regression-dataset-for-docling-parse / tools /test_groundtruth_diff.py
PeterStaar's picture
adding new regression documents and much more precise word and line bbox
3826b7b
Raw History Blame Contribute Delete
3.94 kB
from __future__ import annotations
import subprocess
import sys
import tempfile
import unittest
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent))
import groundtruth_diff # noqa: E402
class GroundtruthDiffTest(unittest.TestCase):
def test_artifact_name(self) -> None:
self.assertEqual(
groundtruth_diff.artifact_name("document.pdf.page_no_2.py.json.word.txt"),
("document.pdf.page_no_2.py.json", "word"),
)
self.assertIsNone(groundtruth_diff.artifact_name("document.py.json.gz"))
def test_measure_modified_added_and_deleted(self) -> None:
modified = groundtruth_diff.measure(
"page", "line", groundtruth_diff.ContentPair("same\nold\n", "same\nnew\nextra\n")
)
self.assertEqual((modified.status, modified.added, modified.removed), ("modified", 2, 1))
self.assertGreater(modified.score, 0)
coordinate_change = groundtruth_diff.measure(
"page",
"char",
groundtruth_diff.ContentPair("(1, 2) /F1 text\n", "(1, 3) /F1 text\n"),
)
self.assertGreater(coordinate_change.score, 0)
self.assertLess(coordinate_change.score, 1)
self.assertEqual(
groundtruth_diff.measure("page", "line", groundtruth_diff.ContentPair("", "new\n")).status,
"added",
)
self.assertEqual(
groundtruth_diff.measure("page", "line", groundtruth_diff.ContentPair("old\n", "")).status,
"deleted",
)
def test_directory_comparison_groups_and_sorts_pages(self) -> None:
with tempfile.TemporaryDirectory() as directory:
root = Path(directory)
old, new = root / "old" / "parser", root / "new" / "parser"
old.mkdir(parents=True)
new.mkdir(parents=True)
(old / "a.pdf.page_no_1.py.json.word.txt").write_text("one\ntwo\n", encoding="utf-8")
(new / "a.pdf.page_no_1.py.json.word.txt").write_text("changed\n", encoding="utf-8")
(old / "b.pdf.page_no_1.py.json.line.txt").write_text("same\n", encoding="utf-8")
(new / "b.pdf.page_no_1.py.json.line.txt").write_text("same\n", encoding="utf-8")
(new / "c.pdf.page_no_1.py.json.char.txt").write_text("added\n", encoding="utf-8")
pages = groundtruth_diff.analyze(groundtruth_diff.Comparison(old, new))
self.assertEqual([page.name for page in pages], ["a.pdf.page_no_1.py.json", "c.pdf.page_no_1.py.json"])
self.assertEqual(pages[1].artifacts["char"].status, "added")
def test_git_comparison_uses_ref_and_finds_untracked_files(self) -> None:
with tempfile.TemporaryDirectory() as directory:
root = Path(directory)
parser = root / "groundtruth" / "parser"
parser.mkdir(parents=True)
subprocess.run(["git", "init", "-q"], cwd=root, check=True)
subprocess.run(["git", "config", "user.email", "test@example.com"], cwd=root, check=True)
subprocess.run(["git", "config", "user.name", "Test"], cwd=root, check=True)
tracked = parser / "a.pdf.page_no_1.py.json.word.txt"
tracked.write_text("before\n", encoding="utf-8")
subprocess.run(["git", "add", "groundtruth/parser"], cwd=root, check=True)
subprocess.run(
["git", "-c", "core.hooksPath=/dev/null", "commit", "-qm", "baseline"],
cwd=root,
check=True,
)
tracked.write_text("after\n", encoding="utf-8")
(parser / "b.pdf.page_no_1.py.json.line.txt").write_text("new\n", encoding="utf-8")
pages = groundtruth_diff.analyze(groundtruth_diff.Comparison(parser, root=root))
self.assertEqual({page.name for page in pages}, {"a.pdf.page_no_1.py.json", "b.pdf.page_no_1.py.json"})
if __name__ == "__main__":
unittest.main()