#!/usr/bin/env python3
"""从 schema 生成一个项目的空白档案骨架，并自动预填可机器获取的字段。

用法
----
    python3 scripts/new_project.py --repo Devin-AXIS/iPolloWork
    python3 scripts/new_project.py --repo owner/name --id custom-id --status watch

生成后需要人工补完的内容（脚本不猜这些）：
  identity / business / scenario / tech / risk / verdict
其中 **risk.claim_vs_reality 是必填**，它承载"宣称 vs 实测"的落差分析，
是整套档案里最不能空着的字段。
"""

from __future__ import annotations

import argparse
import re
import sys
from pathlib import Path

sys.path.insert(0, str(Path(__file__).resolve().parent))
from _common import (  # noqa: E402
    DATA,
    PROJECTS,
    SCHEMA,
    classify_license,
    gh_get,
    load_yaml,
    today,
)

LEAF_KEYS = {"type", "required", "desc", "enum", "auto", "note"}


def render(node, indent=0, out=None, prefix="", idx=None):
    """把 schema 的 fields 树渲染成带注释的 YAML 文本。

    同时返回 {字段路径: 行号} 索引。回填值必须走这个索引按**路径**定位，
    不能靠硬编码缩进 —— schema 一旦加减层级，缩进常量就会悄悄错位，
    而且像 `name` 这种同名不同层的字段（项目名 vs 许可证名）会互相覆盖。
    """
    out = out if out is not None else []
    idx = idx if idx is not None else {}
    pad = " " * indent
    for key, spec in node.items():
        if not isinstance(spec, dict):
            continue
        path = f"{prefix}{key}"
        if "type" in spec:                                    # 叶子字段
            default = {
                "str": '""', "int": "0", "bool": "false",
                "list": "[]", "dict": "{}",
            }.get(spec["type"], '""')
            comment = ""
            if spec.get("enum"):
                # fields 里的 enum 是 enums 段的键名引用（字符串），不是字面量数组；
                # 直接 join 会把 "vendor_type" 拆成 v|e|n|d|o|r|_|t|y|p|e
                e = spec["enum"]
                comment = f"  # {'|'.join(e) if isinstance(e, list) else e}"
            elif spec.get("desc"):
                d = re.sub(r"\s+", " ", spec["desc"].replace("\n", " "))
                comment = f"  # {d[:60]}"
            idx[path] = len(out)
            out.append(f"{pad}{key}: {default}{comment}")
        elif spec and not (set(spec) & LEAF_KEYS):            # 嵌套 section
            out.append(f"{pad}{key}:")
            render(spec, indent + 2, out, path + ".", idx)
    return out, idx


def slug(repo: str) -> str:
    name = repo.rstrip("/").split("/")[-1]
    name = re.sub(r"\.git$", "", name)
    return re.sub(r"[^a-z0-9]+", "-", name.lower()).strip("-")


def main() -> int:
    ap = argparse.ArgumentParser()
    ap.add_argument("--repo", required=True, help="GitHub 仓库 owner/name")
    ap.add_argument("--id", help="档案 id，默认由仓库名生成")
    ap.add_argument("--status", default="active", choices=["active", "watch", "archived"])
    args = ap.parse_args()

    pid = args.id or slug(args.repo)
    ypath = PROJECTS / f"{pid}.yaml"
    if ypath.exists():
        print(f"档案已存在：{ypath}")
        return 1

    print(f"==> 生成档案 {pid}  <-  {args.repo}")

    # ---- 抓可自动获取的部分 ----
    info = gh_get(f"repos/{args.repo}") or {}
    lic = info.get("license") or {}
    spdx = lic.get("spdx_id") or ""
    if spdx == "NOASSERTION":
        spdx = "NOASSERTION"

    cls = classify_license(spdx, lic.get("name", ""))
    print(f"    许可证 spdx={spdx or '(无)'} → {cls['level']} / 可见性 {cls['visibility']}")

    schema = load_yaml(SCHEMA / "project.schema.yaml")
    lines, idx = render(schema.get("fields", {}))

    def setval(path: str, val) -> None:
        """按字段路径回填值，保留该行原有的缩进与键名。"""
        i = idx.get(path)
        if i is None:
            print(f"    [warn] schema 中无字段 {path}，跳过回填")
            return
        line = lines[i]
        indent = " " * (len(line) - len(line.lstrip()))
        key = path.split(".")[-1]
        lines[i] = f"{indent}{key}: {val}"

    setval("id", pid)
    setval("name", f'"{info.get("name", "")}"')
    setval("repo", f'"{args.repo}"')
    setval("homepage", f'"{info.get("homepage") or ""}"')
    setval("added_at", f'"{today()}"')
    setval("status", args.status)

    if info:
        setval("identity.tagline", f'"{info.get("description") or ""}"')
        setval("identity.first_seen", f'"{(info.get("created_at") or "")[:10]}"')

    setval("business.license.spdx", f'"{spdx or "NOASSERTION"}"')
    setval("business.license.name", f'"{lic.get("name") or "自定义许可"}"')
    setval("business.license.openness", cls["level"])
    setval("business.license.redistributable", "true" if cls["redistributable"] else "false")

    setval("mirror.gitea_repo", f'"gitinbox/{info.get("name") or pid}"')
    setval("mirror.visibility", cls["visibility"])
    setval("mirror.upstream", f'"https://github.com/{args.repo}"')

    text = "\n".join(lines)
    header = (
        f"# {info.get('name') or pid} · 项目档案\n"
        f"# 生成于 {today()} · 由 scripts/new_project.py 从 schema 生成\n"
        "# 人工补完重点：business / scenario / tech / risk.claim_vs_reality / verdict\n"
        "# 字段规范见 schema/project.schema.yaml\n"
        "\n"
    )

    PROJECTS.mkdir(parents=True, exist_ok=True)
    ypath.write_text(header + text + "\n", encoding="utf-8")
    print(f"    已写入 {ypath.relative_to(PROJECTS.parent.parent)}")

    # ---- 登记到 registry ----
    reg_path = DATA / "registry.yaml"
    reg_text = reg_path.read_text(encoding="utf-8")
    if f"id: {pid}" not in reg_text:
        entry = (
            f"  - id: {pid}\n"
            f"    repo: {args.repo}\n"
            f"    status: {args.status}\n"
            f"    added_at: {today()}\n"
            f"    license_level: {cls['level']}"
            + ("        # source-available，镜像私有\n" if cls["level"] == "L3" else "\n")
            + f"    mirror_visibility: {cls['visibility']}\n"
        )
        reg_text = reg_text.rstrip("\n") + "\n" + entry
        reg_path.write_text(reg_text, encoding="utf-8")
        print(f"    已登记到 data/registry.yaml")

    print(f"\n下一步：\n  1. 补完档案（重点 risk.claim_vs_reality）\n"
          f"  2. python3 scripts/fetch_stats.py --id {pid}\n"
          f"  3. bash scripts/mirror_sync.sh --id {pid}")
    return 0


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