def marc21_to_contribution(self, key, value):
"""Get contribution."""
if key[4] == '2' or key[:3] not in ['100', '700', '710', '711']:
return None
agent = {'type': 'bf:Person'}
if value.get('a'):
name = utils.force_list(value.get('a'))[0]
agent['preferred_name'] = remove_trailing_punctuation(name)
# 100|700 Person
if key[:3] in ['100', '700']:
if value.get('b'):
numeration = utils.force_list(value.get('b'))[0]
agent['numeration'] = remove_trailing_punctuation(numeration)
if value.get('c'):
qualifier = utils.force_list(value.get('c'))[0]
agent['qualifier'] = remove_trailing_punctuation(qualifier)
if value.get('d'):
date = utils.force_list(value.get('d'))[0]
date = date.rstrip(',')
dates = remove_trailing_punctuation(date).split('-')
with contextlib.suppress(Exception):
if date_of_birth := dates[0].strip():
agent['date_of_birth'] = date_of_birth
with contextlib.suppress(Exception):
if date_of_death := dates[1].strip():
agent['date_of_death'] = date_of_death
def marc21_to_author(self, key, value):
"""Get author.
authors: loop:
authors.name: 100$a [+ 100$b if it exists] or
[700$a (+$b if it exists) repetitive] or
[ 710$a repetitive (+$b if it exists, repetitive)]
authors.date: 100 $d or 700 $d (facultatif)
authors.qualifier: 100 $c or 700 $c (facultatif)
authors.type: if 100 or 700 then person, if 710 then organisation
"""
if not (key[4] == '2' and (key[:3] == '710' or key[:3] == '700')):
author = {}
author['type'] = 'person'
author['name'] = remove_trailing_punctuation(value.get('a'))
author_subs = utils.force_list(value.get('b'))
if author_subs:
for author_sub in author_subs:
author['name'] += ' ' + remove_trailing_punctuation(author_sub)
if key[:3] == '710':
author['type'] = 'organisation'
else:
if value.get('c'):
author['qualifier'] = remove_trailing_punctuation(
value.get('c'))
if value.get('d'):
author['date'] = remove_trailing_punctuation(value.get('d'))
return author
def build_place_or_agent_data(code, label, index, link, add_country):
type_per_code = {'a': 'bf:Place', 'b': 'bf:Agent'}
place_or_agent_data = {
'type': type_per_code[code],
'label': [{
'value': remove_trailing_punctuation(label)
}]
}
if add_country:
if marc21tojson.cantons:
place_or_agent_data['canton'] = marc21tojson.cantons
if marc21tojson.country:
place_or_agent_data['country'] = marc21tojson.country
try:
alt_gr = marc21tojson.alternate_graphic['264'][link]
subfield = \
marc21tojson.get_subfields(alt_gr['field'])[index]
place_or_agent_data['label'].append({
'value':
remove_trailing_punctuation(subfield),
'language':
'-'.join((marc21tojson.lang_from_008, alt_gr['script']))
})
except Exception:
pass
return place_or_agent_data
def marc21_to_title(self, key, value):
"""Get title.
title: 245$a
without the punctuaction. If there's a $b, then 245$a : $b without the ' /'
"""
main_title = remove_trailing_punctuation(value.get('a'))
sub_title = value.get('b')
# responsability = value.get('c')
if sub_title:
main_title += ' : ' + ' : '.join(
utils.force_list(remove_trailing_punctuation(sub_title)))
return main_title
def build_edition_data(code, label, index, link):
data = [{'value': remove_trailing_punctuation(label)}]
try:
alt_gr = marc21.alternate_graphic['250'][link]
subfield = \
marc21.get_subfields(alt_gr['field'])[index]
data.append({
'value': remove_trailing_punctuation(subfield),
'language': get_language_script(alt_gr['script'])
})
except Exception as err:
pass
return data
def edition_format_text(edition):
"""Format edition for _text."""
designations = edition.get('editionDesignation', [])
responsibilities = edition.get('responsibility', [])
designation_output = {}
for designation in designations:
language = designation.get('language', 'default')
value = designation.get('value', '')
designation_output[language] = value
responsibility_output = {}
for responsibility in responsibilities:
language = responsibility.get('language', 'default')
value = responsibility.get('value', '')
responsibility_output[language] = value
edition_text = []
for key, value in designation_output.items():
value = remove_trailing_punctuation(
'{designation} / {responsibility}'.format(
designation=designation_output.get(key),
responsibility=responsibility_output.get(key, ''),
)
)
if display_alternate_graphic_first(key):
edition_text.insert(0, {'value': value, 'language': key})
else:
edition_text.append({'value': value, 'language': key})
return edition_text
def marc21_to_description(self, key, value):
"""Get extent.
