import t, c
from stdint import *
import memhub
import string
import stdio
import stdlib
import w32.win32file as win32file
import w32.win32base as win32base
import hashlib
import viperlib
import lib.Projectrans.Utils as Utils
import lib.core.VLogger as VLogger


# ============================================================
# IncludesScanner - includes 目录递归扫描（Win32 FindFirstFileA/FindNextFileA）
# 递归收集 .vp 文件路径与 SHA1，供 Phase1 stub 生成 / Phase2 stub 合并使用
# ============================================================

# 全局 mbuddy 指针
_mbuddy: memhub.MemBuddy*

# 文件路径最大长度
MAX_PATH_LEN: t.CDefine = 512
# 单次扫描最大文件数
MAX_FILES: t.CDefine = 1024


# ============================================================
# FileEntry - 文件条目（路径 + SHA1）
# ============================================================
@t.NoVTable
class FileEntry:
    """扫描到的文件条目"""
    Path: str           # 文件绝对路径
    Sha1: str           # SHA1 前16字符（16字节+null）
    RelPath: str        # 相对于 includes 根目录的路径
    ModuleName: str     # 模块名（如 "ast.parser"）
    IncDir: str         # 所属 includes 根目录（多 includes 目录支持）


# ============================================================
# ScanResult - 扫描结果
# ============================================================
@t.NoVTable
class ScanResult:
    """扫描结果集合"""
    Entries: FileEntry | t.CPtr   # FileEntry 数组
    Count: int
    Capacity: int


# ============================================================
# create_scan_result - 创建扫描结果容器
# ============================================================
def create_scan_result(pool: memhub.MemBuddy*) -> ScanResult | t.CPtr:
    """创建扫描结果容器，预分配 MAX_FILES 个槽位"""
    if pool is None:
        return None
    size: sizet = MAX_FILES * FileEntry.__sizeof__()
    # 使用 stdlib.malloc 避免 mbuddy 池耗尽（与 StubMerger 保持一致）
    entries: FileEntry | t.CPtr = stdlib.malloc(size)
    if entries is None:
        return None
    string.memset(entries, 0, size)

    result: ScanResult | t.CPtr = stdlib.malloc(ScanResult.__sizeof__())
    if result is None:
        return None
    string.memset(result, 0, ScanResult.__sizeof__())
    result.Entries = entries
    result.Count = 0
    result.Capacity = MAX_FILES
    return result


# ============================================================
# add_file_entry - 向扫描结果添加文件条目
# ============================================================
def add_file_entry(result: ScanResult | t.CPtr,
                   pool: memhub.MemBuddy*,
                   abs_path: str, rel_path: str,
                   sha1: str, inc_dir: str) -> int:
    """添加文件条目到扫描结果，返回 0 成功（inc_dir 为所属 includes 根目录）"""
    if result is None or pool is None:
        return 1
    if result.Count >= result.Capacity:
        return 1

    # 计算模块名：rel_path 中 / 替换为 .，去掉 .vp/.py 扩展名（__init__.vp 特判由 Utils 统一处理）
    mod_buf: str = Utils.PathToModuleName(rel_path, 0)
    if mod_buf is None:
        return 1

    # 获取条目地址
    entry_size: sizet = FileEntry.__sizeof__()
    entry_addr: uint64 = uint64(result.Entries) + result.Count * entry_size
    entry: FileEntry | t.CPtr = (FileEntry | t.CPtr)(entry_addr)
    if entry is None:
        return 1

    # 复制路径字符串
    abs_len: sizet = string.strlen(abs_path)
    abs_buf: str = stdlib.malloc(abs_len + 1)
    if abs_buf is None:
        return 1
    string.strcpy(abs_buf, abs_path)
    entry.Path = abs_buf

    # 复制 SHA1
    sha1_buf: str = stdlib.malloc(17)
    if sha1_buf is None:
        return 1
    string.strcpy(sha1_buf, sha1)
    entry.Sha1 = sha1_buf

    # 复制相对路径
    rel_buf: str = stdlib.malloc(string.strlen(rel_path) + 1)
    if rel_buf is None:
        return 1
    string.strcpy(rel_buf, rel_path)
    entry.RelPath = rel_buf

