[epub] output improvements - clean xhtml instead of html tag soup
[librarian.git] / librarian / dcparser.py
index aacbdbe..12bb24f 100644 (file)
@@ -6,14 +6,23 @@
 from xml.parsers.expat import ExpatError
 from datetime import date
 import time
+import re
+from librarian.util import roman_to_int
 
 from librarian import (ValidationError, NoDublinCore, ParseError, DCNS, RDFNS,
-                       WLURI)
+                       XMLNS, WLURI, WLNS, PLMETNS)
 
 import lxml.etree as etree # ElementTree API using libxml2
 from lxml.etree import XMLSyntaxError
 
 
+class TextPlus(unicode):
+    pass
+
+class DatePlus(date):
+    pass
+
+
 # ==============
 # = Converters =
 # ==============
@@ -61,12 +70,47 @@ class Person(object):
         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])
+        # check out the "N. poł X w." syntax
+        if isinstance(text, str): 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(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
+
+        return DatePlus(t[0], t[1], t[2])
     except ValueError, e:
         raise ValueError("Unrecognized date format. Try YYYY-MM-DD or YYYY.")
 
@@ -77,46 +121,76 @@ def as_unicode(text):
     if isinstance(text, unicode):
         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.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("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])
+                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):
+    def validate(self, fdict, fallbacks=None, strict=False):
+        if fallbacks is None:
+            fallbacks = {}
         if not fdict.has_key(self.uri):
             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:
@@ -126,7 +200,7 @@ class Field(object):
 
 class DCInfo(type):
     def __new__(meta, classname, bases, class_dict):
-        fields = class_dict['FIELDS']
+        fields = list(class_dict['FIELDS'])
 
         for base in bases[::-1]:
             if hasattr(base, 'FIELDS'):
@@ -134,9 +208,9 @@ class DCInfo(type):
                     try:
                         fields.index(field)
                     except ValueError:
-                        fields = (field,) + fields
+                        fields.insert(0, field)
 
-        class_dict['FIELDS'] = fields
+        class_dict['FIELDS'] = tuple(fields)
         return super(DCInfo, meta).__new__(meta, classname, bases, class_dict)
 
 
@@ -144,7 +218,7 @@ class WorkInfo(object):
     __metaclass__ = DCInfo
 
     FIELDS = (
-        Field( DCNS('creator'), 'author', as_person),
+        Field( DCNS('creator'), 'authors', as_person, salias='author', multiple=True),
         Field( DCNS('title'), 'title'),
         Field( DCNS('type'), 'type', required=False, multiple=True),
 
@@ -152,6 +226,9 @@ class WorkInfo(object):
             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),
@@ -162,19 +239,21 @@ class WorkInfo(object):
 
         Field( DCNS('source'), 'source_name', required=False),
         Field( DCNS('source.URL'), 'source_url', required=False),
-        Field( DCNS('identifier.url'), 'url', WLURI),
-
+        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, default=[]),
+        Field( WLNS('digitisationSponsorNote'), 'sponsor_note', required=False),
     )
 
     @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, xmlfile):
+    def from_file(cls, xmlfile, *args, **kwargs):
         desc_tag = None
         try:
             iter = etree.iterparse(xmlfile, ['start', 'end'])
@@ -195,14 +274,14 @@ class WorkInfo(object):
             # 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
-    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'))
@@ -210,15 +289,28 @@ class WorkInfo(object):
         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, 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
 
-        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. """
@@ -227,16 +319,12 @@ class WorkInfo(object):
         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
 
-        self.validate()
-
-    def validate(self):
-        self.url.validate_language(self.language)
-
     def __getattribute__(self, name):
         try:
             field = object.__getattribute__(self, 'fmap')[name]
@@ -247,7 +335,7 @@ class WorkInfo(object):
                 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)
 
@@ -346,18 +434,26 @@ class BookInfo(WorkInfo):
         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('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, 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('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),
     )