YAO
31ce04c655
Merge the beta-ready Yao Meta Skill architecture, report, evidence gate, and release-boundary updates.\n\nRelease boundary: beta/public testing is allowed; formal world-class, fully reviewed, or superiority claims remain blocked until the pending evidence gates are accepted.
322 lines
12 KiB
Python
322 lines
12 KiB
Python
#!/usr/bin/env python3
|
|
import argparse
|
|
import hashlib
|
|
import json
|
|
import re
|
|
from datetime import datetime, timezone
|
|
from pathlib import Path
|
|
from typing import Any
|
|
|
|
|
|
ROOT = Path(__file__).resolve().parent.parent
|
|
SCRIPT_INTERFACE = "cli"
|
|
SCRIPT_INTERFACE_REASON = "Scans an explicit local source file and summarizes redacted repeated user preference signals."
|
|
|
|
TEXT_FIELDS = ("text", "message", "content", "excerpt", "prompt", "note", "body")
|
|
HISTORY_FILENAMES = {".zsh_history", ".bash_history", ".fish_history", "History"}
|
|
SECRET_PATTERNS = [
|
|
re.compile(r"sk-[A-Za-z0-9_-]{12,}"),
|
|
re.compile(r"AKIA[0-9A-Z]{12,}"),
|
|
re.compile(r"(?i)\b(api[_-]?key|token|password|secret)\b\s*[:=]\s*[^\s,;]+"),
|
|
]
|
|
EMAIL_RE = re.compile(r"\b[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}\b")
|
|
LOCAL_PATH_RE = re.compile(r"/Users/[^\s'\"<>]+")
|
|
|
|
PATTERN_RULES = [
|
|
{
|
|
"pattern_id": "language_default",
|
|
"label": "Default language preference",
|
|
"signal_type": "report-language",
|
|
"keywords": ["中文", "简体", "默认中文", "英文", "双语", "language", "bilingual", "chinese", "english"],
|
|
"recommended_action": "Keep generated reports Chinese-first with an English switch where user-facing.",
|
|
},
|
|
{
|
|
"pattern_id": "report_ui",
|
|
"label": "Report UI and visualization preference",
|
|
"signal_type": "artifact-design",
|
|
"keywords": ["报告", "html", "图表", "排版", "ui", "kami", "白底", "模块", "导航", "report", "chart", "layout"],
|
|
"recommended_action": "Prioritize white-background Kami-style reports with readable charts and stable navigation.",
|
|
},
|
|
{
|
|
"pattern_id": "approval_safety",
|
|
"label": "Approval and privacy boundary",
|
|
"signal_type": "governance",
|
|
"keywords": ["审批", "授权", "不要扫描", "隐私", "私人", "日志", "明确路径", "回滚", "提案", "批准", "approval", "privacy", "private", "proposal", "rollback"],
|
|
"recommended_action": "Keep adaptive work proposal-only until a reviewer approves an allowlisted patch path.",
|
|
},
|
|
{
|
|
"pattern_id": "delivery_format",
|
|
"label": "Delivery format preference",
|
|
"signal_type": "artifact-format",
|
|
"keywords": ["markdown", "pdf", "word", "docx", "html", "地址", "路径", "打开", "输出", "交付"],
|
|
"recommended_action": "Surface stable artifact paths and formats in CLI output and generated summaries.",
|
|
},
|
|
{
|
|
"pattern_id": "evidence_testing",
|
|
"label": "Evidence and testing preference",
|
|
"signal_type": "quality-gate",
|
|
"keywords": ["测试", "验证", "ci", "证据", "覆盖", "github", "push", "evidence", "review"],
|
|
"recommended_action": "Attach focused tests and refreshed evidence reports to every non-trivial skill upgrade.",
|
|
},
|
|
]
|
|
|
|
|
|
def utc_now() -> str:
|
|
return datetime.now(timezone.utc).replace(microsecond=0).isoformat().replace("+00:00", "Z")
|
|
|
|
|
|
def display_path(path: Path, skill_dir: Path) -> str:
|
|
try:
|
|
return str(path.resolve().relative_to(skill_dir.resolve()))
|
|
except ValueError:
|
|
return f"[external-explicit-source]/{path.name}"
|
|
|
|
|
|
def resolve_output(skill_dir: Path, value: str) -> Path:
|
|
path = Path(value)
|
|
return path if path.is_absolute() else skill_dir / path
|
|
|
|
|
|
def source_fingerprint(path: Path) -> str:
|
|
digest = hashlib.sha256()
|
|
with path.open("rb") as handle:
|
|
for chunk in iter(lambda: handle.read(1024 * 1024), b""):
|
|
digest.update(chunk)
|
|
return digest.hexdigest()
|
|
|
|
|
|
def redact_text(text: str) -> str:
|
|
redacted = text
|
|
for pattern in SECRET_PATTERNS:
|
|
redacted = pattern.sub("[REDACTED_SECRET]", redacted)
|
|
redacted = EMAIL_RE.sub("[REDACTED_EMAIL]", redacted)
|
|
redacted = LOCAL_PATH_RE.sub("[LOCAL_PATH]", redacted)
|
|
redacted = re.sub(r"\s+", " ", redacted).strip()
|
|
if len(redacted) > 240:
|
|
return redacted[:237].rstrip() + "..."
