#!/usr/bin/env python3
"""
宠小它 本体—代码漂移审计脚本。

扫描 backend/src/main/java/com/petstore/controller/*.java，提取所有 HTTP 端点
（类级 @RequestMapping + 方法级 @GetMapping/@PostMapping/@PutMapping/@DeleteMapping），
对比 docs/graph/ontology.jsonl 中 Action 的 entrypoints，报告：

1. code-only：代码里有但本体未录入的端点（drift：应补录本体或确认废弃）
2. ontology-only：本体里有但代码里找不到的端点（drift：可能是已删除接口或路径写错）
3. entity-coverage：扫描 backend/src/main/java/com/petstore/entity/*.java，对比 ontology.jsonl 的 Entity，
   报告未录入的实体类。

用法：
    python3 docs/graph/audit_drift.py [repo_root]

默认 repo_root = . （当前工作目录，应能访问 backend/ 和 docs/）。
退出码：0 = 无漂移；1 = 有漂移（warning）；2 = 解析错误。
"""

import json
import os
import re
import sys
from collections import defaultdict

VALID_TYPES = {"Entity", "Action"}


def find_controller_files(repo_root):
    ctrl_dir = os.path.join(repo_root, "backend", "src", "main", "java", "com", "petstore", "controller")
    if not os.path.isdir(ctrl_dir):
        return []
    return sorted(
        os.path.join(ctrl_dir, f) for f in os.listdir(ctrl_dir) if f.endswith(".java")
    )


def find_entity_files(repo_root):
    ent_dir = os.path.join(repo_root, "backend", "src", "main", "java", "com", "petstore", "entity")
    if not os.path.isdir(ent_dir):
        return []
    return sorted(
        os.path.join(ent_dir, f) for f in os.listdir(ent_dir) if f.endswith(".java")
    )


# 类级 @RequestMapping("/api/xxx")
CLASS_MAPPING_RE = re.compile(r'@RequestMapping\s*\(\s*(?:value\s*=\s*)?"([^"]+)"', re.MULTILINE)
# 方法级 mapping 注解
METHOD_MAPPING_RE = re.compile(
    r'@(GetMapping|PostMapping|PutMapping|DeleteMapping)\s*\(\s*(?:value\s*=\s*)?"([^"]*)"',
    re.MULTILINE,
)
# 方法签名（紧跟 mapping 注解后的 public 方法）
METHOD_SIG_RE = re.compile(
    r'public\s+\S+\s+(\w+)\s*\(',
    re.MULTILINE,
)
# 类名
CLASS_NAME_RE = re.compile(r'class\s+(\w+)\b')


def extract_endpoints(java_path):
    """从单个 controller 文件提取 HTTP 端点：返回 [(http_method, full_path, java_method_name), ...]"""
    with open(java_path, "r", encoding="utf-8") as f:
        content = f.read()

    class_name_match = CLASS_NAME_RE.search(content)
    class_name = class_name_match.group(1) if class_name_match else os.path.basename(java_path)

    # 类级基础路径
    class_path = ""
    cm = CLASS_MAPPING_RE.search(content)
    if cm:
        class_path = cm.group(1)

    endpoints = []

    # 按方法级 mapping 注解切分
    # 找所有方法级 mapping 注解的位置
    for m in METHOD_MAPPING_RE.finditer(content):
        annotation = m.group(1)
        method_path = m.group(2)
        http_method = {
            "GetMapping": "GET",
            "PostMapping": "POST",
            "PutMapping": "PUT",
            "DeleteMapping": "DELETE",
        }[annotation]

        # 在该注解之后找最近的方法签名
        rest = content[m.end():]
        sig = METHOD_SIG_RE.search(rest)
        java_method = sig.group(1) if sig else "?"

