def load_dict(meta_dict):
if "rec_class" not in meta_dict:
return Common(meta_dict)
elif meta_dict["rec_class"] == "Document":
return Document(meta_dict)
elif meta_dict["rec_class"] == "Person":
return Person(meta_dict)
elif meta_dict["rec_class"] == "Orgunit":
return Orgunit(meta_dict)
elif meta_dict["rec_class"] == "Project":
return Project(meta_dict)
elif meta_dict["rec_class"] == "Event":
return Event(meta_dict)
elif meta_dict["rec_class"] == "Family":
return Family(meta_dict)
elif meta_dict["rec_class"] == "Field":
return Field(meta_dict)
elif meta_dict["rec_class"] == "Resource":
return Resource(meta_dict)
elif meta_dict["rec_class"] == "Target":
return Target(meta_dict)
elif meta_dict["rec_class"] == "Type":
return Type(meta_dict)
elif meta_dict["rec_class"] == "Collection":
return Collection(meta_dict)
else:
logging.debug(jsonbson.dumps_bson(meta_dict))
logging.warning("Unknown rec_class: {O}".format(meta_dict["rec_class"]))
return Common(meta_dict)
def get_rml_relationships(rml):
""" relationship -> relationships """
result = {}
rml_relationships = rml.findall(xmletree.prefixtag("rml", "relationship"))
if rml_relationships is not None:
relationships = []
for rml_relationship in rml_relationships:
if rml_relationship is not None:
# name -> agent.name
name = get_rml_element_text(rml_relationship, "name")
relationship = creator_service.formatted_name_to_creator(name, constants.REC_CLASS_PERSON, None)
if relationship is None:
relationship = {}
relationship["agent"] = Person()
# identifier -> agent.rec_id & agent.identifiers
relationship["agent"].update(get_rml_identifiers(rml_relationship))
# relationType -> relation_type
relationship.update(get_rml_element_text_and_set_key(rml_relationship, "relationType", "relation_type"))
# descriptions -> descriptions
relationship.update(get_rml_textlangs_and_set_key(rml_relationship, "description", "descriptions"))
if relationship is not None:
relationships.append(relationship)
if relationships:
result["relationships"] = relationships
return result
def convert_creators(sum_doc, sum_authors_key, sum_affiliations_key, creator_type, role):
creators = []
if sum_authors_key in sum_doc:
sum_authors = sum_doc[sum_authors_key]
if sum_authors:
affiliations_dict = {}
if sum_affiliations_key in sum_doc:
sum_affiliations = sum_doc[sum_affiliations_key]
if sum_affiliations:
for index, affiliation in enumerate(sum_affiliations):
affiliations_dict[index + 1] = affiliation
for author in sum_authors:
creator = Creator()
if "surname" in author and "givenname" in author:
person = Person()
person.set_key_if_not_none("name_family", author["surname"])
person.set_key_if_not_none("name_given", author["givenname"])
creator["agent"] = person
if role:
creator["roles"] = [role]
else:
formatted_name = ""
if "fullname" in author:
formatted_name = author["fullname"]
elif "name" in author:
formatted_name = author["name"]
creator = creator_service.formatted_name_to_creator(formatted_name, creator_type, role)
if "sequence" in author and author["sequence"] in affiliations_dict:
creator["affiliation"] = affiliations_dict[author["sequence"]]
# todo : location dans le cas des DissertationSchool_xml
if creator:
creators.append(creator)
return creators
def formatted_name_to_creator(formatted_name, rec_class, role):
#logging.debug("formatted_name_to_creator formatted_name: {}".format(formatted_name))
if formatted_name:
formatted_name = formatted_name.strip()
event = None
family = None
orgunit = None
person = None
#logging.debug("formatted_name: {}".format(formatted_name))
# rec_class determination
if rec_class is None or rec_class not in [
constants.REC_CLASS_EVENT, constants.REC_CLASS_FAMILY,
constants.REC_CLASS_ORGUNIT, constants.REC_CLASS_PERSON
]:
for event_term in creator_event_terms:
if event_term in formatted_name.lower():
rec_class = constants.REC_CLASS_EVENT
break
for orgunit_term in creator_orgunit_terms:
if orgunit_term in formatted_name.lower():
rec_class = constants.REC_CLASS_ORGUNIT
break
if rec_class is None:
rec_class = constants.REC_CLASS_PERSON
creator = Creator()
if role:
creator["roles"] = [role]
if rec_class == constants.REC_CLASS_EVENT:
event = Event()
event["title"] = formatted_name
creator["agent"] = event
elif rec_class == constants.REC_CLASS_ORGUNIT:
orgunit = Orgunit()
orgunit["name"] = formatted_name
creator["agent"] = orgunit
elif rec_class == constants.REC_CLASS_PERSON or rec_class == constants.REC_CLASS_FAMILY:
# class is "Person" or "Family"
name_given = ""
name_middle = ""
name_family = ""
name_prefix = ""
name_terms_of_address = ""
date_birth = ""
date_death = ""
parenthesis_index = formatted_name.rfind("(")
if parenthesis_index != -1:
#may be like: name (date_birth-date_death)
dates_part = formatted_name[parenthesis_index + 1:-1].strip()
date_birth = dates_part[:4]
date_death = dates_part[5:]
if date_death == "....":
date_death = ""
formatted_name = formatted_name[:parenthesis_index].strip()
slash_index = formatted_name.find("/")
if slash_index != -1:
#like: name/affiliation
affiliation_name = formatted_name[slash_index + 1:].strip()
formatted_name = formatted_name[:slash_index].strip()
commaspacejrdot_index = formatted_name.rfind(", Jr.")
if (commaspacejrdot_index != -1):
#like "Paul B. Harvey, Jr."
formatted_name = formatted_name[:commaspacejrdot_index].strip()
name_middle = "Jr."
#Is it formatted like "Family, Given" or "Given Family" ?
comma_index = formatted_name.find(",")
if comma_index == -1:
space_index = formatted_name.rfind(" ")
#logging.debug(formatted_name)
#logging.debug(space_index)
if space_index != -1:
#like Given Family
name_given = formatted_name[:space_index].strip()
name_family = formatted_name[space_index + 1:].strip()
else:
#like Family
name_family = formatted_name.strip()
else:
#like Family, Given
name_family = formatted_name[:comma_index].strip()
name_given = formatted_name[comma_index + 1:].strip()
# manage the terms_of_address and particule
for term_of_address in creator_person_terms_of_address:
if name_family and name_family.lower().startswith(
term_of_address + " "):
name_terms_of_address = name_family[:len(term_of_address)]
name_family = name_family[len(term_of_address):].strip()
if name_given:
if name_given.lower().endswith(" " + term_of_address):
name_terms_of_address = name_given[-len(term_of_address
):]
name_given = name_given[:-len(term_of_address)].strip()
if name_given.lower().startswith(term_of_address + " "):
name_terms_of_address = name_given[:len(term_of_address
)]
name_given = name_given[len(term_of_address):].strip()
if name_given.lower() == term_of_address:
name_terms_of_address = name_given
name_given = None
# Be careful with a particule inside the name like: Viveiros de Castro, Eduardo
for particule in creator_particule:
if name_family and name_family.lower().startswith(particule +
" "):
name_prefix = name_family[0:len(particule)]
name_family = name_family[len(particule):].strip()
if name_given:
if name_given.lower().endswith(" " + particule):
name_prefix = name_given[-len(particule):]
name_given = name_given[:-len(particule)].strip()
if name_given.lower().startswith(particule + " "):
name_prefix = name_given[:len(particule)]
name_given = name_given[len(particule):].strip()
if name_given.lower() == particule:
name_prefix = name_given
name_given = None
if rec_class == constants.REC_CLASS_PERSON:
person = Person()
person.set_key_if_not_none("name_family", name_family)
person.set_key_if_not_none("name_given", name_given)
person.set_key_if_not_none("name_middle", name_middle)
person.set_key_if_not_none("name_terms_of_address",
name_terms_of_address)
person.set_key_if_not_none("name_prefix", name_prefix)
person.set_key_if_not_none("date_birth", date_birth)
person.set_key_if_not_none("date_death", date_death)
creator["agent"] = person
if 'affiliation_name' in vars() and affiliation_name:
#todo manage as an object
affiliation = Orgunit()
affiliation["name"] = affiliation_name
creator["affiliation"] = affiliation
elif rec_class == constants.REC_CLASS_FAMILY:
family = Family()
family.set_key_if_not_none("name_family", name_family)
creator["agent"] = family
#logging.debug(jsonbson.dumps_json(creator, True))
return creator