extent: 300$a (the first one if many)
"""
if value.get('a') and not self.get('extent', None):
self['extent'] = remove_trailing_punctuation(
utils.force_list(value.get('a'))[0])
return None
def build_place_or_agent_data(code, label, index):
type_per_code = {'a': 'bf:Place', 'c': 'bf:Agent'}
place_or_agent_data = {
'type': type_per_code[code],
'label': [{
'value': remove_trailing_punctuation(label)
}]
}
return place_or_agent_data
def marc21_to_edition_statement(self, key, value):
"""Get edition statement data.
editionDesignation: 250 [$a non repetitive] (without trailing ponctuation)
responsibility: 250 [$b non repetitive]
"""
edition_data = {}
if subfields_a := utils.force_list(value.get('a')):
subfield_a = remove_trailing_punctuation(subfields_a[0])
edition_data['editionDesignation'] = [{'value': subfield_a}]
def marc21_to_author(self, key, value):
"""Get author.
authors: loop:
authors.name: 100$a [+ 100$b if it exists] or
[700$a (+$b if it exists) repetitive] or
[ 710$a repetitive (+$b if it exists, repetitive)]
authors.date: 100 $d or 700 $d (facultatif)
authors.qualifier: 100 $c or 700 $c (facultatif)
authors.type: if 100 or 700 then person, if 710 then organisation
"""
if not key[4] == '2':
author = {}
author['type'] = 'person'
if value.get('0'):
refs = utils.force_list(value.get('0'))
for ref in refs:
ref = get_person_link(marc21.bib_id, ref, key, value)
if ref:
author['$ref'] = ref
# we do not have a $ref
if not author.get('$ref'):
author['name'] = ''
if value.get('a'):
data = not_repetitive(marc21.bib_id, key, value, 'a')
author['name'] = remove_trailing_punctuation(data)
author_subs = utils.force_list(value.get('b'))
if author_subs:
for author_sub in author_subs:
author['name'] += ' ' + \
remove_trailing_punctuation(author_sub)
if key[:3] == '710':
author['type'] = 'organisation'
else:
if value.get('c'):
data = not_repetitive(marc21.bib_id, key, value, 'c')
author['qualifier'] = remove_trailing_punctuation(data)
if value.get('d'):
data = not_repetitive(marc21.bib_id, key, value, 'd')
author['date'] = remove_trailing_punctuation(data)
return author
else:
return None
def build_place_or_agent_data(code, label):
place_or_agent_data = None
type_per_code = {'a': 'bf:Place', 'b': 'bf:Agent'}
value = remove_trailing_punctuation(label)
if value:
place_or_agent_data = {
'type': type_per_code[code],
'label': [{
'value': value
}]
}
return place_or_agent_data
def marc21_to_description(self, key, value):
"""Get extent, otherMaterialCharacteristics, formats.
extent: 300$a (the first one if many)
otherMaterialCharacteristics: 300$b (the first one if many)
formats: 300 [$c repetitive]
"""
if value.get('a'):
if not self.get('extent', None):
self['extent'] = remove_trailing_punctuation(
utils.force_list(value.get('a'))[0])
if value.get('b'):
if self.get('otherMaterialCharacteristics', []) == []:
self['otherMaterialCharacteristics'] = remove_trailing_punctuation(
utils.force_list(value.get('b'))[0])
if value.get('c'):
formats = self.get('formats', None)
if not formats:
data = value.get('c')
formats = list(utils.force_list(data))
return formats
def build_agent_data(code, label, index, link):
type_per_code = {'a': 'bf:Place', 'b': 'bf:Agent'}
agent_data = {
'type': type_per_code[code],
'label': [{
'value': remove_trailing_punctuation(label)
}]
}
try:
alt_gr = marc21.alternate_graphic['264'][link]
subfield = \
marc21.get_subfields(alt_gr['field'])[index]
agent_data['label'].append({
'value':
remove_trailing_punctuation(subfield),
'language':
marc21.get_language_script(alt_gr['script'])
})
except Exception as err:
pass
return agent_data
def build_place_or_agent_data(code, label, add_country):
place_or_agent_data = None
type_per_code = {'a': 'bf:Place', 'b': 'bf:Agent'}
value = remove_trailing_punctuation(label)
if value:
place_or_agent_data = {
'type': type_per_code[code],
'label': [{
'value': value
}]
}
if add_country and marc21.country:
place_or_agent_data['country'] = marc21.country
return place_or_agent_data
def build_place_or_agent_data(code, label, index, add_country):
type_per_code = {'a': 'bf:Place', 'c': 'bf:Agent'}
place_or_agent_data = {
'type': type_per_code[code],
'label': [{
'value': remove_trailing_punctuation(label)
}]
}
if add_country:
# country from 102
field_102 = unimarctojson.get_fields(tag='102')
if field_102:
field_102 = field_102[0]
country_codes = unimarctojson.get_subfields(field_102, 'a')
if country_codes:
place_or_agent_data['country'] = country_codes[0].lower()
return place_or_agent_data
def publication_statement_text(provision_activity):
"""Create publication statement from place, agent and date values."""