    # 复制模块名
    entry.ModuleName = mod_buf

    # 复制所属 includes 根目录（多 includes 目录支持，Phase2 拼路径用）
    if inc_dir is not None:
        inc_len: sizet = string.strlen(inc_dir)
        inc_buf: str = stdlib.malloc(inc_len + 1)
        if inc_buf is None:
            return 1
        string.strcpy(inc_buf, inc_dir)
        entry.IncDir = inc_buf

    result.Count += 1
    return 0


# ============================================================
# compute_file_sha1 - 读取文件内容并计算 SHA1（CRLF→LF 后计算，与 Projectrans.py 一致）
# ============================================================
def compute_file_sha1(pool: memhub.MemBuddy*,
                      file_path: str) -> str:
    """读取文件内容并计算 SHA1 前16字符（CRLF→LF 后计算，与 Projectrans.py 一致）"""
    if pool is None or file_path is None:
        return None

    # 读取文件内容（stdlib.malloc 分配，避免 mbuddy 池耗尽）；BUF_SIZE 须与翻译侧 SRC_BUF_SIZE（1MB）一致，否则尾部变更不改 SHA1 导致缓存永不失效
    BUF_SIZE: sizet = 1048576
    buf: bytes = Utils.read_file_to_buf(file_path, BUF_SIZE)
    if buf is None:
        return None

    # 原地去除 \r（CRLF → LF），与 Projectrans.py 文本模式读取一致
    buf_len: sizet = string.strlen(buf)
    write_pos: sizet = 0
    for ch in buf:
        if ch != '\r':
            buf[write_pos] = ch
            write_pos += 1

    # 添加 null 终止符
    buf[write_pos] = 0

    # 计算 SHA1
    ctx: rawptr[hashlib.sha1] = hashlib.sha1()
    if ctx is None:
        stdlib.free(buf)
        return None
    ctx.update(buf)

    digest: bytes = stdlib.malloc(hashlib.SHA1_DIGEST_LEN)
    if digest is None:
        stdlib.free(buf)
        return None
    ctx.final(digest)

    # 转为十六进制字符串（取前 8 字节 = 16 个十六进制字符，stdlib 分配保持原行为）
    hex_buf: str = Utils.bytes_to_hex(pool, digest, 8, 0, 1)
    stdlib.free(buf)
    stdlib.free(digest)
    return hex_buf


# ============================================================
# scan_directory_recursive - 递归扫描目录（FindFirstFileA/FindNextFileA 遍历目录树）
# ============================================================
def scan_directory_recursive(pool: memhub.MemBuddy*,
                             root_dir: str,
                             rel_prefix: str,
                             result: ScanResult | t.CPtr,
                             inc_dir: str) -> int:
    """递归扫描目录，收集 .vp/.py 文件（inc_dir 为所属 includes 根目录）"""
    if pool is None or root_dir is None or result is None:
        return 1
    # 子目录递归失败计数（不静默忽略，汇总处 fail-fast 终止）
    subdir_failed: int = 0

    # 构造搜索模式: root_dir/*
    root_len: sizet = string.strlen(root_dir)
    pattern: bytes = stdlib.malloc(root_len + 4)
    if pattern is None:
        return 1
    viperlib.snprintf(pattern, root_len + 4, "%s/*", root_dir)

    # find_data 缓冲 +16 余量：TPV 的 WIN32_FIND_DATAA.__sizeof__() 小于 Windows 实际写入，越界写会破坏相邻堆块
    fd_size_s: sizet = win32file.WIN32_FIND_DATAA.__sizeof__()
    find_data: rawptr[win32file.WIN32_FIND_DATAA] = stdlib.malloc(fd_size_s + 16)
    if find_data is None:
        return 1
    string.memset(find_data, 0, fd_size_s + 16)

    handle: win32base.HANDLE = win32file.FindFirstFileA(pattern, find_data)
    if handle == win32base.INVALID_HANDLE_VALUE:
        return 1