|
|
return redacted
|
|
|
|
|
|
def extract_text(raw: Any) -> str:
|
|
if isinstance(raw, str):
|
|
return raw
|
|
if not isinstance(raw, dict):
|
|
return ""
|
|
for field in TEXT_FIELDS:
|
|
value = raw.get(field)
|
|
if isinstance(value, str) and value.strip():
|
|
return value
|
|
messages = raw.get("messages")
|
|
if isinstance(messages, list):
|
|
parts = []
|
|
for item in messages:
|
|
if isinstance(item, dict):
|
|
content = item.get("content")
|
|
if isinstance(content, str):
|
|
parts.append(content)
|
|
elif isinstance(item, str):
|
|
parts.append(item)
|
|
return "\n".join(parts)
|
|
return ""
|
|
|
|
|
|
def load_records(source: Path) -> tuple[list[dict[str, str]], list[str]]:
|
|
records: list[dict[str, str]] = []
|
|
failures: list[str] = []
|
|
text = source.read_text(encoding="utf-8", errors="replace")
|
|
if source.suffix.lower() == ".jsonl":
|
|
for index, line in enumerate(text.splitlines(), start=1):
|
|
if not line.strip():
|
|
continue
|
|
try:
|
|
raw = json.loads(line)
|
|
except json.JSONDecodeError as exc:
|
|
failures.append(f"line {index}: invalid JSONL source: {exc.msg}")
|
|
continue
|
|
extracted = extract_text(raw)
|
|
if not extracted.strip():
|
|
failures.append(f"line {index}: no supported text field found")
|
|
continue
|
|
records.append({"record_id": f"line-{index}", "excerpt": redact_text(extracted)})
|
|
else:
|
|
for index, line in enumerate(text.splitlines(), start=1):
|
|
if line.strip():
|
|
records.append({"record_id": f"line-{index}", "excerpt": redact_text(line)})
|
|
return records, failures
|
|
|
|
|
|
def classify_patterns(records: list[dict[str, str]], min_support: int) -> tuple[list[dict[str, Any]], list[dict[str, Any]]]:
|
|
patterns: list[dict[str, Any]] = []
|
|
discarded: list[dict[str, Any]] = []
|
|
for rule in PATTERN_RULES:
|
|
matches = []
|
|
for record in records:
|
|
lowered = record["excerpt"].lower()
|
|
if any(keyword.lower() in lowered for keyword in rule["keywords"]):
|
|
matches.append(record)
|
|
if not matches:
|
|
continue
|
|
item = {
|
|
"pattern_id": rule["pattern_id"],
|
|
"label": rule["label"],
|
|
"signal_type": rule["signal_type"],
|
|
"support_count": len(matches),
|
|
"confidence": min(0.95, round(0.55 + (0.12 * len(matches)), 2)),
|
|
"reason": f"{len(matches)} redacted records matched repeated {rule['signal_type']} signals.",
|
|
"recommended_action": rule["recommended_action"],
|
|
"evidence": matches[:3],
|
|
}
|
|
if len(matches) >= min_support:
|
|
patterns.append(item)
|
|
else:
|
|
discarded.append({**item, "discard_reason": f"support_count below min_support {min_support}"})
|
|
return patterns, discarded
|
|
|
|
|
|
def build_report(
|
|
skill_dir: Path,
|
|
source: Path,
|
|
min_support: int,
|
|
generated_at: str,
|
|
allow_history_source: bool,
|
|
) -> dict[str, Any]:
|
|
skill_dir = skill_dir.resolve()
|
|
source = source.resolve()
|
|
failures: list[str] = []
|
|
records: list[dict[str, str]] = []
|
|
fingerprint = ""
|
|
if not source.exists():
|
|
failures.append(f"Explicit source does not exist: {display_path(source, skill_dir)}")
|
|
elif not source.is_file():
|
|
failures.append(f"Explicit source must be a file: {display_path(source, skill_dir)}")
|
|
elif source.name in HISTORY_FILENAMES and not allow_history_source:
|
|
failures.append(f"Refusing private history source by default: {source.name}")
|
|
else:
|
|
fingerprint = source_fingerprint(source)
|
|
records, load_failures = load_records(source)
|
|
failures.extend(load_failures)
|
|
patterns, discarded = classify_patterns(records, min_support) if not failures else ([], [])
|
|
return {
|
|
"ok": not failures,
|
|
"schema_version": "1.0",
|
|
"generated_at": generated_at,
|
|
"skill_dir": display_path(skill_dir, skill_dir),
|
|
"source": {
|
|
"label": source.name,
|
|
"path": display_path(source, skill_dir),
|
|
"fingerprint_sha256": fingerprint,
|
|
"explicit_source": True,
|
|
"record_count": len(records),
|
|
},
|
|
"privacy_contract": {
|
|
"local_only": True,
