marcpatterns.py

'''
Declarations used to elucidate MARC model
'''
#Just set up some flags
#BOUND_TO_WORK = object()
#BOUND_TO_INSTANCE = object()

#Full MARC field list: http://www.loc.gov/marc/bibliographic/ecbdlist.html

#This line must be included
from bibframe.reader.util import *

#from bibframe.reader.marcpatterns import *
#sorted([ (m, MATERIALIZE[m]) for m in MATERIALIZE if [ wf for wf in WORK_FIELDS if m[:2] == wf[:2]] ])

#    '100': fromwork.materialize('Agent', 'creator', unique=all_subfields, mr_properties={'a': 'label'}),

TRANSFORMS = {
    #Link to the 010a value, naming the relationship 'lccn'
    '010a': fromwork.rename(rel='lccn'),
    '017a': fromwork.rename(rel='legalDeposit'),
    '020a': frominstance.rename(rel='isbn'),
    '022a': frominstance.rename(rel='issn'),
    '024a': fromwork.rename(rel='otherControlNumber'),
    '025a': fromwork.rename(rel='lcOverseasAcq'),

    '034a': fromwork.rename(rel='cartographicMathematicalDataScaleStatement'),  #Rebecca & Sally suggested this should effectively be a merge with 034a
    '034b': fromwork.rename(rel='cartographicMathematicalDataProjectionStatement'),
    '034c': fromwork.rename(rel='cartographicMathematicalDataCoordinateStatement'),
    '035a': fromwork.rename(rel='systemControlNumber'),
    '037a': fromwork.rename(rel='stockNumber'),

    '040a': fromwork.rename(rel='catalogingSource'),
    '041a': fromwork.rename(rel='language'),
    '050a': fromwork.rename(rel='lcCallNumber'),
    '050b': fromwork.rename(rel='lcItemNumber'),
    '0503': fromwork.rename(rel='material'),
    '060a': frominstance.rename(rel='nlmCallNumber'),
    '060b': frominstance.rename(rel='nlmItemNumber'),
    '061a': fromwork.rename(rel='nlmCopyStatement'),
    '070a': frominstance.rename(rel='nalCallNumber'),
    '070b': frominstance.rename(rel='nalItemNumber'),
    '071a': fromwork.rename(rel='nalCopyStatement'),
    '082a': fromwork.rename(rel='deweyNumber'),

    '130l': fromwork.rename(rel='language'),
    '210a': frominstance.rename(rel='abbreviatedTitle'),
    '222a': frominstance.rename(rel='keyTitle'),
    '240d': fromwork.rename(rel='legalDate'),
    '240h': fromwork.rename(rel='medium'),
    '240m': fromwork.rename(rel='musicMedium'),
    '240r': fromwork.rename(rel='musicKey'),
    '245a': fromwork.rename(rel='title'),
    '245b': fromwork.rename(rel='subtitle'),
    '245c': fromwork.rename(rel='statement'),
    '245f': fromwork.rename(rel='inclusiveDates'),
    '245h': fromwork.rename(rel='medium'),
    '245k': fromwork.rename(rel='formDesignation'),
    '246a': fromwork.rename(rel='titleVariation'),
    '246f': fromwork.rename(rel='titleVariationDate'),
    '247a': fromwork.rename(rel='formerTitle'),
    '250a': frominstance.rename(rel='edition'),
    '250b': frominstance.rename(rel='edition'),
    '254a': frominstance.rename(rel='musicalPresentation'),
    '255a': frominstance.rename(rel='cartographicMathematicalDataScaleStatement'),
    '255b': frominstance.rename(rel='cartographicMathematicalDataProjectionStatement'),
    '255c': frominstance.rename(rel='cartographicMathematicalDataCoordinateStatement'),
    '256a': frominstance.rename(rel='computerFilecharacteristics'),
#    '260a': frominstance.rename(rel='providerPlace'),
#    '260b': frominstance.rename(rel='providerAgent'),
    '260c': frominstance.rename(rel='providerDate'),
    '260g': frominstance.rename(rel='providerDate'),
    '264c': fromwork.rename(rel='providerDate'),
    '730a': fromwork.rename(rel='label'),
    '830a': fromwork.rename(rel='title'),

    #Field 100 has a creator relationship to a new Agent object (only created as a new object if all subfields are unique)
    '100': fromwork.materialize('Agent', 'creator', unique=all_subfields, mr_properties={'name': subfield('a'), 'numeration': subfield('b'), 'titles': subfield('c'), 'date': subfield('d')}),
    '110': fromwork.materialize('Organization', 'creator', unique=all_subfields, mr_properties={'name': subfield('a'), 'date': subfield('d')}),
    '111': fromwork.materialize('Meeting', 'creator', unique=all_subfields, mr_properties={'name': subfield('a'), 'date': subfield('d')}),