    # 遍历所有文件和子目录
    while True:
        # 跳过 . 和 ..
        fname: str = find_data.cFileName
        if fname is not None:
            fname0: char = fname[0]
            if fname0 == '.':
                fname1: char = fname[1]
                if fname1 == '\0':
                    # "."
                    if win32file.FindNextFileA(handle, find_data) == 0:
                        break
                    continue
                elif fname1 == '.':
                    fname2: char = fname[2]
                    if fname2 == '\0':
                        # ".."
                        if win32file.FindNextFileA(handle, find_data) == 0:
                            break
                        continue

        # 检查是否为目录
        is_dir: int = find_data.dwFileAttributes & win32file.FILE_ATTRIBUTE_DIRECTORY

        # 构造完整路径
        fname_len: sizet = string.strlen(fname)
        full_path: bytes = Utils.JoinPath(root_dir, fname)
        if full_path is None:
            break

        # 构造相对路径
        prefix_len: sizet = 0
        if rel_prefix is not None:
            prefix_len = string.strlen(rel_prefix)
        rel_path: bytes = None
        if rel_prefix is not None and prefix_len > 0:
            rel_path = Utils.JoinPath(rel_prefix, fname)
        else:
            rel_path = stdlib.malloc(fname_len + 1)
            if rel_path is None:
                break
            string.strcpy(rel_path, fname)
        if rel_path is None:
            break

        if is_dir != 0:
            # 递归扫描子目录
            if scan_directory_recursive(pool, full_path, rel_path, result, inc_dir) != 0:
                subdir_failed += 1
        else:
            # 检查是否为 .vp 源文件（Viper 源仅 .vp，不再识别 .py）
            is_py: int = 0
            if Utils.is_vp_at(fname, 0, fname_len) != 0:
                is_py = 1

            if is_py != 0:
                # 跳过 __pycache__ 目录下的文件
                is_pycache: int = 0
                if rel_prefix is not None:
                    if string.strcmp(rel_prefix, "__pycache__") == 0:
                        is_pycache = 1

                if is_pycache == 0:
                    # 计算 SHA1
                    sha1: str = compute_file_sha1(pool, full_path)
                    if sha1 is not None:
                        add_file_entry(result, pool, full_path, rel_path, sha1, inc_dir)
                        VLogger.debug(f"扫描: {rel_path} -> {sha1}", "scan")
                    else:
                        VLogger.warning(f"SHA1 计算失败: {full_path}", "scan")

        # 继续搜索下一个文件
        if win32file.FindNextFileA(handle, find_data) == 0:
            break

    win32file.FindClose(handle)
    if subdir_failed > 0:
        VLogger.error(f"子目录扫描失败 {subdir_failed} 个", "scan")

    return 0


# ============================================================
# scan_includes - 扫描 includes 目录入口（返回所有 .vp 文件路径与 SHA1）
# ============================================================
def scan_includes(pool: memhub.MemBuddy*,
                  includes_dirs: str) -> ScanResult | t.CPtr:
    """扫描 includes 目录（支持空格分隔的多目录），返回所有 .vp/.py 文件的扫描结果"""
    if pool is None or includes_dirs is None:
        return None

    result: ScanResult | t.CPtr = create_scan_result(pool)
    if result is None:
        return None

    # 解析空格分隔的多 includes 目录（如 "../../includes ./includes"），逐目录递归扫描合并
    dirs_total: sizet = string.strlen(includes_dirs)
    d_i: sizet = 0
    while d_i < dirs_total:
        while d_i < dirs_total and includes_dirs[d_i] == ' ':
            d_i += 1
        if d_i >= dirs_total:
            break
        d_start: sizet = d_i
        while d_i < dirs_total and includes_dirs[d_i] != ' ':
            d_i += 1
        d_len: sizet = d_i - d_start
        if d_len == 0:
            continue
        dir_buf: bytes = stdlib.malloc(d_len + 1)
        if dir_buf is None:
            VLogger.error("includes 目录分配失败，终止扫描", "scan")
        string.strncpy(dir_buf, includes_dirs + d_start, d_len)
        dir_buf[d_len] = '\0'
        scan_directory_recursive(pool, dir_buf, None, result, dir_buf)
        stdlib.free(dir_buf)

