def get_existing_before(self, profile_to_check, link_string, prof_id):
'''
This function will detect if the profile has a link existing before
it will create a task informing that profile should not be avoided
'''
#We first get all webs of the current profile
current_webs = profile_to_check.get_all_webs()
for web in current_webs:
#This will mean that we have an existing match before
if web["name"] == link_string:
profile = self.db_input.get_profile_by_ID(prof_id)
details = MATCH_POTENTIAL_INFO_EXISTING + profile_to_check.nameLifespan() + " " + str(profile_to_check.get_id())
include_task_no_duplicate(profile, MATCH_POTENTIAL_EXISTING, 1, details)
print_out("- AVOIDING INTRODUCTION OF EXISTING LINK IN "+ str(profile_to_check.nameLifespan()) )
return True
return False
def _conflict_storing(self, profile_rm, conflicted_profiles_ids, db_conflict):
'''
Internal function to avoid duplicates. It stores a conflict of matches deviation
profile_rm is a profile kind
conflicted_profile_ids is a list of ids with conflict
'''
conflict_str = ""
for prof_conf_id in conflicted_profiles_ids:
prof_conf = db_conflict.get_profile_by_ID(prof_conf_id)
conflict_str += str(prof_conf.nameLifespan()) + " "
print_out("- CONFLICT of profile " + str(profile_rm.nameLifespan()) + " WITH PROFILE(S) " + conflict_str)
#This is a conflict, we should have a single match!!!
self.conflict_profiles[profile_rm.get_id()] = conflicted_profiles_ids
details_info = MATCH_CONFLICT_INFO
for ids_geni in conflicted_profiles_ids:
details_info += self.database_geni.get_profile_by_ID(ids_geni).get_this_profile_url() + " "
include_task_no_duplicate(profile_rm, MATCH_CONFLICT_TASK, 1, details_info)
def process(self,
profiles_2_analyze="all",
storage=False,
threshold=360,
avoid_import_living_from=["input", "check"]):
'''
Determines the full database and detects potential matches
profiles_2_analyze = an array of profiles for the first database which shall be analyzed, in case of no
input it will analyze all profiles.
avoid_import_living_from will be a list with either input or check, it will not import the living from that database
'''
kind_match = self.db_check.get_db_kind()
match_str = str(kind_match) + MATCH
matcher_profiles = match_single_profile(
self.db_input,
self.db_check,
data_language=self.data_language,
name_convention=self.name_convention)
linked_profiles = {}
for prof in self.db_input.get_all_profiles():
if prof.get_specific_web(kind_match):
linked_profiles[prof.get_id()] = prof
print_out(PROCESS_MATCH_NUMBER_OF_IMPACTS_BEGIN + str(kind_match) +
PROCESS_MATCH_NUMBER_OF_IMPACTS_END +
str(len(linked_profiles)))
for prof_id in linked_profiles:
if (profiles_2_analyze == "all") or (prof_id
in profiles_2_analyze):
prof_in_study = linked_profiles[prof_id]
if continue_match(prof_in_study,
match_str,
threshold=threshold):
prof_linked_id = prof_in_study.get_specific_web(
kind_match)["url"]
prof_linked = self.db_check.get_profile_by_ID(
prof_linked_id)
#In order to get the marriage, first we get the family from the other database
_, family_profile = self.db_check.get_family_from_child(
prof_linked.get_id())
_, family_input = self.db_input.get_family_from_child(
prof_id)
#We will only analyze the profiles which have been introduced
loc_research = get_research_log_id(prof_in_study,
storage=storage)
non_matched_profiles_input, non_matched_profiles_check, conflict_profiles, matched_profiles = matcher_profiles.match(
prof_id)
#We create the following slots that will be used for the introduction of confirmed candidates inside the database
check_non_matches_existing_in_input = {}
input_non_matches_existing_in_check = {}
#REMOVAL OF LIVING
#If the option is selected to not publish data that is restricted like living we remove teh matched characters
temp_data = {
"input": {
"non_matches": non_matched_profiles_input,
"database": self.db_input,
"other_db": self.db_check,
"family_is_child": family_input,
"current_id_to_add": prof_linked_id,
"current_id_matched": prof_id,
"existing_prof":
input_non_matches_existing_in_check
},
"check": {
"non_matches": non_matched_profiles_check,
"database": self.db_check,
"other_db": self.db_input,
"family_is_child": family_profile,
"current_id_to_add": prof_id,
