refactor: 提取 is_vp_or_py_at/JoinPath/read_file_to_buf 公共工具，消除 IncludesScanner/Phase2 重复代码

App/lib/Projectrans/Utils.vp

@@ -7,6 +7,7 @@ import viperlib
 import hashlib
 import w32.win32file
 import w32.win32base
+import w32.fileio as fileio
 
 
 # ============================================================
@@ -91,6 +92,122 @@ def compute_sha1(pool: memhub.MemBuddy | t.CPtr, content: str) -> str:
     return bytes_to_hex(pool, digest, 8, 0, 0)
 
 
+# ============================================================
+# is_vp_or_py_at - 检查路径段是否以 .vp/.py 后缀结尾
+# ============================================================
+def is_vp_or_py_at(p: str, start: t.CSizeT, plen: t.CSizeT) -> int:
+    """检查 p[start, start+plen) 是否以 .vp/.py 后缀结尾（1=是，0=否）"""
+    if p is None or plen <= 3:
+        return 0
+    base: t.CSizeT = start + plen - 3
+    if p[base] == '.':
+        if p[base + 1] == 'v' and p[base + 2] == 'p':
+            return 1
+        if p[base + 1] == 'p' and p[base + 2] == 'y':
+            return 1
+    return 0
+
+
+# ============================================================
+# JoinPath - 拼接两个路径段为 a/b
+# ============================================================
+def JoinPath(a: str, b: str) -> str:
+    """拼接路径段为 "a/b"（stdlib.malloc 分配，调用方负责 free），失败返回 None"""
+    if a is None or b is None:
+        return None
+    a_len: t.CSizeT = string.strlen(a)
+    b_len: t.CSizeT = string.strlen(b)
+    total_len: t.CSizeT = a_len + b_len + 2
+    buf: str = stdlib.malloc(total_len)
+    if buf is None:
+        return None
+    viperlib.snprintf(buf, total_len, "%s/%s", a, b)
+    return buf
+
+
+# ============================================================
+# read_file_to_buf - 读取整个文件到缓冲区（null 终止）
+# ============================================================
+def read_file_to_buf(path: str, buf_size: t.CSizeT) -> bytes:
+    """读取文件到 stdlib.malloc 缓冲区并保证 null 终止
+
+    返回缓冲区（调用方负责 free）；打开失败/空文件/分配失败返回 None。
+    合并自 Phase2/StubMerger/IncludesScanner 的重复"打开+读取+终止"序列。
+    """
+    if path is None or buf_size <= 0:
+        return None
+    sf: fileio.File | t.CPtr = fileio.File(path, fileio.MODE.R)
+    if sf.closed:
+        return None
+    buf: bytes = stdlib.malloc(buf_size)
+    if buf is None:
+        sf.close()
+        return None
+    br: t.CInt64T = sf.read_all(buf, buf_size)
+    sf.close()
+    if br <= 0:
+        stdlib.free(buf)
+        return None
+    if br < buf_size:
+        buf[br] = 0
+    else:
+        buf[buf_size - 1] = 0
+    return buf
+
+
+# ============================================================
+# PathToModuleName - 文件路径转模块名
+# ============================================================
+def PathToModuleName(path: str, strip_includes_prefix: int) -> str:
+    """将文件路径转换为模块名（stdlib.malloc 分配，调用方负责 free）
+
+    strip_includes_prefix=1 时跳过 includes/ 前缀（如 includes/ast/__init__.vp → ast）。
+    特殊处理 __init__.vp：pkg/__init__.vp → pkg（而非 pkg.__init__）。
+    合并自 StubMerger._PathToModuleName / IncludesScanner 内联模块名计算。
+    """
+    if path is None:
+        return None
+    plen: t.CSizeT = string.strlen(path)
+    # 跳过 includes/ 或 includes\ 前缀（Windows 反斜杠兼容）
+    prefix_len: t.CSizeT = 9
+    path_start: t.CSizeT = 0
+    if strip_includes_prefix != 0 and plen > prefix_len and string.strncmp(path, "includes", 8) == 0:
+        psep_ch: t.CChar = path[8]
+        if psep_ch == '/' or psep_ch == '\\':
+            path_start = prefix_len
+            plen = plen - prefix_len
+    # 去掉 .vp 或 .py 后缀（.vp 为 Viper 源，.py 向后兼容）
+    if is_vp_or_py_at(path, path_start, plen) != 0:
+        plen -= 3
+    if plen == 0:
+        return None
+    # 检查是否以 /__init__ 或 \__init__ 结尾（包的 __init__.vp）
+    # 同时支持 / 和 \ 两种分隔符（TPC 写入的路径可能混合使用）
+    if plen > 9:
+        tail_off: t.CSizeT = path_start + plen - 9
+        sep_char: t.CChar = path[tail_off]
+        if (sep_char == '/' or sep_char == '\\') \
+           and path[tail_off + 1] == '_' and path[tail_off + 2] == '_' \
+           and path[tail_off + 3] == 'i' and path[tail_off + 4] == 'n' and path[tail_off + 5] == 'i' \
+           and path[tail_off + 6] == 't' and path[tail_off + 7] == '_' and path[tail_off + 8] == '_':
+            # 去掉 分隔符+__init__ 后缀，plen 就是包名部分长度
+            plen = tail_off - path_start
+    if plen == 0:
+        return None
+    # 复制并替换 / 为 .
+    result: str = stdlib.malloc(plen + 1)
+    if result is None:
+        return None
+    for ri in range(plen):
+        ch: t.CChar = path[path_start + ri]
+        if ch == '/' or ch == '\\':
+            result[ri] = '.'
+        else:
+            result[ri] = ch
+    result[plen] = '\0'
+    return result
+
+
 # ============================================================
 # CleanDir - 递归删除目录中所有文件和子目录（保留 dir_path 本身）
 #
@@ -131,10 +248,8 @@ def CleanDir(dir_path: str) -> int:
                 elif fname[1] == '.' and fname[2] == '\0':
                     fname = None
         if fname is not None:
-            fname_len: t.CSizeT = string.strlen(fname)
-            full_path: bytes = stdlib.malloc(dir_len + fname_len + 2)
+            full_path: bytes = JoinPath(dir_path, fname)
             if full_path is not None:
-                viperlib.snprintf(full_path, dir_len + fname_len + 2, "%s/%s", dir_path, fname)
                 # 检查是否为目录（dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY）
                 is_dir: t.CUInt32T = find_data.dwFileAttributes & w32.win32file.FILE_ATTRIBUTE_DIRECTORY
                 if is_dir != 0:

