def test_extract_report_owner_form3_collection(test_form3_collection,
doc_num: int):
"""
Validate Form3 extraction code against a random sample of documents
:param test_form3_collection:
:return:
"""
file = list(test_form3_collection.glob("*.txt"))[doc_num]
doc = Form3(file)
assert doc.filename == file.name
fields_list = doc.report_owners
assert len(fields_list) > 0
for idx, fields in enumerate(fields_list):
assert (len(fields)) == 19
assert fields["filename"] == file.name
assert fields["order"] == f"{idx+1}"
assert fields["type"] == "reportingOwner"
assert fields["index"] == f"reportingOwner{idx+1}"
assert validate(file, fields["rpt_owner_cik"], r"\d{10}")
assert validate(file, fields["rpt_owner_name"], r".+")
assert validate(file, fields["rpt_owner_city"], r".+")
assert validate(file,
fields["is_director"],
r"[10]",
none_allowed=True)
assert validate(file, fields["is_officer"], r"[10]", none_allowed=True)
assert validate(file,
fields["is_ten_percent_owner"],
r"[10]",
none_allowed=True)
assert validate(file, fields["is_other"], r"[10]", none_allowed=True)
def test_extract_report_owner_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields_list = doc.report_owners
assert len(fields_list) == 1
assert len(fields_list[0]) == 19
assert fields_list[0]["filename"] == test_form3.name
assert fields_list[0]["accession_num"] == doc.accession_num
assert fields_list[0]["order"] == "1"
assert fields_list[0]["type"] == "reportingOwner"
assert fields_list[0]["index"] == "reportingOwner1"
assert fields_list[0]["rpt_owner_cik"] == "0001676526"
assert fields_list[0]["rpt_owner_name"] == "Lawler John T."
assert fields_list[0]["rpt_owner_street1"] == "ONE AMERICAN ROAD"
assert fields_list[0]["rpt_owner_street2"] is None
assert fields_list[0]["rpt_owner_city"] == "DEARBORN"
assert fields_list[0]["rpt_owner_state"] == "MI"
assert fields_list[0]["rpt_owner_zip_code"] == "48126"
assert fields_list[0]["rpt_owner_state_descr"] is None
assert fields_list[0]["is_director"] == "0"
assert fields_list[0]["is_officer"] == "1"
assert fields_list[0]["is_ten_percent_owner"] == "0"
assert fields_list[0]["is_other"] == "0"
assert fields_list[0]["officer_title"] == "Vice President, CFO"
assert fields_list[0]["other_text"] is None
def test_extract_derivative_trans_form3_collection(test_form3_collection, doc_num: int):
"""
Validate Form3 extraction code against a random sample of documents
:param test_form3_collection:
:return:
"""
file = list(test_form3_collection.glob("*.txt"))[doc_num]
doc = Form3(file)
assert doc.filename == file.name
fields_list = doc.derivatives
assert len(fields_list) >= 0
for idx, fields in enumerate(fields_list):
assert (len(fields)) == 12
assert fields["filename"] == file.name
assert fields["accession_num"] == doc.accession_num
assert fields["order"] == f"{idx+1}"
assert fields["type"] == "derivHolding"
assert fields["index"] == f"derivHolding{idx+1}"
assert validate(file, fields["accession_num"], r"[\d-]+")
assert validate(file, fields["security_title"], r".+")
assert validate(file, fields["conversion_or_exercise_price"], r"[\d+\.]*")
assert validate(file, fields["exercise_date"], r"(\d\d\d\d-\d\d-\d\d)?")
assert validate(file, fields["expiration_date"], r"(\d\d\d\d-\d\d-\d\d)?")