"current_id_matched": prof_linked.get_id(),
"existing_prof":
check_non_matches_existing_in_input
}
}
for kind_db in temp_data:
#We create a temporal copy to remove the data
temp_input = list(temp_data[kind_db]["non_matches"])
for prof in temp_input:
potential_profile = temp_data[kind_db][
"database"].get_profile_by_ID(prof)
#REMOVAL OF LIVING
#If the option is selected to not publish data that is restricted like living we remove teh matched characters
if (kind_db in avoid_import_living_from
) and potential_profile.getLiving():
del temp_data[kind_db]["non_matches"][prof]
print_out(
"- AVOIDING INTRODUCTION OF LIVING " +
str(potential_profile.nameLifespan()) +
" FROM THE DATABASE " + temp_data[kind_db]
["database"].get_db_kind())
#Prior to starting the overall copy/match, we should check first if potential candidates exists in the input database
temp_checking = dict(non_matched_profiles_check)
for prof in temp_checking:
potential_profile = self.db_check.get_profile_by_ID(
prof)
#MATCHING: we look if the profile also exists before
matches = self.db_input.get_potential_profile_match(
potential_profile,
data_language=self.data_language,
name_convention=self.name_convention)
#Profiles might have been matched before
Existing_found = False
duplicate_names = ""
for candidate_prof in matches.keys():
existing_prof = self.db_input.get_profile_by_ID(
candidate_prof)
if (existing_prof.getName() != NOT_KNOWN_VALUE
) and (existing_prof.getSurname() !=
NOT_KNOWN_VALUE):
#We only continue looking for alternatives if there is no uncertain information int he profile
duplicate_names += existing_prof.nameLifespan(
) + " "
for web_data in existing_prof.get_all_webs():
#We now know that the profiles are the same in this database, we can proceed to add this existing one
if web_data[
"url"] == potential_profile.get_this_profile_url(
):
#We store the link of a profile id in check to a profile existing in input (profile class)
check_non_matches_existing_in_input[
prof] = existing_prof
Existing_found = True
if matches[candidate_prof][
"score*factor"] >= FACTOR_DUPLICATE:
check_non_matches_existing_in_input[
prof] = existing_prof
Existing_found = True
if (len(matches.keys()) > 0) and not Existing_found:
#In this case we have a certain potential match. We should not add and rather, leave for checking an human being
del non_matched_profiles_check[prof]
print_out("- POTENTIAL DUPLICATE of profile " +
str(potential_profile.nameLifespan()) +
" WITH PROFILE(S) ID(S) " +
str(duplicate_names))
#This is a conflict, we should avoid duplicating job of checking and solving
conflict_profiles[prof] = matches
details_info = (
"Potential existing duplicates for profile " +
potential_profile.nameLifespan() +
" with web " +
potential_profile.get_this_profile_url() +
" in the profiles: ")
for ids_check in matches:
temp_prof = self.db_input.get_profile_by_ID(
ids_check)
details_info += str(
ids_check
) + " : " + temp_prof.nameLifespan(
) + " "
include_task_no_duplicate(
prof_in_study, MATCH_POTENTIAL_DUPLICATE, 1,
details_info)
###################
#MATCH INTRODUCTION
###################
for db_kind in temp_data:
#CASE MATCH: Missing a parents or several parents.
non_match_now = temp_data[db_kind]["non_matches"]
db_now = temp_data[db_kind]["database"]
db_other = temp_data[db_kind]["other_db"]
family_is_child = temp_data[db_kind]["family_is_child"]
current_id_to_add = temp_data[db_kind][
"current_id_to_add"]
if (("father" in non_match_now.values())
or ("mother" in non_match_now.values())):
father_id = None
mother_id = None
father_profile = None
mother_profile = None
#We shall first obtain the id of the profiles from the check database (i.e. geni)
if ("father" in non_match_now.values()):
father_id = list(non_match_now.keys())[list(
non_match_now.values()).index("father")]
father_profile = db_now.get_profile_by_ID(
father_id)
if ("mother" in non_match_now.values()):
mother_id = list(non_match_now.keys())[list(
non_match_now.values()).index("mother")]
mother_profile = db_now.get_profile_by_ID(
mother_id)
Intro_sentence = MATCH_ADDING_PROFILES
if (father_id
in temp_data[kind_db]["existing_prof"]
) or (mother_id
in temp_data[kind_db]["existing_prof"]):
Intro_sentence = MATCH_EXISTING_PROFILES
#We also add the link!
