def div_before_death(fam):
for key, family in fam.items():
if family.divorced[0] not in ['N/A', '', 'N']:
divor_day = datetime.strptime(family.divorced[0], "%d %b %Y")
wife_death = family.wife[0].death[0]
hus_death = family.husband[0].death[0]
if wife_death[0] not in ['N/A', '', 'N']:
wife_deathday = datetime.strptime(wife_death, "%d %b %Y")
if wife_deathday < divor_day:
new_error = GedcomError((
"ERROR", "FAMILY", "US06", family.divorced[1], key
), f"divorce date {divor_day} after wife's death {wife_deathday}"
)
family.error_list = new_error
if hus_death[0] not in ['N/A', '', 'N']:
hus_deathday = datetime.strptime(hus_death, "%d %b %Y")
if hus_deathday < divor_day:
new_error = GedcomError((
"ERROR", "FAMILY", "US06", family.divorced[1], key
), f"divorce date {divor_day} after husband's death {hus_deathday}"
)
family.error_list = new_error
def birth_b_marriage_us02(fam):
""" US02: birth before marriage"""
for key, value in fam.items():
id_h = value.husband[0]
birth_h = id_h.birthday
h_birth_date = datetime.strptime(birth_h[0], "%d %b %Y")
id_w = value.wife[0]
birth_w = id_w.birthday
w_birth_date = datetime.strptime(birth_w[0], "%d %b %Y")
date_married = value.married
marriage_date = datetime.strptime(date_married[0], "%d %b %Y")
if h_birth_date >= marriage_date:
new_error = GedcomError((
"ERROR", "FAMILY", "US02", birth_h[1], key
), f"husband {id_h.indi_id[0]} birthday {birth_h[0]} isn't before married date {date_married[0]}"
)
value.error_list = new_error
if w_birth_date >= marriage_date:
new_error = GedcomError((
"ERROR", "FAMILY", "US02", birth_w[1], key
), f" wife {id_w.indi_id[0]} birthday {birth_w[0]} isn't before married date {date_married[0]}"
)
value.error_list = new_error
else:
continue
def US05(fam):
for key, value in fam.items():
date_married = value.married
if date_married in {'N/A', ''}:
raise ValueError(f"Lost:{value.id[0]} family marriage date lost")
else:
mar_date = datetime.strptime(date_married[0], "%d %b %Y")
hus_id = value.husband[0]
wif_id = value.wife[0]
hus_death = hus_id.death
wif_death = wif_id.death
if hus_death not in ('N/A', ''):
hus_death_date = datetime.strptime(hus_death[0], "%d %b %Y")
if hus_death_date < mar_date:
new_error = GedcomError(("Error", "FAMILY", "US05", date_married[1], key),
f"Married {date_married[0]} after husband's ({hus_id.indi_id[0]}) death on {hus_death[0]}")
value.error_list = new_error
elif wif_death not in ('N/A', ''):
wif_death_date = datetime.strptime(wif_death[0], "%d %b %Y")
if wif_death_date < mar_date:
new_error = GedcomError(("Error", "FAMILY", "US05", date_married[1], key),
f"Married {date_married[0]} after wife's ({wif_id.indi_id[0]}) death on {wif_death[0]}")
value.error_list = new_error
else:
continue
def check_existence_in_family(role, type, indi, key, value, offset):
if role.indi_id[0] not in indi.keys():
new_error = GedcomError((
"Error", "Family", "US26", role.husband[1], key
), f"Corresponding issue happened. {role.indi_id[0]} not found in individual list."
)
value.error_list = new_error
elif key not in [id_tuple[0] for id_tuple in indi[role.indi_id[0]].spouse
] and type == "husband":
new_error = GedcomError(
("Error", "Family", "US26", value.husband[1], key),
f"Corresponding issue happened. Family {key}'s spouse "
f"{role.indi_id[0]} has wrong family Id.")
value.error_list = new_error
elif key not in [id_tuple[0] for id_tuple in indi[role.indi_id[0]].spouse
] and type == "wife":
new_error = GedcomError(
("Error", "Family", "US26", value.wife[1], key),
f"Corresponding issue happened. Family {key}'s spouse "
f"{role.indi_id[0]} has wrong family Id.")
