gitingest/tests/test_ingestion.py
Romain Courtois d6cb920660
Refactor/ingestion (#209)
Co-authored-by: Filip Christiansen <22807962+filipchristiansen@users.noreply.github.com>
2025-03-04 01:11:54 +01:00

46 lines
1.6 KiB
Python

"""
Tests for the `query_ingestion` module.
These tests validate directory scanning, file content extraction, notebook handling, and the overall ingestion logic,
including filtering patterns and subpaths.
"""
from pathlib import Path
from gitingest.ingestion import ingest_query
from gitingest.query_parsing import ParsedQuery
def test_run_ingest_query(temp_directory: Path, sample_query: ParsedQuery) -> None:
"""
Test `ingest_query` to ensure it processes the directory and returns expected results.
Given a directory with .txt and .py files:
When `ingest_query` is invoked,
Then it should produce a summary string listing the files analyzed and a combined content string.
"""
sample_query.local_path = temp_directory
sample_query.subpath = "/"
sample_query.type = None
summary, _, content = ingest_query(sample_query)
assert "Repository: test_user/test_repo" in summary
assert "Files analyzed: 8" in summary
# Check presence of key files in the content
assert "src/subfile1.txt" in content
assert "src/subfile2.py" in content
assert "src/subdir/file_subdir.txt" in content
assert "src/subdir/file_subdir.py" in content
assert "file1.txt" in content
assert "file2.py" in content
assert "dir1/file_dir1.txt" in content
assert "dir2/file_dir2.txt" in content
# TODO: Additional tests:
# - Multiple include patterns, e.g. ["*.txt", "*.py"] or ["/src/*", "*.txt"].
# - Edge cases with weird file names or deep subdirectory structures.
# TODO : def test_include_txt_pattern
# TODO : def test_include_nonexistent_extension