if father_profile and mother_profile:
Intro_sentence += father_profile.nameLifespan(
) + AND_STRING + mother_profile.nameLifespan(
) + TO_STRING + db_other.get_db_kind()
elif father_profile:
Intro_sentence += father_profile.nameLifespan(
) + TO_STRING + db_other.get_db_kind()
elif mother_profile:
Intro_sentence += mother_profile.nameLifespan(
) + TO_STRING + db_other.get_db_kind()
#We inform of the inclusion of the new profiles
print_out(Intro_sentence)
marriage_event = family_is_child.getMarriage()
#If the parent was existing before, we avoid double introduction
father_to_add = temp_data[kind_db][
"existing_prof"].get(father_id, father_profile)
mother_to_add = temp_data[kind_db][
"existing_prof"].get(mother_id, mother_profile)
if father_to_add and mother_to_add:
#So.. we add the new profiles
new_father_id, new_mother_id, _ = db_other.add_parents(
child_profile_id=current_id_to_add,
father_profile=father_to_add,
mother_profile=mother_to_add,
marriage_event=marriage_event)
elif father_to_add:
#So.. we just add the father
new_father_id, new_mother_id, _ = db_other.add_parents(
child_profile_id=current_id_to_add,
father_profile=father_to_add,
marriage_event=marriage_event)
elif mother_to_add:
#So.. we just add the father
new_father_id, new_mother_id, _ = db_other.add_parents(
child_profile_id=current_id_to_add,
mother_profile=mother_to_add,
marriage_event=marriage_event)
if db_kind == "check":
if father_profile:
self.add_match_to_prof(
new_father_id, father_profile)
if mother_profile:
self.add_match_to_prof(
new_mother_id, mother_profile)
#We remove the profiles, as will be added
if father_id: del non_match_now[father_id]
if mother_id: del non_match_now[mother_id]
#################
# CONTINUE HERE
#################
#PARTNERS: Review of partners for inclusion
partners_input = self.db_input.get_partners_from_profile(
prof_id)
matched_partners = {}
for partner_input in partners_input:
if partner_input in matched_profiles:
matched_partners[partner_input] = matched_profiles[
partner_input]
for kind_db in temp_data:
#We create a temporal copy to remove the data
temp_2_use = list(temp_data[kind_db]["non_matches"])
for prof in temp_2_use:
#We might delete some profiles (children) in the middle, that's why we check first if the profile is in the list
if (prof in temp_data[kind_db]["non_matches"]
) and temp_data[kind_db]["non_matches"][
prof] == "partner":
partner_profile = temp_data[kind_db][
"database"].get_profile_by_ID(prof)
#Checking is already having a link to the database
existing_link = self.get_existing_before(
partner_profile,
self.db_check.get_db_kind(), prof_id)
#Now, if the partner is not accessible due to data restriction, we will skip this step
if partner_profile.get_accessible(
) and not existing_link:
#Good, we will need now to add the new partner to the INPUT area
Intro_sentence = PROCESS_ADD_PROFILE_BEGIN
if prof in temp_data[kind_db][
"existing_prof"]:
Intro_sentence = PROCESS_LINK_PROFILE_BEGIN
print_out(Intro_sentence +
partner_profile.nameLifespan() +
TO_STRING + temp_data[kind_db]
["other_db"].get_db_kind() +
PROCESS_ADD_PROFILE_END +
temp_data[kind_db]["non_matches"]
[prof])
family_check = temp_data[kind_db][
"database"].get_family_from_partners(
temp_data[kind_db]
["current_id_matched"],
partner_profile.get_id())
marriage_event = temp_data[kind_db][
"database"].get_family_by_ID(
family_check).getMarriage()
#The profile might be existing before and we do not need to add it again, so we have a potential deviation
#If we store it before, in this case, we will use the existing profile, if not, we continue we the one in the other database
partner_to_introduce = temp_data[kind_db][
"existing_prof"].get(
prof, partner_profile)
id_partner, _ = temp_data[kind_db][
"other_db"].add_partner(
temp_data[kind_db]
["current_id_to_add"],
partner_to_introduce,
marriage=marriage_event)
if kind_db == "input":
self.add_match_to_prof(
prof,
temp_data[kind_db]["other_db"].