punctuation = {
'bf:Place': ' ; ',
'bf:Agent': ' ; ',
'Date': ', '
}
statement_with_language = {'default': ''}
last_statement_type = None
# Perform each statement entries to build the best possible string
for statement in provision_activity.get('statement', []):
for label in statement['label']:
language = label.get('language', 'default')
statement_with_language.setdefault(language, '')
if statement_with_language[language]:
if last_statement_type == statement['type']:
statement_with_language[language] += punctuation[
last_statement_type
]
elif statement['type'] == 'bf:Place':
statement_with_language[language] += ' ; '
elif statement['type'] == 'Date':
statement_with_language[language] += ', '
else:
statement_with_language[language] += ' : '
statement_with_language[language] += label['value']
last_statement_type = statement['type']
# date field: remove ';' and append
statement_text = []
for key, value in statement_with_language.items():
value = remove_trailing_punctuation(value)
if display_alternate_graphic_first(key):
statement_text.insert(0, {'value': value, 'language': key})
else:
statement_text.append({'value': value, 'language': key})
return statement_text
def unimarc_to_contribution(self, key, value):
"""Get contribution.
contribution: loop:
700 Nom de personne – Responsabilité principale
701 Nom de personne – Autre responsabilité principale
702 Nom de personne – Responsabilité secondaire
710 Nom de collectivité – Responsabilité principale
711 Nom de collectivité – Autre responsabilité principale
712 Nom de collectivité – Responsabilité secondaire
"""
agent = {}
agent['preferred_name'] = ', '.join(utils.force_list(value.get('a', '')))
agent['type'] = 'bf:Person'
if agent['preferred_name']:
if value.get('b'):
agent['preferred_name'] += \
', ' + ', '.join(utils.force_list(value.get('b')))
if key[:3] in ['700', '701', '702', '703']:
if value.get('d'):
agent['numeration'] = value.get('d')
if value.get('c'):
agent['qualifier'] = value.get('c')
if value.get('f'):
date = utils.force_list(value.get('f'))[0]
date = date.replace('-....', '-')
dates = date.split('-')
try:
date_of_birth = dates[0].strip()
if date_of_birth:
agent['date_of_birth'] = date_of_birth
except Exception:
pass
try:
date_of_death = dates[1].strip()
if date_of_death:
agent['date_of_death'] = date_of_death
except Exception:
pass
if key[:3] in ['710', '711', '712']:
agent['type'] = 'bf:Organisation'
agent['conference'] = key[3] == '1'
if agent['preferred_name']:
if value.get('c'):
agent['preferred_name'] += \
', ' + ', '.join(utils.force_list(value.get('c')))
if value.get('d'):
conference_number = utils.force_list(value.get('d'))[0]
agent['conference_number'] = remove_trailing_punctuation(
conference_number
).lstrip('(').rstrip(')')
if value.get('e'):
conference_place = utils.force_list(value.get('e'))[0]
agent['conference_place'] = remove_trailing_punctuation(
conference_place
).lstrip('(').rstrip(')')
if value.get('f'):
conference_date = utils.force_list(value.get('f'))[0]
agent['conference_date'] = remove_trailing_punctuation(
conference_date
).lstrip('(').rstrip(')')
IDREF_ROLE_CONV = {
"070": "aut",
"230": "cmp",
"205": "ctb",
"340": "edt",
"420": "hnr",
"440": "ill",
"600": "pht",
"590": "prf",
"730": "trl",
"080": "aui",
"160": "bsl",
"220": "com",
"300": "drt",
"430": "ilu",
"651": "pbd",
"350": "egr",
"630": "pro",
"510": "ltg",
"365": "exp",
"727": "dgs",
"180": "ctg",
"220": "com",
"210": "cmm",
"200": "chr",
"110": "bnd",
"720": "ato",
"030": "arr",
"020": "ann",
"632": "adi",
"005": "act",
"390": "fmo",
"545": "mus"
}
roles = []
if value.get('4'):
for role in utils.force_list(value.get('4')):
role_conv = IDREF_ROLE_CONV.get(role)
if role_conv:
roles.append(role_conv)
roles = list(set(roles))
if not roles:
roles = ['aut']
return {
'agent': agent,
'role': roles
}
def marc21_to_work_access_point(self, key, value):
"""Get work access point."""