App/lib/core/IncludesScanner.vp

@@ -6,7 +6,6 @@ import stdio
 import stdlib
 import w32.win32file as win32file
 import w32.win32base as win32base
-import w32.fileio as fileio
 import hashlib
 import viperlib
 import lib.Projectrans.Utils as Utils
@@ -91,21 +90,10 @@ def add_file_entry(result: ScanResult | t.CPtr,
     if result.Count >= result.Capacity:
         return 1
 
-    # 计算模块名：rel_path 中的 / 替换为 .，去掉 .vp/.py 扩展名
-    rel_len: t.CSizeT = string.strlen(rel_path)
-    mod_buf: str = stdlib.malloc(rel_len + 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
-    string.strcpy(mod_buf, rel_path)
-    # 替换 / 为 .
-    for i in range(rel_len):
-        ch: t.CChar = mod_buf[i]
-        if ch == '/' or ch == '\\':
-            mod_buf[i] = '.'
-    # 去掉 .vp 或 .py 扩展名（.vp 为 Viper 源，.py 向后兼容）
-    if rel_len >= 3:
-        if (mod_buf[rel_len - 3] == '.' and mod_buf[rel_len - 2] == 'v' and mod_buf[rel_len - 1] == 'p') or (mod_buf[rel_len - 3] == '.' and mod_buf[rel_len - 2] == 'p' and mod_buf[rel_len - 1] == 'y'):
-            mod_buf[rel_len - 3] = '\0'
 
