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.
scripts/persist_chunks.py
View on GitHub"""Persist segmenter output to disk as physical chunk files + index.json.
Phase 2 Step 2.2. Consumes a validated segmenter output JSON + the raw text
used as its input, slices the text by (start_char, end_char), writes one
`chunks/NN-<slug>.txt` file per chunk, and emits an index.json that matches
the legacy extract_and_chunk.py shape (index, label, char_count, chunk_file,
primary_date placeholder) so downstream legacy steps keep working during the
transition.
Invariants enforced at write time (raises PersistError):
- segmenter output schema valid (delegated to agent_io.validate)
- raw_text length matches raw_text_length declared by segmenter
- chunks cover the entire raw_text without gaps or overlap
- persisted chunks pass integrity_gate.assert_chunks_consistent
"""
from __future__ import annotations
import argparse
import json
import re
import sys
import unicodedata
from pathlib import Path
_THIS_DIR = Path(__file__).resolve().parent
if str(_THIS_DIR) not in sys.path:
sys.path.insert(0, str(_THIS_DIR))
from agent_io import validate_agent_output # noqa: E402
from integrity_gate import assert_chunks_consistent # noqa: E402
class PersistError(Exception):
"""Raised when the segmenter output cannot be persisted safely."""
def _slug(label: str) -> str:
norm = unicodedata.normalize("NFKD", label)
norm = norm.encode("ascii", "ignore").decode("ascii")
norm = re.sub(r"[^a-zA-Z0-9]+", "-", norm).strip("-").lower()
return norm or "peca"
def _verify_coverage(chunks: list[dict], raw_len: int) -> None:
if not chunks:
raise PersistError("segmenter returned zero chunks")
if chunks[0]["start_char"] != 0:
raise PersistError(
f"first chunk does not start at 0: start_char={chunks[0]['start_char']}"
)
if chunks[-1]["end_char"] != raw_len:
raise PersistError(
f"last chunk does not end at raw_len: "
f"end_char={chunks[-1]['end_char']}, raw_len={raw_len}"
)
for i, ch in enumerate(chunks):
if ch["end_char"] <= ch["start_char"]:
raise PersistError(
f"chunk {ch['id']} has non-positive length: "
f"{ch['start_char']}..{ch['end_char']}"
)
if i > 0 and ch["start_char"] != chunks[i - 1]["end_char"]:
raise PersistError(
f"gap/overlap between chunk {chunks[i-1]['id']} "
f"(end={chunks[i-1]['end_char']}) and {ch['id']} "
f"(start={ch['start_char']})"
)
def persist(
segmenter_output_path: Path,
raw_text_path: Path,
output_dir: Path,
*,
skip_validate: bool = False,
) -> Path:
"""Materialize chunks on disk. Returns path to the written index.json."""
# 1. Validate the segmenter JSON against its schema.
if not skip_validate:
ok, errors = validate_agent_output("segmenter", segmenter_output_path)
if not ok:
raise PersistError(
"segmenter output failed schema validation:\n - "
+ "\n - ".join(errors)
)
seg = json.loads(segmenter_output_path.read_text(encoding="utf-8"))
raw_text = raw_text_path.read_text(encoding="utf-8")
raw_len = len(raw_text)
declared_len = seg["raw_text_length"]
if declared_len != raw_len:
raise PersistError(
f"raw_text_length mismatch: segmenter declared {declared_len}, "
f"file has {raw_len} chars (stale input?)"
)
chunks = seg["chunks"]
_verify_coverage(chunks, raw_len)
chunks_dir = output_dir / "chunks"
chunks_dir.mkdir(parents=True, exist_ok=True)
# Remove any stale *.txt in chunks/ to avoid orphans from previous runs.
for stale in chunks_dir.glob("*.txt"):
stale.unlink()
index_chunks: list[dict] = []
for i, ch in enumerate(chunks):
start, end = ch["start_char"], ch["end_char"]
text_slice = raw_text[start:end]
slug = _slug(ch["tipo_provavel"])
filename = f"{i:02d}-{slug}.txt"
file_path = chunks_dir / filename
file_path.write_text(text_slice, encoding="utf-8")
index_chunks.append({
"index": i,
"id": ch["id"],
"label": ch["tipo_provavel"],
"char_count": len(text_slice),
"chunk_file": f"chunks/{filename}",
"start_char": start,
"end_char": end,
"confianca": ch["confianca"],
"evidencia": ch.get("evidencia"),
"page_range": ch.get("page_range"),
})
index = {
"generated_at": None,
"source": "juriscan-segmenter",
"schema_version": seg["schema_version"],
"segmenter_notes": seg.get("notes"),
"raw_text_length": raw_len,
"total_chunks": len(index_chunks),
"chunks": index_chunks,
}
index_path = output_dir / "index.json"
index_path.write_text(
json.dumps(index, ensure_ascii=False, indent=2),
encoding="utf-8",
)
# 4. Final check: integrity gate (defense in depth).
assert_chunks_consistent(index_path)
return index_path
def main(argv: list[str] | None = None) -> int:
parser = argparse.ArgumentParser(
description="Persist juriscan-segmenter output to disk as chunks/*.txt + index.json"
)
parser.add_argument("--segmenter-output", required=True,
help="Path to the JSON produced by juriscan-segmenter")
parser.add_argument("--raw-text", required=True,
help="Path to the raw_text.txt the segmenter saw")
parser.add_argument("--output-dir", required=True,
help="Directory where chunks/ and index.json will be written")
parser.add_argument("--skip-validate", action="store_true",
help="Skip jsonschema validation of the segmenter output")
args = parser.parse_args(argv)
try:
index_path = persist(
Path(args.segmenter_output),
Path(args.raw_text),
Path(args.output_dir),
skip_validate=args.skip_validate,
)
except PersistError as e:
print(f"[FAIL] persist_chunks: {e}", file=sys.stderr)
return 1
print(f"[OK] wrote {index_path}")
return 0
if __name__ == "__main__":
raise SystemExit(main())
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 |