"""
* "date_of_work": "[130$f|730$f]"
* "miscellaneous_information": "[130$g|130$s|730$g|730$s]"
* "language": "[130$l|730$l]"
* "form_subdivision": ["[130$k|730$k]"]
* "medium_of_performance_for_music": ["[130$m|730$m]"]
* "arranged_statement_for_music": "[130$o|730$o]"
* "key_for_music": "[130$r|730$r]"
[1] Nettoyer la chaîne: supprimer la ponctuation finale "/:;.,=",
supprimer en particulier la chaine ". - "
"""
work = {}
tag_key = key[:3]
part_list = TitlePartList(part_number_code='n', part_name_code='p')
part_selection = {'n', 'p'}
items = get_field_items(value)
if tag_key in ['130', '730']:
work_selection = {'a', 'f', 'k', 'l', 'm', 'o', 'r'}
miscellaneous_selection = {'g', 's'}
miscellaneous_parts = []
# _WORK_ACCESS_POINT.get(subfield_code)
for blob_key, blob_value in items:
if blob_key in work_selection:
if blob_key in {'k', 'm'}:
datas = work.get(_WORK_ACCESS_POINT.get(blob_key), [])
datas.append(blob_value)
work[_WORK_ACCESS_POINT.get(blob_key)] = datas
else:
work[_WORK_ACCESS_POINT.get(blob_key)] = blob_value
if blob_key in miscellaneous_selection:
miscellaneous_parts.append(blob_value)
if blob_key in part_selection:
part_list.update_part(blob_value, blob_key, blob_value)
if miscellaneous_parts:
work['miscellaneous_information'] = '. '.join(miscellaneous_parts)
if tag_key == '240':
for blob_key, blob_value in items:
if blob_key in {'a'}:
# work[_WORK_ACCESS_POINT.get('a')] = value.get('a')
work[_WORK_ACCESS_POINT.get(blob_key)] = blob_value
if blob_key in part_selection:
part_list.update_part(blob_value, blob_key, blob_value)
field_100 = marc21.get_fields('100')
if field_100:
agent = {}
for blob_key, blob_value in field_100[0].get('subfields').items():
agent['type'] = 'bf:Person'
if blob_key == 'a':
# numeration = not_repetitive(
# marc21.bib_id, marc21.bib_id, blob_key, blob_value, 'b')
agent['preferred_name'] = remove_trailing_punctuation(
blob_value)
if blob_key == 'b':
# numeration = not_repetitive(
# marc21.bib_id, marc21.bib_id, blob_key, blob_value, 'b')
agent['numeration'] = remove_trailing_punctuation(
blob_value)
if blob_key == 'c':
# qualifier = not_repetitive(
# marc21.bib_id, marc21.bib_id, blob_key, blob_value, 'c')
agent['qualifier'] = remove_trailing_punctuation(
blob_value)
if blob_key == 'd':
# date = not_repetitive(
# marc21.bib_id, marc21.bib_id, blob_key, blob_value, 'd')
date = blob_value.rstrip(',')
dates = remove_trailing_punctuation(date).split('-')
try:
date_of_birth = dates[0].strip()
if date_of_birth:
agent['date_of_birth'] = date_of_birth
except Exception:
pass
try:
date_of_death = dates[1].strip()
if date_of_death:
agent['date_of_death'] = date_of_death
except Exception:
pass
if blob_key == 'q':
# fuller_form_of_name = not_repetitive(
# marc21.bib_id, marc21.bib_id, blob_key, blob_value, 'q')
agent['fuller_form_of_name'] = remove_trailing_punctuation(
blob_value).lstrip('(').rstrip(')')
work['agent'] = agent
the_part_list = part_list.get_part_list()
if the_part_list:
work['part'] = the_part_list
if work:
work_access_points = self.get('work_access_point', [])
work_access_points.append(work)
self['work_access_point'] = work_access_points
def build_place_or_agent_data(code, label):
type_per_code = {'a': 'bf:Place', 'b': 'bf:Agent'}
return {'type': type_per_code[code], 'label': [{'value': value}]} \
if (value := remove_trailing_punctuation(label)) else None
def marc21_to_contribution(self, key, value):
"""Get contribution."""
