93         self.name = attr_name
  94         self.validator = validator
  95         self.strict = strict
  96         self.multiple = multiple
  97         self.salias = salias
  98
  99         self.required = kwargs.get('required', True) and 'default' not in kwargs
 100         self.default = kwargs.get('default', [] if multiple else [None])
 101
 102     def validate_value(self, val, strict=False):
 103         val = [v.strip() if v is not None else v for v in val]
 104         if strict and self.strict is not None:
 105             validator = self.strict
 106         else:
 107             validator = self.validator
 108         try:
 109             if self.multiple:
 110                 if validator is None:
 111                     return val
 112                 return [validator(v) if v is not None else v for v in val]
 113             elif len(val) > 1:
 114                 raise ValidationError("Multiple values not allowed for field '%s'" % self.uri)
 115             elif len(val) == 0:
 116                 raise ValidationError("Field %s has no value to assign. Check your defaults." % self.uri)
 117             else:
 118                 if validator is None or val[0] is None:
 119                     return val[0]
 120                 return validator(val[0])
 121         except ValueError, e:
 122             raise ValidationError("Field '%s' - invald value: %s" % (self.uri, e.message))
 123
 124     def validate(self, fdict, fallbacks=None, strict=False):
 125         if fallbacks is None:
 126             fallbacks = {}
 127         if self.uri not in fdict:
 128             if not self.required:
 129                 # Accept single value for single fields and saliases.
 130                 if self.name in fallbacks:
 131                     if self.multiple:
 132                         f = fallbacks[self.name]
 133                     else:
 134                         f = [fallbacks[self.name]]
 135                 elif self.salias and self.salias in fallbacks:
 136                     f = [fallbacks[self.salias]]
 137                 else:
 138                     f = self.default
 139             else:
 140                 raise ValidationError("Required field %s not found" % self.uri)
 141         else:
 142             f = fdict[self.uri]
 143
 144         return self.validate_value(f, strict=strict)
 145
 146     def __eq__(self, other):
 147         if isinstance(other, Field) and other.name == self.name and other.uri == self.uri:
 148             return True
 149         return False
 150
 151
 152 class DCInfo(type):
 153     def __new__(mcs, classname, bases, class_dict):
 154         fields = list(class_dict['FIELDS'])
 155
 156         for base in bases[::-1]:
 157             if hasattr(base, 'FIELDS'):
 158                 for field in base.FIELDS[::-1]:
 159                     try:
 160                         fields.index(field)
 161                     except ValueError:
 162                         fields.insert(0, field)
 163
 164         class_dict['FIELDS'] = tuple(fields)
 165         return super(DCInfo, mcs).__new__(mcs, classname, bases, class_dict)
 166
 167
 168 class WorkInfo(object):
 169     __metaclass__ = DCInfo
 170
 171     FIELDS = (
 172         Field(DCNS('creator.expert'), 'authors_expert', as_person, salias='author', required=False, multiple=True),
 173         Field(DCNS('creator.methodologist'), 'authors_methodologist', as_person, salias='author', required=False,
 174               multiple=True),
 175         Field(DCNS('creator.scenario'), 'authors_scenario', as_person, salias='author', required=False, multiple=True),
 176         Field(DCNS('creator.textbook'), 'authors_textbook', as_person, salias='author', required=False, multiple=True),
 177         Field(DCNS('requires'), 'requires', required=False, multiple=True),
 178         Field(DCNS('title'), 'title'),
 179         Field(DCNS('type'), 'type', required=False),
 180
 181         Field(DCNS('contributor.editor'), 'editors', as_person, salias='editor', multiple=True, default=[]),
 182         Field(DCNS('contributor.technical_editor'), 'technical_editors', as_person, salias='technical_editor',
 183               multiple=True, default=[]),
 184
 185         Field(DCNS('date'), 'created_at', as_date),
 186         Field(DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False),
 187         Field(DCNS('publisher'), 'publisher'),
 188
 189         Field(DCNS('subject.competence'), 'competences', multiple=True, required=False),
 190         Field(DCNS('subject.curriculum'), 'curriculum', multiple=True, required=False),
 191         Field(DCNS('subject.curriculum.new'), 'curriculum_new', multiple=True, required=False),
 192
