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_teachings(rml):
""" teaching -> teachings """
result = {}
rml_teachings = rml.findall(xmletree.prefixtag("rml", "teaching"))
if rml_teachings is not None:
teachings = []
for rml_teaching in rml_teachings:
if rml_teaching is not None:
teaching = {}
# dateBegin -> date_begin
teaching.update(get_rml_element_text_and_set_key(rml_teaching, "dateBegin", "date_begin"))
# dateEnd -> date_end
teaching.update(get_rml_element_text_and_set_key(rml_teaching, "dateEnd", "date_end"))
# description -> descriptions[i]
teaching.update(get_rml_textlangs_and_set_key(rml_teaching, "description", "descriptions"))
# level -> level
teaching.update(get_rml_element_text_and_set_key(rml_teaching, "level", "level"))
# title -> title
teaching.update(get_rml_element_text_and_set_key(rml_teaching, "title", "title"))
# creators
# name -> creators[0].agent.name
name = get_rml_element_text(rml_teaching, "name")
creator = creator_service.formatted_name_to_creator(name, constants.REC_CLASS_ORGUNIT, "dgg")
if creator is None:
creator = Creator()
creator["agent"] = Orgunit()
creator["roles"] = "dgg"
# identifiers -> creators[0].agent.rec_id or creators[0].agent.identifiers
creator["agent"].update(get_rml_identifiers(rml_teaching))
if "name" in creator["agent"] or "rec_id" in creator["agent"] or "identifiers" in creator["agent"]:
teaching["creators"] = [creator]
if teaching is not None:
teachings.append(teaching)
if teachings:
result["teachings"] = teachings
return result
def get_rml_degrees(rml):
""" degree -> degrees """
result = {}
rml_degrees = rml.findall(xmletree.prefixtag("rml", "degree"))
if rml_degrees is not None:
degrees = []
for rml_degree in rml_degrees:
if rml_degree is not None:
degree = {}
# dateBegin -> date_begin
degree.update(get_rml_element_text_and_set_key(rml_degree, "dateBegin", "date_begin"))
# dateEnd -> date_end
degree.update(get_rml_element_text_and_set_key(rml_degree, "dateEnd", "date_end"))
# description -> descriptions
degree.update(get_rml_textlangs_and_set_key(rml_degree, "description", "descriptions"))
# level -> level
degree.update(get_rml_element_text_and_set_key(rml_degree, "level", "level"))
# title -> title
degree.update(get_rml_element_text_and_set_key(rml_degree, "title", "title"))
# creators
# name -> creators[0].agent.name
name = get_rml_element_text(rml_degree, "name")
creator = creator_service.formatted_name_to_creator(name, constants.REC_CLASS_ORGUNIT, "dgg")
if creator is None:
creator = Creator()
creator["agent"] = Orgunit()
creator["roles"] = ["dgg"]
# identifiers -> creators[0].agent.rec_id or creators[0].agent.identifiers
creator["agent"].update(get_rml_identifiers(rml_degree))
if "name" in creator["agent"] or "rec_id" in creator["agent"] or "identifiers" in creator["agent"]:
degree["creators"] = [creator]
if degree is not None:
degrees.append(degree)
if degrees:
result["degrees"] = degrees
return result
def get_rml_call(rml):
""" call -> call """
result = {}
rml_call = rml.find(xmletree.prefixtag("rml", "call"))
if rml_call is not None:
call = Call()
# funding -> funding
rml_funding = rml_call.find(xmletree.prefixtag("rml", "funding"))
if rml_funding is not None:
# name -> agent.name
name = get_rml_element_text(rml_funding, "name")
funding = creator_service.formatted_name_to_creator(name, constants.REC_CLASS_ORGUNIT, None)
if funding is None:
funding = Creator()
funding["agent"] = Orgunit()
# identifier -> agent.rec_id & agent.identifiers
funding["agent"].update(get_rml_identifiers(rml_funding))
# programme -> programme
funding.update(get_rml_element_text_and_set_key(rml_funding, "programme", "programme"))
