9f97f3abbe
CI - Node Bindings / Build darwin-arm64 (push) Waiting to run
CI - Node Bindings / Build darwin-x64 (push) Waiting to run
CI - Node Bindings / Test win32-x64-msvc (push) Blocked by required conditions
CI - Node Bindings / Build linux-arm64-gnu (push) Waiting to run
CI - Node Bindings / Build linux-x64-gnu (push) Waiting to run
CI - Node Bindings / Build linux-x64-musl (push) Waiting to run
CI - Node Bindings / Build win32-arm64-msvc (push) Waiting to run
CI - Node Bindings / Build win32-x64-msvc (push) Waiting to run
CI - Node Bindings / Test darwin-arm64 (push) Blocked by required conditions
CI - Node Bindings / Test darwin-x64 (push) Blocked by required conditions
CI - Node Bindings / Test linux-x64-gnu (push) Blocked by required conditions
CI - Node Bindings / Test linux-x64-musl (push) Blocked by required conditions
CI - Node Bindings / Test win32-arm64-msvc (push) Blocked by required conditions
CI - Python Bindings / Build aarch64-pc-windows-msvc (push) Waiting to run
CI - Python Bindings / Build x86_64-pc-windows-msvc (push) Waiting to run
CI - Python Bindings / Build x86_64-apple-darwin (push) Waiting to run
CI - Python Bindings / Build aarch64-apple-darwin (push) Waiting to run
CI - Python Bindings / Build aarch64-unknown-linux-gnu (push) Waiting to run
CI - Python Bindings / Test x86_64-apple-darwin (push) Blocked by required conditions
CI - Python Bindings / Test aarch64-apple-darwin (push) Blocked by required conditions
CI - Python Bindings / Test x86_64-unknown-linux-gnu (push) Blocked by required conditions
CI - Python Bindings / Test x86_64-unknown-linux-musl (push) Blocked by required conditions
CI - Python Bindings / Test aarch64-pc-windows-msvc (push) Blocked by required conditions
CI - Python Bindings / Test x86_64-pc-windows-msvc (push) Blocked by required conditions
CI / build-and-test (macos-26-intel) (push) Waiting to run
CI / build-and-test (macos-latest) (push) Waiting to run
CI / build-and-test (windows-11-arm) (push) Waiting to run
CI / build-and-test (windows-latest) (push) Waiting to run
CI - Python Bindings / sdist (push) Failing after 1s
CI - Python Bindings / Build x86_64-unknown-linux-musl (push) Failing after 1s
CI / fmt (push) Failing after 1s
E2E Output Validation / compare-outputs (push) Failing after 1s
Sync Docs to Developer Hub / sync-docs (push) Failing after 1s
CI - Python Bindings / Build x86_64-unknown-linux-gnu (push) Failing after 1s
CI - WASM Bindings / Build WASM (push) Failing after 0s
CI / clippy (push) Failing after 1s
CI / build-and-test (ubuntu-latest) (push) Failing after 0s
CI - WASM Bindings / Edge runtime PDF parse test (push) Has been skipped
CI - WASM Bindings / Browser PDF parse test (push) Has been skipped
E2E Output Validation / upload-dataset (push) Waiting to run
Deploy Demo to GitHub Pages / deploy (push) Failing after 1s
CI / build-docker-image (push) Failing after 3s
89 lines
2.9 KiB
Python
89 lines
2.9 KiB
Python
"""E2E tests for LiteParse.batch_parse() — validates Python types match CLI output."""
|
|
|
|
import tempfile
|
|
from pathlib import Path
|
|
|
|
import pytest
|
|
|
|
from liteparse import (
|
|
BatchResult,
|
|
LiteParse,
|
|
OutputFormat,
|
|
)
|
|
|
|
|
|
class TestBatchParseBasic:
|
|
"""Basic batch parsing functionality."""
|
|
|
|
def test_batch_parse_returns_batch_result(
|
|
self, parser: LiteParse, invoice_pdf: Path
|
|
):
|
|
input_dir = invoice_pdf.parent
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
result = parser.batch_parse(
|
|
input_dir,
|
|
tmpdir,
|
|
ocr_enabled=False,
|
|
extension_filter=".pdf",
|
|
)
|
|
assert isinstance(result, BatchResult)
|
|
assert result.output_dir == tmpdir
|
|
|
|
def test_batch_parse_creates_output_files(
|
|
self, parser: LiteParse, invoice_pdf: Path
|
|
):
|
|
input_dir = invoice_pdf.parent
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
parser.batch_parse(
|
|
input_dir,
|
|
tmpdir,
|
|
output_format=OutputFormat.TEXT,
|
|
ocr_enabled=False,
|
|
extension_filter=".pdf",
|
|
)
|
|
output_files = list(Path(tmpdir).rglob("*"))
|
|
# Should have produced at least one output file
|
|
assert len(output_files) > 0
|
|
|
|
def test_batch_parse_json_format(self, parser: LiteParse, invoice_pdf: Path):
|
|
input_dir = invoice_pdf.parent
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
parser.batch_parse(
|
|
input_dir,
|
|
tmpdir,
|
|
output_format="json",
|
|
ocr_enabled=False,
|
|
extension_filter=".pdf",
|
|
)
|
|
json_files = list(Path(tmpdir).rglob("*.json"))
|
|
assert len(json_files) > 0
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_batch_parse_async(self, parser: LiteParse, invoice_pdf: Path):
|
|
input_dir = invoice_pdf.parent
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
result = await parser.batch_parse_async(
|
|
input_dir,
|
|
tmpdir,
|
|
output_format="json",
|
|
ocr_enabled=False,
|
|
extension_filter=".pdf",
|
|
)
|
|
assert isinstance(result, BatchResult)
|
|
assert result.output_dir == tmpdir
|
|
|
|
|
|
class TestBatchParseErrors:
|
|
"""Test error handling."""
|
|
|
|
def test_input_dir_not_found(self, parser: LiteParse):
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
with pytest.raises(FileNotFoundError):
|
|
parser.batch_parse("/nonexistent/dir", tmpdir)
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_input_dir_not_found_async(self, parser: LiteParse):
|
|
with tempfile.TemporaryDirectory() as tmpdir:
|
|
with pytest.raises(FileNotFoundError):
|
|
await parser.batch_parse_async("/nonexistent/dir", tmpdir)
|