    return result


# ============================================================
# find_entry_by_module - 按模块名查找文件条目
# ============================================================
def find_entry_by_module(result: ScanResult | t.CPtr,
                         module_name: str) -> FileEntry | t.CPtr:
    """按模块名查找文件条目"""
    if result is None or module_name is None:
        return None
    entry_size: sizet = FileEntry.__sizeof__()
    for i in range(result.Count):
        entry_addr: uint64 = uint64(result.Entries) + i * entry_size
        entry: FileEntry | t.CPtr = (FileEntry | t.CPtr)(entry_addr)
        if entry is not None:
            if entry.ModuleName is not None:
                if string.strcmp(entry.ModuleName, module_name) == 0:
                    return entry
    return None


# ============================================================
# _PathExistsA - 路径存在性检查（GetFileAttributesA）
# ============================================================
def _PathExistsA(path: str) -> int:
    """路径是否存在（1=存在，0=不存在/参数错误）

    用 GetFileAttributesA（无输出缓冲）替代 FindFirstFileA：后者写
    WIN32_FIND_DATAA 结构，而 TPV 该结构 sizeof 小于 Windows 实际写入
    （约 592 字节，差距 >270 字节），会越界写坏相邻堆块（BFS 输出槽
    0x2e 崩溃根因）。GetFileAttributesA 只返回属性值，无越界风险。
    """
    if path is None:
        return 0
    attr: ULONG = w32.win32file.GetFileAttributesA(path)
    if attr == 0xFFFFFFFF:  # INVALID_FILE_ATTRIBUTES
        return 0
    return 1


# ============================================================
# _TryResolveModuleFile - 尝试单个候选路径并写输出
#
# Args:
#   inc_dir: includes 根目录
#   rel_seg: 模块名转出的相对路径段（ast/parser）
#   rel_len: rel_seg 长度
#   suffix: 候选后缀（".vp" 或 "/__init__.vp"）
#   out_abs_path/out_rel_path/out_inc_dir: 输出槽（t.CPtr 指针）
#
# Returns:
#   1 命中（已写输出），0 未命中
# ============================================================
def _TryResolveModuleFile(inc_dir: str, rel_seg: bytes, rel_len: sizet,
                          suffix: str,
                          out_abs_path: t.CPtr, out_rel_path: t.CPtr,
                          out_inc_dir: t.CPtr) -> int:
    if inc_dir is None or rel_seg is None or suffix is None:
        return 0
    inc_len: sizet = string.strlen(inc_dir)
    suf_len: sizet = string.strlen(suffix)
    # 候选绝对路径: {inc}/{rel}{suffix}（手写拼接，避免 %. *s 精度修饰符）
    cand_len: sizet = inc_len + 1 + rel_len + suf_len
    cand: bytes = stdlib.malloc(cand_len + 1)
    if cand is None:
        return 0
    string.strcpy(cand, inc_dir)
    cand[inc_len] = '/'
    string.strncpy(cand + inc_len + 1, rel_seg, rel_len)
    string.strcpy(cand + inc_len + 1 + rel_len, suffix)
    if _PathExistsA(cand) == 0:
        stdlib.free(cand)
        return 0
    # 相对 includes 根目录的路径: {rel}{suffix}
    rel_path: bytes = stdlib.malloc(rel_len + suf_len + 1)
    if rel_path is None:
        stdlib.free(cand)
        return 0
    string.strncpy(rel_path, rel_seg, rel_len)
    rel_path[rel_len] = '\0'
    string.strcat(rel_path, suffix)
    # 所属 includes 根目录深拷贝：先全部分配就绪再统一写槽，
    # 避免 inc 分配失败时 abs/rel 已转移所有权造成的半写状态
    inc_buf: bytes = None
    if out_inc_dir is not None:
        inc_buf = stdlib.malloc(inc_len + 1)
        if inc_buf is None:
            stdlib.free(cand)
            stdlib.free(rel_path)
            return 0
        string.strcpy(inc_buf, inc_dir)
    # 写输出槽（stdlib 深拷贝，调用方负责 free）
    # 槽联合必须用 (sizet*)：t.CPtr|t.CPtr 会被折叠成 t.CPtr，
    # [0] 变成 1 字节访问 → 指针被截断（0x2e 崩溃根因）。CSizeT 成员为
    # 8 字节，写入用 sizet(ptr) 直接转换；读取用 t\.CVoid(u64, t.CPtr)
    # 转回指针（t\.CVoid 仅配合 t.CPtr 表示 void* 强转）。
    if out_abs_path is not None:
        abs_slot: sizet* = (sizet*)(out_abs_path)
        abs_slot[0] = sizet(cand)
    else:
        stdlib.free(cand)
    if out_rel_path is not None:
        rel_slot: sizet* = (sizet*)(out_rel_path)
        rel_slot[0] = sizet(rel_path)
    else:
        stdlib.free(rel_path)
    if out_inc_dir is not None and inc_buf is not None:
        inc_slot: sizet* = (sizet*)(out_inc_dir)
        inc_slot[0] = sizet(inc_buf)
    return 1


