Smart Data Solutions Software Engineer
Interview Guide

import csv
import re
from datetime import date

# Assumed sample layout: (column, 1-based start, width); positions count bytes
LAYOUT = [("patient_id", 1, 8), ("last_name", 9, 15), ("first_name", 24, 12),
          ("dob", 36, 8), ("sex", 44, 1)]
RULES = {"patient_id": re.compile(r"\d{8}"), "sex": re.compile(r"[MFU]?"),
         "dob": re.compile(r"(\d{4}(0[1-9]|1[0-2])(0[1-9]|[12]\d|3[01]))?")}

def parse_record(line, layout=LAYOUT, encoding="utf-8"):
    """Slice a line of bytes by position into trimmed, validated values."""
    record_len = max(start - 1 + width for _, start, width in layout)
    if len(line.rstrip()) > record_len:
        raise ValueError(f"expected <= {record_len} bytes, got {len(line.rstrip())}")
    values = []
    for name, start, width in layout:
        try:  # a short line gives b'' here: trailing blanks are often stripped
            value = line[start - 1:start - 1 + width].decode(encoding).strip()
            if name in RULES and not RULES[name].fullmatch(value):
                raise ValueError(f"invalid value {value!r}")
            if name == "dob" and value:  # shape is fine; now check the calendar
                date(int(value[:4]), int(value[4:6]), int(value[6:]))
        except ValueError as exc:  # UnicodeDecodeError is a ValueError too
            raise ValueError(f"{name}: {exc}") from None
        values.append(value)
    return values

def convert(dat_path, csv_path, layout=LAYOUT, encoding="utf-8"):
    """Stream a .dat file into a .csv file. Returns (rows_written, rejects)."""
    written, rejects = 0, []
    with open(dat_path, "rb") as src, \
            open(csv_path, "w", encoding="utf-8", newline="") as dst:
        out = csv.writer(dst)  # quotes values that contain commas or quotes
        out.writerow([name for name, _, _ in layout])
        for line_no, raw in enumerate(src, start=1):
            line = raw.removeprefix(b"\xef\xbb\xbf") if line_no == 1 else raw  # BOM
            if not line.strip():
                continue  # blank line; strip() trims CR/LF off data lines
            try:
                out.writerow(parse_record(line, layout, encoding))
                written += 1
            except ValueError as exc:
                rejects.append((line_no, str(exc)))  # report, keep going
    return written, rejects

Scroll sideways to view long lines.

import csv
import io
import unittest

def convert_line(line, widths):  # one fixed-width line -> CSV row, None if blank
    line = line.rstrip("\r\n")
    if not line.strip():
        return None
    if line[sum(widths):].strip():  # spaces past the record are fine, data is not
        raise ValueError(f"unexpected data after column {sum(widths)}")
    starts = [sum(widths[:i]) for i in range(len(widths))]
    fields = [line[s:s + w].strip() for s, w in zip(starts, widths)]
    buf = io.StringIO()
    csv.writer(buf, lineterminator="").writerow(fields)  # quotes 'Doe, Jr'
    return buf.getvalue()

class ConvertLineBoundaryTests(unittest.TestCase):
    WIDTHS = [5, 10, 3]  # id, name, code

    def test_boundaries(self):
        cases = [  # (what the case pins down, input line, expected output)
            ("field fills exact width", "12345Maximilian042", "12345,Maximilian,042"),
            ("pad inside field trimmed", "12345Jo        042", "12345,Jo,042"),
            ("trailing spaces after record", "12345Jo        042    ", "12345,Jo,042"),
            ("trailing field stripped away", "12345Jo", "12345,Jo,"),
            ("all-blank middle field kept", "12345          042", "12345,,042"),
            ("right-aligned field trimmed", "  345Jo         42", "345,Jo,42"),
            ("CRLF on a short line", "12345Jo\r\n", "12345,Jo,"),
            ("comma in a value quoted", "12345Doe, Jr   042", '12345,"Doe, Jr",042'),
            ("quote in a value doubled", '12345Al "Big"  042', '12345,"Al ""Big""",042'),
            ("empty line skipped", "\n", None),
            ("whitespace-only line skipped", "      \t  ", None),
        ]
        for name, line, expected in cases:
            with self.subTest(name):
                self.assertEqual(convert_line(line, self.WIDTHS), expected)

    def test_data_after_record_rejected(self):
        self.assertRaises(ValueError, convert_line, "12345Jo        042XYZ", self.WIDTHS)

    def test_column_count_never_changes(self):
        for line in ["1", "12345Jo", "12345          042", "12345Doe, Jr   042"]:
            with self.subTest(line=line):
                row = next(csv.reader([convert_line(line, self.WIDTHS)]))
                self.assertEqual(len(row), 3)

Scroll sideways to view long lines.