turns out properties are not callable

[librarian.git] / librarian / dcparser.py

diff --git a/librarian/dcparser.py b/librarian/dcparser.py
index 534cd27..747ac86 100644 (file)
--- a/librarian/dcparser.py
+++ b/librarian/dcparser.py
@@ -10,7 +10,7 @@ import time
 from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS,
                        WLURI)
 
 from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS,
                        WLURI)
 

-import lxml.etree as etree # ElementTree API using libxml2
+import lxml.etree as etree  # ElementTree API using libxml2

 from lxml.etree import XMLSyntaxError
 
 
 from lxml.etree import XMLSyntaxError
 
 


@@ -25,7 +25,7 @@ class Person(object):
 
     @classmethod
     def from_text(cls, text):
 
     @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 = []
         if len(parts) == 1:
             surname = parts[0]
             names = []


@@ -36,7 +36,7 @@ class Person(object):
             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))
             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 = [name for name in parts[1].split() if len(name)]  # all non-whitespace tokens

         return cls(surname, *names)
 
     def readable(self):
         return cls(surname, *names)
 
     def readable(self):


@@ -60,6 +60,7 @@ class Person(object):
     def __repr__(self):
         return 'Person(last_name=%r, first_names=*%r)' % (self.last_name, self.first_names)
 
     def __repr__(self):
         return 'Person(last_name=%r, first_names=*%r)' % (self.last_name, self.first_names)
 

+

 def as_date(text):
     try:
         try:
 def as_date(text):
     try:
         try:


@@ -70,62 +71,86 @@ def as_date(text):
     except ValueError, e:
         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
 
     except ValueError, e:
         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
 

+

 def as_person(text):
     return Person.from_text(text)
 
 def as_person(text):
     return Person.from_text(text)
 

+

 def as_unicode(text):
     if isinstance(text, unicode):
         return text
     else:
         return text.decode('utf-8')
 
 def as_unicode(text):
     if isinstance(text, unicode):
         return text
     else:
         return text.decode('utf-8')
 

+
+def as_wluri_strict(text):
+    return WLURI.strict(text)
+
+

 class Field(object):
 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.uri = uri
         self.name = attr_name

-        self.validator = type
+        self.validator = validator
+        self.strict = strict

         self.multiple = multiple
         self.salias = salias
 
         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])
 
         self.default = kwargs.get('default', [] if multiple else [None])
 

-    def validate_value(self, val):
+    def validate_value(self, val, strict=False):
+        val = [v.strip() if v is not None else v for v in val]
+        if strict and self.strict is not None:
+            validator = self.strict
+        else:
+            validator = self.validator

         try:
             if self.multiple:
         try:
             if self.multiple:

-                if self.validator is None:
+                if validator is None:

                     return val
                     return val

-                return [ self.validator(v) if v is not None else v for v in val ]
+                return [validator(v) if v is not None else v for v in val]

             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:
             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 self.validator is None or val[0] is None:
+                if validator is None or val[0] is None:

                     return val[0]
                     return val[0]

-                return self.validator(val[0])
+                return validator(val[0])

         except ValueError, e:
             raise ValidationError("Field '%s' - invald value: %s" % (self.uri, e.message))
 
         except ValueError, 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:
             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]
 
             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):
 
     def __eq__(self, other):

-        if isinstance(other, Field) and other.name == self.name:
+        if isinstance(other, Field) and other.name == self.name and other.uri == self.uri:

             return True
         return False
 
 
 class DCInfo(type):
             return True
         return False
 
 
 class DCInfo(type):

-    def __new__(meta, classname, bases, class_dict):
+    def __new__(mcs, classname, bases, class_dict):

         fields = list(class_dict['FIELDS'])
 
         for base in bases[::-1]:
         fields = list(class_dict['FIELDS'])
 
         for base in bases[::-1]:


@@ -137,47 +162,54 @@ class DCInfo(type):
                         fields.insert(0, field)
 
         class_dict['FIELDS'] = tuple(fields)
                         fields.insert(0, field)
 
         class_dict['FIELDS'] = tuple(fields)

-        return super(DCInfo, meta).__new__(meta, classname, bases, class_dict)
+        return super(DCInfo, mcs).__new__(mcs, classname, bases, class_dict)

 
 
 class WorkInfo(object):
     __metaclass__ = DCInfo
 
     FIELDS = (
 
 
 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('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('rights.license'), 'license', required=False),
-        Field( DCNS('rights'), 'license_description'),
+        Field(DCNS('creator.expert'), 'authors_expert', as_person, salias='author', required=False, multiple=True),
+        Field(DCNS('creator.methodologist'), 'authors_methodologist', as_person, salias='author', required=False,
+              multiple=True),
+        Field(DCNS('creator.scenario'), 'authors_scenario', as_person, salias='author', required=False, multiple=True),
+        Field(DCNS('creator.textbook'), 'authors_textbook', as_person, salias='author', required=False, multiple=True),
+        Field(DCNS('requires'), 'requires', required=False, multiple=True),
+        Field(DCNS('title'), 'title'),
+        Field(DCNS('type'), 'type', required=False),
+
+        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('date'), 'created_at', as_date),
+        Field(DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False),
+        Field(DCNS('publisher'), 'publisher'),
+
+        Field(DCNS('subject.competence'), 'competences', multiple=True, required=False),
+        Field(DCNS('subject.curriculum'), 'curriculum', multiple=True, required=False),
+
+        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
     )
 