get_profile_by_ID(id_partner),
adding=False)
matched_partners[prof] = id_partner
elif kind_db == "check":
self.add_match_to_prof(
id_partner, partner_profile)
matched_partners[id_partner] = prof
elif not partner_profile.get_accessible():
print_out(
"- AVOIDING INTRODUCTION OF LIVING "
+ str(partner_profile.nameLifespan()) +
" FROM THE DATABASE " +
temp_data[kind_db]
["database"].get_db_kind())
family_eliminate = temp_data[kind_db][
"database"].get_family_from_partners(
temp_data[kind_db]
["current_id_matched"],
partner_profile.get_id())
#As the partner is not known and we might have access issues,
#we stop the review of those children in the non accesible partner
for child in family_eliminate.getChildren(
):
if child in temp_data[kind_db][
"non_matches"]:
del temp_data[kind_db][
"non_matches"][child]
#We remove also from the non-matching pending those profiles that have been skipped due to privacy
if not existing_link:
del temp_data[kind_db]["non_matches"][
partner_profile.get_id()]
#We go ahead looking first for each matched partner
for partner_input in matched_partners:
partner_check = matched_partners[partner_input]
#We need the family which will be the input for the children
family_part_input = self.db_input.get_family_from_partners(
prof_id, partner_input)
family_part_check = self.db_check.get_family_from_partners(
prof_linked.get_id(), partner_check)
family_part = {
"input": family_part_check,
"check": family_part_input
}
family_current = {
"input": family_part_input,
"check": family_part_check
}
#INTRODUCTION: CHILDREN in the family
for kind_db in temp_data:
temp = dict(temp_data[kind_db]["non_matches"])
for prof in temp.keys():
#We will only select those profiles which are children
if (temp_data[kind_db]["non_matches"][prof]
== "child") and (
prof
in temp_data[kind_db]["database"].
get_children_from_family(
family_current[kind_db])):
child_profile = temp_data[kind_db][
"database"].get_profile_by_ID(prof)
#Ok, if the profile is accessible, we go ahead for creation
if (child_profile.get_accessible()):
Intro_sentence = PROCESS_ADD_PROFILE_BEGIN
if prof in temp_data[kind_db][
"existing_prof"]:
Intro_sentence = PROCESS_LINK_PROFILE_BEGIN
print_out(
Intro_sentence +
child_profile.nameLifespan() +
TO_STRING + temp_data[kind_db]
["other_db"].get_db_kind() +
PROCESS_ADD_PROFILE_END +
temp_data[kind_db]["non_matches"]
[prof])
#If the child was existing before, we add it directly with the right profile, avoiding double introduction
child_to_add = temp_data[kind_db][
"existing_prof"].get(
prof, child_profile)
child_new_ids = temp_data[kind_db][
"other_db"].add_child(
family_part[kind_db],
[child_to_add])
if kind_db == "input":
child_new_prof = self.db_check.get_profile_by_ID(
child_new_ids[0])
self.add_match_to_prof(
prof,
child_new_prof,
adding=False)
elif kind_db == "check":
self.add_match_to_prof(
child_new_ids[0],
child_profile)
else:
print_out(
child_profile.nameLifespan() +
PROCESS_NO_ACCESS)
del temp_data[kind_db]["non_matches"][prof]
################################################################
if len({
**non_matched_profiles_input,
**non_matched_profiles_check,
**conflict_profiles
}) == 0:
#In this case, we have achieved the full matching, we store the information
today = datetime.date.today().toordinal()
notes_toadd = STATUS_MATCHED + str(today)
record_research_log(prof_in_study, match_str,
loc_research, prof_linked_id,
notes_toadd)
else:
notes_toadd = (STATUS_TO_CHECK + " " * 10 +
"--Missing match input " +
str(non_matched_profiles_input) +
" " * 10 + "--Missing match check " +
str(non_matched_profiles_check) +
" " * 10 + "--Pending conflicts " +
str(conflict_profiles))
record_research_log(prof_in_study, match_str,
loc_research, prof_linked_id,
notes_toadd)
else:
print_out("SKIPPING " + prof_in_study.nameLifespan(),
log_level=15)
def match(self, profile_ID):
'''
It executes the match, assumes contains a GENI link in the profile_ID
It will:
- Generate a web link to the matched profile
- Return the following:
- A list of non-matched profiles with the relationship according to standard namings requesting review.
- A dictonary of conflicts linked to lists of profile in score with request to review
- A dictionary of matched profiles.
