settled-reach/tooling/pql-migrate/restructure_decisions.py

#!/usr/bin/env python3
"""Restructure decisions/ into the pql governance/ DQR tree (pql migration, Phase 1).

Moves the flat decisions/*.md layout into governance/{decisions,questions,rejected}/<domain>.md,
which is what `pql decisions sync` parses (type from subdir, domain from filename stem).

Link rewrites are TOKEN-PRESERVING: only the relative `foo.md` path portion of a
markdown link changes; the `[D-NNN]` bracket text and `#d-nnn-...` anchor stay
byte-identical, so pql's reference extraction is unaffected regardless of whether it
reads bracket text, anchor, or path.

Default is a dry run (prints every planned move + link change). Pass --apply to execute.
Idempotent-ish: re-running after apply is a no-op for moves (sources already gone).
"""
import re
import subprocess
import sys
from pathlib import Path

ROOT = Path("/var/mnt/data/projects/settled-reach/main")
APPLY = "--apply" in sys.argv

DSTEMS = {"architecture", "content", "economics", "perception", "process", "scope"}
QSTEMS = ["architecture", "content", "perception", "process", "scope"]

# R-record -> domain (R-001..R-010 platform/engine = architecture; R-011 economics; R-012 perception)
def rdomain(n: int) -> str:
    if n == 11:
        return "economics"
    if n == 12:
        return "perception"
    return "architecture"

REJECTED_DOMAINS = sorted({rdomain(n) for n in range(1, 13)})  # architecture, economics, perception

LINK = re.compile(r"\]\((?P<path>[A-Za-z0-9_-]+\.md)(?P<frag>#[^)]*)?\)")


def new_target_path(path: str, frag: str, src_subdir: str):
    """Return the rewritten relative path for a link target, or None to leave unchanged."""
    stem = path[:-3]
    if path == "rejected.md":
        m = re.match(r"#r-0*(\d+)", frag or "")
        rd = rdomain(int(m.group(1))) if m else "architecture"
        ttype, tdom = "rejected", rd
    elif path.startswith("questions-"):
        ttype, tdom = "questions", stem[len("questions-"):]
    elif path in ("questions.md", "README.md"):
        ttype, tdom = "index", None
    elif stem in DSTEMS:
        ttype, tdom = "decisions", stem
    else:
        return None  # unknown stem; leave as-is

    if ttype == "index":
        return "README.md" if src_subdir == "root" else "../README.md"
    if ttype == src_subdir:
        # same subdir: bare filename
        return f"{tdom}.md"
    # cross-subdir
    prefix = "" if src_subdir == "root" else "../"
    return f"{prefix}{ttype}/{tdom}.md"


def rewrite_links(text: str, src_subdir: str):
    changes = []

    def repl(mo):
        path, frag = mo.group("path"), mo.group("frag") or ""
        np = new_target_path(path, frag, src_subdir)
        if np is None or np == path:
            return mo.group(0)
        changes.append((f"{path}{frag}", f"{np}{frag}"))
        return f"]({np}{frag})"

    return LINK.sub(repl, text), changes


def git(*args):
    subprocess.run(["git", "-C", str(ROOT), *args], check=True)


def log_changes(rel, changes):
    if changes:
        print(f"  {rel}: {len(changes)} link(s) rewritten")
        for old, new in changes:
            print(f"      {old}  ->  {new}")


def move_and_rewrite(src_rel, dst_rel, src_subdir):
    src, dst = ROOT / src_rel, ROOT / dst_rel
    text = src.read_text(encoding="utf-8")
    new_text, changes = rewrite_links(text, src_subdir)
    print(f"MOVE {src_rel}  ->  {dst_rel}")
    log_changes(dst_rel, changes)
    if APPLY:
        dst.parent.mkdir(parents=True, exist_ok=True)
        git("mv", src_rel, dst_rel)
        dst.write_text(new_text, encoding="utf-8")


def split_rejected():
    src_rel = "decisions/rejected.md"
    text = (ROOT / src_rel).read_text(encoding="utf-8")
    # split into records on '### R-' headings
    parts = re.split(r"(?m)^(?=### R-\d)", text)
    head = parts[0]  # title + intro before first record
    records = parts[1:]
    buckets = {d: [] for d in REJECTED_DOMAINS}
    for rec in records:
        m = re.match(r"### R-0*(\d+)", rec)
        n = int(m.group(1))
        # strip a trailing '---' / footer that may cling to the last record
        body = re.split(r"(?m)^---\s*$", rec)[0].rstrip() + "\n"
        buckets[rdomain(n)].append(body)
    print(f"SPLIT {src_rel}  ->  {len(REJECTED_DOMAINS)} domain files")
    for dom, recs in buckets.items():
        dst_rel = f"governance/rejected/{dom}.md"
        title = f"# Rejected Alternatives — {dom.capitalize()}\n\nRejected proposals in the **{dom}** domain, rationale preserved for the audit trail.\n\n"
        content = title + "\n".join(recs).rstrip() + "\n"
        content, changes = rewrite_links(content, "rejected")
        print(f"  -> {dst_rel}: {len(recs)} record(s)")
        log_changes(dst_rel, changes)
        if APPLY:
            (ROOT / dst_rel).parent.mkdir(parents=True, exist_ok=True)
            (ROOT / dst_rel).write_text(content, encoding="utf-8")
    if APPLY:
        git("rm", src_rel)


def main():
    print(f"=== restructure_decisions.py ({'APPLY' if APPLY else 'DRY-RUN'}) ===\n")
    # 1. D-domain files -> governance/decisions/
    for s in sorted(DSTEMS):
        move_and_rewrite(f"decisions/{s}.md", f"governance/decisions/{s}.md", "decisions")
    print()
    # 2. questions-<domain>.md -> governance/questions/<domain>.md
    for q in QSTEMS:
        move_and_rewrite(f"decisions/questions-{q}.md", f"governance/questions/{q}.md", "questions")
    print()
    # 3. rejected.md -> governance/rejected/<domain>.md (split by domain)
    split_rejected()
    print()
    # 4. index files (README.md, questions.md) are folded into governance/README.md by hand;
    #    remove the originals here.
    for idx in ("decisions/README.md", "decisions/questions.md"):
        print(f"REMOVE {idx}  (folded into governance/README.md)")
        if APPLY:
            git("rm", idx)
    print("\nDone." + ("" if APPLY else "  (dry run — pass --apply to execute)"))


if __name__ == "__main__":
    main()