修正了一些错误

This commit is contained in:
2026-07-26 20:33:17 +08:00
parent 909792bc8f
commit 03d0bba534
40 changed files with 6939 additions and 827 deletions

View File

@@ -3,6 +3,7 @@ from stdint import *
import ast
import memhub
import string
import stdlib
import viperlib
import stdio
import lib.core.Handles.HandlesBase as HandlesBase
@@ -88,14 +89,24 @@ def is_module_imported(imported_modules: str, name: str) -> int:
def add_from_import(pool: memhub.MemBuddy | t.CPtr,
from_imports: str,
local_name: str,
module_name: str) -> str:
"""添加 from-import 映射 "name:module",返回新的 from_imports 字符串"""
module_name: str,
original_name: str = None) -> str:
"""添加 from-import 映射,返回新的 from_imports 字符串
格式:
无别名: "local_name:module_name"
有别名: "local_name:module_name:original_name"
original_name 为源模块中的真实函数名,用于跨模块 SHA1 修饰)
"""
if local_name is None or module_name is None:
return from_imports
entry: t.CChar | t.CPtr = pool.alloc(128)
entry: t.CChar | t.CPtr = pool.alloc(256)
if entry is None:
return from_imports
viperlib.snprintf(entry, 128, "%s:%s", local_name, module_name)
if original_name is not None and original_name != local_name:
viperlib.snprintf(entry, 256, "%s:%s:%s", local_name, module_name, original_name)
else:
viperlib.snprintf(entry, 256, "%s:%s", local_name, module_name)
if from_imports is None:
return entry
else:
@@ -111,6 +122,75 @@ def add_from_import(pool: memhub.MemBuddy | t.CPtr,
return from_imports
# ============================================================
# 查找 from-import 名称 → 返回原始函数名(别名场景)
#
# 对于 from X import Y as Z返回 Y源模块中的真实函数名
# 无别名时返回 Nonelocal_name 即为原始名)。
# ============================================================
def lookup_from_import_original(from_imports: str,
local_name: str) -> str:
"""查找 from-import 别名对应的原始函数名,返回 None=无别名或未找到"""
if local_name is None or from_imports is None:
return None
name_len: t.CSizeT = string.strlen(local_name)
cur: t.CChar | t.CPtr = from_imports
ci: t.CSizeT = 0
total_len: t.CSizeT = string.strlen(from_imports)
while ci < total_len:
# 跳过前导空格
while ci < total_len and cur[ci] == ' ':
ci += 1
if ci >= total_len:
break
# 找到第一个 ':' 的位置
colon1: t.CSizeT = ci
while colon1 < total_len and cur[colon1] != ':' and cur[colon1] != ' ':
colon1 += 1
if colon1 >= total_len or cur[colon1] != ':':
break
entry_name_len: t.CSizeT = colon1 - ci
# 跳过 star import
if entry_name_len == 1 and cur[ci] == '*':
ci = colon1
while ci < total_len and cur[ci] != ' ':
ci += 1
continue
# 比较名称
if entry_name_len == name_len:
match: int = 1
ei: t.CSizeT = 0
while ei < name_len:
if cur[ci + ei] != local_name[ei]:
match = 0
break
ei += 1
if match == 1:
# 找到匹配,检查是否有第三个字段(原始名)
pos: t.CSizeT = colon1 + 1
# 跳过模块名
while pos < total_len and cur[pos] != ':' and cur[pos] != ' ' and cur[pos] != '\0':
pos += 1
if pos < total_len and cur[pos] == ':':
# 有第三个字段: original_name
orig_start: t.CSizeT = pos + 1
orig_end: t.CSizeT = orig_start
while orig_end < total_len and cur[orig_end] != ' ' and cur[orig_end] != '\0':
orig_end += 1
orig_len: t.CSizeT = orig_end - orig_start
if orig_len > 0:
orig_buf: str = cur + orig_start
# 返回指向内部的指针(调用方需在使用期内保持 from_imports 有效)
return orig_buf