if not key[4] == '2' and key[:3] in ['100', '700', '710', '711']:
agent = {}
if value.get('0'):
refs = utils.force_list(value.get('0'))
for ref in refs:
ref = get_person_link(marc21.bib_id, marc21.rero_id, ref, key,
value)
if ref:
agent['$ref'] = ref
# we do not have a $ref
if not agent.get('$ref') and value.get('a'):
agent = {'type': 'bf:Person'}
if value.get('a'):
name = not_repetitive(marc21.bib_id, marc21.rero_id, key,
value, 'a').rstrip('.')
if name:
agent['preferred_name'] = name
# 100|700 Person
if key[:3] in ['100', '700']:
if value.get('b'):
numeration = not_repetitive(marc21.bib_id, marc21.rero_id,
key, value, 'b')
numeration = remove_trailing_punctuation(numeration)
if numeration:
agent['numeration'] = numeration
if value.get('c'):
qualifier = not_repetitive(marc21.bib_id, marc21.rero_id,
key, value, 'c')
agent['qualifier'] = remove_trailing_punctuation(qualifier)
if value.get('d'):
date = not_repetitive(marc21.bib_id, marc21.rero_id, key,
value, 'd')
date = date.rstrip(',')
dates = remove_trailing_punctuation(date).split('-')
try:
date_of_birth = dates[0].strip()
if date_of_birth:
agent['date_of_birth'] = date_of_birth
except Exception:
pass
try:
date_of_death = dates[1].strip()
if date_of_death:
agent['date_of_death'] = date_of_death
except Exception:
pass
if value.get('q'):
fuller_form_of_name = not_repetitive(
marc21.bib_id, marc21.rero_id, key, value, 'q')
fuller_form_of_name = remove_trailing_punctuation(
fuller_form_of_name).lstrip('(').rstrip(')')
if fuller_form_of_name:
agent['fuller_form_of_name'] = fuller_form_of_name
# 710|711 Organisation
elif key[:3] in ['710', '711']:
agent['type'] = 'bf:Organisation'
if key[:3] == '711':
agent['conference'] = True
else:
agent['conference'] = False
if value.get('b'):
subordinate_units = []
for subordinate_unit in utils.force_list(value.get('b')):
subordinate_units.append(subordinate_unit.rstrip('.'))
agent['subordinate_unit'] = subordinate_units
if value.get('e'):
subordinate_units = agent.get('subordinate_unit', [])
for subordinate_unit in utils.force_list(value.get('e')):
subordinate_units.append(subordinate_unit.rstrip('.'))
agent['subordinate_unit'] = subordinate_units
if value.get('n'):
conference_number = not_repetitive(marc21.bib_id,
marc21.rero_id, key,
value, 'n')
conference_number = remove_trailing_punctuation(
conference_number).lstrip('(').rstrip(')')
if conference_number:
agent['conference_number'] = conference_number
if value.get('d'):
conference_date = not_repetitive(marc21.bib_id,
marc21.rero_id, key,
value, 'd')
conference_date = remove_trailing_punctuation(
conference_date).lstrip('(').rstrip(')')
if conference_date:
agent['conference_date'] = conference_date
if value.get('c'):
conference_place = not_repetitive(marc21.bib_id,
marc21.rero_id, key,
value, 'c')
conference_place = remove_trailing_punctuation(
conference_place).lstrip('(').rstrip(')')
if conference_place:
agent['conference_place'] = conference_place
if value.get('4'):
roles = []
for role in utils.force_list(value.get('4')):
if len(role) != 3:
error_print('WARNING CONTRIBUTION ROLE LENGTH:',
marc21.bib_id, marc21.rero_id, role)
role = role[:3]
if role == 'sce':
error_print('WARNING CONTRIBUTION ROLE SCE:',
marc21.bib_id, marc21.rero_id, 'sce --> aus')
role = 'aus'
role = role.lower()
if role not in _CONTRIBUTION_ROLE:
error_print('WARNING CONTRIBUTION ROLE DEFINITION:',
marc21.bib_id, marc21.rero_id, role)
role = 'ctb'
roles.append(role)
else:
if key[:3] == '100':
roles = ['cre']
elif key[:3] == '711':
roles = ['aut']
else:
roles = ['ctb']
if agent:
return {'agent': agent, 'role': list(set(roles))}
return None
def marc21_to_subjects(self, key, value):
"""Get subjects.