|
|
"explicit_source_required": True,
|
|
"implicit_private_log_scan": False,
|
|
"raw_content_stored": False,
|
|
"redacted_excerpts_only": True,
|
|
"redacted_excerpt_limit": 240,
|
|
"writes_repository_files": False,
|
|
},
|
|
"summary": {
|
|
"record_count": len(records),
|
|
"pattern_count": len(patterns),
|
|
"discarded_signal_count": len(discarded),
|
|
"min_support": min_support,
|
|
"failure_count": len(failures),
|
|
},
|
|
"patterns": patterns,
|
|
"discarded_signals": discarded,
|
|
"failures": failures,
|
|
"artifacts": {
|
|
"json": "reports/user_patterns.json",
|
|
"markdown": "reports/user_patterns.md",
|
|
},
|
|
}
|
|
|
|
|
|
def render_markdown(report: dict[str, Any]) -> str:
|
|
lines = [
|
|
"# User Pattern Summary",
|
|
"",
|
|
f"- Generated at: `{report['generated_at']}`",
|
|
f"- Local only: `{str(report['privacy_contract']['local_only']).lower()}`",
|
|
f"- Explicit source: `{report['source']['path']}`",
|
|
f"- Records: `{report['summary']['record_count']}`",
|
|
f"- Patterns: `{report['summary']['pattern_count']}`",
|
|
f"- Discarded signals: `{report['summary']['discarded_signal_count']}`",
|
|
"",
|
|
"## Privacy Contract",
|
|
"",
|
|
"- No implicit private log scan.",
|
|
"- No unredacted raw content stored.",
|
|
"- Scan and proposal stages do not write source files.",
|
|
"",
|
|
"## Patterns",
|
|
"",
|
|
]
|
|
if not report["patterns"]:
|
|
lines.append("- No repeated pattern met the support threshold.")
|
|
for pattern in report["patterns"]:
|
|
lines.extend(
|
|
[
|
|
f"### {pattern['label']}",
|
|
"",
|
|
f"- Pattern: `{pattern['pattern_id']}`",
|
|
f"- Support: `{pattern['support_count']}`",
|
|
f"- Confidence: `{pattern['confidence']}`",
|
|
f"- Reason: {pattern['reason']}",
|
|
f"- Recommended action: {pattern['recommended_action']}",
|
|
"- Redacted evidence:",
|
|
]
|
|
)
|
|
for item in pattern["evidence"]:
|
|
lines.append(f" - `{item['record_id']}`: {item['excerpt']}")
|
|
lines.append("")
|
|
if report["discarded_signals"]:
|
|
lines.extend(["## Discarded Signals", ""])
|
|
for item in report["discarded_signals"]:
|
|
lines.append(f"- `{item['pattern_id']}`: {item['discard_reason']}")
|
|
if report["failures"]:
|
|
lines.extend(["", "## Failures", ""])
|
|
lines.extend(f"- {failure}" for failure in report["failures"])
|
|
return "\n".join(lines).rstrip() + "\n"
|
|
|
|
|
|
def main() -> None:
|
|
parser = argparse.ArgumentParser(description="Summarize repeated user preference signals from one explicit local source file.")
|
|
parser.add_argument("skill_dir", nargs="?", default=".")
|
|
parser.add_argument("--source", required=True)
|
|
parser.add_argument("--output-json", default="reports/user_patterns.json")
|
|
parser.add_argument("--output-md", default="reports/user_patterns.md")
|
|
parser.add_argument("--min-support", type=int, default=2)
|
|
parser.add_argument("--generated-at", default=utc_now())
|
|
parser.add_argument("--allow-history-source", action="store_true")
|
|
args = parser.parse_args()
|
|
|
|
skill_dir = Path(args.skill_dir).resolve()
|
|
report = build_report(
|
|
skill_dir,
|
|
Path(args.source),
|
|
min_support=max(2, args.min_support),
|
|
generated_at=args.generated_at,
|
|
allow_history_source=args.allow_history_source,
|
|
)
|
|
if report["ok"]:
|
|
output_json = resolve_output(skill_dir, args.output_json)
|
|
output_md = resolve_output(skill_dir, args.output_md)
|
|
output_json.parent.mkdir(parents=True, exist_ok=True)
|
|
output_md.parent.mkdir(parents=True, exist_ok=True)
|
|
report["artifacts"] = {
|
|
"json": display_path(output_json, skill_dir),
|
|
"markdown": display_path(output_md, skill_dir),
|
|
}
|
|
output_json.write_text(json.dumps(report, ensure_ascii=False, indent=2) + "\n", encoding="utf-8")
|
|
output_md.write_text(render_markdown(report), encoding="utf-8")
|
|
print(json.dumps(report, ensure_ascii=False, indent=2))
|
|
if not report["ok"]:
|
|
raise SystemExit(2)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|