src/p4l/mapping/parsers.py
author ymh <ymh.work@gmail.com>
Fri, 20 Sep 2013 10:34:49 +0200
changeset 107 48440ff95906
child 119 ece69ca3ac24
permissions -rw-r--r--
small code reorg
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
107
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     1
# -*- coding: utf-8 -*-
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     2
'''
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     3
Created on Sep 20, 2013
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     4
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     5
@author: ymh
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     6
'''
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     7
from rdflib.plugins.sparql.processor import prepareQuery
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     8
from rdflib.term import URIRef
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
     9
from p4l.models.data import Language, Record
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    10
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    11
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    12
class QueryCache(object):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    13
    def __init__(self, *args, **kwargs):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    14
        self.__query_cache = {}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    15
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    16
    def get_sparql_query(self, query, namespaces_dict):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    17
        return self.__query_cache.get(query, False) \
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    18
            or self.__query_cache.setdefault(query, prepareQuery(query, initNs=namespaces_dict))
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    19
    
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    20
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    21
def convert_bool(val):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    22
    if val == True or val == False:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    23
        return val
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    24
    if val is None:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    25
        return False
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    26
    if isinstance(val, basestring):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    27
        if len(val) == 0:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    28
            return False
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    29
        if val[0].lower() in ['t','y','1','o']:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    30
            return True
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    31
        else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    32
            return False        
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    33
    return bool(val)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    34
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    35
class RecordParser(object):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    36
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    37
    
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    38
    def __init__(self, query_cache = None):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    39
        self.query_cache = None
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    40
        if self.query_cache is None:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    41
            self.query_cache = QueryCache()        
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    42
    
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    43
    def extract_single_value_form_graph(self, graph, q, bindings={}, index=0, convert=lambda v:unicode(v) if v is not None else None, default=None):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    44
        return next(self.extract_multiple_values_from_graph(graph, q, bindings, index, convert), default)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    45
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    46
    def extract_multiple_values_from_graph(self, graph, q, bindings={}, index=0, convert=lambda v:unicode(v) if v is not None else None):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    47
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    48
        index_list = index
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    49
        if isinstance(index, int):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    50
            index_list = range(index+1)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    51
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    52
        if hasattr(convert, '__call__'):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    53
            convert_dict = dict((k, convert) for k in index_list)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    54
        else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    55
            convert_dict = convert
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    56
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    57
        convert_dict = dict((k, f if hasattr(f,'__call__') else lambda v:unicode(v) if v is not None else None) for k,f in convert_dict.iteritems())
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    58
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    59
        for row in graph.query(self.query_cache.get_sparql_query(q, dict(graph.namespaces())), initBindings=bindings):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    60
            if len(row) < len(index_list):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    61
                break
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    62
            else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    63
                res = dict([ (k, convert_dict.get(k, lambda v:unicode(v) if v is not None else None)(v)) for k, v in zip(index_list, row)])
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    64
                if isinstance(index, int):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    65
                    yield res[index]
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    66
                else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    67
                    yield res
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    68
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    69
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    70
    def convert_bool(self, val):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    71
        if val == True or val == False:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    72
            return val
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    73
        if val is None:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    74
            return False
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    75
        if isinstance(val, basestring):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    76
            if len(val) == 0:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    77
                return False
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    78
            if val[0].lower() in ['t','y','1','o']:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    79
                return True
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    80
            else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    81
                return False        
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    82
        return bool(val)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    83
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    84
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    85
    def add_to_related_collection(self, coll, graph, fields, q, bindings={},  convert=lambda v: unicode(v) if v is not None else None, through_fields=None):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    86
        
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    87
        for val in self.extract_multiple_values_from_graph(graph, q, bindings=bindings, index=fields, convert=convert):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    88
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    89
            if through_fields:                
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    90
                new_obj_val = dict([(k,v) for k,v in val.iteritems() if k not in through_fields])
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    91
            else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    92
                new_obj_val = val
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    93
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    94
            if hasattr(coll, 'through'):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    95
                new_obj_rel, _ = coll.model.objects.get_or_create(**new_obj_val)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    96
                if through_fields:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    97
                    through_vals = {coll.source_field_name: coll.instance, coll.target_field_name: new_obj_rel}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    98
                    through_vals.update(dict([(k,v) for k,v in val.iteritems() if k in through_fields]))
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
    99
                    coll.through.objects.create(**through_vals)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   100
                    new_obj = None
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   101
                else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   102
                    new_obj = new_obj_rel
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   103
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   104
            else:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   105
                new_obj = coll.create(**new_obj_val)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   106
            
