def test_extract_fails_with_invalid_date(ai_code: str, bad_value: str) -> None:
ai = GS1ApplicationIdentifier.extract(ai_code)
with pytest.raises(ParseError) as exc_info:
GS1ElementString.extract(f"{ai_code}{bad_value}")
assert (str(exc_info.value) ==
f"Failed to parse GS1 AI {ai} date from {bad_value!r}.")
def test_filter_element_strings_by_ai_instance() -> None:
ai = GS1ApplicationIdentifier.extract("01")
msg = GS1Message.parse("010703206980498815210526100329")
element_string = msg.get(ai=ai)
assert element_string is not None
assert element_string.value == "07032069804988"
def test_extract_fails_when_not_matching_pattern(ai_code: str,
bad_value: str) -> None:
ai = GS1ApplicationIdentifier.extract(ai_code)
with pytest.raises(ParseError) as exc_info:
GS1ElementString.extract(bad_value)
assert (
str(exc_info.value) ==
f"Failed to match {bad_value!r} with GS1 AI {ai} pattern '{ai.pattern}'."
)
def extract(
cls,
value: str,
*,
rcn_region: Optional[RcnRegion] = None,
separator_chars: Iterable[str] = DEFAULT_SEPARATOR_CHARS,
) -> GS1ElementString:
"""Extract the first GS1 Element String from the given value.
Args:
value: The string to extract an Element String from. May contain
more than one Element String.
rcn_region: The geographical region whose rules should be used to
interpret Restricted Circulation Numbers (RCN).
Needed to extract e.g. variable weight/price from GTIN.
separator_chars: Characters used in place of the FNC1 symbol.
Defaults to `<GS>` (ASCII value 29).
If variable-length fields are not terminated with a separator
character, the parser might greedily consume later fields.
Returns:
A data class with the Element String's parts and data extracted from it.
Raises:
ValueError: If the ``separator_char`` isn't exactly 1 character long.
ParseError: If the parsing fails.
"""
if any(len(char) != 1 for char in separator_chars):
raise ValueError(
"All separator characters must be exactly 1 character long, "
f"got {list(separator_chars)!r}.")
ai = GS1ApplicationIdentifier.extract(value)
for separator_char in separator_chars:
value = value.split(separator_char, maxsplit=1)[0]
pattern = ai.pattern[:-1] if ai.pattern.endswith("$") else ai.pattern
matches = re.match(pattern, value)
if not matches:
raise ParseError(
f"Failed to match {value!r} with GS1 AI {ai} pattern '{ai.pattern}'."
)
pattern_groups = list(matches.groups())
value = "".join(pattern_groups)
element = cls(ai=ai, value=value, pattern_groups=pattern_groups)
element._set_gln()
element._set_gtin(rcn_region=rcn_region)
element._set_sscc()
element._set_date()
element._set_decimal()
return element
def parse(html_content: str) -> List[GS1ApplicationIdentifier]:
"""Parse the data from HTML to GS1ApplicationIdentifier objects."""
result = []
page = BeautifulSoup(html_content, "html.parser")
datatable = page.find("table", {"class": ["datatable"]})
tbody = datatable.find("tbody")
for row in tbody.find_all("tr"):
columns = row.find_all("td")
result.append(
GS1ApplicationIdentifier(
ai=columns[0].text.strip(),
description=columns[1].text.strip(),
format=columns[2].text.strip(),
data_title=columns[3].text.strip(),
fnc1_required=columns[4].text.strip() == "Yes",
pattern=columns[5].text.strip(),
))
return result
import pytest
from biip import ParseError
from biip.gln import Gln
from biip.gs1 import GS1ApplicationIdentifier, GS1ElementString, GS1Prefix
from biip.gtin import Gtin, GtinFormat
from biip.sscc import Sscc
@pytest.mark.parametrize(
"value, expected",
[
(
"00373400306809981733",
GS1ElementString(
ai=GS1ApplicationIdentifier.extract("00"),
value="373400306809981733",
pattern_groups=["373400306809981733"],
sscc=Sscc(
value="373400306809981733",
prefix=GS1Prefix(value="734", usage="GS1 Sweden"),
extension_digit=3,
payload="37340030680998173",
check_digit=3,
),
),
),
(
"0107032069804988",
GS1ElementString(
ai=GS1ApplicationIdentifier.extract("01"),
from biip.gtin import Gtin, GtinFormat
@pytest.mark.parametrize(
"value, expected",
[
(
"010703206980498815210526100329",
GS1Message(
value="010703206980498815210526100329",
element_strings=[
GS1ElementString(
ai=GS1ApplicationIdentifier(
ai="01",
description="Global Trade Item Number (GTIN)",
data_title="GTIN",
fnc1_required=False,
format="N2+N14",
pattern="^01(\\d{14})$",
),
value="07032069804988",
pattern_groups=["07032069804988"],
gtin=Gtin(
value="07032069804988",
format=GtinFormat.GTIN_13,
prefix=GS1Prefix(value="703", usage="GS1 Norway"),
payload="703206980498",
check_digit=8,
),
),
GS1ElementString(
ai=GS1ApplicationIdentifier(
def test_extract_unknown_gs1_ai(unknown_ai: str) -> None:
with pytest.raises(ParseError) as exc_info:
GS1ApplicationIdentifier.extract(unknown_ai)
assert (str(exc_info.value) ==
f"Failed to get GS1 Application Identifier from {unknown_ai!r}.")
def test_is_hashable() -> None:
ai = GS1ApplicationIdentifier.extract("01")
assert hash(ai) is not None
def test_gs1_ai_object_len_is_ai_str_len() -> None:
ai = GS1ApplicationIdentifier.extract("01")
assert len(ai) == len("01")
def test_extract_gs1_ai(value: str,
expected: GS1ApplicationIdentifier) -> None:
assert GS1ApplicationIdentifier.extract(value) == expected
with pytest.raises(ParseError) as exc_info:
GS1ApplicationIdentifier.extract(unknown_ai)
assert (str(exc_info.value) ==
f"Failed to get GS1 Application Identifier from {unknown_ai!r}.")
@pytest.mark.parametrize(
"value, expected",
[
(
"0195012345678903",
GS1ApplicationIdentifier(
ai="01",
description="Global Trade Item Number (GTIN)",
data_title="GTIN",
fnc1_required=False,
format="N2+N14",
pattern=r"^01(\d{14})$",
),
),
(
"90SE011\x1d2501611262",
GS1ApplicationIdentifier(
ai="90",
description=
"Information mutually agreed between trading partners",
data_title="INTERNAL",
fnc1_required=True,
format="N2+X..30",
pattern=(r"^90([\x21-\x22\x25-\x2F\x30-\x39\x3A-\x3F\x41-"
r"\x5A\x5F\x61-\x7A]{0,30})$"),
# GTIN-8
"96385074",
ParseResult(
value="96385074",
gtin=Gtin(
value="96385074",
format=GtinFormat.GTIN_8,
prefix=GS1Prefix(value="00009", usage="GS1 US"),
payload="9638507",
check_digit=4,
),
gs1_message=GS1Message(
value="96385074",
element_strings=[
GS1ElementString(
ai=GS1ApplicationIdentifier.extract("96"),
value="385074",
pattern_groups=["385074"],
)
],
),
sscc_error=
("Failed to parse '96385074' as SSCC: Expected 18 digits, got 8."
),
),
),
(
# GTIN-12
"123601057072",
ParseResult(
value="123601057072",