value.error_list = new_error
elif key not in [id_tuple[0] for id_tuple in indi[role.indi_id[0]].child
] and type == "child":
new_error = GedcomError(
("Error", "Family", "US26", value.wife[1] + 1, key),
f"Corresponding issue happened. Family {key}'s children "
f"{role.indi_id[0]} has wrong family Id.")
value.error_list = new_error
def check_existence_in_indi(family_id_list, type, fam, key, value):
for offset, family_id in enumerate(
[id_tuple[0] for id_tuple in family_id_list]):
if family_id not in fam.keys():
new_error = GedcomError((
"Error", "Individual", "US26", family_id_list[offset][1], key
), f"Corresponding issue happened. {family_id} not found in family list."
)
value.error_list = new_error
elif type == "child" and key not in [
child.indi_id[0] for child in fam[family_id].children
]:
new_error = GedcomError((
"Error", "Individual", "US26", family_id_list[offset][1], key
), f"Corresponding issue happened. Family {family_id} does not have individual {key}."
)
value.error_list = new_error
elif type == "spouse" and (key != fam[family_id].husband[0].indi_id[0]
and
key != fam[family_id].wife[0].indi_id[0]):
new_error = GedcomError((
"Error", "Individual", "US26", family_id_list[offset][1], key
), f"Corresponding issue happened. Family {family_id} does not have individual {key}."
)
value.error_list = new_error
def unique_family_24(fam):
person = dict()
for key, family in fam.items():
wife = family.wife[0].indi_id[0]
husband = family.husband[0].indi_id[0]
if wife in person.keys():
marry_date = family.married[0]
if marry_date in person[wife]:
new_error = GedcomError(("ANOMALY", "FAMILY", "US24", family.married[1], key),
f"wife should have same marry date in different family")
family.error_list = new_error
else:
person[wife].append(marry_date)
else:
person[wife] = [family.married[0]]
if husband in person.keys():
marry_date = family.married[0]
if marry_date in person[husband]:
new_error = GedcomError(("ANOMALY", "FAMILY", "US24", family.married[1], key),
f"husband should have same marry date in different family")
family.error_list = new_error
else:
person[husband].append(marry_date)
else:
person[husband] = [family.married[0]]
def birth_before_marriage(fam):
for key, family in fam.items():
div_date = ""
marr_date = datetime.strptime(family.married[0], "%d %b %Y")
if_div = False
if family.divorced != "N/A":
div_date = datetime.strptime(family.divorced[0], "%d %b %Y")
if_div = True
for child in family.children:
birth = datetime.strptime(child.birthday[0], "%d %b %Y")
bir_line = child.birthday[1]
child_id = child.indi_id[0]
if birth <= marr_date:
"""{"error_type": self.__error_info[0],
"indi_or_fam": self.__error_info[1],
"user_story": self.__error_info[2],
"error_line": self.__error_info[3],
"entity_id": self.__error_info[4]}
"""
new_error = GedcomError((
"ANOMALY", "FAMILY", "US08", bir_line, key
), f"Child {child_id} born {birth} before marriage on {marr_date}"
)
family.error_list = new_error
elif if_div and birth > div_date + timedelta(days=270):
new_error = GedcomError((
"ANOMALY", "FAMILY", "US08", bir_line, key
), f"Child {child_id} born {birth} after divorce on {div_date}"
)
family.error_list = new_error
if family.wife[0].death != "N/A":
mom_death = datetime.strptime(family.wife[0].death[0],
"%d %b %Y")
if birth > mom_death:
new_error = GedcomError((
"ANOMALY", "FAMILY", "US09", bir_line, key
), f"Child {child_id} born {birth} after mother's death on {mom_death}"
)
family.error_list = new_error
print(
f"ANOMALY: FAMILY: US09: {bir_line}: {key}: "
f"Child {child_id} born {birth} after mother's death on {mom_death}"
)
if family.husband[0].death != "N/A":
dad_death = datetime.strptime(family.husband[0].death[0],
"%d %b %Y")
if birth > dad_death + timedelta(days=270):
