impl/python.py

import re
from typing import List, Optional, Tuple

from .validation_uk_modulus_table_types import UkModulusRow, UkModulusTable, UkSortCodeSubstitution

_SORT_CODE = re.compile(r"[0-9]{6}")
_WEIGHT = re.compile(r"-?[0-9]{1,4}")
_EXCEPTION = re.compile(r"[0-9]{1,2}")
_SEPARATOR = re.compile(r"[ \t]+")
_ALGORITHMS = ("MOD10", "MOD11", "DBLAL")


def _lines(text: str) -> List[Tuple[int, List[str]]]:
    """The fields of each non-blank line. Only ASCII spaces and tabs separate
    fields, and a trailing carriage return is dropped, so a file saved with
    Windows line endings reads the same and every language splits alike."""
    out: List[Tuple[int, List[str]]] = []
    body = text[1:] if text.startswith("") else text
    for i, raw in enumerate(body.split("\n")):
        line = raw[:-1] if raw.endswith("\r") else raw
        fields = [f for f in _SEPARATOR.split(line) if f != ""]
        if fields:
            out.append((i + 1, fields))
    return out


def _sort_code(file: str, line: int, value: str) -> str:
    if not _SORT_CODE.fullmatch(value):
        raise ValueError('%s line %d: "%s" is not a six-digit sort code' % (file, line, value))
    return value


def _parse_row(line: int, fields: List[str]) -> UkModulusRow:
    where = "VALACDOS.txt line %d" % line
    if len(fields) not in (17, 18):
        raise ValueError(
            "%s: expected 17 or 18 fields (start, end, algorithm, 14 weights, optional exception), found %d"
            % (where, len(fields))
        )
    start = _sort_code("VALACDOS.txt", line, fields[0])
    end = _sort_code("VALACDOS.txt", line, fields[1])
    if end < start:
        raise ValueError("%s: range %s to %s ends before it starts" % (where, start, end))
    algorithm = fields[2]
    if algorithm not in _ALGORITHMS:
        raise ValueError('%s: unknown algorithm "%s"; expected MOD10, MOD11 or DBLAL' % (where, algorithm))
    weights: List[int] = []
    for w in fields[3:17]:
        if not _WEIGHT.fullmatch(w):
            raise ValueError('%s: weight "%s" is not a whole number' % (where, w))
        weights.append(int(w))
    # A digit sum of a negative product means different things in different
    # languages' remainder rules; the specification never needs one.
    if algorithm == "DBLAL" and any(w < 0 for w in weights):
        raise ValueError("%s: a DBLAL row cannot have a negative weight" % where)
    exception: Optional[int] = None
    if len(fields) == 18:
        e = fields[17]
        if not _EXCEPTION.fullmatch(e) or not 1 <= int(e) <= 14:
            raise ValueError('%s: exception "%s" is not a number from 1 to 14' % (where, e))
        exception = int(e)
    return UkModulusRow(start=start, end=end, algorithm=algorithm, weights=weights, exception=exception)


def _parse_substitution(line: int, fields: List[str]) -> UkSortCodeSubstitution:
    if len(fields) != 2:
        raise ValueError("SCSUBTAB.txt line %d: expected two sort codes, found %d fields" % (line, len(fields)))
    return UkSortCodeSubstitution(
        original=_sort_code("SCSUBTAB.txt", line, fields[0]),
        substitute=_sort_code("SCSUBTAB.txt", line, fields[1]),
    )


def parse_uk_modulus_table(valacdos: str, scsubtab: str) -> UkModulusTable:
    """Parse the text of Vocalink's VALACDOS.txt (the modulus weight table)
    and SCSUBTAB.txt (exception 5's sort code substitutions) into the table
    validate_uk_sort_code_account checks against.

    Rows keep their file order, which matters: where a sort code falls in two
    rows, the first is the first check. Blank lines are skipped; anything else
    malformed is an error naming the file and the line, because a silently
    dropped row turns a checkable sort code into an unchecked one.
    """
    rows = [_parse_row(number, fields) for number, fields in _lines(valacdos)]
    if not rows:
        raise ValueError("VALACDOS.txt has no rows")
    substitutions = [_parse_substitution(number, fields) for number, fields in _lines(scsubtab)]
    return UkModulusTable(rows=rows, substitutions=substitutions)