    '130': fromwork.materialize('Collection', 'creator', unique=all_subfields, mr_properties={'title': subfield('a'), 'workSection': subfield('n')}),
    '240a': fromwork.materialize('Collection', 'creator', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '243a': fromwork.materialize('Collection', 'creator', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '730': fromwork.materialize('Collection', 'creator', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '830': fromwork.materialize('Collection', 'creator', unique=all_subfields, mr_properties={'name': subfield('a')}),

    '260': frominstance.materialize('ProviderEvent', 'publication', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264': fromwork.materialize('ProviderEvent', 'publication', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264-x3': fromwork.materialize('ProviderEvent', 'manufacture', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264-x2': fromwork.materialize('ProviderEvent', 'distribution', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264-x1': fromwork.materialize('ProviderEvent', 'publication', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264-x0': fromwork.materialize('ProviderEvent', 'production', unique=all_subfields, mr_properties={'name': subfield('a')}),

    '260a': frominstance.materialize('Place', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '260b': frominstance.materialize('Agent', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '260e': frominstance.materialize('Place', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '260f': frominstance.materialize('Agent', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264a': fromwork.materialize('Place', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '264b': fromwork.materialize('Agent', 'providerAgent', unique=all_subfields, mr_properties={'name': subfield('a')}),

    #'300': fromwork.materialize('Measurement', 'physicalDescription', unique=all_subfields, mr_properties={'name': subfield('a')}),

    '600': fromwork.materialize('Person', 'subject', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '610': fromwork.materialize('Organization', 'subject', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '611': fromwork.materialize('Meeting', 'subject', unique=all_subfields, mr_properties={'name': subfield('a')}),

    '630': fromwork.materialize('Title', 'uniformTitle', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '650': fromwork.materialize('Topic', 'subject', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '651': fromwork.materialize('Geographic', 'subject', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '655': fromwork.materialize('Genre', 'genre', unique=all_subfields, mr_properties={'name': subfield('a')}),

    '700': fromwork.materialize('Person', 'contributor', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '710': fromwork.materialize('Organization', 'contributor', unique=all_subfields, mr_properties={'name': subfield('a')}),
    '711': fromwork.materialize('Meeting', 'contributor', unique=all_subfields, mr_properties={'name': subfield('a')}),

    #HeldItem is a refinement of Annotation
    '852': frominstance.materialize('HeldItem', 'institution', unique=all_subfields, mr_properties={'holderType': 'Library', 'location': subfield('a'), 'subLocation': subfield('b'), 'callNumber': subfield('h'), 'code': subfield('n'), 'link': subfield('u'), 'streetAddress': subfield('e')}),

    #A refinement is a relationship from one mapping to another in order to refine
    #Replaces the 700a value by looking up 700e
    '700a': fromwork.refinement(normalizeparse, rel=subfield('e')),

    '880a': fromwork.rename(rel='title'),
    '856u': frominstance.rename(rel='link'),
}