return None
return None
# 跳到下一个条目
ci = colon1
while ci < total_len and cur[ci] != ' ':
ci += 1
return None
# ============================================================
# 查找 from-import 名称 → 返回模块名或 None
#
@@ -150,7 +230,7 @@ def lookup_from_import(from_imports: str, name: str,
if entry_name_len == 1 and cur[ci] == '*':
mod_start: t.CSizeT = colon_pos + 1
mod_end: t.CSizeT = mod_start
while mod_end < total_len and cur[mod_end] != ' ' and cur[mod_end] != '\0':
while mod_end < total_len and cur[mod_end] != ' ' and cur[mod_end] != '\0' and cur[mod_end] != ':':
mod_end += 1
star_mod = cur + mod_start
# 比较名称
@@ -165,7 +245,8 @@ def lookup_from_import(from_imports: str, name: str,
if match == 1:
mod_start2: t.CSizeT = colon_pos + 1
mod_end2: t.CSizeT = mod_start2
while mod_end2 < total_len and cur[mod_end2] != ' ' and cur[mod_end2] != '\0':
# 模块名结束于: 空格、null、或第二个':'(别名格式的分隔符)
while mod_end2 < total_len and cur[mod_end2] != ' ' and cur[mod_end2] != '\0' and cur[mod_end2] != ':':
mod_end2 += 1
return cur + mod_start2
# 跳到下一个条目
@@ -178,6 +259,93 @@ def lookup_from_import(from_imports: str, name: str,
return star_mod
# ============================================================
# 全局 re-export 映射表module_sha1 + func_name → source_module
#
# 当 ast/__init__.py 中有 from .lexer import _lexer_init 时,
# _lexer_init 被 re-export 为 ast._lexer_init。
# 跨模块调用 ast._lexer_init(...) 需使用 ast.lexer 的 SHA1 混淆,
# 而非 ast 的 SHA1。
#
# 数据布局:
# _g_reexport_sha1s: 每条 17 字节SHA1 16字符 + null
# _g_reexport_funcs: 每条 64 字节(函数名 + null
# _g_reexport_srcs: 每条 64 字节(源模块名 + null
# ============================================================
MAX_REEXPORT: t.CDefine = 512
_g_reexport_sha1s: bytes = None
_g_reexport_funcs: bytes = None
_g_reexport_srcs: bytes = None
_g_reexport_count: int = 0
# ============================================================
# register_reexport - 注册 re-export 映射
#
# 在 HandleImportFromNames 中调用:当处理 from .X import Y 时,
# 记录当前模块(通过 ModuleSha1re-export 了函数 Y源模块为 X。
#
# 幂等:重复注册相同映射不会增加条目。
# 内存用 stdlib.malloc 分配(全局存储器,跨 Phase 持久化)。
# ============================================================
def register_reexport(mod_sha1: str, func_name: str, source_module: str) -> int:
"""注册 re-export 映射(幂等),返回 0 成功,-1 失败"""
global _g_reexport_sha1s, _g_reexport_funcs, _g_reexport_srcs, _g_reexport_count
if mod_sha1 is None or func_name is None or source_module is None:
return -1
# 懒初始化
if _g_reexport_sha1s is None:
_g_reexport_sha1s = stdlib.malloc(MAX_REEXPORT * 17)
_g_reexport_funcs = stdlib.malloc(MAX_REEXPORT * 64)
_g_reexport_srcs = stdlib.malloc(MAX_REEXPORT * 64)
if _g_reexport_sha1s is None or _g_reexport_funcs is None or _g_reexport_srcs is None:
return -1
string.memset(_g_reexport_sha1s, 0, MAX_REEXPORT * 17)
string.memset(_g_reexport_funcs, 0, MAX_REEXPORT * 64)
string.memset(_g_reexport_srcs, 0, MAX_REEXPORT * 64)
if _g_reexport_count >= MAX_REEXPORT:
return -1
# 幂等检查:查找是否已存在相同映射
for i in range(_g_reexport_count):
sidx: t.CSizeT = t.CSizeT(i) * 17
fidx: t.CSizeT = t.CSizeT(i) * 64
if string.strcmp(_g_reexport_sha1s + sidx, mod_sha1) == 0:
if string.strcmp(_g_reexport_funcs + fidx, func_name) == 0:
# 已存在,更新 source_module以防变化
string.strcpy(_g_reexport_srcs + fidx, source_module)
return 0
# 添加新条目
idx2: t.CSizeT = t.CSizeT(_g_reexport_count) * 17
fidx2: t.CSizeT = t.CSizeT(_g_reexport_count) * 64
string.strcpy(_g_reexport_sha1s + idx2, mod_sha1)
string.strcpy(_g_reexport_funcs + fidx2, func_name)
string.strcpy(_g_reexport_srcs + fidx2, source_module)
_g_reexport_count += 1
return 0
# ============================================================
# lookup_reexport - 查找 re-export 映射
#
# 给定模块 SHA1 和函数名,返回源模块名(如 "ast.lexer")或 None。
# 调用方通过 _lookup_module_sha1(source_module) 获取源模块的 SHA1。
# ============================================================
def lookup_reexport(mod_sha1: str, func_name: str) -> str:
"""查找 re-export 映射,返回源模块名或 None"""
if mod_sha1 is None or func_name is None:
return None
if _g_reexport_sha1s is None or _g_reexport_count <= 0:
return None
for i in range(_g_reexport_count):
sidx: t.CSizeT = t.CSizeT(i) * 17
fidx: t.CSizeT = t.CSizeT(i) * 64
if string.strcmp(_g_reexport_sha1s + sidx, mod_sha1) == 0:
if string.strcmp(_g_reexport_funcs + fidx, func_name) == 0:
return _g_reexport_srcs + fidx
return None
# ============================================================
# _resolve_relative_module - 解析相对导入为完整模块名
#
@@ -335,6 +503,24 @@ class ImportsHandle(HandlesBase.Mixin):
if alias.asname is not None:
self.Trans._imported_modules = add_imported_module(
pool_val, self.Trans._imported_modules, alias.asname)
# 别名也加入 from_importsasname:fullname
# 使 HT.func() 跨模块调用能通过 from_imports 查找别名→完整模块名→SHA1
# 否则 _lookup_module_sha1("HT") 找不到mod_arr 只存文件名 HandlesTranslator
self.Trans._from_imports = add_from_import(
pool_val, self.Trans._from_imports, alias.asname, alias.name)
else:
# 无别名时import a.b.c将顶层包名 "a" 也加入 _imported_modules
# 使 a.b.c 作为属性访问的 base 被翻译时能通过 is_module_imported("a") 检查
dot_ptr: str = string.strstr(alias.name, '.')
if dot_ptr is not None:
pkg_len: t.CSizeT = t.CSizeT(t.CUInt64T(dot_ptr) - t.CUInt64T(alias.name))
if pkg_len > 0 and pkg_len < 256:
pkg_buf: bytes = pool_val.alloc(pkg_len + 1)
if pkg_buf is not None:
string.strncpy(pkg_buf, alias.name, pkg_len)
pkg_buf[pkg_len] = '\0'
self.Trans._imported_modules = add_imported_module(
pool_val, self.Trans._imported_modules, pkg_buf)
return 0
# ============================================================
@@ -381,13 +567,19 @@ class ImportsHandle(HandlesBase.Mixin):
alias: ast.Alias | t.CPtr = (ast.Alias | t.CPtr)(names.get(ni))
if alias is not None and alias.name is not None:
local_name: str = alias.name
orig_name: str = None
if alias.asname is not None:
local_name = alias.asname
orig_name = alias.name
self.Trans._from_imports = add_from_import(
self.Trans.Pool, self.Trans._from_imports,
local_name, resolved)
local_name, resolved, orig_name)
# 命名空间隔离from-import 的名称标记为可见结构体
HandlesStruct.add_visible_struct(self.Trans.Pool, local_name)
# 注册 re-export 映射:当前模块 re-export 了 local_name源模块为 resolved
# 使跨模块调用 module.func() 能解析到正确的源模块 SHA1
if self.Trans.ModuleSha1 is not None:
register_reexport(self.Trans.ModuleSha1, local_name, resolved)
return 0