- create an object :
genreForm : for the field 655
subjects : for 6xx with $2 rero
subjects_imported : for 6xx having indicator 2 '0' or '2'
"""
type_per_tag = {
'600': 'bf:Person',
'610': 'bf:Organization',
'611': 'bf:Organization',
'600t': 'bf:Work',
'610t': 'bf:Work',
'611t': 'bf:Work',
'630': 'bf:Work',
'650': 'bf:Topic', # or bf:Temporal, changed by code
'651': 'bf:Place',
'655': 'bf:Topic'
}
ref_link_per_tag = {
'600': 'IdRef agent',
'610': 'IdRef agent',
'611': 'IdRef agent',
'600t': 'IdRef work',
'610t': 'IdRef work',
'611t': 'IdRef work',
'630': 'IdRef work',
'650': 'RERO RAMEAU concept',
'651': 'Idref place',
'655': 'RERO RAMEAU concept'
}
field_data_per_tag = {
'600': 'preferred_name',
'610': 'preferred_name',
'611': 'preferred_name',
'600t': 'title',
'610t': 'title',
'611t': 'title',
'630': 'title',
'650': 'term',
'651': 'preferred_name',
'655': 'term'
}
subfield_code_per_tag = {
'600': 'abcd',
'610': 'ab',
'611': 'acden',
'600t': 'tpn',
'610t': 'tpn',
'611t': 't',
'630': 'apn',
'650': 'a',
'651': 'a',
'655': 'a'
}
conference_per_tag = {'610': False, '611': True}
source_per_indicator_2 = {'0': 'LCSH', '2': 'MeSH'}
indicator_2 = key[4]
tag_key = key[:3]
subfields_2 = utils.force_list(value.get('2'))
subfield_2 = None
if subfields_2:
subfield_2 = subfields_2[0]
subfields_a = utils.force_list(value.get('a', []))
if subfield_2 == 'rero':
has_dollar_t = value.get('t')
if tag_key in ('600', '610', '611') and has_dollar_t:
tag_key += 't'
data_type = type_per_tag[tag_key]
start_with_digit = False
if tag_key == '650':
for subfield_a in subfields_a:
start_with_digit_regexp = re.compile(r'^\d')
match = start_with_digit_regexp.search(subfield_a)
if match:
data_type = 'bf:Temporal'
break
subject = {
'type': data_type,
}
string_build = build_string_from_subfields(
value, subfield_code_per_tag[tag_key])
if (tag_key == '655'):
# remove the square brackets
string_build = re.sub(r'^\[(.*)\]$', r'\1', string_build)
subject[field_data_per_tag[tag_key]] = string_build
if tag_key in ('610', '611'):
subject['conference'] = conference_per_tag[tag_key]
if tag_key in ('600t', '610t', '611t'):
creator_tag_key = tag_key[:3] # to keep only tag: 600, 610, 611
subject['creator'] = remove_trailing_punctuation(
build_string_from_subfields(
value, subfield_code_per_tag[creator_tag_key]), '.', '.')