        # 拼接完整路径
        full = class_path.rstrip("/") + "/" + method_path.lstrip("/") if method_path else class_path
        # 去掉多余斜杠
        full = re.sub(r"/+", "/", full)
        # 处理通配符路径（如 /api/upload/image/**）——保留通配符用于匹配
        endpoints.append((http_method, full, java_method, class_name))

    return endpoints


def extract_entity_classes(java_path):
    """从实体文件提取类名"""
    with open(java_path, "r", encoding="utf-8") as f:
        content = f.read()
    m = re.search(r'@Entity\b', content)
    if not m:
        return None
    cm = CLASS_NAME_RE.search(content)
    return cm.group(1) if cm else None


def load_ontology(repo_root):
    """加载 ontology.jsonl，返回 actions 与 entities 的端点/类名集合"""
    jsonl_path = os.path.join(repo_root, "docs", "graph", "ontology.jsonl")
    if not os.path.exists(jsonl_path):
        print(f"ERROR: {jsonl_path} not found")
        sys.exit(2)

    action_endpoints = set()  # (HTTP_METHOD, path_pattern)
    action_endpoints_by_id = defaultdict(list)
    entity_class_names = set()  # PascalCase 类名
    entity_ids = {}  # class_name -> entity_id
    non_jpa_entity_class_names = set()  # 本体里标 jpa:false 的（不要求代码有 @Entity 类）

    with open(jsonl_path, "r", encoding="utf-8") as f:
        for line in f:
            line = line.strip()
            if not line:
                continue
            try:
                obj = json.loads(line)
            except json.JSONDecodeError:
                continue
            if obj.get("type") == "Action":
                entrypoints = obj.get("entrypoints", [])
                eid = obj.get("id", "?")
                for ep in entrypoints:
                    if isinstance(ep, str) and (" " in ep or ep.startswith(("GET", "POST", "PUT", "DELETE"))):
                        # 格式 "POST /api/report/create" 或 "ReportController#create"
                        parts = ep.split(" ", 1)
                        if len(parts) == 2 and parts[0] in ("GET", "POST", "PUT", "DELETE"):
                            action_endpoints.add((parts[0], parts[1]))
                            action_endpoints_by_id[eid].append((parts[0], parts[1]))
            elif obj.get("type") == "Entity":
                eid = obj.get("id", "")
                # 从 entity:store 推 PascalCase Store
                if eid.startswith("entity:"):
                    class_name = "".join(p.capitalize() for p in eid[7:].split("_"))
                    # jpa:false 表示本体内建模为领域实体但代码里不是 JPA @Entity（如字段承载或 record）
                    if obj.get("jpa") is False:
                        non_jpa_entity_class_names.add(class_name)
                    else:
                        entity_class_names.add(class_name)
                        entity_ids[class_name] = eid

    return action_endpoints, action_endpoints_by_id, entity_class_names, entity_ids, non_jpa_entity_class_names


def path_matches_ontology_pattern(code_path, ontology_path):
    """代码路径是否匹配本体里的路径模式（支持 ** 通配符）"""
    if code_path == ontology_path:
        return True
    # 处理 ** 通配符
    if "**" in ontology_path:
        # 转换为正则：** → .*, * → [^/]*
        pattern = re.escape(ontology_path).replace(r"\*\*", ".*").replace(r"\*", r"[^/]*")
        return re.fullmatch(pattern, code_path) is not None
    # 处理单 * 通配符（如 /api/report/*/suggestion）
    if "*" in ontology_path:
        pattern = re.escape(ontology_path).replace(r"\*", r"[^/]*")
        return re.fullmatch(pattern, code_path) is not None
    return False


def main():
    repo_root = sys.argv[1] if len(sys.argv) > 1 else "."
    repo_root = os.path.abspath(repo_root)

    action_endpoints, action_endpoints_by_id, entity_class_names, entity_ids, non_jpa_entity_class_names = load_ontology(repo_root)

    # === 1. Controller 端点漂移 ===
    controller_files = find_controller_files(repo_root)
    code_endpoints = []  # (http_method, path, java_method, class_name, file)
    for cf in controller_files:
        for http_method, path, java_method, class_name in extract_endpoints(cf):
            code_endpoints.append((http_method, path, java_method, class_name, cf))

    # 对比
    code_only = []  # 代码里有，本体没有
    for http_method, path, java_method, class_name, cf in code_endpoints:
        found = False
        for ont_method, ont_path in action_endpoints:
            if ont_method == http_method and path_matches_ontology_pattern(path, ont_path):
                found = True
                break
        if not found:
            code_only.append((http_method, path, java_method, class_name, cf))

    ontology_only = []  # 本体里有，代码没有
    for ont_method, ont_path in action_endpoints:
        found = False
        for http_method, path, java_method, class_name, cf in code_endpoints:
            if http_method == ont_method and path_matches_ontology_pattern(path, ont_path):
                found = True
                break
        if not found:
            ontology_only.append((ont_method, ont_path))

