ThomasMoreAI/legal-skills-open / br/litigation/skills/juriscan-brunobracaioli
juriscan-brunobracaioli
VerifiedAnálise forense de processos judiciais brasileiros. Extrai, classifica e cruza peças processuais, detecta contradições, calcula prazos CPC e gera relatórios de risco com exportação para Obsidian. Use quando mencionar: processo judicial, petição, sentença, acórdão, contradições, prazos, timeline, análise forense, Obsidian.
tests/test_golden_regression.py
View on GitHub"""Golden regression suite for juriscan.
Runs both the legacy and (eventually) agents pipelines against synthetic
fixtures in tests/golden/ and asserts structural invariants.
Phase 0 scope: only legacy pipeline. Invariants are intentionally conservative
so they pass with the current regex chunker. Agents-only invariants (e.g.
art. 942 detection) live in separate `expected_art_942.json` files and are
gated by the pipeline mode.
"""
from __future__ import annotations
import json
import shutil
import subprocess
import sys
from pathlib import Path
import pytest
REPO_ROOT = Path(__file__).resolve().parents[1]
GOLDEN_ROOT = REPO_ROOT / "tests" / "golden"
EXTRACT_SCRIPT = REPO_ROOT / "scripts" / "extract_and_chunk.py"
def _discover_fixtures() -> list[Path]:
fixtures = []
for child in sorted(GOLDEN_ROOT.iterdir()):
if not child.is_dir():
continue
if child.name.startswith("_"):
continue
if (child / "input.pdf").exists() and (child / "expected_invariants.json").exists():
fixtures.append(child)
return fixtures
FIXTURES = _discover_fixtures()
@pytest.fixture(scope="module")
def legacy_outputs(tmp_path_factory) -> dict[str, Path]:
"""Run the legacy extract_and_chunk pipeline against every fixture.
Returns a map { fixture_name: output_dir }.
"""
out: dict[str, Path] = {}
root = tmp_path_factory.mktemp("golden_legacy")
for fixture in FIXTURES:
out_dir = root / fixture.name
out_dir.mkdir(parents=True, exist_ok=True)
result = subprocess.run(
[
sys.executable,
str(EXTRACT_SCRIPT),
"--input",
str(fixture / "input.pdf"),
"--output",
str(out_dir),
],
capture_output=True,
text=True,
)
assert result.returncode == 0, (
f"extract_and_chunk failed for {fixture.name}:\n"
f"stdout={result.stdout}\nstderr={result.stderr}"
)
out[fixture.name] = out_dir
return out
def _load_index(out_dir: Path) -> dict:
index_path = out_dir / "index.json"
assert index_path.exists(), f"missing index.json in {out_dir}"
return json.loads(index_path.read_text(encoding="utf-8"))
def _load_all_chunks_text(out_dir: Path) -> str:
chunks_dir = out_dir / "chunks"
assert chunks_dir.exists(), f"missing chunks/ in {out_dir}"
parts = []
for chunk_file in sorted(chunks_dir.iterdir()):
if chunk_file.suffix == ".txt":
parts.append(chunk_file.read_text(encoding="utf-8"))
return "\n".join(parts)
@pytest.mark.parametrize("fixture", FIXTURES, ids=lambda f: f.name)
def test_golden_legacy_pipeline(fixture: Path, legacy_outputs: dict[str, Path]) -> None:
"""Run legacy pipeline and validate against expected_invariants.json."""
invariants = json.loads((fixture / "expected_invariants.json").read_text(encoding="utf-8"))
if "legacy" not in invariants.get("pipeline_compat", ["legacy"]):
pytest.skip(f"{fixture.name} not marked legacy-compatible")
out_dir = legacy_outputs[fixture.name]
index = _load_index(out_dir)
all_text = _load_all_chunks_text(out_dir)
# Extraction lower bound
min_chars = invariants.get("extraction", {}).get("min_chars")
if min_chars is not None:
assert len(all_text) >= min_chars, (
f"extracted text too short: {len(all_text)} < {min_chars}"
)
# Chunk count range
chunk_min = invariants.get("chunks", {}).get("min_count")
chunk_max = invariants.get("chunks", {}).get("max_count")
chunks = index.get("chunks") or index.get("pieces") or []
chunk_count = len(chunks)
if chunk_min is not None:
assert chunk_count >= chunk_min, (
f"too few chunks: {chunk_count} < {chunk_min}"
)
if chunk_max is not None:
assert chunk_count <= chunk_max, (
f"too many chunks: {chunk_count} > {chunk_max}"
)
# Piece types present
required_types = invariants.get("piece_types_present", [])
observed_types = {
(c.get("tipo_peca") or c.get("label") or "").upper()
for c in chunks
}
for required in required_types:
# Normalize accents for comparison — legacy labels may vary
required_upper = required.upper()
matched = any(required_upper in obs for obs in observed_types)
assert matched, (
f"required piece type not found: {required} "
f"(observed: {sorted(observed_types)})"
)
# Text must contain
for needle in invariants.get("text_must_contain", []):
assert needle in all_text, f"text missing required substring: {needle!r}"
# Text must NOT contain
for needle in invariants.get("text_must_not_contain", []):
assert needle not in all_text, (
f"text contains forbidden substring: {needle!r}"
)
@pytest.mark.parametrize("fixture", FIXTURES, ids=lambda f: f.name)
def test_golden_art942_gate_is_skipped_in_legacy(fixture: Path) -> None:
"""Ensure agents-only invariants (art_942) are properly gated.
In Phase 0 we only run the legacy pipeline. Any fixture that has an
expected_art_942.json file MUST declare skip_when_pipeline=legacy,
otherwise the gate is misconfigured.
"""
art942_path = fixture / "expected_art_942.json"
if not art942_path.exists():
pytest.skip("no art942 invariant for this fixture")
invariant = json.loads(art942_path.read_text(encoding="utf-8"))
assert invariant.get("skip_when_pipeline") == "legacy", (
f"{fixture.name}: expected_art_942.json must declare "
f"skip_when_pipeline=legacy during Phase 0 "
f"(got: {invariant.get('skip_when_pipeline')})"
)
assert invariant.get("required_when_pipeline") == "agents", (
f"{fixture.name}: expected_art_942.json must declare "
f"required_when_pipeline=agents"
)
def test_golden_fixtures_exist() -> None:
"""Sanity check: make sure the golden suite is not empty."""
assert len(FIXTURES) >= 2, (
f"expected at least 2 golden fixtures, found {len(FIXTURES)}: "
f"{[f.name for f in FIXTURES]}"
)
names = {f.name for f in FIXTURES}
assert "processo_01_sintetico_simples" in names
assert "processo_02_sintetico_art942" in names
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| slack-gif-creatoranthropics/skillsKnowledge and utilities for creating animated GIFs optimized for Slack. Provides constraints, validation tools, and animation concepts. Use when users request animated GIFs for Slack like "make me a GIF of X doing Y for Slack." | Stars: 198k | Updated | Claude CodeCodex | Verified | Install |
| ponytailDietrichGebert/ponytailLazy senior dev mode for any coding task: the smallest change that fully solves it, and a reply a busy human understands in one read. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailHarvest every shortcut: comment into one debt ledger, so deferrals get tracked instead of forgotten. One-shot report. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-debtDietrichGebert/ponytailList every `shortcut:` comment (and older `ponytail:` ones) as a debt ledger. One-shot report, changes nothing. Use for "ponytail debt", "what did ponytail defer", "list the shortcuts", /ponytail-debt. | Stars: 196k | Updated | Claude CodeCodex | Verified | Install |