Files
X-Financial/server/tests/test_ocr_service.py

84 lines
2.6 KiB
Python
Raw Normal View History

from __future__ import annotations
import stat
from pathlib import Path
from app.core.config import get_settings
from app.services.ocr import OcrService
def test_ocr_service_uses_worker_runtime_and_keeps_unsupported_files_as_warnings(
monkeypatch,
tmp_path: Path,
) -> None:
fake_python = tmp_path / "fake-ocr-python.py"
fake_python.write_text(
"""#!/usr/bin/env python3
import json
import sys
inputs = []
for index, arg in enumerate(sys.argv):
if arg == "--input" and index + 1 < len(sys.argv):
input_path = sys.argv[index + 1]
inputs.append(
{
"input_path": input_path,
"engine": "paddleocr_mobile",
"model": "PP-OCRv5_mobile",
"text": "发票金额 100 元",
"summary": "发票金额 100 元",
"avg_score": 0.98,
"line_count": 1,
"page_count": 1,
"warnings": [],
"lines": [
{
"text": "发票金额 100 元",
"score": 0.98,
"box": [[1, 2], [10, 2], [10, 8], [1, 8]],
"page_index": 0,
}
],
}
)
payload = {
"engine": "paddleocr_mobile",
"model": "PP-OCRv5_mobile",
"documents": inputs,
}
print("__OCR_JSON__=" + json.dumps(payload, ensure_ascii=False))
""",
encoding="utf-8",
)
fake_python.chmod(fake_python.stat().st_mode | stat.S_IEXEC)
monkeypatch.setenv("OCR_PYTHON_BIN", str(fake_python))
monkeypatch.setenv("STORAGE_ROOT_DIR", str(tmp_path / "storage"))
get_settings.cache_clear()
try:
result = OcrService().recognize_files(
[
("invoice.png", b"fake-image", "image/png"),
("notes.txt", b"plain-text", "text/plain"),
]
)
finally:
get_settings.cache_clear()
assert result.engine == "paddleocr_mobile"
assert result.model == "PP-OCRv5_mobile"
assert result.total_file_count == 2
assert result.success_count == 1
assert len(result.documents) == 2
recognized = next(item for item in result.documents if item.filename == "invoice.png")
assert recognized.summary == "发票金额 100 元"
assert recognized.line_count == 1
assert recognized.lines[0].text == "发票金额 100 元"
skipped = next(item for item in result.documents if item.filename == "notes.txt")
assert skipped.line_count == 0
assert skipped.warnings == ["当前仅支持图片和 PDF 文件进行 OCR。"]