document-haness/scripts/fold-studio-contract-into-index.py

#!/usr/bin/env python3
"""분해 작업 재료를 `tech-log-tree.json` 으로 옮기고 폴더에서 지운다.

`analysis/` 를 `final/document.md` 로 옮긴 것과 같은 일을 `tech-log-studio/` 에서 한다.
끝난 프로젝트의 `tech-log-studio/` 에는 `tech-log-tree.json` 과 기록 폴더만 있다.

옮기는 것

    root-tree.md                     사람이 읽는 트리 + Node Specifications → topics
    candidate-ledger.json            후보와 처분                          → candidates
    root-tree-source-manifest.json   원본 해시                            → ssotSha256
    _meta/state.json                 생성·편집·검증 이력                   → history
    _meta/**                         편집 과정 기록                        → 지운다 (git 에 남는다)

    python3 scripts/fold-studio-contract-into-index.py <프로젝트> [--dry-run] [--keep]

**옮기는 것이지 요약하는 것이 아니다.** 노드의 칸은 하나도 버리지 않는다.
"""
from __future__ import annotations

import argparse
import datetime
import hashlib
import json
import os
import re
import shutil
import sys

sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import techlog  # noqa: E402
from techlog import KINDS  # noqa: E402

ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))

KIND_LABELS = {"CASE": "case", "CONCEPT": "concept", "REFERENCE": "reference",
               "OPEN QUESTION": "question", "QUESTION": "question", "DECISION": "decision"}
FIELD_NAMES = ("slug", "readiness", "disposition", "source", "code", "evidence",
               "classification", "missing-verification", "relations", "scope", "exceptions",
               "known", "unknown", "next-verification", "decision-criterion",
               "decision-status", "decision-evidence", "grounds", "basis-version")
MULTI = {"source", "code", "evidence", "relations", "grounds", "decision-evidence",
         "known", "unknown", "scope", "exceptions"}

_TOPIC_HEAD = re.compile(r"^##\s+TOPIC(?:\s+\d+)?\s+[—-]\s+(.+?)\s*$")
_SPEC_HEAD = re.compile(r"^###\s+(CASE|CONCEPT|REFERENCE|OPEN QUESTION|QUESTION|DECISION)\s+[—-]\s+(.+?)\s*$")
_BRANCH = re.compile(r"^[├└]──\s+(CASE|CONCEPT|REFERENCE|OPEN QUESTION|QUESTION|DECISION)\s*$")
_ITEM = re.compile(r"^(?:│|\s)\s{2,}[├└]──\s+(.+?)\s*$")
_FIELD = re.compile(r"^-\s+([a-z][a-z-]*):\s*(.*)$")
_CONT = re.compile(r"^\s{2,}-\s+(.+?)\s*$")
_INLINE = re.compile(r"\s+·\s+(" + "|".join(FIELD_NAMES) + r"):\s*")
_READER_Q = re.compile(r"^독자\s*질문\s*[—:-]\s*(.+?)\s*$")
EMPTY = ("(추가 없음)", "(없음)", "(none)", "-")


def _set(fields: dict, key: str, value: str) -> None:
    value = value.strip()
    if not value:
        fields.setdefault(key, [])
        return
    items = [v.strip() for v in value.split(" · ")] if key in MULTI else [value]
    fields[key] = [v for v in items if v]


def parse_root_tree(path: str) -> dict:
    """사람이 읽는 트리와 Node Specifications 를 하나로 읽는다."""
    text = open(path, encoding="utf-8").read()
    header, body = {}, text
    if text.startswith("---"):
        end = text.find("\n---", 3)
        for line in text[3:end].splitlines():
            m = re.match(r"^([A-Za-z][A-Za-z0-9_]*):\s*(.*)$", line)
            if m:
                header[m.group(1)] = m.group(2).strip().strip('"')
        body = text[end + 4:]

    topics, specs, prose = [], [], []
    in_specs = False
    topic = branch = spec = None
    spec_topic = ""
    pending = None
    expect = 0

