obrabotka_kiz/.agents/skills/kiz-testing/SKILL.md

name, description

name	description
kiz-testing	How to test the KIZ -> CSV scripts (VBS and Python) with fake invoice and KIZ files, and what to check in the resulting CSV and report.

Testing the KIZ -> CSV scripts

There is no automated test suite. Test by generating fake input files, running the script, and checking the CSV and the report by eye and with a short script.

Where things run

• Python (obrabotka_kiz_python/kiz_to_csv.py) runs on any OS. It needs openpyxl. If it is not installed, use the copy shipped in the portable kit: PYTHONPATH=obrabotka_kiz_python/KIZ_portable_Win7/python/Lib/site-packages python3 ...
• VBS (obrabotka_kiz_vb/xlsx_to_csv_utf8.vbs) runs only on Windows with Excel installed (it drives Excel through COM). On Linux/Mac it cannot be run: test the Python version and give the user the same fake files and the expected CSV to compare on Windows.

Both scripts must produce the same CSV content for the same input (the VBS file starts with a UTF-8 BOM, the Python file does not). Any change to the algorithm goes into both, and into the copy obrabotka_kiz_python/KIZ_portable_Win7/kiz_to_csv.py (it must stay identical to obrabotka_kiz_python/kiz_to_csv.py).

Fake input files

Put them in a temporary folder, never in the repo. The sample files in the repo root (Куропятник.xlsx, Книга1.xlsx) show the real layouts, but Куропятник.xlsx has no GTINs in column B, so nothing matches with it.

import openpyxl, sys
d = sys.argv[1]
# invoice (file 1): goods from row 19, B = "name GTIN", K = qty, L = price, N = VAT
wb = openpyxl.Workbook(); ws = wb.active
rows = [("Мишка с сердцем 22см 04640720990098", 3, 358.18, 10),
        ("Кролик в шарфе 4640720990111", 1, 436.37, 20)]
for i, (b, k, l, n) in enumerate(rows, start=19):
    ws.cell(i, 1, i - 18); ws.cell(i, 2, b); ws.cell(i, 11, k); ws.cell(i, 12, l); ws.cell(i, 14, n)
wb.save(d + "/nakl.xlsx")
# KIZ file (file 2): codes in the last filled column (H), text format
wb = openpyxl.Workbook(); ws = wb.active
kiz = ["01046407209900982151B+-h6cMAOOb91EE1292vCB/qXHt4iD3Rilqesef3ITNwAAaBVALGwGuFASGU+E=",
       '01046407209900982154Jcb6ZtF"NPK91EE1292fUJhZxhh7qKJZaHjU+Em+7hIA4u4Ei64ilySLnpMyDc=',
       "0104640720990098215KK,kKKkKkKKK91EE1292abcdefghijklmnopqrstuvwxyz0123456789ABCDEFGH=",
       "0104640720990111215ZZZZZZZZZZZZ91EE1292zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz=",
       "01046407209901112155HORT"]
for i, k in enumerate(kiz, start=1):
    ws.cell(i, 1, i); ws.cell(i, 8, k).number_format = "@"
wb.save(d + "/kiz.xlsx")

The set covers: a normal code, a code with a " in the serial, a code with a , in the serial, a second product with a quantity mismatch, and a code shorter than 31 characters.

Run: python3 obrabotka_kiz_python/kiz_to_csv.py <dir>/nakl.xlsx <dir>/kiz.xlsx (two arguments = no file dialogs).

What to check

• <dir>/kiz.csv: UTF-8, CRLF line ends, comma separator; B..G filled (name, price with a dot, 1, 796, VAT with %, КИЗ).
• Last column: each KIZ is cut to its first 31 characters (01 + 14-digit GTIN + 21 + 13-character serial); a code with a comma is wrapped in double quotes; a shorter code is left unchanged.
• <dir>/kiz_report.txt: an OK line for the matching product, a "НЕ СОВПАДАЕТ" (quantity mismatch) line for the other, and a "КИЗ короче 31 символов" line for the short code.

Quick length check:

python3 -c "import csv; print([len(r[-1]) for r in csv.reader(open('<dir>/kiz.csv', encoding='utf-8'))])"

Expected for the set above: [31, 31, 31, 31, 24].

Delete the temporary folder when done.