'''
#Initialization of the different matcher functions
self._init_tracking_logs()
profile_rm = self.database.get_profile_by_ID(profile_ID)
print_out(str(profile_ID) + " = " + profile_rm.nameLifespan())
url = None
#We confirm is a valid profile, should contain a match
confirmed = False
for web_ref in profile_rm.get_all_webs():
if web_ref["name"] == self.database_geni.get_db_kind():
confirmed = True
url = web_ref["url"]
if not confirmed:
logging.error(MATCH_PROFILE_ERROR)
return False
#This is the profile for analysis
profile_geni = self.database_geni.get_profile_by_ID(url)
#We might have an address that has been updated, we double check for updating it in DB
if url != profile_geni.get_this_profile_url():
profile_rm.update_web_ref(url = profile_geni.get_this_profile_url(), name = self.database_geni.get_db_kind())
print_out("UPDATING " + self.database_geni.get_db_kind() + " LINK to " + profile_geni.get_this_profile_url())
#Starting checking of the parents
#FATHER
_, father_rm = self.database.get_father_from_child(profile_rm.get_id())
_, father_geni = self.database_geni.get_father_from_child(profile_geni.get_id())
#First case is a potential match between the profiles
if (father_rm and father_geni):
self._match_single_pair(father_rm, father_geni)
#We can only have the father_rm
elif father_rm:
self.non_matched_profiles_rm[father_rm.get_id()] = FATHER
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " + str(father_rm.nameLifespan()) + " Relation = " + FATHER)
elif father_geni:
self.non_matched_profiles_geni[father_geni.get_id()] = FATHER
print_out("- NO MATCH of profile in " + self.database.get_db_kind() + " " + str(father_geni.nameLifespan()) + " Relation = " + FATHER)
#Notice that we do not consider the case of no parents at all identified, no match needed.
#MOTHER
_, mother_rm = self.database.get_mother_from_child(profile_rm.get_id())
_, mother_geni = self.database_geni.get_mother_from_child(profile_geni.get_id())
#First case is a potential match between the profiles
if (mother_rm and mother_geni):
self._match_single_pair(mother_rm, mother_geni)
#We can only have the father_rm
elif mother_rm:
self.non_matched_profiles_rm[mother_rm.get_id()] = MOTHER
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " + str(mother_rm.nameLifespan()) + " Relation = " + MOTHER)
elif mother_geni:
self.non_matched_profiles_geni[mother_geni.get_id()] = MOTHER
print_out("- NO MATCH of profile in " + self.database.get_db_kind() + " " + str(mother_geni.nameLifespan()) + " Relation = " + MOTHER)
#Notice that we do not consider teh case of no parents at all identified, no match needed.
#PARTNERS
partners_rm = self.database.get_partners_from_profile(profile_rm.get_id())
partners_geni = self.database_geni.get_partners_from_profile(profile_geni.get_id())
self._track_2_lists(partners_rm, partners_geni, PARTNER)
#CHILDREN
children_rm = self.database.get_all_children(profile_ID)
children_geni = self.database_geni.get_all_children(url)
self._track_2_lists(children_rm, children_geni, CHILD)
return self.non_matched_profiles_rm, self.non_matched_profiles_geni, self.conflict_profiles, self.matched_profiles
def _track_2_lists(self, profiles_rm, profiles_geni, kind_of_match):
'''
Function used for both partners and children as is a common function
'''
#We store here the profiles that have been identified in profiles_geni
profiles_not_identified = list(profiles_geni)
conflict_potential_dictionary = {}
for rm_id in profiles_rm:
#We might be in a situation where very small similarities might create confusion of profiles, we store the previous score
previous_score = 0
conflict_match = False
profile_rm = self.database.get_profile_by_ID(rm_id)
geni_matches = []
for geni_id in list(profiles_not_identified):
profile_geni = self.database_geni.get_profile_by_ID(geni_id)
score, factor = profile_rm.comparison_score(profile_geni, self.data_language, self.name_convention)
if score*factor > self.threshold:
#OPTIONS:
# 1.New profile is the right one
# 2.New profile is the first one
# 3.New profile is not the right one.but is the previous
# 4.New profile is as bad as the others.
#Option 1
if score*factor > 3*previous_score:
recover_profs = list(geni_matches)
geni_matches = [geni_id]
profiles_not_identified += recover_profs
previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
#Option 2
elif len(geni_matches) == 0:
geni_matches.append(geni_id)
previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
#Option 3
elif previous_score >= 3*score*factor:
#In this case we ignore... we keep the previous one
pass
#Option 4
else:
if score*factor > previous_score: previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
elif score >= 3*self.threshold:
conflict_match = True
#This is a common case, where profiles have a minimum difference but still relevant, user to check
if rm_id in conflict_potential_dictionary:
conflict_potential_dictionary[rm_id].append(geni_id)
else:
conflict_potential_dictionary[rm_id] = [geni_id]
#If there is a single match, whatever other conditions, we introduce as a match
if len(geni_matches) == 1:
self._match_single_pair(profile_rm, self.database_geni.get_profile_by_ID(geni_matches[0]))
#In case there has been found also a conflict, the conflict is no longer needed, as we do have a match.
if rm_id in conflict_potential_dictionary: del conflict_potential_dictionary[rm_id]
else:
#If there is no single match, we can have several options...
if (len(geni_matches) == 0) and (not conflict_match):
self.non_matched_profiles_rm[rm_id] = kind_of_match
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " +
str(profile_rm.nameLifespan()) + " Relation = " + kind_of_match)
#Or we have more than one match... that is a conflict
elif len(geni_matches) > 1:
self._conflict_storing(profile_rm, geni_matches, self.database_geni)
#We perform another loop with those profiles which were conflicted as some of the might have been identified on the other side
temp_conflicted = conflict_potential_dictionary.copy()
for rm_id_conflicted in temp_conflicted.keys():
#We might have some profiles that have been already identified but stored as potential conflicts. We shall remove them
for geni_conflict_key in temp_conflicted[rm_id_conflicted]:
if geni_conflict_key in self.matched_profiles.values(): conflict_potential_dictionary[rm_id_conflicted].remove(geni_conflict_key)
#The profiles left, are either a match, or we have found again gaps of profiles not found
for rm_id_final_confliced in conflict_potential_dictionary.keys():
#If the list in the dictionary is empty, we do have a missing proifle
profile_rm_new = self.database.get_profile_by_ID(rm_id_final_confliced)
if conflict_potential_dictionary[rm_id_final_confliced] == []:
#This profile was having a potential conflict that ended being actually an empty profile, it is a missing match
self.non_matched_profiles_rm[rm_id_final_confliced] = kind_of_match
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " +
str(profile_rm_new.nameLifespan()) + " Relation = " + kind_of_match)
else:
#Ok, in this case we have a potential match with an actual conflict
details_info = "- CONFLICT POTENTIAL MATCH " + str(profile_rm_new.nameLifespan()) + " with the following: "
address_list = []
for profile_id in conflict_potential_dictionary[rm_id_final_confliced]:
#We remove conflicted profiles from the matching step
if profile_id in profiles_not_identified: profiles_not_identified.remove(profile_id)
profile = self.database_geni.get_profile_by_ID(profile_id)
address_list.append(profile.get_this_profile_url())
details_info += str(profile.nameLifespan()) + " Relation = " + kind_of_match
include_task_no_duplicate(profile_rm_new, MATCH_CONFLICT_TASK, 1, details_info)
print_out(details_info)
self.conflict_profiles[rm_id_final_confliced] = address_list
#Now, we are able to detect those children on the "RIGHT" side. Not linked to other
if len(profiles_not_identified) > 0:
for missing_prof in profiles_not_identified:
self.non_matched_profiles_geni[missing_prof] = kind_of_match
prof = self.database_geni.get_profile_by_ID(missing_prof)
print_out("- NO MATCH of profile in " + self.database.get_db_kind() + " " + str(prof.nameLifespan()) + " Relation = " + kind_of_match)
def _track_2_lists(self, profiles_rm, profiles_geni, kind_of_match):
'''
Function used for both partners and children as is a common function
'''
#We store here the profiles that have been identified in profiles_geni
profiles_not_identified = list(profiles_geni)
#We will create here the dictionary of the addresses
dict_address = {}
total_prof = self.database_geni.get_several_profile_by_ID(profiles_not_identified)
for prof in total_prof:
dict_address[total_prof[prof].get_this_profile_url()] = total_prof[prof].get_id()
conflict_potential_dictionary = {}
for rm_id in profiles_rm:
#We might be in a situation where very small similarities might create confusion of profiles, we store the previous score
previous_score = 0
conflict_match = False
geni_matches = []
profile_rm = self.database.get_profile_by_ID(rm_id)
url_rm_now = profile_rm.get_specific_web(self.database_geni.get_db_kind()).get("url", None)
if url_rm_now in dict_address:
#In this case we have a match, and potentially will be the same profile, we avoid several checks by using one
profile_geni = total_prof[dict_address[url_rm_now]]
score, factor = profile_rm.comparison_score(profile_geni, self.data_language, self.name_convention)
if score*factor > self.threshold:
#We confirm that this link is correct, so we avoid doing the complete loop...