     # 获取条目地址
     entry_size: t.CSizeT = FileEntry.__sizeof__()
@@ -130,7 +118,7 @@ def add_file_entry(result: ScanResult | t.CPtr,
     entry.Sha1 = sha1_buf
 
     # 复制相对路径
-    rel_buf: str = stdlib.malloc(rel_len + 1)
+    rel_buf: str = stdlib.malloc(string.strlen(rel_path) + 1)
     if rel_buf is None:
         return 1
     string.strcpy(rel_buf, rel_path)
@@ -156,30 +144,17 @@ def compute_file_sha1(pool: memhub.MemBuddy | t.CPtr,
     if pool is None or file_path is None:
         return None
 
-    # 打开文件
-    f: fileio.File | t.CPtr = fileio.File(file_path, fileio.MODE.R)
-    if f.closed:
-        return None
-
-    # 分配读取缓冲区（128KB，足够大多数 .vp/.py 文件）
-    # 使用 stdlib.malloc 避免 mbuddy 池耗尽（每文件 128KB，60+ 文件会耗尽 16MB 池）
+    # 读取文件内容（stdlib.malloc 分配，避免 mbuddy 池耗尽）
     BUF_SIZE: t.CSizeT = 131072
-    buf: bytes = stdlib.malloc(BUF_SIZE)
+    buf: bytes = Utils.read_file_to_buf(file_path, BUF_SIZE)
     if buf is None:
-        f.close()
-        return None
-
-    # 读取文件内容
-    bytes_read: t.CInt64T = f.read_all(buf, BUF_SIZE)
-    f.close()
-    if bytes_read <= 0:
-        stdlib.free(buf)
         return None
 
     # 原地去除 \r（CRLF → LF），与 Projectrans.py 文本模式读取一致
+    buf_len: t.CSizeT = string.strlen(buf)
     write_pos: t.CSizeT = 0
     read_pos: t.CSizeT = 0
-    while read_pos < bytes_read:
+    while read_pos < buf_len:
         ch: t.CChar = buf[read_pos]
         if ch != '\r':
             buf[write_pos] = ch
@@ -266,22 +241,24 @@ def scan_directory_recursive(pool: memhub.MemBuddy | t.CPtr,
 
         # 构造完整路径
         fname_len: t.CSizeT = string.strlen(fname)
-        full_path: bytes = stdlib.malloc(root_len + fname_len + 2)
+        full_path: bytes = Utils.JoinPath(root_dir, fname)
         if full_path is None:
             break
-        viperlib.snprintf(full_path, root_len + fname_len + 2, "%s/%s", root_dir, fname)
 
         # 构造相对路径
         prefix_len: t.CSizeT = 0
         if rel_prefix is not None:
             prefix_len = string.strlen(rel_prefix)
-        rel_path: bytes = stdlib.malloc(prefix_len + fname_len + 2)
+        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 rel_prefix is not None and prefix_len > 0:
-            viperlib.snprintf(rel_path, prefix_len + fname_len + 2, "%s/%s", rel_prefix, fname)
-        else:
-            string.strcpy(rel_path, fname)
 
         if is_dir != 0:
             # 递归扫描子目录
@@ -289,9 +266,8 @@ def scan_directory_recursive(pool: memhub.MemBuddy | t.CPtr,
         else:
             # 检查是否为 .vp 或 .py 文件（.vp 为 Viper 源，.py 向后兼容）
             is_py: int = 0
-            if fname_len >= 3:
-                if (fname[fname_len - 3] == '.' and fname[fname_len - 2] == 'v' and fname[fname_len - 1] == 'p') or (fname[fname_len - 3] == '.' and fname[fname_len - 2] == 'p' and fname[fname_len - 1] == 'y'):
-                    is_py = 1
+            if Utils.is_vp_or_py_at(fname, 0, fname_len) != 0:
+                is_py = 1
 