# scheme -> scheme
funding.update(get_rml_element_text_and_set_key(rml_funding, "scheme", "scheme"))
# contribution -> contribution
funding.update(get_rml_money_and_set_key(rml_funding, "contribution", "contribution"))
if funding:
call["funding"] = funding
# identifier -> rec_id
call.update(get_rml_element_text_and_set_key(rml_call, "identifier", "rec_id"))
# title -> title
call.update(get_rml_element_text_and_set_key(rml_call, "title", "title"))
# year -> date_issued
call.update(get_rml_element_text_and_set_key(rml_call, "year", "date_issued"))
if call:
result["call"] = call
return result
def create_affiliation(rec_id, name, role=None, date_begin=None, date_end=None, preferred=False, descriptions=None):
affiliation = {}
if date_begin:
affiliation["date_begin"] = date_begin
if date_end:
affiliation["date_end"] = date_end
if descriptions:
affiliation["descriptions"] = descriptions
if preferred:
affiliation["preferred"] = preferred
if role:
affiliation["role"] = role
# agent
agent = Orgunit()
if rec_id:
agent["rec_id"] = rec_id
if name:
agent["name"] = name
affiliation["agent"] = agent
return affiliation
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 org_laboratory_to_metajson(org_laboratory):
orgunit = Orgunit()
return orgunit
def rml_orgunit_to_metajson(rml_orgunit, source, rec_id_prefix):
""" orgUnit -> orgunit """
orgunit = Orgunit()
# source
if source:
orgunit["rec_source"] = source
# acronym -> acronym
orgunit.update(get_rml_element_text_and_set_key(rml_orgunit, "acronym", "acronym"))
# address -> addresses
orgunit.update(get_rml_addresses(rml_orgunit))
# affiliation -> affiliations
orgunit.update(get_rml_affiliations(rml_orgunit))
# award -> awards
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "award", "awards"))
# ckbData -> self_archiving_policy
orgunit.update(get_rml_self_archiving_policy(rml_orgunit))
# dateOfDissolution -> date_dissolution
orgunit.update(get_rml_element_text_and_set_key(rml_orgunit, "dateOfDissolution", "date_dissolution"))
# dateOfFoundation -> date_foundation
orgunit.update(get_rml_element_text_and_set_key(rml_orgunit, "dateOfFoundation", "date_foundation"))
# description -> descriptions
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "description", "descriptions"))
# email -> emails
orgunit.update(get_rml_emails(rml_orgunit))
# headcount -> headcounts
orgunit.update(get_rml_headcounts(rml_orgunit))
# identifier -> identifiers
orgunit.update(get_rml_identifiers(rml_orgunit))
# image -> resources[0]
orgunit.update(get_rml_images(rml_orgunit, "logo"))
# name -> name
orgunit.update(get_rml_element_text_and_set_key(rml_orgunit, "name", "name"))
# nameAlternative -> name_alternatives
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "nameAlternative", "name_alternatives"))
# nationality -> nationality
orgunit.update(get_rml_element_text_and_set_key(rml_orgunit, "nationality", "nationality"))
# note -> notes
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "note", "notes"))
# olDescription -> descriptions_short
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "olDescription", "descriptions_short"))
# ongoingResearch -> ongoing_researches
orgunit.update(get_rml_ongoing_researches(rml_orgunit))
# phone -> phones
orgunit.update(get_rml_phones(rml_orgunit))
# researchCoverage -> research_coverages
orgunit.update(get_rml_research_coverages(rml_orgunit))
# skill -> skills
orgunit.update(get_rml_textlangs_and_set_key(rml_orgunit, "skill", "skills"))
# turnover -> turnovers
orgunit.update(get_rml_turnovers(rml_orgunit))
# @type -> rec_type
orgunit.update(xmletree.get_element_attribute_and_set_key(rml_orgunit, "type", "rec_type"))
# uri -> urls
orgunit.update(get_rml_uris(rml_orgunit))
return orgunit