assert validate(file, fields["underlying_security_title"], r".+")
assert validate(file, fields["underlying_security_shares"], r"[\d+\.]+")
assert validate(file, fields["direct_or_indirect_ownership"], r"[DI]")
def test_extract_doclevel_form3_collection(test_form3_collection):
"""
Validate Form3 extraction code against a random sample of documents
:param test_form3_collection:
:return:
"""
for file in test_form3_collection.glob("*.txt"):
doc = Form3(file)
assert doc.filename == file.name
fields = doc.doc_info
assert len(fields) == 19
assert fields["filename"] == file.name
assert fields["schema_version"] == "X0206"
assert fields["document_type"] == "3"
assert validate(file, fields["accession_num"], r"[\d-]+")
assert validate(file, fields["sec_accept_datetime"], r"\d{14,14}")
assert validate(file, fields["sec_file_num"], r"[\d-]+")
assert validate(file, fields["doc_count"], r"\d+")
assert validate(file, fields["filed_date"], r"\d{8,8}")
assert validate(file, fields["conformed_period_of_report"], r"\d{8,8}")
assert validate(file, fields["change_date"], r"\d{8,8}")
assert validate(file, fields["period_of_report"], r"\d\d\d\d-\d\d-\d\d")
assert validate(
file, fields["not_subject_to_section_16"], r"[10]", none_allowed=True
)
assert validate(file, fields["issuer_cik"], r"\d+")
assert validate(file, fields["issuer_name"], r".+")
assert validate(file, fields["issuer_trading_symbol"], r"[A-Z\.]+")
assert validate(file, fields["regcik"], r"\d+")
assert validate(file, fields["regsic"], r"\d\d\d\d", none_allowed=True)
assert validate(file, fields["no_securities_owned"], r"[01]")
def test_extract_doclevel_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields = doc.doc_info
assert len(fields) == 19
assert fields["filename"] == test_form3.name
assert fields["accession_num"] == doc.accession_num
assert fields["sec_accept_datetime"] == "20201007163300"
assert fields["sec_file_num"] == "001-03950"
assert fields["doc_count"] == "2"
assert fields["filed_date"] == "20201007"
assert fields["conformed_period_of_report"] == "20201001"
assert fields["change_date"] == "20201007"
assert fields["schema_version"] == "X0206"
assert fields["document_type"] == "3"
assert fields["period_of_report"] == "2020-10-01"
assert fields["not_subject_to_section_16"] is None
assert fields["issuer_cik"] == "0000037996"
assert fields["issuer_name"] == "FORD MOTOR CO"
assert fields["issuer_trading_symbol"] == "F"
assert fields["remarks"] is None
assert fields["regcik"] == "0000037996"
assert fields["regsic"] == "3711"
assert fields["no_securities_owned"] == "0"
def test_extract_footnotes_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields_list = doc.footnotes
assert len(fields_list) == 19
assert len(fields_list[0]) == 6
assert fields_list[0]["filename"] == test_form3.name
assert fields_list[0]["accession_num"] == doc.accession_num
assert fields_list[0]["footnote"] == "F1"
assert fields_list[0]["index"] == "derivHolding1"
assert fields_list[0]["field"] == "conversionOrExercisePrice"
assert fields_list[0]["text"][:27] == "These Ford Stock Fund Units"
assert fields_list[18]["filename"] == test_form3.name
assert fields_list[18]["accession_num"] == doc.accession_num
assert fields_list[18]["footnote"] == "F9"
assert fields_list[18]["index"] == "derivHolding9"
assert fields_list[18]["field"] == "expirationDate"
assert fields_list[18]["text"][:27] == "These Ford Restricted Stock"
def create_doc(file: Path) -> Document:
xmlpath = Document.xml_document_fields["document_type"]
regex = re.compile(f"<{xmlpath}>(.+)</{xmlpath}>")
form_type = regex.findall(file.read_text())[0]
if form_type == "3":
return Form3(file)
elif form_type == "4":
return Form4(file)
elif form_type == "5":
return Form5(file)
else:
typer.secho(f"WARNING: {form_type} not a supported form type",
fg=typer.colors.RED)
return None
def test_extract_derivative_trans_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields_list = doc.derivatives
assert len(fields_list) == 9
assert len(fields_list[0]) == 12
assert fields_list[0]["filename"] == test_form3.name
assert fields_list[0]["accession_num"] == doc.accession_num
assert fields_list[0]["order"] == "1"
assert fields_list[0]["type"] == "derivHolding"
assert fields_list[0]["index"] == "derivHolding1"
assert fields_list[0]["security_title"] == "BEP Ford Stock Fund Units"
assert fields_list[0]["conversion_or_exercise_price"] == ""
assert fields_list[0]["exercise_date"] == ""
assert fields_list[0]["expiration_date"] == ""
assert (
fields_list[0]["underlying_security_title"] == "Common Stock, $0.01 par value"
)
assert fields_list[0]["underlying_security_shares"] == "65"
assert fields_list[0]["direct_or_indirect_ownership"] == "D"
assert len(fields_list[8]) == 12
assert fields_list[8]["filename"] == test_form3.name
assert fields_list[8]["accession_num"] == doc.accession_num
assert fields_list[8]["order"] == "9"
assert fields_list[8]["type"] == "derivHolding"
assert fields_list[8]["index"] == "derivHolding9"
assert fields_list[8]["security_title"] == "Ford Stock Units"
assert fields_list[8]["conversion_or_exercise_price"] == ""
assert fields_list[8]["exercise_date"] == ""
assert fields_list[8]["expiration_date"] == ""
assert (
fields_list[8]["underlying_security_title"] == "Common Stock, $0.01 par value"
)
assert fields_list[8]["underlying_security_shares"] == "247533"
assert fields_list[8]["direct_or_indirect_ownership"] == "D"
def create_doc(file: Path) -> Union[None, Document]:
"""
A factory function returns a Document object with the right subtype.