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   107
            if new_obj:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   108
                coll.add(new_obj)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   109
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   110
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   111
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   112
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   113
    def build_record(self, graph, delete=True):
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   114
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   115
        record_uri = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?s WHERE { ?s rdf:type iiep:Record .}")
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   116
        record_identifier = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s dct:identifier ?o .}", bindings={'s':URIRef(record_uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   117
        
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   118
        if delete:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   119
            Record.objects.filter(identifier=record_identifier).delete()
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   120
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   121
        record = Record()
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   122
        record.uri = record_uri
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   123
        record.identifier = record_identifier
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   124
        record.notes = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:notes ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   125
        record.recordType = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s dct:type ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   126
        record.isDocumentPart = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:isDocumentPart ?o .}", bindings={'s':URIRef(record.uri)}, convert=self.convert_bool, default=False)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   127
        record.hidden = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:hidden ?o .}", bindings={'s':URIRef(record.uri)}, convert=self.convert_bool, default=False)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   128
        record.restricted = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:restricted ?o .}", bindings={'s':URIRef(record.uri)}, convert=self.convert_bool, default=False)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   129
        record.editionStatement = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:editionStatement ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   130
        record.corporateAuthorLabel = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s iiep:corporateAuthorLabel ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   131
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   132
        language = self.extract_single_value_form_graph(graph,"SELECT DISTINCT ?o WHERE { ?s dct:language ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   133
        if language:
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   134
            record.language, _ = Language.objects.get_or_create(uri=language)
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   135
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   136
        record.save()
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   137
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   138
        self.add_to_related_collection(record.otherLanguages, graph,  ['uri'], "SELECT ?o WHERE { ?s iiep:otherLanguage ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   139
        self.add_to_related_collection(record.subjects, graph, ['uri'], "SELECT ?o WHERE { ?s dct:subject ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   140
        self.add_to_related_collection(record.themes, graph, ['uri'], "SELECT ?o WHERE { ?s iiep:theme ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   141
        self.add_to_related_collection(record.countries, graph,  ['uri'], "SELECT ?o WHERE { ?s iiep:country ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   142
        self.add_to_related_collection(record.authors, graph, ['name'], "SELECT ?o WHERE { ?s iiep:author ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   143
        self.add_to_related_collection(record.subjectPersons, graph, ['name'], "SELECT ?o WHERE { ?s iiep:subjectPerson ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   144
        self.add_to_related_collection(record.projectNames, graph, ['uri'], "SELECT ?o WHERE { ?s iiep:projectName ?o . }")
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   145
        self.add_to_related_collection(record.audiences, graph,  ['uri'], "SELECT ?o WHERE { ?s dct:audience ?o .}", bindings={'s':URIRef(record.uri)})
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   146
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   147
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   148
            record.periodicals,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   149
            graph, 
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   150
            ['label','lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   151
            "SELECT DISTINCT ?o  ( lang(?o) as ?l) WHERE { ?s iiep:periodical ?o .}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   152
            bindings={'s':URIRef(record.uri)}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   153
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   154
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   155
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   156
            record.meetings,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   157
            graph, 
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   158
            ['label', 'meetingNumber', 'meetingPlace', 'meetingDate', 'meetingYear', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   159
            "SELECT ?l ?mn ?mp ?md ?my (lang(COALESCE(?l,?nm, ?mp,?md,?my)) as ?lang) WHERE { [iiep:meeting ?bnode]. OPTIONAL { ?bnode rdfs:label ?l }. OPTIONAL { ?bnode iiep:meetingNumber ?mn }. OPTIONAL { ?bnode iiep:meetingPlace ?mp }.  OPTIONAL { ?bnode iiep:meetingDate ?md }. OPTIONAL { ?bnode iiep:meetingYear ?my }}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   160
            convert={'meetingYear' : lambda y: int(y) if y is not None else None}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   161
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   162
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   163
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   164
            record.series,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   165
            graph, 
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   166
            ['title', 'volume', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   167
            "SELECT ?t ?vol (lang(COALESCE(?t,?vol)) as ?lang) WHERE { [iiep:serie ?bnode]. OPTIONAL { ?bnode dct:title ?t }. OPTIONAL { ?bnode iiep:volume ?vol } }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   168
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   169
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   170
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   171
            record.subjectCorporateBodies,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   172
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   173
            ['uri'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   174
            "SELECT ?o WHERE { ?s iiep:subjectCorporateBody ?o. }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   175
            bindings={'s':URIRef(record.uri)}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   176
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   177
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   178
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   179
            record.subjectMeetings,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   180
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   181
            ['label', 'meetingNumber', 'meetingPlace', 'meetingDate', 'meetingYear'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   182
