def test_initialization():
object_element = ObjectElement(level=-1,
pointer="",
tag=gedcom.tags.GEDCOM_TAG_OBJECT,
value="")
assert isinstance(object_element, Element)
assert isinstance(object_element, ObjectElement)
def new_child_element(self, tag, pointer="", value=""):
"""Creates and returns a new child element of this element
:type tag: str
:type pointer: str
:type value: str
:rtype: Element
"""
from gedcom.element.family import FamilyElement
from gedcom.element.file import FileElement
from gedcom.element.individual import IndividualElement
from gedcom.element.object import ObjectElement
# Differentiate between the type of the new child element
if tag == gedcom.tags.GEDCOM_TAG_FAMILY:
child_element = FamilyElement(self.get_level() + 1, pointer, tag, value, self.__crlf)
elif tag == gedcom.tags.GEDCOM_TAG_FILE:
child_element = FileElement(self.get_level() + 1, pointer, tag, value, self.__crlf)
elif tag == gedcom.tags.GEDCOM_TAG_INDIVIDUAL:
child_element = IndividualElement(self.get_level() + 1, pointer, tag, value, self.__crlf)
elif tag == gedcom.tags.GEDCOM_TAG_OBJECT:
child_element = ObjectElement(self.get_level() + 1, pointer, tag, value, self.__crlf)
else:
child_element = Element(self.get_level() + 1, pointer, tag, value, self.__crlf)
self.add_child_element(child_element)
return child_element
def __parse_line(line_number, line, last_element, strict=True):
"""Parse a line from a GEDCOM 5.5 formatted document
Each line should have the following (bracketed items optional):
level + ' ' + [pointer + ' ' +] tag + [' ' + line_value]
:type line_number: int
:type line: str
:type last_element: Element
:type strict: bool
:rtype: Element
"""
# Level must start with non-negative int, no leading zeros.
level_regex = '^(0|[1-9]+[0-9]*) '
# Pointer optional, if it exists it must be flanked by `@`
pointer_regex = '(@[^@]+@ |)'
# Tag must be an alphanumeric string
tag_regex = '([A-Za-z0-9_]+)'
# Value optional, consists of anything after a space to end of line
value_regex = '( [^\n\r]*|)'
# End of line defined by `\n` or `\r`
end_of_line_regex = '([\r\n]{1,2})'
# Complete regex
gedcom_line_regex = level_regex + pointer_regex + tag_regex + value_regex + end_of_line_regex
regex_match = regex.match(gedcom_line_regex, line)
if regex_match is None:
if strict:
error_message = (
"Line %d of document violates GEDCOM format 5.5" %
line_number +
"\nSee: https://chronoplexsoftware.com/gedcomvalidator/gedcom/gedcom-5.5.pdf"
)
raise GedcomFormatViolationError(error_message)
else:
# Quirk check - see if this is a line without a CRLF (which could be the last line)
last_line_regex = level_regex + pointer_regex + tag_regex + value_regex
regex_match = regex.match(last_line_regex, line)
if regex_match is not None:
line_parts = regex_match.groups()
level = int(line_parts[0])
pointer = line_parts[1].rstrip(' ')
tag = line_parts[2]
value = line_parts[3][1:]
crlf = '\n'
else:
# Quirk check - Sometimes a gedcom has a text field with a CR.
# This creates a line without the standard level and pointer.
# If this is detected then turn it into a CONC or CONT.
line_regex = '([^\n\r]*|)'
cont_line_regex = line_regex + end_of_line_regex
regex_match = regex.match(cont_line_regex, line)
line_parts = regex_match.groups()
level = last_element.get_level()
tag = last_element.get_tag()
pointer = None
value = line_parts[0][1:]
crlf = line_parts[1]
if tag != gedcom.tags.GEDCOM_TAG_CONTINUED and tag != gedcom.tags.GEDCOM_TAG_CONCATENATION:
# Increment level and change this line to a CONC
level += 1
tag = gedcom.tags.GEDCOM_TAG_CONCATENATION
else:
line_parts = regex_match.groups()
level = int(line_parts[0])
pointer = line_parts[1].rstrip(' ')
tag = line_parts[2]
value = line_parts[3][1:]
crlf = line_parts[4]
# Check level: should never be more than one higher than previous line.
if level > last_element.get_level() + 1:
error_message = (
"Line %d of document violates GEDCOM format 5.5" % line_number
+
"\nLines must be no more than one level higher than previous line."
+
"\nSee: https://chronoplexsoftware.com/gedcomvalidator/gedcom/gedcom-5.5.pdf"
)
raise GedcomFormatViolationError(error_message)
# Create element. Store in list and dict, create children and parents.
if tag == gedcom.tags.GEDCOM_TAG_INDIVIDUAL:
element = IndividualElement(level,
pointer,
tag,
value,
crlf,
multi_line=False)
elif tag == gedcom.tags.GEDCOM_TAG_FAMILY:
element = FamilyElement(level,
pointer,
tag,
value,
crlf,
multi_line=False)
elif tag == gedcom.tags.GEDCOM_TAG_FILE:
element = FileElement(level,
pointer,
tag,
value,
crlf,
multi_line=False)
elif tag == gedcom.tags.GEDCOM_TAG_OBJECT:
element = ObjectElement(level,
pointer,
tag,
value,
crlf,
multi_line=False)
else:
element = Element(level,
pointer,
tag,
value,
crlf,
multi_line=False)
# Start with last element as parent, back up if necessary.
parent_element = last_element
while parent_element.get_level() > level - 1:
parent_element = parent_element.get_parent_element()
# Add child to parent & parent to child.
parent_element.add_child_element(element)
return element