    for raw in body.splitlines():
        line = raw.rstrip()
        stripped = line.strip()
        if line.startswith("# Node Specifications"):
            in_specs = True
            topic = branch = None
            continue
        if not in_specs:
            if stripped.startswith(">"):
                prose.append(stripped.lstrip("> ").rstrip())
                continue
            if stripped == "---":
                continue
            if stripped == "PROJECT":
                expect = -1
                continue
            if expect == -1:
                if stripped:
                    expect = 0
                continue
            if stripped == "TOPIC":
                topic = {"topic": "", "title": "", "readerQuestion": "", "nodes": []}
                topics.append(topic)
                branch, expect = None, 1
                continue
            if topic is not None and expect in (1, 2, 3):
                if expect == 1 and stripped:
                    topic["title"] = stripped
                    expect = 2
                    continue
                if expect == 2 and stripped:
                    topic["topic"] = stripped
                    expect = 3
                    continue
                if expect == 3:
                    m = _READER_Q.match(stripped)
                    if m:
                        topic["readerQuestion"] = m.group(1)
                        expect = 0
                        continue
                    if stripped:
                        expect = 0
            m = _BRANCH.match(stripped)
            if m:
                branch = KIND_LABELS[m.group(1)]
                continue
            m = _ITEM.match(line)
            if m and topic is not None and branch:
                title = m.group(1).strip()
                if title not in EMPTY:
                    topic["nodes"].append({"kind": branch, "title": title})
            continue

        m = _TOPIC_HEAD.match(line)
        if m:
            spec_topic, spec, pending = m.group(1).strip(), None, None
            continue
        m = _SPEC_HEAD.match(line)
        if m:
            spec = {"topic": spec_topic, "kind": KIND_LABELS[m.group(1)],
                    "title": m.group(2).strip(), "fields": {}}
            specs.append(spec)
            pending = None
            continue
        if spec is None:
            continue
        m = _FIELD.match(line)
        if m:
            key, value = m.group(1), m.group(2).strip()
            parts = _INLINE.split(value)
            _set(spec["fields"], key, parts[0])
            rest = parts[1:]
            while rest:
                _set(spec["fields"], rest[0], rest[1])
                rest = rest[2:]
            pending = key if not rest else None
            continue
        m = _CONT.match(line)
        if m and pending:
            spec["fields"].setdefault(pending, []).append(m.group(1).strip())
            continue
        if not stripped:
            pending = None

    return {"header": header, "topics": topics, "specs": specs,
            "prose": [p for p in prose if p]}


def node_from_spec(spec: dict) -> dict:
    node = {"title": spec["title"], "kind": spec["kind"]}
    for key in ("slug", "readiness"):
        values = spec["fields"].get(key) or []
        node[key] = values[0].strip("`").strip() if values else ""
    node["readiness"] = node["readiness"].upper()
    for key, values in spec["fields"].items():
        if key in ("slug", "readiness") or not values:
            continue
        node[key] = values if key in MULTI or len(values) > 1 else values[0]
    return node


def main() -> int:
    ap = argparse.ArgumentParser(description="분해 작업 재료를 tech-log-tree.json 으로 옮긴다.")
    ap.add_argument("project")
    ap.add_argument("--dry-run", action="store_true")
    ap.add_argument("--keep", action="store_true", help="옮기기만 하고 지우지 않는다")
    args = ap.parse_args()

    base = os.path.join(ROOT, "docs", args.project)
    studio = os.path.join(base, "tech-log-studio")
    tree_path = os.path.join(studio, "root-tree.md")
    index_path = os.path.join(studio, "tech-log-tree.json")
    if not os.path.exists(tree_path):
        print(f"{args.project}: root-tree.md 가 없다 — 이미 옮겼거나 분해 계약을 쓴 적이 없다")
        return 0

    parsed = parse_root_tree(tree_path)
    header = parsed["header"]
    by_key = {(s["topic"], s["kind"], s["title"]): s for s in parsed["specs"]}