new_error = GedcomError((
"ANOMALY", "FAMILY", "US09", bir_line, key
), f"Child {child_id} born {birth} after daddy's death on {dad_death}"
)
family.error_list = new_error
def corrent_gender_us21(fam):
for key, family in fam.items():
husband = family.husband[0]
wife = family.wife[0]
if husband.gender[0] == 'F':
new_error = GedcomError(("ANOMALY", "FAMILY", "US21", family.husband[1], key),
f"husband's gender should be male")
family.error_list = new_error
if wife.gender[0] == 'M':
new_error = GedcomError(("ANOMALY", "FAMILY", "US21", family.wife[1], key),
f"wife's gender should be female")
family.error_list = new_error
def unique_first_names_in_families_us25(fam):
"""
US25: Unique first names in families
"""
for key, value in fam.items():
dict_c_name = defaultdict(lambda: [])
if value.children == ['N/A']:
continue
else:
for child in value.children:
id_c = child.indi_id[0]
name_c = child.name
first_name_c = name_c[0].split(' ')[0]
if not dict_c_name[first_name_c]:
dict_c_name[first_name_c] = [id_c, name_c[1]]
else:
l_id = dict_c_name[first_name_c]
l_id.extend([id_c, name_c[1]])
dict_c_name[first_name_c] = l_id
for name1, value1 in dict_c_name.items():
if len(value1) != 2:
l_child = []
for n in range(0, len(value1), 2):
l_child.append(value1[n])
new_error = GedcomError((
"ANOMALY", "FAMILY", "US25", value1[1], key
), f"children {l_child} of family {key} have the same first name '{name1}'"
)
value.error_list = new_error
else:
continue
def multiple_siblings(fam):
for key, family in fam.items():
if len(family.children) > 15:
new_error = GedcomError(("ERROR", "FAMILY", "US15",
family.children[15].indi_id[1], key),
f"More than 15 siblings in one family.")
family.error_list = new_error
def multiple_birth(fam):
for key, family in fam.items():
if len(family.children) > 5 and \
len(set([datetime.strptime(child.birthday[0], "%d %b %Y") for child in family.children])) == 1:
new_error = GedcomError(
("ERROR", "FAMILY", "US14", family.children[4].birthday[1],
key), f"More than 5 siblings born at the same day.")
family.error_list = new_error
def no_marriage_to_children_us17(fam):
""" US17: no_marriage_to_children"""
dict_h = defaultdict(lambda: 'N/A')
dict_w = defaultdict(lambda: 'N/A')
for key, value in fam.items():
l_c = []
id_h = value.husband[0]
id_w = value.wife[0]
if value.children == ['N/A']:
continue
else:
for child in value.children:
id_c = child.indi_id[0]
l_c.append(id_c)
if dict_h[id_h.indi_id[0]] == 'N/A':
dict_h[id_h.indi_id[0]] = l_c
else:
dict_h[id_h.indi_id[0]] = l_c + dict_h[id_h.indi_id[0]]
if dict_w[id_w.indi_id[0]] == 'N/A':
dict_w[id_w.indi_id[0]] = l_c
else:
dict_w[id_w.indi_id[0]] = l_c + dict_w[id_w.indi_id[0]]
for key1, value1 in fam.items():
husband_id = value1.husband[0]
wife_id = value1.wife[0]
if wife_id.indi_id[0] in dict_h[husband_id.indi_id[0]]:
new_error = GedcomError((
"ERROR", "FAMILY", "US17", wife_id.indi_id[1], key1
), f"father {husband_id.indi_id[0]} married to his child {wife_id.indi_id[0]}"
)
value1.error_list = new_error
if husband_id.indi_id[0] in dict_w[wife_id.indi_id[0]]:
new_error = GedcomError((
"ERROR", "FAMILY", "US17", husband_id.indi_id[1], key1
), f"mother {wife_id.indi_id[0]} married to her child {husband_id.indi_id[0]}"
)
value1.error_list = new_error
else:
continue
def less_than_150(indi):
for key, people in indi.items():
birthday = datetime.strptime(people.birthday[0], "%d %b %Y")
deathday = ""
if people.death != 'N/A':
deathday = datetime.strptime(people.death[0], "%d %b %Y")
if people.alive is False and birthday + timedelta(days=150 *
365) < deathday:
new_error = GedcomError(
("ERROR", "INDIVIDUAL", "US07", people.death[1], key),
f"{people.name[0]}'s age should be less than 150 years.")
