Jason Thistlethwaite
82 lines
3.0 KiB
Python
82 lines
3.0 KiB
Python
#!/usr/bin/env python3
|
|
from __future__ import annotations
|
|
|
|
import tempfile
|
|
import unittest
|
|
from pathlib import Path
|
|
|
|
from redmine_ticket_ocr import (
|
|
build_ocr_note,
|
|
extract_text_from_file,
|
|
looks_like_generic_scan_subject,
|
|
suggest_subject_from_text,
|
|
)
|
|
from redmine_tickets import KNOWN_COMMANDS, build_parser, inject_default_sync_command, is_closed_status
|
|
|
|
|
|
class OcrSuggestTest(unittest.TestCase):
|
|
def test_generic_subject(self):
|
|
self.assertTrue(looks_like_generic_scan_subject("Scanned document from HP"))
|
|
self.assertFalse(looks_like_generic_scan_subject("Invoice — Acme Corp"))
|
|
|
|
def test_suggest_w2(self):
|
|
text = "Form W-2 Wage and Tax Statement\nName: Jane Doe\nCompany: Acme LLC\nJanuary 15, 2026"
|
|
subj = suggest_subject_from_text(text, original_subject="Scanned document")
|
|
self.assertIn("W-2", subj)
|
|
self.assertIn("Jane Doe", subj)
|
|
|
|
def test_suggest_vendor_invoice_skips_fee_bullet(self):
|
|
text = (
|
|
"- Return Check Fee up to $45.00\n"
|
|
"- A 1.5% Service Charge, 18% Annually ($1.00 Minimum) will be charged\n"
|
|
"Best Way Disposal\n"
|
|
"Invoice Number: 2404778\n"
|
|
"Invoice Due Date 9/25/2026\n"
|
|
"Total Due $307.52\n"
|
|
)
|
|
subj = suggest_subject_from_text(text, original_subject="Scanned document from LDR Office")
|
|
self.assertIn("Invoice 2404778", subj)
|
|
self.assertIn("Best Way Disposal", subj)
|
|
self.assertNotIn("Return Check Fee", subj)
|
|
|
|
def test_txt_extract(self):
|
|
with tempfile.TemporaryDirectory() as td:
|
|
p = Path(td) / "a.txt"
|
|
p.write_text("hello ocr world " * 5, encoding="utf-8")
|
|
text, method = extract_text_from_file(p)
|
|
self.assertEqual("read", method)
|
|
self.assertIn("hello", text)
|
|
|
|
def test_note_builder(self):
|
|
note = build_ocr_note(
|
|
issue_id=1,
|
|
files=[{"filename": "a.pdf", "method": "pdftotext", "text": "body"}],
|
|
proposed_subject="Invoice — X",
|
|
)
|
|
self.assertIn("proposed_subject: Invoice — X", note)
|
|
self.assertIn("a.pdf", note)
|
|
|
|
|
|
class CliRegistrationTest(unittest.TestCase):
|
|
def test_commands(self):
|
|
self.assertIn("ocr-rename", KNOWN_COMMANDS)
|
|
self.assertIn("categories", KNOWN_COMMANDS)
|
|
args = build_parser().parse_args(
|
|
inject_default_sync_command(
|
|
["ocr-rename", "--issue-id", "1", "--no-queue-note", "--no-queue-subject"]
|
|
)
|
|
)
|
|
self.assertEqual("ocr-rename", args.command)
|
|
args2 = build_parser().parse_args(
|
|
inject_default_sync_command(["categories", "list", "--refresh"])
|
|
)
|
|
self.assertEqual("categories", args2.command)
|
|
args3 = build_parser().parse_args(
|
|
inject_default_sync_command(["walk", "--open-only", "--max-tickets", "3"])
|
|
)
|
|
self.assertTrue(args3.open_only)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|