    topics = {}
    used = set()
    for t in parsed["topics"]:
        entry = {"topic": t["topic"], "title": t["title"],
                 "readerQuestion": t["readerQuestion"], "kinds": {k: [] for k in KINDS}}
        for n in t["nodes"]:
            key = (t["topic"], n["kind"], n["title"])
            spec = by_key.get(key)
            entry["kinds"][n["kind"]].append(
                node_from_spec(spec) if spec else {"title": n["title"], "kind": n["kind"],
                                                   "slug": "", "readiness": ""})
            used.add(key)
        topics[t["topic"]] = entry
    # 사람이 읽는 트리에 줄이 없던 Node Specification 도 잃지 않는다
    orphans = 0
    for key, spec in by_key.items():
        if key in used:
            continue
        orphans += 1
        entry = topics.setdefault(spec["topic"], {
            "topic": spec["topic"], "title": "", "readerQuestion": "",
            "kinds": {k: [] for k in KINDS}})
        node = node_from_spec(spec)
        node["listedInTree"] = False
        entry["kinds"][spec["kind"]].append(node)

    ledger_path = os.path.join(studio, "candidate-ledger.json")
    ledger = json.load(open(ledger_path, encoding="utf-8")) if os.path.exists(ledger_path) else {}
    meta_state_path = os.path.join(studio, "_meta", "state.json")
    meta_state = json.load(open(meta_state_path, encoding="utf-8")) \
        if os.path.exists(meta_state_path) else {}

    index = json.load(open(index_path, encoding="utf-8")) if os.path.exists(index_path) else {}
    ssot = header.get("sourceDocument", "final/document.md")
    out = {
        "schemaVersion": 4,
        "project": args.project,
        "ssot": ssot,
        "ssotSha256": techlog.sha256_of(os.path.join(base, ssot)),
        "sourceRevision": header.get("sourceRevision", ""),
        "generatedAt": datetime.date.today().isoformat(),
        "note": ("이 프로젝트의 글감 전부다. 분해 계약이자 색인이고, 이 파일이 정본이다. "
                 "노드의 칸(readiness·source·classification·relations…)은 사람이 적고, "
                 "file·publication·status 는 기록 파일에서 읽어 채운다 — "
                 "python3 scripts/build-tech-log-tree.py <프로젝트>"),
        "contract": {
            "decomposition": parsed["prose"],
            "readinessValues": techlog.READINESS,
            "dispositionValues": ledger.get("dispositionValues", {}),
        },
        "counts": {},
        "topics": topics,
        "candidates": ledger.get("candidates", []),
        "history": {k: v for k, v in meta_state.items()
                    if k not in ("schemaVersion", "project", "rootTreePath")},
    }
    for key in ("counts", "unmapped", "explicitAnalysisCandidates", "cycle2", "cycle3",
                "conceptRecall", "migration", "fold"):
        if key in ledger:
            out["history"].setdefault("ledger", {})[key] = ledger[key]

    total = sum(len(v) for t in topics.values() for v in t["kinds"].values())
    out["counts"] = {"topics": len(topics), "nodes": total,
                     "candidates": len(out["candidates"])}

    print(f"{args.project}: 주제 {len(topics)} · 글감 {total} · 후보 {len(out['candidates'])}")
    if orphans:
        print(f"  사람이 읽는 트리에 줄이 없던 노드 {orphans}건은 listedInTree=false 로 옮겼다")
    if args.dry_run:
        print("  (--dry-run: 쓰지 않았다)")
        return 0

    with open(index_path, "w", encoding="utf-8") as fh:
        json.dump(out, fh, ensure_ascii=False, indent=2)
        fh.write("\n")
    print(f"  tech-log-tree.json {os.path.getsize(index_path):,} bytes")

    if not args.keep:
        for name in ("root-tree.md", "candidate-ledger.json", "root-tree-source-manifest.json"):
            path = os.path.join(studio, name)
            if os.path.exists(path):
                os.remove(path)
        shutil.rmtree(os.path.join(studio, "_meta"), ignore_errors=True)
        print("  root-tree.md · candidate-ledger.json · manifest · _meta/ 를 지웠다")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())