people.error_list = new_error
elif people.alive is True and birthday < datetime.today() - timedelta(
days=150 * 365):
new_error = GedcomError(
("ERROR", "INDIVIDUAL", "US07", people.birthday[1], key),
f"{people.name[0]} should no more than 150 years old.")
people.error_list = new_error
def parents_not_too_old_us12(fam):
""" US12: parents_not_too_old"""
for key, value in fam.items():
if value.children == ['N/A']:
continue
else:
id_h = value.husband[0]
birth_h = id_h.birthday
h_birth_date = datetime.strptime(birth_h[0], "%d %b %Y")
id_w = value.wife[0]
birth_w = id_w.birthday
w_birth_date = datetime.strptime(birth_w[0], "%d %b %Y")
for child in value.children:
id_c = child.indi_id[0]
birth_c = child.birthday
c_birth_date = datetime.strptime(birth_c[0], "%d %b %Y")
num_hc = c_birth_date - h_birth_date
l_hc = str(num_hc).split(' ')
num_wc = c_birth_date - w_birth_date
l_wc = str(num_wc).split(' ')
if int(l_hc[0]) >= 80 * 365.25:
new_error = GedcomError(
("ERROR", "FAMILY", "US12", birth_h[1], key),
f"father {id_h.indi_id[0]} {birth_h[0]} "
f"is not less than 80 years old than child {id_c} {birth_c[0]}"
)
value.error_list = new_error
if int(l_wc[0]) >= 60 * 365.25:
new_error = GedcomError(
("ERROR", "FAMILY", "US12", birth_w[1], key),
f"mother {id_w.indi_id[0]} {birth_w[0]} "
f"is not less than 60 years old than child {id_c} {birth_c[0]}"
)
value.error_list = new_error
else:
continue
def marry_after_14(fam):
for key, family in fam.items():
marr_date = datetime.strptime(family.married[0], "%d %b %Y")
husband_id = family.husband[0]
wife_id = family.wife[0]
hus_birth = datetime.strptime(husband_id.birthday[0], "%d %b %Y")
wife_birth = datetime.strptime(wife_id.birthday[0], "%d %b %Y")
days1 = marr_date - hus_birth
days2 = marr_date - wife_birth
if days1.days <= 14 * 365.25:
new_error = GedcomError(
("ANOMALY", "FAMILY", "US10", family.married[1], key),
f"Child {husband_id.indi_id[0]}'s marry date is {marr_date},"
f"less than 14 years after {hus_birth} ")
family.error_list = new_error
if days2.days <= 14 * 365.25:
new_error = GedcomError(
("ANOMALY", "FAMILY", "US10", family.married[1], key),
f"Child {wife_id.indi_id[0]}'s marry date is {marr_date},"
f"less than 14 years after {wife_birth} ")
family.error_list = new_error
def date_before_current(indi, fam):
for key, family in fam.items():
marry_day = datetime.strptime(family.married[0], "%d %b %Y")
if marry_day > today:
new_error = GedcomError(
("ERROR", "FAMILY", "US01", family.married[1], key),
f"marry date {marry_day} before today")
family.error_list = new_error
if family.divorced[0] not in ['N/A', '', 'N']:
divor_day = datetime.strptime(family.divorced[0], "%d %b %Y")
if divor_day > today:
new_error = GedcomError(
("ERROR", "FAMILY", "US01", family.divorced[1], key),
f"divorce date {divor_day} before today")
family.error_list = new_error
for key, individual in indi.items():
birth_day = datetime.strptime(individual.birthday[0], "%d %b %Y")
if birth_day > today:
new_error = GedcomError(
("ERROR", "INDIVIDUAL", "US01", individual.birthday[1], key),
f"birthday date {birth_day} before today")
individual.error_list = new_error
if individual.death[0] not in ['N/A', '', 'N']:
death_day = datetime.strptime(individual.death[0], "%d %b %Y")
if death_day > today:
new_error = GedcomError(
("ERROR", "INDIVIDUAL", "US01", individual.death[1], key),
f"death date {death_day} before today")
individual.error_list = new_error
def first_cousin_should_not_marry_us19(fam):
grandparents_family_id_lis = []
for key, value in fam.items():
hus_id = value.husband[0]
wif_id = value.wife[0]
hus_family = hus_id.child
wif_family = wif_id.child