     @classmethod

-    def from_string(cls, xml):
+    def from_string(cls, xml, *args, **kwargs):

         from StringIO import StringIO
         from StringIO import StringIO

-        return cls.from_file(StringIO(xml))
+        return cls.from_file(StringIO(xml), *args, **kwargs)

 
     @classmethod
 
     @classmethod

-    def from_file(cls, xmlfile):
+    def from_file(cls, xmlfile, *args, **kwargs):

         desc_tag = None
         try:
         desc_tag = None
         try:

-            iter = etree.iterparse(xmlfile, ['start', 'end'])
-            for (event, element) in iter:
+            elements = etree.iterparse(xmlfile, ['start', 'end'])
+            for (event, element) in elements:

                 if element.tag == RDFNS('RDF') and event == 'start':
                     desc_tag = element
                     break
                 if element.tag == RDFNS('RDF') and event == 'start':
                     desc_tag = element
                     break


@@ -187,21 +219,21 @@ class WorkInfo(object):
                     Check if there are rdf:RDF and rdf:Description tags.")
 
             # continue 'till the end of RDF section
                     Check if there are rdf:RDF and rdf:Description tags.")
 
             # continue 'till the end of RDF section

-            for (event, element) in iter:
+            for (event, element) in elements:

                 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
                 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)
+            return cls.from_element(desc_tag, *args, **kwargs)

         except XMLSyntaxError, e:
             raise ParseError(e)
         except ExpatError, e:
             raise ParseError(e)
 
     @classmethod
         except XMLSyntaxError, e:
             raise ParseError(e)
         except ExpatError, 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'))
         # the tree is already parsed, so we don't need to worry about Expat errors
         field_dict = {}
         desc = rdf_tag.find(".//" + RDFNS('Description'))


@@ -214,10 +246,9 @@ class WorkInfo(object):
             fv.append(e.text)
             field_dict[e.tag] = fv
 
             fv.append(e.text)
             field_dict[e.tag] = fv
 

-        print field_dict
-        return cls(desc.attrib, field_dict)
+        return cls(desc.attrib, field_dict, *args, **kwargs)

 
 

-    def __init__(self, rdf_attrs, dc_fields):
+    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. """
         """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. """


@@ -226,15 +257,13 @@ class WorkInfo(object):
         self.fmap = {}
 
         for field in self.FIELDS:
         self.fmap = {}
 
         for field in self.FIELDS:

-            value = field.validate( dc_fields )
+            value = field.validate(dc_fields, fallbacks=fallbacks, strict=strict)
+            if field.multiple:
+                value = getattr(self, 'prop_' + field.name, []) + value

             setattr(self, 'prop_' + field.name, value)
             self.fmap[field.name] = field
             setattr(self, 'prop_' + field.name, value)
             self.fmap[field.name] = field

-            if field.salias: self.fmap[field.salias] = field
-
-        self.validate()
-
-    def validate(self):
-        self.url.validate_language(self.language)
+            if field.salias:
+                self.fmap[field.salias] = field

 
     def __getattribute__(self, name):
         try:
 
     def __getattribute__(self, name):
         try:


@@ -242,11 +271,12 @@ class WorkInfo(object):
             value = object.__getattribute__(self, 'prop_'+field.name)
             if field.name == name:
                 return value
             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
 
                 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)
 
         except (KeyError, AttributeError):
             return object.__getattribute__(self, name)
 


@@ -255,7 +285,8 @@ class WorkInfo(object):
             field = object.__getattribute__(self, 'fmap')[name]
             if field.name == name:
                 object.__setattr__(self, 'prop_'+field.name, newvalue)
             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
 
                 if not field.multiple:
                     raise "OUCH! while setting field %s" % name
 


@@ -267,13 +298,13 @@ class WorkInfo(object):
         """Update using field_dict. Verify correctness, but don't check if all
         required fields are present."""
         for field in self.FIELDS:
         """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])
 
                 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."""
         """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'))
 
         if parent is None:
             root = etree.Element(RDFNS('RDF'))


@@ -289,7 +320,8 @@ class WorkInfo(object):
             v = getattr(self, field.name, None)
             if v is not None:
                 if field.multiple:
             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)
                         if x is not None:
                     for x in v:
                         e = etree.Element(field.uri)
                         if x is not None:


@@ -303,16 +335,16 @@ class WorkInfo(object):
         return root
 
     def serialize(self):
         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:
 
         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 ]
+                    if len(v) == 0:
+                        continue
+                    v = [unicode(x) for x in v if x is not None]

                 else:
                     v = unicode(v)
 
                 else:
                     v = unicode(v)
 


@@ -327,35 +359,38 @@ class WorkInfo(object):
 
             if v is not None:
                 if field.multiple:
 
             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 ]
+                    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)
                 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)
+                if v is not None:
+                    result[field.salias] = unicode(v)

 
         return result
 
 
 class BookInfo(WorkInfo):
     FIELDS = (
 
         return result
 
 
 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),
+        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'), 'relations', WLURI, strict=as_wluri_strict, multiple=True, 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),

     )
 
 
     )