field_key = 'subjects'
if tag_key == '655':
field_key = 'genreForm'
subfields_0 = utils.force_list(value.get('0'))
if data_type in ['bf:Person', 'bf:Organisation'] and subfields_0:
ref = get_contribution_link(marc21.bib_id, marc21.rero_id,
subfields_0[0], key)
if ref:
subject = {
'$ref': ref,
'type': data_type,
}
if not subject.get('$ref'):
identifier = build_identifier(value)
if identifier:
subject['identifiedBy'] = identifier
if subject.get('$ref') or subject.get(field_data_per_tag[tag_key]):
subjects = self.get(field_key, [])
subjects.append(subject)
self[field_key] = subjects
elif subfield_2 == 'rerovoc' or indicator_2 in ['0', '2']:
term_string = build_string_from_subfields(
value, 'abcdefghijklmnopqrstuvwxyz', ' - ')
if term_string:
if subfield_2 == 'rerovoc':
source = 'rerovoc'
else:
source = source_per_indicator_2[indicator_2]
subject_imported = {
'type': type_per_tag[tag_key],
'source': source
}
subject_imported[field_data_per_tag[tag_key]] = term_string
if tag_key in ('610', '611'):
subject_imported['conference'] = conference_per_tag[tag_key]
subjects_imported = self.get('subjects_imported', [])
if subject_imported:
subjects_imported.append(subject_imported)
self['subjects_imported'] = subjects_imported
if value.get('c'):
qualifier = utils.force_list(value.get('c'))[0]
agent['qualifier'] = remove_trailing_punctuation(qualifier)
if value.get('d'):
date = utils.force_list(value.get('d'))[0]
date = date.rstrip(',')
dates = remove_trailing_punctuation(date).split('-')
with contextlib.suppress(Exception):
if date_of_birth := dates[0].strip():
agent['date_of_birth'] = date_of_birth
with contextlib.suppress(Exception):
if date_of_death := dates[1].strip():
agent['date_of_death'] = date_of_death
if value.get('q'):
fuller_form_of_name = utils.force_list(value.get('q'))[0]
agent['fuller_form_of_name'] = remove_trailing_punctuation(
fuller_form_of_name).lstrip('(').rstrip(')')
elif key[:3] in ['710', '711']:
agent['type'] = 'bf:Organisation'
agent['conference'] = key[:3] == '711'
if value.get('e'):
subordinate_units = [
subordinate_unit.rstrip('.')
for subordinate_unit in utils.force_list(value.get('e'))
]
agent['subordinate_unit'] = subordinate_units
if value.get('n'):
numbering = utils.force_list(value.get('n'))[0]
agent['numbering'] = remove_trailing_punctuation(numbering).lstrip(
'(').rstrip(')')
def marc21_to_contribution(self, key, value):
"""Get contribution."""
if not key[4] == '2' and key[:3] in ['100', '700', '710', '711']:
agent = {'type': 'bf:Person'}
if value.get('a'):
name = utils.force_list(value.get('a'))[0]
agent['preferred_name'] = remove_trailing_punctuation(name)
# 100|700 Person
if key[:3] in ['100', '700']:
if value.get('b'):
numeration = utils.force_list(value.get('b'))[0]
agent['numeration'] = remove_trailing_punctuation(numeration)
if value.get('c'):
qualifier = utils.force_list(value.get('c'))[0]
agent['qualifier'] = remove_trailing_punctuation(qualifier)
if value.get('d'):
date = utils.force_list(value.get('d'))[0]
date = date.rstrip(',')
dates = remove_trailing_punctuation(date).split('-')
try:
date_of_birth = dates[0].strip()
if date_of_birth:
agent['date_of_birth'] = date_of_birth
except Exception:
pass
try:
date_of_death = dates[1].strip()
if date_of_death:
agent['date_of_death'] = date_of_death
except Exception:
pass
if value.get('q'):
fuller_form_of_name = utils.force_list(value.get('q'))[0]
agent['fuller_form_of_name'] = remove_trailing_punctuation(
fuller_form_of_name).lstrip('(').rstrip(')')
# 710|711 Organisation
elif key[:3] in ['710', '711']:
agent['type'] = 'bf:Organisation'
if key[:3] == '711':
agent['conference'] = True
else:
agent['conference'] = False
if value.get('e'):
subordinate_units = []
for subordinate_unit in utils.force_list(value.get('e')):
subordinate_units.append(subordinate_unit.rstrip('.'))
agent['subordinate_unit'] = subordinate_units
if value.get('n'):
conference_number = utils.force_list(value.get('n'))[0]
agent['conference_number'] = remove_trailing_punctuation(
conference_number).lstrip('(').rstrip(')')
if value.get('d'):
conference_date = utils.force_list(value.get('d'))[0]
agent['conference_date'] = remove_trailing_punctuation(
conference_date).lstrip('(').rstrip(')')
if value.get('c'):
conference_place = utils.force_list(value.get('c'))[0]
agent['conference_place'] = remove_trailing_punctuation(
conference_place).lstrip('(').rstrip(')')
roles = ['aut']
if value.get('4'):
roles = []
for role in utils.force_list(value.get('4')):
roles.append(role)
else:
if key[:3] == '100':
roles = ['cre']
elif key[:3] == '711':
roles = ['aut']
else:
roles = ['ctb']
return {'agent': agent, 'role': roles}
else:
return None
def marc21_to_subjects_6XX(self, key, value):
"""Get subjects.