FIELD_RENAMINGS = {
# where do we put LDR info, e.g. LDR 07 / 19 positions = mode of issuance
#Don't do a simple field renaming of ISBN because
'300a': 'extent',
'300b': 'otherPhysicalDetails',
'300c': 'dimensions',
'300e': 'accompanyingMaterial',
'300f': 'typeOfunit',
'300g': 'size',
'3003': 'materials',
'310a': 'publicationFrequency',
'310b': 'publicationDateFrequency',
'336a': 'contentCategory',
'336b': 'contentTypeCode',
'3362': 'contentTypeRDAsource',
'337a': 'mediaCategory',
'337b': 'mediaTypeCode',
'3372': 'medaiRDAsource',
'338a': 'carrierCategory',
'338b': 'carrierCategoryCode',
'3382': 'carrierRDASource',
'340a': 'physicalSubstance',
'340b': 'dimensions',
'340c': 'materialsApplied',
'340d': 'recordingTechnique',
'340e': 'physicalSupport',
'351a': 'orgazationMethod',
'351b': 'arrangement',
'351c': 'hierarchy',
'3513': 'materialsSpec',
'490a': 'seriesStatement',
'490v': 'seriesVolume',
'500a': 'note',
'501a': 'note',
'502a': 'dissertationNote',
'502b': 'degree',
'502c': 'grantingInstitution',
'502d': 'dissertationYear',
'502g': 'dissertationNote', 
'502o': 'dissertationID',
'504a': 'bibliographyNote',
'505a': 'contentsNote',
'506a': 'governingAccessNote',
'506b': 'jurisdictionNote',
'506c': 'physicalAccess',
'506d': 'authorizedUsers',
'506e': 'authorization',
'506u': 'uriNote',
'507a': 'representativeFractionOfScale',
'507b': 'remainderOfScale',
'508a': 'creditsNote',
'510a': 'citationSource', 
'510b': 'citationCoverage',
'510c': 'citationLocationWithinSource',
'510u': 'citationUri',
'511a': 'performerNote',
'513a': 'typeOfReport',
'513b': 'periodCoveredn',
'514a': 'dataQuality',
'515a': 'numberingPerculiarities', 
'516a': 'typeOfComputerFile',
'518a': 'dateTimePlace',
'518d': 'dateOfEvent',
'518o': 'otherEventInformation', 
'518p': 'placeOfEvent',
'520a': 'summary',
'520b': 'summaryExpansion',
'520c': 'assigningSource',
'520u': 'summaryURI',
'521a': 'intendedAudience',
'521b': 'intendedAudienceSource', 
'522a': 'geograhpicCoverage',
'525a': 'supplement',
'538a': 'systemDetails',
'526a': 'studyProgramName',
'526b': 'interestLevel',
'526c': 'readingLevel',
'530a': 'additionalPhysicalForm',
'533a': 'reproductionNote',
'534a': 'originalVersionNote',
'535a': 'locationOfOriginalsDuplicates',
'536a': 'fundingInformation',
'538a': 'systemDetails',
'540a': 'termsGoverningUse',
'541a': 'immediateSourceOfAcquisition',
'542a': 'informationRelatingToCopyrightStatus',
'544a': 'locationOfOtherArchivalMaterial',
'545a': 'biographicalOrHistoricalData',
'546a': 'languageNote',
'547a': 'formerTitleComplexity',
'550a': 'issuingBody',
'552a': 'entityAndAttributeInformation',
'555a': 'cumulativeIndexFindingAids',
'556a': 'informationAboutDocumentation',
'561a': 'ownership', 
'583a': 'action',
'600a': 'label',
'600d': 'date',
'610a': 'label',
'610d': 'date',  #Note: there has been discussion about removing this, but we are not sure we get reliable ID.LOC lookups without it.  If it is removed, update augment.py 
'650a': 'label',
'650d': 'date',
'651a': 'label',
'651d': 'date',
'630a': 'uniformTitle',
'630l': 'language',
'630a': 'label',
'630h': 'medium',
'630v': 'formSubdivision',
'630x': 'generalSubdivision',
'630y': 'chronologicalSubdivision',
'630z': 'geographicSubdivision',
'650a': 'label',
'650c': 'locationOfEvent',
'650v': 'formSubdivision',
'650x': 'generalSubdivision',
'650y': 'chronologicalSubdivision',
'650z': 'geographicSubdivision',
'651v': 'formSubdivision',
'651x': 'generalSubdivision',
'651y': 'chronologicalSubdivision',
'651z': 'geographicSubdivision',
'655a': 'label',
'6552': 'source', #Note: use this to trigger link authority lookup
'700a': 'label',
'700b': 'numeration',
'700c': 'titles',
'700d': 'date',  #Note: there has been discussion about removing this, but we are not sure we get reliable ID.LOC lookups without it.  If it is removed, update augment.py 
'710a': 'label',
'710d': 'date',
'711a': 'label',
'711d': 'date',
}


WORK_FIELDS = set([
'010',
'028',
'035',
'040',
'041',
'050a', #Note: should be able to link directly to authority @ id.loc.gov authority/classification/####
'082',
'100',
'110',
'111',
'130',
'210',
'222',
'240',
'243',
'245',
'246',
'264',
'247',
'310',
'310',
'321',
'321',
'362',
'490',
'500',
'502',
'504',
'510',
'511',
'513',
'514',
'518',
'520',
'521',
'522',
'583',
'600',
'610',
'611',
'630',
'650',
'651',
'700',
'710',
'711',
'730',
'740',
'880',
])


INSTANCE_FIELDS = set([
'020',
'022',
'055',
'060',
'070',
'086',
'210',
'222',
'250',
'254',
'255',
'256',
'257',
'260',
'263',
'300',
'310',
'340',
'351',
'306',
'340',
'336',
'337',
'338',
'351',
'505',
'506',
'507',
'508',
'515',
'516',
'525',
'530',
'538',
'561',
'850',
'852',
'856',
])

ANNOTATIONS_FIELDS = set([
'852a',
'852b',
'852h',
'852n',
'852u',
'852e',
])

PROVIDER_EVENT_FIELDS = set([
'260a',
'260b',
'260c',
'260e',
'260f',
'260g',
'264a',
'264b',
'264c',
])

HOLDINGS_FIELDS = set([
'852',
])