            "SELECT ?l ?mn ?mp ?md ?my WHERE { [iiep:subjectMeeting ?bnode]. OPTIONAL { ?bnode rdfs:label ?l }. OPTIONAL { ?bnode iiep:meetingNumber ?mn }. OPTIONAL { ?bnode iiep:meetingPlace ?mp }.  OPTIONAL { ?bnode iiep:meetingDate ?md }. OPTIONAL { ?bnode iiep:meetingYear ?my }}",            
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   183
            convert={'meetingYear' : lambda y: int(y) if y is not None else None}
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   184
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   185
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   186
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   187
            record.corporateAuthors,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   188
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   189
            ['uri'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   190
            "SELECT ?o WHERE { ?s iiep:corporateAuthor ?o.}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   191
            bindings={'s':URIRef(record.uri)}            
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   192
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   193
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   194
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   195
            record.issns,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   196
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   197
            ['issn', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   198
            "SELECT ?issn (lang(COALESCE(?issn)) as ?lang) WHERE { ?s iiep:issn ?issn . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   199
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   200
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   201
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   202
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   203
            record.isbns,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   204
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   205
            ['isbn', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   206
            "SELECT ?isbn (lang(COALESCE(?isbn)) as ?lang) WHERE { ?s iiep:isbn ?isbn . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   207
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   208
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   209
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   210
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   211
            record.documentCodes,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   212
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   213
            ['documentCode', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   214
            "SELECT ?c (lang(COALESCE(?c)) as ?lang) WHERE { ?s iiep:documentCode ?c . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   215
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   216
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   217
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   218
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   219
            record.titles,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   220
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   221
            ['title', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   222
            "SELECT ?t (lang(COALESCE(?t)) as ?lang) WHERE { ?s dct:title ?t . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   223
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   224
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   225
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   226
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   227
            record.abstracts,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   228
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   229
            ['abstract', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   230
            "SELECT ?t (lang(COALESCE(?t)) as ?lang) WHERE { ?s dct:abstract ?t . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   231
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   232
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   233
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   234
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   235
            record.addedTitles,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   236
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   237
            ['title', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   238
            "SELECT ?t (lang(COALESCE(?t)) as ?lang) WHERE { ?s iiep:addedTitle ?t . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   239
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   240
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   241
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   242
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   243
            record.titlesMainDocument,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   244
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   245
            ['title', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   246
            "SELECT ?t (lang(COALESCE(?t)) as ?lang) WHERE { ?s iiep:titleMainDocument ?t . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   247
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   248
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   249
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   250
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   251
            record.imprints,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   252
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   253
            ['imprintCity', 'publisher', 'imprintDate', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   254
            "SELECT ?c ?p ?d (lang(COALESCE(?c, ?p, ?d)) as ?lang) WHERE { [ iiep:imprint ?bnode ]. OPTIONAL { ?bnode iiep:imprintCity ?c }. OPTIONAL { ?bnode dct:publisher ?p }. OPTIONAL { ?bnode iiep:imprintDate ?d }}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   255
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   256
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   257
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   258
            record.collations,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   259
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   260
            ['collation', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   261
            "SELECT ?c (lang(COALESCE(?c)) as ?lang) WHERE { ?s iiep:collation ?c . }",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   262
            bindings={'s':URIRef(record.uri)},
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   263
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   264
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   265
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   266
            record.volumeIssues,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   267
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   268
            ['volume', 'number', 'lang'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   269
            "SELECT ?v ?n (lang(COALESCE(?v, ?n)) as ?lang) WHERE { [ iiep:volumeIssue ?bnode ]. OPTIONAL { ?bnode iiep:volume ?v }. OPTIONAL { ?bnode iiep:number ?n }}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   270
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   271
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   272
        self.add_to_related_collection(
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   273
            record.urls,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   274
            graph,
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   275
            ['address', 'display'],
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   276
            "SELECT ?a ?d WHERE { [ iiep:url ?bnode ]. OPTIONAL { ?bnode iiep:address ?a }. OPTIONAL { ?bnode iiep:display ?d }.}",
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   277
        )
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   278
48440ff95906 small code reorg
ymh <ymh.work@gmail.com>
parents:
diff changeset
   279
        return record