- create an object :
genreForm : for the field 655
subjects : for 6xx with $2 rero
subjects_imported : for 6xx having indicator 2 '0' or '2'
"""
def perform_subdivisions(field):
"""Perform subject subdivisions from MARC field."""
subdivisions = {
'v': 'genreForm_subdivisions',
'x': 'topic_subdivisions',
'y': 'temporal_subdivisions',
'z': 'place_subdivisions'
}
for code, subdivision in subdivisions.items():
for subfield_value in utils.force_list(value.get(code, [])):
field.setdefault(subdivision, []).append(subfield_value)
type_per_tag = {
'600': DocumentSubjectType.PERSON,
'610': DocumentSubjectType.ORGANISATION,
'611': DocumentSubjectType.ORGANISATION,
'600t': DocumentSubjectType.WORK,
'610t': DocumentSubjectType.WORK,
'611t': DocumentSubjectType.WORK,
'630': DocumentSubjectType.WORK,
'650': DocumentSubjectType.TOPIC, # or bf:Temporal, changed by code
'651': DocumentSubjectType.PLACE,
'655': DocumentSubjectType.TOPIC
}
field_data_per_tag = {
'600': 'preferred_name',
'610': 'preferred_name',
'611': 'preferred_name',
'600t': 'title',
'610t': 'title',
'611t': 'title',
'630': 'title',
'650': 'term',
'651': 'preferred_name',
'655': 'term'
}
subfield_code_per_tag = {
'600': 'abcd',
'610': 'ab',
'611': 'acden',
'600t': 'tpn',
'610t': 'tpn',
'611t': 't',
'630': 'apn',
'650': 'a',
'651': 'a',
'655': 'a'
}
conference_per_tag = {
'610': False,
'611': True
}
source_per_indicator_2 = {
'0': 'LCSH',
'2': 'MeSH'
}
indicator_2 = key[4]
tag_key = key[:3]
subfields_2 = utils.force_list(value.get('2'))
subfield_2 = subfields_2[0] if subfields_2 else None
subfields_a = utils.force_list(value.get('a', []))
if subfield_2 in ['rero', 'gnd', 'idref']:
if tag_key in ['600', '610', '611'] and value.get('t'):
tag_key += 't'
data_type = type_per_tag[tag_key]
# `data_type` is Temporal if tag is 650 and a $a start with digit.
if tag_key == '650':
for subfield_a in subfields_a:
if subfield_a[0].isdigit():
data_type = 'bf:Temporal'
break
subject = {
'type': data_type,
}
string_build = build_string_from_subfields(
value, subfield_code_per_tag[tag_key])
if tag_key == '655':
# remove the square brackets
string_build = re.sub(r'^\[(.*)\]$', r'\1', string_build)
subject[field_data_per_tag[tag_key]] = string_build
if tag_key in ['610', '611']:
subject['conference'] = conference_per_tag[tag_key]
if tag_key in ['600t', '610t', '611t']:
creator_tag_key = tag_key[:3] # to keep only tag: 600, 610, 611
subject['creator'] = remove_trailing_punctuation(
build_string_from_subfields(
value, subfield_code_per_tag[creator_tag_key]), '.', '.')
field_key = 'genreForm' if tag_key == '655' else 'subjects'
subfields_0 = utils.force_list(value.get('0'))
if data_type in [DocumentSubjectType.PERSON,
DocumentSubjectType.ORGANISATION] and subfields_0:
ref = get_contribution_link(marc21.bib_id, marc21.rero_id,
subfields_0[0], key)
if ref:
subject = {
'$ref': ref,
'type': data_type,
}
if not subject.get('$ref'):
identifier = build_identifier(value)
if identifier:
subject['identifiedBy'] = identifier
perform_subdivisions(subject)
if subject.get('$ref') or subject.get(field_data_per_tag[tag_key]):
subjects = self.get(field_key, [])
subjects.append(subject)
self[field_key] = subjects
elif indicator_2 in ['0', '2']:
term_string = build_string_from_subfields(
value, 'abcdefghijklmnopqrstuw', ' - ')
if term_string:
subject_imported = {
'type': type_per_tag[tag_key],
'source': source_per_indicator_2[indicator_2],
field_data_per_tag[tag_key]: term_string.rstrip('.')
}
perform_subdivisions(subject_imported)
if tag_key in ['610', '611']:
subject_imported['conference'] = conference_per_tag[tag_key]
subjects_imported = self.get('subjects_imported', [])
if subject_imported:
subjects_imported.append(subject_imported)
self['subjects_imported'] = subjects_imported