# -*- coding: utf-8 -*-
"""取证脚本（第 2 棒）：Postiz ↔ PostSider 文件树差异量化。

输入：api/postiz.tree.json 与 api/postsider.tree.json（第 1 棒已拉，本脚本只读复用）。
输出：api/fork/fork-diff.json —— 逐路径比对结果 + 子树级统计，供分析文档引用（可复算）。
口径：GitHub tree API 的 blob `sha`（内容哈希）。
      sha 相同 ⇒ 两个文件**逐字节相同**；sha 不同 ⇒ 内容不同（可能只是换牌字符串，也可能是真改动）。
      ⛔ 本脚本**不判**"改动是好是坏"，只给计数与清单；⛔ 也不做"改动行数"——tree API 不给行数。
"""
import json, os, sys, time, collections

sys.stdout.reconfigure(encoding="utf-8", errors="replace")
HERE = os.path.dirname(os.path.abspath(__file__))
API = os.path.join(HERE, "api")
OUT = os.path.join(API, "fork")
os.makedirs(OUT, exist_ok=True)


def blobs(name):
    with open(os.path.join(API, name), encoding="utf-8") as f:
        d = json.load(f)
    return {e["path"]: e["sha"] for e in d["tree"] if e["type"] == "blob"}


P = blobs("postiz.tree.json")
S = blobs("postsider.tree.json")
common = set(P) & set(S)
identical = sorted(k for k in common if P[k] == S[k])
content_diff = sorted(k for k in common if P[k] != S[k])
only_postiz = sorted(set(P) - set(S))
only_postsider = sorted(set(S) - set(P))


def subtree(pref):
    p = {k for k in P if k.startswith(pref)}
    s = {k for k in S if k.startswith(pref)}
    b = p & s
    return {
        "pref": pref,
        "postiz": len(p),
        "postsider": len(s),
        "common": len(b),
        "byte_identical": len([k for k in b if P[k] == S[k]]),
        "content_diff": len([k for k in b if P[k] != S[k]]),
        "only_postiz": len(p - s),
        "only_postsider": len(s - p),
    }


PREF = [
    "apps/", "apps/backend/", "apps/frontend/", "apps/orchestrator/", "apps/sdk/",
    "apps/commands/", "apps/extension/", "apps/mcp/",
    "libraries/", "libraries/nestjs-libraries/src/integrations/",
    "libraries/nestjs-libraries/src/database/prisma/", "libraries/helpers/src/",
    "libraries/react-shared-libraries/",
]

social = "libraries/nestjs-libraries/src/integrations/social/"


def providers(m):
    return sorted(k[len(social):] for k in m if k.startswith(social) and k.endswith(".provider.ts"))


def cnt(m, pred):
    return sum(1 for k in m if pred(k))


res = {
    "generated_at": time.strftime("%Y-%m-%dT%H:%M:%S"),
    "source_trees": {
        "postiz": {"branch": "main", "tree_sha": "3128e870e653b1763c47392784df20f5e58cc087", "blobs": len(P)},
        "postsider": {"branch": "main", "tree_sha": "d5ecaf019acf2b8b38da1768c8b27bfd376de8eb", "blobs": len(S)},
    },
    "totals": {
        "postiz_blobs": len(P),
        "postsider_blobs": len(S),
        "common_paths": len(common),
        "byte_identical": len(identical),
        "content_diff": len(content_diff),
        "only_postiz": len(only_postiz),
        "only_postsider": len(only_postsider),
    },
    "subtrees": [subtree(p) for p in PREF],
    "signals": {
        "test_files": {
            "postiz_spec_ts": cnt(P, lambda k: k.endswith((".spec.ts", ".test.ts"))),
            "postsider_spec_ts": cnt(S, lambda k: k.endswith((".spec.ts", ".test.ts"))),
        },
        "root_jest_configs": {
            "postiz": cnt(P, lambda k: k.startswith("jest.") and "/" not in k),
            "postsider": cnt(S, lambda k: k.startswith("jest.") and "/" not in k),
        },
        "social_providers": {
            "postiz": providers(P),
            "postsider": providers(S),
            "only_postsider": sorted(set(providers(S)) - set(providers(P))),
            "only_postiz": sorted(set(providers(P)) - set(providers(S))),
        },
        "prisma_migrations": {
            "postiz_count": cnt(P, lambda k: "/database/prisma/migrations/" in k),
            "postsider_count": cnt(S, lambda k: "/database/prisma/migrations/" in k),
        },
    },
    "paths": {
        "only_postsider_top_dirs": sorted({k.split("/")[0] for k in only_postsider}),
        "only_postiz_top_dirs": sorted({k.split("/")[0] for k in only_postiz}),
        "only_postsider_root_files": [k for k in only_postsider if "/" not in k],
        "content_diff_by_top": dict(collections.Counter(k.split("/")[0] for k in content_diff).most_common()),
    },
}

with open(os.path.join(OUT, "fork-diff.json"), "w", encoding="utf-8") as f:
    json.dump(res, f, ensure_ascii=False, indent=1)

print(json.dumps(res["totals"], ensure_ascii=False))
print(json.dumps(res["signals"]["test_files"], ensure_ascii=False))
print("DONE -> " + os.path.join(OUT, "fork-diff.json"))