if hus_family not in ['N/A', ''] and wif_family not in ['N/A', '']:
for family_hus in hus_family:
family1 = fam.get(family_hus[0])
husband_father_id = family1.husband[0]
husband_mother_id = family1.wife[0]
father_child = husband_father_id.child
mother_child = husband_mother_id.child
for child in father_child:
if father_child not in ['N/A', '']:
grandparents_family_id_lis.append(child[0])
for child in mother_child:
if mother_child not in ['N/A', '']:
grandparents_family_id_lis.append(child[0])
for family_wif in wif_family:
family2 = fam.get(family_wif[0])
wife_father_id = family2.husband[0]
wife_mother_id = family2.wife[0]
father_child = wife_father_id.child
mother_child = wife_mother_id.child
for child in father_child:
if father_child not in ['N/A', '']:
grandparents_family_id_lis.append(child[0])
for child in mother_child:
if mother_child not in ['N/A', '']:
grandparents_family_id_lis.append(child[0])
if len(grandparents_family_id_lis) == set(
grandparents_family_id_lis):
grandparents_family_id_lis = []
else:
new_error = GedcomError((
"Error", "FAMILY", "US19", value.id[1], key
), f"First cousins marry one another happened in family {value.id[0]}"
)
value.error_list = new_error
else:
continue
def sibi_not_marry(fam):
fami = {}
for key, family in fam.items():
husband = family.husband[0]
wife = family.wife[0]
if husband.child != 'N/A' and wife.child != 'N/A':
fami[key] = {
'hus_fam': [],
'wife_fam': [],
'hus_par': set(),
'wife_par': set(),
'family': family
}
for i in husband.child:
fami[key]['hus_fam'].append(i[0])
for i in wife.child:
fami[key]['wife_fam'].append(i[0])
for key, family in fam.items():
for i, j in fami.items():
if key in j['hus_fam']:
j['hus_par'].add(family.husband[0].indi_id)
j['hus_par'].add(family.wife[0].indi_id)
if key in j['wife_fam']:
j['wife_par'].add(family.husband[0].indi_id)
j['wife_par'].add(family.wife[0].indi_id)
global iserror
for key, value in fami.items():
for num in value['wife_par']:
if num in value['hus_par']:
iserror = 1
new_error = GedcomError(
("ANOMALY", "FAMILY", "US18", value['family'].id[1], key),
f"Siblings should not marry one another")
value['family'].error_list = new_error
break
def US11(indi, fam):
for key, value in indi.items():
lis_spouse = value.spouse
if len(lis_spouse) > 1 and lis_spouse not in ['N/A', '']:
for i in range(len(lis_spouse)):
fam_id_1 = lis_spouse[i][0]
date_married_1 = fam[fam_id_1].married
marriage_date_1 = datetime.strptime(date_married_1[0],
"%d %b %Y")
date_divorced_1 = fam[fam_id_1].divorced
if date_divorced_1 in ('N/A', ''):
divorced_date_1 = datetime.strptime(
'21 MAR 2020', "%d %b %Y")
else:
divorced_date_1 = datetime.strptime(
date_divorced_1[0], "%d %b %Y")
for j in range(i + 1, len(lis_spouse)):
fam_id_2 = lis_spouse[j][0]
date_married_2 = fam[fam_id_2].married
marriage_date_2 = datetime.strptime(
date_married_2[0], "%d %b %Y")
divorced_date_2 = fam[fam_id_2].divorced
if divorced_date_2 in ('N/A', ''):
divorced_date_2 = datetime.strptime(
'21 MAR 2020', "%d %b %Y")
else:
divorced_date_2 = datetime.strptime(
divorced_date_2[0], "%d %b %Y")
if marriage_date_1 >= divorced_date_2 or marriage_date_2 >= divorced_date_1:
continue
else:
new_error = GedcomError((
"Error", "INDIVIDUAL", "US11", date_married_1[1],
key
), f"bigamy happened on family {fam_id_1} and family {fam_id_2}"
)
value.error_list = new_error
else:
continue
def build_family(fam, indi_dict):
fam_dict = defaultdict(Family)
for family in fam:
if family["FAM"][0] in fam_dict.keys():
new_error = GedcomError(("ERROR", "FAMILY", "US22", family["FAM"][1], family["FAM"][0]),
f"Family ID {family['FAM'][0]} should be unique!")