             if is_py != 0:
                 # 跳过 __pycache__ 目录下的文件

App/lib/core/Phase2.vp

@@ -111,44 +111,33 @@ def _ScanDirForPyFiles(mb: memhub.MemBuddy | t.CPtr,
                         is_dir = 1
                     if is_dir != 0:
                         # 递归扫描子目录
-                        sub_dir: bytes = stdlib.malloc(dir_len + fname_len + 2)
+                        sub_dir: bytes = Utils.JoinPath(dir_path, fname)
                         if sub_dir is not None:
-                            viperlib.snprintf(sub_dir, dir_len + fname_len + 2, "%s/%s", dir_path, fname)
                             file_count = _ScanDirForPyFiles(mb, sub_dir, entries, entry_size, file_count, max_files)
                             stdlib.free(sub_dir)
                     else:
                         # 检查是否是 .vp 或 .py 文件（.vp 为 Viper 源，.py 向后兼容）
                         if fname_len > 3:
-                            if (fname[fname_len - 3] == '.' and fname[fname_len - 2] == 'v' and fname[fname_len - 1] == 'p') or (fname[fname_len - 3] == '.' and fname[fname_len - 2] == 'p' and fname[fname_len - 1] == 'y'):
+                            if Utils.is_vp_or_py_at(fname, 0, fname_len) != 0:
                                 if file_count < max_files:
-                                    full_path: bytes = stdlib.malloc(dir_len + fname_len + 2)
+                                    full_path: bytes = Utils.JoinPath(dir_path, fname)
                                     if full_path is not None:
-                                        viperlib.snprintf(full_path, dir_len + fname_len + 2, "%s/%s", dir_path, fname)
-                                        sf: fileio.File | t.CPtr = fileio.File(full_path, fileio.MODE.R)
-                                        if not sf.closed:
-                                            sbuf: bytes = stdlib.malloc(SRC_BUF_SIZE)
-                                            if sbuf is not None:
-                                                br: LONG = sf.read_all(sbuf, SRC_BUF_SIZE)
-                                                sf.close()
-                                                if br > 0:
-                                                    if br < SRC_BUF_SIZE:
-                                                        sbuf[br] = 0
-                                                    else:
-                                                        sbuf[SRC_BUF_SIZE - 1] = 0
-                                                    sha1_val: str = Utils.compute_sha1(mb, sbuf)
-                                                    if sha1_val is not None:
-                                                        ea: t.CUInt64T = t.CUInt64T(entries) + file_count * entry_size
-                                                        ent: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea, t.CPtr))
-                                                        if ent is not None:
-                                                            ent.Path = full_path
-                                                            ent.Sha1 = sha1_val
-                                                            file_count += 1
-                                                            fb: t.CChar | t.CPtr = VLogger.fmt_buf()
-                                                            if fb is not None:
-                                                                viperlib.snprintf(fb, 1024, "%s (sha1=%s)", fname, sha1_val)
-                                                                VLogger.info(fb, "project")
-                                                stdlib.free(sbuf)
-                                        # full_path 不释放：ent.Path 引用它
+                                        sbuf: bytes = Utils.read_file_to_buf(full_path, SRC_BUF_SIZE)
+                                        if sbuf is not None:
+                                            sha1_val: str = Utils.compute_sha1(mb, sbuf)
+                                            if sha1_val is not None:
+                                                ea: t.CUInt64T = t.CUInt64T(entries) + file_count * entry_size
+                                                ent: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea, t.CPtr))
+                                                if ent is not None:
+                                                    ent.Path = full_path
+                                                    ent.Sha1 = sha1_val
+                                                    file_count += 1
+                                                    fb: t.CChar | t.CPtr = VLogger.fmt_buf()
+                                                    if fb is not None:
+                                                        viperlib.snprintf(fb, 1024, "%s (sha1=%s)", fname, sha1_val)
+                                                        VLogger.info(fb, "project")
+                                            stdlib.free(sbuf)
+                                    # full_path 不释放：ent.Path 引用它
 
         if w32.win32file.FindNextFileA(handle, find_data) == 0:
             break