def formatted_name_to_creator(formatted_name, rec_class, role):
#logging.debug("formatted_name_to_creator formatted_name: {}".format(formatted_name))
if formatted_name:
formatted_name = formatted_name.strip()
event = None
family = None
orgunit = None
person = None
#logging.debug("formatted_name: {}".format(formatted_name))
# rec_class determination
if rec_class is None or rec_class not in [constants.REC_CLASS_EVENT, constants.REC_CLASS_FAMILY, constants.REC_CLASS_ORGUNIT, constants.REC_CLASS_PERSON]:
for event_term in creator_event_terms:
if event_term in formatted_name.lower():
rec_class = constants.REC_CLASS_EVENT
break
for orgunit_term in creator_orgunit_terms:
if orgunit_term in formatted_name.lower():
rec_class = constants.REC_CLASS_ORGUNIT
break
if rec_class is None:
rec_class = constants.REC_CLASS_PERSON
creator = Creator()
if role:
creator["roles"] = [role]
if rec_class == constants.REC_CLASS_EVENT:
event = Event()
event["title"] = formatted_name
creator["agent"] = event
elif rec_class == constants.REC_CLASS_ORGUNIT:
orgunit = Orgunit()
orgunit["name"] = formatted_name
creator["agent"] = orgunit
elif rec_class == constants.REC_CLASS_PERSON or rec_class == constants.REC_CLASS_FAMILY:
# class is "Person" or "Family"
name_given = ""
name_middle = ""
name_family = ""
name_prefix = ""
name_terms_of_address = ""
date_birth = ""
date_death = ""
parenthesis_index = formatted_name.rfind("(")
if parenthesis_index != -1:
#may be like: name (date_birth-date_death)
dates_part = formatted_name[parenthesis_index + 1:-1].strip()
date_birth = dates_part[:4]
date_death = dates_part[5:]
if date_death == "....":
date_death = ""
formatted_name = formatted_name[:parenthesis_index].strip()
slash_index = formatted_name.find("/")
if slash_index != -1:
#like: name/affiliation
affiliation_name = formatted_name[slash_index + 1:].strip()
formatted_name = formatted_name[:slash_index].strip()
commaspacejrdot_index = formatted_name.rfind(", Jr.")
if (commaspacejrdot_index != -1):
#like "Paul B. Harvey, Jr."
formatted_name = formatted_name[:commaspacejrdot_index].strip()
name_middle = "Jr."
#Is it formatted like "Family, Given" or "Given Family" ?
comma_index = formatted_name.find(",")
if comma_index == -1:
space_index = formatted_name.rfind(" ")
#logging.debug(formatted_name)
#logging.debug(space_index)
if space_index != -1:
#like Given Family
name_given = formatted_name[:space_index].strip()
name_family = formatted_name[space_index+1:].strip()
else:
#like Family
name_family = formatted_name.strip()
else:
#like Family, Given
name_family = formatted_name[:comma_index].strip()
name_given = formatted_name[comma_index+1:].strip()
# manage the terms_of_address and particule
for term_of_address in creator_person_terms_of_address:
if name_family and name_family.lower().startswith(term_of_address+" "):
name_terms_of_address = name_family[:len(term_of_address)]
name_family = name_family[len(term_of_address):].strip()
if name_given:
if name_given.lower().endswith(" "+term_of_address):
name_terms_of_address = name_given[-len(term_of_address):]
name_given = name_given[:-len(term_of_address)].strip()
if name_given.lower().startswith(term_of_address+" "):
name_terms_of_address = name_given[:len(term_of_address)]
name_given = name_given[len(term_of_address):].strip()
if name_given.lower() == term_of_address:
name_terms_of_address = name_given
name_given = None
# Be careful with a particule inside the name like: Viveiros de Castro, Eduardo
for particule in creator_particule:
if name_family and name_family.lower().startswith(particule+" "):
name_prefix = name_family[0:len(particule)]
name_family = name_family[len(particule):].strip()
if name_given:
if name_given.lower().endswith(" "+particule):
name_prefix = name_given[-len(particule):]
name_given = name_given[:-len(particule)].strip()
if name_given.lower().startswith(particule+" "):
name_prefix = name_given[:len(particule)]
name_given = name_given[len(particule):].strip()
if name_given.lower() == particule:
name_prefix = name_given
name_given = None
if rec_class == constants.REC_CLASS_PERSON:
person = Person()