fam_dict[family["FAM"][0]].error_list = new_error
else:
new_fam = Family(family["FAM"], family["MARR"], family["DIV"],
# add husband into family dictionary as an individual class
# which search from individual dictionary
(indi_dict[family["HUSB"][0]], family["HUSB"][1]),
(indi_dict[family["WIFE"][0]], family["WIFE"][1]))
if family["CHIL"] != "N/A":
for people in family["CHIL"]:
new_child = indi_dict[people[0]]
new_fam.children = new_child
fam_dict[family["FAM"][0]] = new_fam
return fam_dict
def USO4(fam):
for key, value in fam.items():
date_divorced = value.divorced
date_married = value.married
if date_divorced in ('N/A', ''):
continue
else:
if date_married in ('N/A', ''):
raise ValueError(f"Lost:{value.id[0]} family marriage date lost")
else:
div_date = datetime.strptime(date_divorced[0], "%d %b %Y")
mar_date = datetime.strptime(date_married[0], "%d %b %Y")
if div_date < mar_date:
new_error = GedcomError(("Error", "FAMILY", "US04", date_divorced[1], key),
f"Divorced {date_divorced[0]} before married {date_married[0]}")
value.error_list = new_error
def unique_name_and_birthday_us23(indi):
lis_individual = []
for key, value in indi.items():
name = value.name
birthday = value.birthday
lis_individual.append((name[0], birthday[0]))
set_individual = set(lis_individual)
if len(lis_individual) != len(set_individual):
for item in set_individual:
lis_individual.remove(item)
set_individual = set(lis_individual)
for i in set_individual:
for key, value in indi.items():
if value.name[0] == i[0]:
new_error = GedcomError(
("Error", "INDIVIDUAL", "US23", value.name[1], key),
f"more than one individual with same name {value.name[0]} and birth date "
+ f"{value.birthday[0]} appear in the GED file")
value.error_list = new_error
def birth_b_death_us03(indi):
""" US03: birth before death"""
for key, value in indi.items():
birth = value.birthday
death = value.death
if death != 'N/A':
death_date = datetime.strptime(death[0], "%d %b %Y")
else:
continue
birth_date = datetime.strptime(birth[0], "%d %b %Y")
if birth_date >= death_date:
new_error = GedcomError((
"ERROR", "INDIVIDUAL", "US03", birth[1], key
), f"{key}'s birthday {birth[0]} isn't before death date {death[0]}"
)
value.error_list = new_error
else:
continue
def build_individual(indi):
indi_dict = defaultdict(Individual)
for people in indi:
if people["INDI"][0] in indi_dict.keys():
new_error = GedcomError(("ERROR", "INDIVIDUAL", "US22", people["INDI"][1], people["INDI"][0]),
f"Individual ID {people['INDI'][0]} should be unique!")
indi_dict[people["INDI"][0]].error_list = new_error
else:
indi_birth_date = datetime.strptime(people["BIRTH"][0], "%d %b %Y")
current_date = datetime.now()
if people["DEAT"] == "N/A":
num_date = current_date - indi_birth_date
l_num_date = str(num_date).split(' ')
age = int(float(l_num_date[0]) / 365.25)
else:
death_date = datetime.strptime(people["DEAT"][0], "%d %b %Y")
num_date = death_date - indi_birth_date
l_num_date = str(num_date).split(' ')
age = f"death age: {int(float(l_num_date[0]) / 365.25)}"
new_indi = Individual(people["INDI"], people["NAME"], people["SEX"],
people["BIRTH"], people["DEAT"], people["FAMC"], people["FAMS"], age)
indi_dict[people["INDI"][0]] = new_indi
return indi_dict