# ============================================================
# ResolveModuleToPath - 模块名→文件路径（按文件系统按需解析，不做全量目录遍历）
#
# 解析规则（多 includes 目录逐个尝试，模块优先于包）：
#   {inc_dir}/{mod.replace('.','/')}.vp          （文件模块）
#   {inc_dir}/{mod.replace('.','/')}/__init__.vp （包模块）
#
# Args:
#   includes_dirs: 空格分隔的多个 includes 根目录
#   mod_name: 绝对模块名
#   out_abs_path: 输出绝对路径（stdlib.malloc，调用方负责 free）
#   out_rel_path: 输出相对 includes 根目录的路径（同上，如 ast/parser.vp）
#   out_inc_dir:  输出所属 includes 根目录（同上）
#
# Returns:
#   0 找到；1 未找到/参数错误
# ============================================================
def ResolveModuleToPath(includes_dirs: str, mod_name: str,
                        out_abs_path: t.CPtr, out_rel_path: t.CPtr,
                        out_inc_dir: t.CPtr) -> int:
    if includes_dirs is None or mod_name is None:
        return 1
    mn_len: sizet = string.strlen(mod_name)
    if mn_len == 0:
        return 1
    # mod_name → 相对路径段（. → /）：ast.parser → ast/parser
    rel_seg: bytes = stdlib.malloc(mn_len + 1)
    if rel_seg is None:
        return 1
    rel_len: sizet = 0
    for mi in range(mn_len):
        ch_m: char = mod_name[mi]
        if ch_m == '.':
            rel_seg[rel_len] = '/'
        else:
            rel_seg[rel_len] = ch_m
        rel_len += 1
    rel_seg[rel_len] = '\0'

    found: int = 0
    d_total: sizet = string.strlen(includes_dirs)
    d_i: sizet = 0
    while d_i < d_total:
        while d_i < d_total and includes_dirs[d_i] == ' ':
            d_i += 1
        if d_i >= d_total:
            break
        d_start: sizet = d_i
        while d_i < d_total and includes_dirs[d_i] != ' ':
            d_i += 1
        d_len: sizet = d_i - d_start
        if d_len == 0:
            continue
        inc_dir_buf: bytes = stdlib.malloc(d_len + 1)
        if inc_dir_buf is None:
            continue
        string.strncpy(inc_dir_buf, includes_dirs + d_start, d_len)
        inc_dir_buf[d_len] = '\0'
        # 候选 1: {inc}/{rel}.vp（文件模块）
        if _TryResolveModuleFile(inc_dir_buf, rel_seg, rel_len, ".vp",
                                 out_abs_path, out_rel_path, out_inc_dir) != 0:
            found = 1
            stdlib.free(inc_dir_buf)
            break
        # 候选 2: {inc}/{rel}/__init__.vp（包模块）
        if _TryResolveModuleFile(inc_dir_buf, rel_seg, rel_len, "/__init__.vp",
                                 out_abs_path, out_rel_path, out_inc_dir) != 0:
            found = 1
        stdlib.free(inc_dir_buf)
        if found != 0:
            break
    stdlib.free(rel_seg)
    if found != 0:
        return 0
    return 1