 193         Field(DCNS('language'), 'language'),
 194         Field(DCNS('description'), 'description', required=False),
 195
 196         Field(DCNS('source'), 'source_name', required=False),
 197         Field(DCNS('source.URL'), 'source_url', required=False),
 198         Field(DCNS('identifier.url'), 'url', WLURI, strict=as_wluri_strict),
 199         Field(DCNS('rights.license'), 'license', required=False),
 200         Field(DCNS('rights'), 'license_description'),
 201     )
 202
 203     @classmethod
 204     def from_string(cls, xml, *args, **kwargs):
 205         from StringIO import StringIO
 206         return cls.from_file(StringIO(xml), *args, **kwargs)
 207
 208     @classmethod
 209     def from_file(cls, xmlfile, *args, **kwargs):
 210         desc_tag = None
 211         try:
 212             elements = etree.iterparse(xmlfile, ['start', 'end'])
 213             for (event, element) in elements:
 214                 if element.tag == RDFNS('RDF') and event == 'start':
 215                     desc_tag = element
 216                     break
 217
 218             if desc_tag is None:
 219                 raise NoDublinCore("DublinCore section not found. \
 220                     Check if there are rdf:RDF and rdf:Description tags.")
 221
 222             # continue 'till the end of RDF section
 223             for (event, element) in elements:
 224                 if element.tag == RDFNS('RDF') and event == 'end':
 225                     break
 226
 227             # if there is no end, Expat should yell at us with an ExpatError
 228
 229             # extract data from the element and make the info
 230             return cls.from_element(desc_tag, *args, **kwargs)
 231         except XMLSyntaxError, e:
 232             raise ParseError(e)
 233         except ExpatError, e:
 234             raise ParseError(e)
 235
 236     @classmethod
 237     def from_element(cls, rdf_tag, *args, **kwargs):
 238         # the tree is already parsed, so we don't need to worry about Expat errors
 239         field_dict = {}
 240         desc = rdf_tag.find(".//" + RDFNS('Description'))
 241
 242         if desc is None:
 243             raise NoDublinCore("No DublinCore section found.")
 244
 245         for e in desc.getchildren():
 246             fv = field_dict.get(e.tag, [])
 247             fv.append(e.text)
 248             field_dict[e.tag] = fv
 249
 250         return cls(desc.attrib, field_dict, *args, **kwargs)
 251
 252     def __init__(self, rdf_attrs, dc_fields, fallbacks=None, strict=False):
 253         """rdf_attrs should be a dictionary-like object with any attributes of the RDF:Description.
 254         dc_fields - dictionary mapping DC fields (with namespace) to list of text values for the
 255         given field. """
 256
 257         self.about = rdf_attrs.get(RDFNS('about'))
 258         self.fmap = {}
 259
 260         for field in self.FIELDS:
 261             value = field.validate(dc_fields, fallbacks=fallbacks, strict=strict)
 262             if field.multiple:
 263                 value = getattr(self, 'prop_' + field.name, []) + value
 264             setattr(self, 'prop_' + field.name, value)
 265             self.fmap[field.name] = field
 266             if field.salias:
 267                 self.fmap[field.salias] = field
 268
 269     def __getattribute__(self, name):
 270         try:
 271             field = object.__getattribute__(self, 'fmap')[name]
 272             value = object.__getattribute__(self, 'prop_'+field.name)
 273             if field.name == name:
 274                 return value
 275             else:
 276                 # singular alias
 277                 if not field.multiple:
 278                     raise "OUCH!! for field %s" % name
 279
 280                 return value[0] if value else None
 281         except (KeyError, AttributeError):
 282             return object.__getattribute__(self, name)
 283
 284     def __setattr__(self, name, newvalue):
 285         try:
 286             field = object.__getattribute__(self, 'fmap')[name]
 287             if field.name == name:
 288                 object.__setattr__(self, 'prop_'+field.name, newvalue)
 289             else:
 290                 # singular alias
 291                 if not field.multiple:
 292                     raise "OUCH! while setting field %s" % name
 293
 294                 object.__setattr__(self, 'prop_'+field.name, [newvalue])
 295         except (KeyError, AttributeError):
 296             return object.__setattr__(self, name, newvalue)
 297
 298     def update(self, field_dict):
 299         """Update using field_dict. Verify correctness, but don't check if all
 300         required fields are present."""