geni_matches = [dict_address[url_rm_now]]
if dict_address[url_rm_now] in profiles_not_identified: profiles_not_identified.remove(dict_address[url_rm_now])
#We only continue if the check of the url was not sucecssful
if len(geni_matches) == 0:
for geni_id in list(profiles_not_identified):
#We have already obtained all profiles above
profile_geni = total_prof[geni_id]
score, factor = profile_rm.comparison_score(profile_geni, self.data_language, self.name_convention)
if score*factor > self.threshold:
#OPTIONS:
# 1.New profile is the right one
# 2.New profile is the first one
# 3.New profile is not the right one.but is the previous
# 4.New profile is as bad as the others.
#Option 1
if score*factor > 2.5*previous_score:
recover_profs = list(geni_matches)
geni_matches = [geni_id]
profiles_not_identified += recover_profs
previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
#Option 2
elif len(geni_matches) == 0:
geni_matches.append(geni_id)
previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
#Option 3
elif previous_score >= 2.5*score*factor:
#In this case we ignore... we keep the previous one
pass
#Option 4
else:
if score*factor > previous_score: previous_score = score*factor
if geni_id in profiles_not_identified: profiles_not_identified.remove(geni_id)
elif score >= 3*self.threshold:
conflict_match = True
#This is a common case, where profiles have a minimum difference but still relevant, user to check
if rm_id in conflict_potential_dictionary:
conflict_potential_dictionary[rm_id].append(geni_id)
else:
conflict_potential_dictionary[rm_id] = [geni_id]
#Options in place with the current match of url
# No existing url => we ignore
# Existing so...
# - The match is the same. => NO ACTION. Covered by code
# - The match is different => NO ACTION. Covered by code double match will be created with warning
# - There is no match => ACTION. Include a conflict warning
#
#If there is a single match, whatever other conditions, we introduce as a match
if len(geni_matches) == 1:
self._match_single_pair(profile_rm, self.database_geni.get_profile_by_ID(geni_matches[0]))
#In case there has been found also a conflict, the conflict is no longer needed, as we do have a match.
if rm_id in conflict_potential_dictionary: del conflict_potential_dictionary[rm_id]
else:
#If there is no single match, we can have several options...
if (len(geni_matches) == 0) and (not conflict_match):
if url_rm_now and (len(geni_matches) == 0):
#This is the only option where we are going to generate a conflict, as existing before!
print_out("- CONFLICT of profile " + str(profile_rm.nameLifespan()) + MATCH_PREVIOUS_MATCH)
details = MATCH_CONFLICT_URL_MESSAGE + self.current_match + " as " + kind_of_match
include_task_no_duplicate(profile_rm, MATCH_CONFLICT_URL_EXISTING, 1, details)
#We move the profile to conflicted ones
self.conflict_profiles[profile_rm.get_id()] = []
else:
self.non_matched_profiles_rm[rm_id] = kind_of_match
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " +
str(profile_rm.nameLifespan()) + " Relation = " + kind_of_match)
#Or we have more than one match... that is a conflict
elif len(geni_matches) > 1:
self._conflict_storing(profile_rm, geni_matches, self.database_geni)
#We perform another loop with those profiles which were conflicted as some of the might have been identified on the other side
temp_conflicted = conflict_potential_dictionary.copy()
for rm_id_conflicted in temp_conflicted.keys():
#We might have some profiles that have been already identified but stored as potential conflicts. We shall remove them
for geni_conflict_key in temp_conflicted[rm_id_conflicted]:
if geni_conflict_key in self.matched_profiles.values(): conflict_potential_dictionary[rm_id_conflicted].remove(geni_conflict_key)
#The profiles left, are either a match, or we have found again gaps of profiles not found
for rm_id_final_confliced in conflict_potential_dictionary.keys():
#If the list in the dictionary is empty, we do have a missing proifle
profile_rm_new = self.database.get_profile_by_ID(rm_id_final_confliced)
if conflict_potential_dictionary[rm_id_final_confliced] == []:
#This profile was having a potential conflict that ended being actually an empty profile, it is a missing match
self.non_matched_profiles_rm[rm_id_final_confliced] = kind_of_match
print_out("- NO MATCH of profile in " + self.database_geni.get_db_kind() + " " +
str(profile_rm_new.nameLifespan()) + " Relation = " + kind_of_match)
else:
#Ok, in this case we have a potential match with an actual conflict
details_info = "- CONFLICT POTENTIAL MATCH " + str(profile_rm_new.nameLifespan()) + " with the following: "
address_list = []
for profile_id in conflict_potential_dictionary[rm_id_final_confliced]:
#We remove conflicted profiles from the matching step
if profile_id in profiles_not_identified: profiles_not_identified.remove(profile_id)
profile = self.database_geni.get_profile_by_ID(profile_id)
address_list.append(profile.get_this_profile_url())
details_info += str(profile.nameLifespan()) + " Relation = " + kind_of_match
include_task_no_duplicate(profile_rm_new, MATCH_CONFLICT_TASK, 1, details_info)
print_out(details_info)
self.conflict_profiles[rm_id_final_confliced] = address_list
#Now, we are able to detect those children on the "RIGHT" side. Not linked to other
if len(profiles_not_identified) > 0:
for missing_prof in profiles_not_identified:
self.non_matched_profiles_geni[missing_prof] = kind_of_match
prof = self.database_geni.get_profile_by_ID(missing_prof)
print_out("- NO MATCH of profile in " + self.database.get_db_kind() + " " + str(prof.nameLifespan()) + " Relation = " + kind_of_match)
def execute_sync(self,
profiles2analyze="all",
threshold=360,
storage=False):
'''
This is the core function, it will execute the global sync of profiles Between
primary and secondary database
'''
list_prof = self.dbp.get_all_profiles()
if profiles2analyze != "all":
list_prof = self.dbp.get_several_profile_by_ID(
profiles2analyze).values()
kind_match = self.dbs.get_db_kind()
match_str = str(kind_match) + MATCH
for prof in list_prof:
if prof.get_specific_web(kind_match) and continue_execution_step(
prof, match_str, STATUS_SYNC, threshold=threshold):
#Obtain the secondary database profile needed
prof_second = self.dbs.get_profile_by_ID(
prof.get_specific_web(kind_match)["url"])
#We obtain the latest update time
update_primary = (datetime.now() - prof.get_update_datetime())
update_secondary = (datetime.now() -
prof_second.get_update_datetime())
minimum_diff = min(update_primary.days, update_secondary.days)
#Now, we only review in case the modification date is recent or there has not been any review before
if (prof.get_research_item_by_name(match_str) is
None) or minimum_diff < threshold:
#We inform by the command line that we are analyzing one profile
print_out(str(prof.get_id()) + " : " + prof.nameLifespan())
#We check all the events of the profile in the dictionary
events_primary = prof.getEventsDict()
events_secondary = prof_second.getEventsDict()
events_in_both = set(events_primary) & set(
events_secondary)
events_only_in_primary = set(events_primary).difference(
set(events_secondary))
events_only_in_secondary = set(
events_secondary).difference(set(events_primary))
sync_data = {
"PRIM": {
"prof_destination": prof,
"events2introduce": events_only_in_secondary,
"events_dict": events_secondary,
"db_destiny": self.dbp
},
"SEC": {
"prof_destination": prof_second,
"events2introduce": events_only_in_primary,
"events_dict": events_primary,
"db_destiny": self.dbs
}
}
for id_sync in sync_data:
#As we know, we iterate on those events we know we need to intoduce
for event_id in sync_data[id_sync]["events2introduce"]:
if event_id in ARRAY_EVENTS:
print_out("NOT IMPLEMENTED FOR " + event_id)
else:
#We obtain the new event class
print_out(SYNC_NEW_EVENT + event_id +
SYNC_IN_DB + sync_data[id_sync]
["db_destiny"].get_db_kind())
event_new = sync_data[id_sync]["events_dict"][
event_id]
sync_data[id_sync][
"prof_destination"].setNewEvent(event_new)
#We store teh exercise performed
loc_research = get_research_log_id(prof, storage=storage)
today = datetime.now().toordinal()
notes_toadd = STATUS_SYNC + str(today)
record_research_log(prof, match_str, loc_research, "",
notes_toadd)