:param file: The file object that is the source of the Document
:return: A Document object
"""
xmlpath = Document.xml_document_fields["document_type"]
regex = re.compile(f"<{xmlpath}>(.+)</{xmlpath}>")
form_type = regex.findall(file.read_text())[0]
if form_type == "3":
return Form3(file)
elif form_type == "4":
return Form4(file)
elif form_type == "5":
return Form5(file)
else:
typer.secho(
f"WARNING: {form_type} not a supported form type", fg=typer.colors.RED
)
return None
def test_extract_signature_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields_list = doc.nonderivatives
assert len(fields_list) == 1
assert len(fields_list[0]) == 8
assert fields_list[0]["filename"] == test_form3.name
assert fields_list[0]["accession_num"] == doc.accession_num
assert fields_list[0]["order"] == "1"
assert fields_list[0]["type"] == "nonDerivHolding"
assert fields_list[0]["index"] == "nonDerivHolding1"
assert fields_list[0]["security_title"] == "Common Stock, $0.01 par value"
assert fields_list[0]["shares_owned_following_transaction"] == "157800"
assert fields_list[0]["direct_or_indirect_ownership"] == "D"
def test_extract_signature_form3_collection(test_form3_collection,
doc_num: int):
"""
Validate Form3 extraction code against a random sample of documents
:param test_form3_collection:
:return:
"""
file = list(test_form3_collection.glob("*.txt"))[doc_num]
doc = Form3(file)
assert doc.filename == file.name
fields_list = doc.signatures
assert len(fields_list) > 0
for idx, fields in enumerate(fields_list):
assert (len(fields)) == 7
assert fields["filename"] == file.name
assert fields["accession_num"] == doc.accession_num
assert fields["order"] == f"{idx+1}"
assert fields["type"] == "signature"
assert fields["index"] == f"signature{idx+1}"
assert validate(file, fields["signature_name"], r".+")
assert validate(file, fields["signature_date"], r"\d\d\d\d-\d\d-\d\d")
def test_extract_footnotes_form4_collection(test_form4_collection,
doc_num: int):
"""
Validate Form4 extraction code against a random sample of documents
:param test_form4_collection:
:return:
"""
file = list(test_form4_collection.glob("*.txt"))[doc_num]
doc = Form3(file)
assert doc.filename == file.name
fields_list = doc.footnotes
assert len(fields_list) >= 0
for idx, fields in enumerate(fields_list):
assert (len(fields)) == 6
assert fields["filename"] == file.name
assert fields["accession_num"] == doc.accession_num
assert validate(file, fields["accession_num"], r"[\d-]+")
assert validate(file, fields["footnote"], r".+")
assert validate(file, fields["index"], r"\w+\d+")
assert validate(file, fields["field"], r"\w+")
assert validate(file, fields["field"], r".+")
def test_extract_signature_form3(test_form3):
"""
Validate Form3 extraction code against a single detailed example
:param test_form3:
:return:
"""
doc = Form3(test_form3)
assert doc.accession_num == "0001209191-20-054135"
assert doc.filename == test_form3.name
fields_list = doc.signatures
assert len(fields_list) == 1
assert len(fields_list[0]) == 7
assert fields_list[0]["filename"] == test_form3.name
assert fields_list[0]["accession_num"] == doc.accession_num
assert fields_list[0]["order"] == "1"
assert fields_list[0]["type"] == "signature"
assert fields_list[0]["index"] == "signature1"
assert fields_list[0][
"signature_name"] == "Jerome F. Zaremba,\nAttorney-in-Fact"
assert fields_list[0]["signature_date"] == "2020-10-07"
def test_extract_nonderivatives_form3_collection(test_form3_collection,
doc_num: int):
"""
Validate Form3 extraction code against a random sample of documents
:param test_form3_collection:
:return:
"""
file = list(test_form3_collection.glob("*.txt"))[doc_num]
doc = Form3(file)
assert doc.filename == file.name
fields_list = doc.nonderivatives
assert len(fields_list) >= 0
for idx, fields in enumerate(fields_list):
assert (len(fields)) == 8
assert fields["filename"] == file.name
assert fields["accession_num"] == doc.accession_num
assert fields["order"] == f"{idx+1}"
assert fields["type"] == "nonDerivHolding"
assert fields["index"] == f"nonDerivHolding{idx+1}"
assert validate(file, fields["security_title"], r".+")
assert validate(file, fields["shares_owned_following_transaction"],
r"[\d\.]+")
assert validate(file, fields["direct_or_indirect_ownership"], r"[DI]")