 301         for field in self.FIELDS:
 302             if field.name in field_dict:
 303                 setattr(self, field.name, field_dict[field.name])
 304
 305     def to_etree(self, parent=None):
 306         """XML representation of this object."""
 307         # etree._namespace_map[str(self.RDF)] = 'rdf'
 308         # etree._namespace_map[str(self.DC)] = 'dc'
 309
 310         if parent is None:
 311             root = etree.Element(RDFNS('RDF'))
 312         else:
 313             root = parent.makeelement(RDFNS('RDF'))
 314
 315         description = etree.SubElement(root, RDFNS('Description'))
 316
 317         if self.about:
 318             description.set(RDFNS('about'), self.about)
 319
 320         for field in self.FIELDS:
 321             v = getattr(self, field.name, None)
 322             if v is not None:
 323                 if field.multiple:
 324                     if len(v) == 0:
 325                         continue
 326                     for x in v:
 327                         e = etree.Element(field.uri)
 328                         if x is not None:
 329                             e.text = unicode(x)
 330                         description.append(e)
 331                 else:
 332                     e = etree.Element(field.uri)
 333                     e.text = unicode(v)
 334                     description.append(e)
 335
 336         return root
 337
 338     def serialize(self):
 339         rdf = {'about': {'uri': RDFNS('about'), 'value': self.about}}
 340
 341         dc = {}
 342         for field in self.FIELDS:
 343             v = getattr(self, field.name, None)
 344             if v is not None:
 345                 if field.multiple:
 346                     if len(v) == 0:
 347                         continue
 348                     v = [unicode(x) for x in v if x is not None]
 349                 else:
 350                     v = unicode(v)
 351
 352                 dc[field.name] = {'uri': field.uri, 'value': v}
 353         rdf['fields'] = dc
 354         return rdf
 355
 356     def to_dict(self):
 357         result = {'about': self.about}
 358         for field in self.FIELDS:
 359             v = getattr(self, field.name, None)
 360
 361             if v is not None:
 362                 if field.multiple:
 363                     if len(v) == 0:
 364                         continue
 365                     v = [unicode(x) for x in v if x is not None]
 366                 else:
 367                     v = unicode(v)
 368                 result[field.name] = v
 369
 370             if field.salias:
 371                 v = getattr(self, field.salias)
 372                 if v is not None:
 373                     result[field.salias] = unicode(v)
 374
 375         return result
 376
 377
 378 class BookInfo(WorkInfo):
 379     FIELDS = (
 380         Field(DCNS('audience'), 'audiences', salias='audience', multiple=True, required=False),
 381
 382         Field(DCNS('subject.period'), 'epochs', salias='epoch', multiple=True, required=False),
 383         Field(DCNS('subject.type'), 'kinds', salias='kind', multiple=True, required=False),
 384         Field(DCNS('subject.genre'), 'genres', salias='genre', multiple=True, required=False),
 385
 386         Field(DCNS('contributor.translator'), 'translators', as_person,  salias='translator', multiple=True,
 387               default=[]),
 388         Field(DCNS('relation.hasPart'), 'parts', WLURI, strict=as_wluri_strict, multiple=True, required=False),
 389         Field(DCNS('relation.isVariantOf'), 'variant_of', WLURI, strict=as_wluri_strict, required=False),
 390         Field(DCNS('relation'), 'relations', WLURI, strict=as_wluri_strict, multiple=True, required=False),
 391
 392         Field(DCNS('relation.coverImage.url'), 'cover_url', required=False),
 393         Field(DCNS('relation.coverImage.attribution'), 'cover_by', required=False),
 394         Field(DCNS('relation.coverImage.source'), 'cover_source', required=False),
 395     )
 396
 397
 398 def parse(file_name, cls=BookInfo):
 399     return cls.from_file(file_name)