    # === 2. Entity 类覆盖 ===
    entity_files = find_entity_files(repo_root)
    code_entity_classes = set()
    for ef in entity_files:
        cn = extract_entity_classes(ef)
        if cn:
            code_entity_classes.add(cn)

    entities_missing_in_ontology = code_entity_classes - entity_class_names
    entities_missing_in_code = entity_class_names - code_entity_classes

    # non-jpa 本体实体（如 HighlightVideo 字段承载、SessionToken record）不算 extra drift
    # 它们在本体里是领域实体，但代码里不是 @Entity 类，单独列出作为信息项
    non_jpa_in_ontology = non_jpa_entity_class_names

    # === 输出报告 ===
    print("=" * 60)
    print("宠小它 本体—代码漂移审计")
    print("=" * 60)

    print(f"\n扫描 controller 文件: {len(controller_files)}")
    print(f"代码 HTTP 端点数: {len(code_endpoints)}")
    print(f"本体 Action 端点数: {len(action_endpoints)}")
    print(f"扫描 entity 文件: {len(entity_files)}")
    print(f"代码 Entity 类数: {len(code_entity_classes)}")
    print(f"本体 Entity 数（JPA）: {len(entity_class_names)}")
    print(f"本体 Entity 数（非 JPA/领域实体）: {len(non_jpa_in_ontology)}")

    print("\n--- Controller 端点漂移 ---")
    if code_only:
        print(f"\n⚠️  code-only（代码里有，本体未录入）: {len(code_only)} 个")
        for http_method, path, java_method, class_name, cf in sorted(code_only):
            rel = os.path.relpath(cf, repo_root)
            print(f"  {http_method:6s} {path:40s} {class_name}#{java_method}  [{rel}]")
    else:
        print("\n✅ 无 code-only 漂移（所有代码端点都已录入本体）")

    if ontology_only:
        print(f"\n⚠️  ontology-only（本体里有，代码找不到）: {len(ontology_only)} 个")
        for ont_method, ont_path in sorted(ontology_only):
            print(f"  {ont_method:6s} {ont_path}")
    else:
        print("\n✅ 无 ontology-only 漂移（所有本体端点都有对应代码）")

    print("\n--- Entity 类覆盖 ---")
    if entities_missing_in_ontology:
        print(f"\n⚠️  代码里有 Entity 类但本体未录入: {len(entities_missing_in_ontology)} 个")
        for cn in sorted(entities_missing_in_ontology):
            print(f"  {cn}")
    else:
        print("\n✅ 所有代码 Entity 类都已录入本体")

    if entities_missing_in_code:
        print(f"\n⚠️  本体里有 Entity 但代码找不到: {len(entities_missing_in_code)} 个")
        for cn in sorted(entities_missing_in_code):
            print(f"  {cn}  (本体 ID: {entity_ids.get(cn, '?')})")
    else:
        print("✅ 所有本体 Entity 都有对应代码类")

    if non_jpa_in_ontology:
        print(f"\nℹ️  本体非 JPA 领域实体（代码里不是 @Entity 类，正常）: {len(non_jpa_in_ontology)} 个")
        for cn in sorted(non_jpa_in_ontology):
            print(f"  {cn}")

    # === 退出码 ===
    has_drift = bool(code_only or ontology_only or entities_missing_in_ontology or entities_missing_in_code)
    if has_drift:
        print("\n" + "=" * 60)
        print("漂移汇总: code-only={} ontology-only={} entity-missing={} entity-extra={}".format(
            len(code_only), len(ontology_only), len(entities_missing_in_ontology), len(entities_missing_in_code)
        ))
        print("=" * 60)
        sys.exit(1)
    else:
        print("\n" + "=" * 60)
        print("✅ 无漂移：本体与代码完全对齐")
        print("=" * 60)
        sys.exit(0)


if __name__ == "__main__":
    main()