person.set_key_if_not_none("name_family", name_family)
person.set_key_if_not_none("name_given", name_given)
person.set_key_if_not_none("name_middle", name_middle)
person.set_key_if_not_none("name_terms_of_address", name_terms_of_address)
person.set_key_if_not_none("name_prefix", name_prefix)
person.set_key_if_not_none("date_birth", date_birth)
person.set_key_if_not_none("date_death", date_death)
creator["agent"] = person
if 'affiliation_name' in vars() and affiliation_name:
#todo manage as an object
affiliation = Orgunit()
affiliation["name"] = affiliation_name
creator["affiliation"] = affiliation
elif rec_class == constants.REC_CLASS_FAMILY:
family = Family()
family.set_key_if_not_none("name_family", name_family)
creator["agent"] = family
#logging.debug(jsonbson.dumps_json(creator, True))
return creator
def rml_person_to_metajson(rml_person, source, rec_id_prefix):
""" person -> person """
person = Person()
# source
if source:
person["rec_source"] = source
# academicTitle, honorificTitle -> titles
person.update(get_rml_titles(rml_person))
# address -> addresses
person.update(get_rml_addresses(rml_person))
# affiliation -> affiliations
person.update(get_rml_affiliations(rml_person))
# award -> awards
person.update(get_rml_textlangs_and_set_key(rml_person, "award", "awards"))
# biography -> biographies
person.update(get_rml_textlangs_and_set_key(rml_person, "biography", "biographies"))
# dateOfBirth -> date_birth
person.update(get_rml_element_text_and_set_key(rml_person, "dateOfBirth", "date_birth"))
# dateOfDeath -> date_death
person.update(get_rml_element_text_and_set_key(rml_person, "dateOfDeath", "date_death"))
# degree -> degrees
person.update(get_rml_degrees(rml_person))
# email -> emails
person.update(get_rml_emails(rml_person))
# @fictitious -> fictitious
person.update(xmletree.get_element_attribute_as_boolean_and_set_key(rml_person, "fictitious", "fictitious"))
# firstname -> name_given
person.update(get_rml_element_text_and_set_key(rml_person, "firstname", "name_given"))
# identifier -> identifiers & rec_id
person.update(get_rml_identifiers(rml_person))
# image -> resources[i]
person.update(get_rml_images(rml_person, "picture"))
# instantMessage -> instant_messages
person.update(get_rml_instant_messages(rml_person))
# languageCapability -> language_capabilities
person.update(get_rml_language_capabilities(rml_person))
# lastname -> name_family
person.update(get_rml_element_text_and_set_key(rml_person, "lastname", "name_family"))
# lastnamePrefix -> name_prefix
person.update(get_rml_element_text_and_set_key(rml_person, "lastnamePrefix", "name_prefix"))
# lastnameSuffix -> name_suffix
person.update(get_rml_element_text_and_set_key(rml_person, "lastnameSuffix", "name_suffix"))
# middlename -> name_middle
person.update(get_rml_element_text_and_set_key(rml_person, "middlename", "name_middle"))
# nationality -> nationality
person.update(get_rml_element_text_and_set_key(rml_person, "nationality", "nationality"))
# nickname -> name_nick
person.update(get_rml_element_text_and_set_key(rml_person, "nickname", "name_nick"))
# note -> notes
person.update(get_rml_textlangs_and_set_key(rml_person, "note", "notes"))
# olBiography -> biographies_short
person.update(get_rml_textlangs_and_set_key(rml_person, "olBiography", "biographies_short"))
# ongoingResearch -> ongoing_researches
person.update(get_rml_ongoing_researches(rml_person))
# phone -> phones
person.update(get_rml_phones(rml_person))
# relationship -> relationships
person.update(get_rml_relationships(rml_person))
# researchCoverage -> research_coverages
person.update(get_rml_research_coverages(rml_person))
# responsability -> responsabilities
person.update(get_rml_textlangs_and_set_key(rml_person, "responsability", "responsabilities"))
# sex -> gender
person.update(get_rml_element_text_and_set_key(rml_person, "sex", "gender"))
# skill -> skills
person.update(get_rml_textlangs_and_set_key(rml_person, "skill", "skills"))
# teaching -> teachings
person.update(get_rml_teachings(rml_person))
# uri -> urls
person.update(get_rml_uris(rml_person))
return person
