Files
ViperC/App/lib/Phase/Phase2/__init__.vp
T
TermiNexus a12dcafed2 fix(tpv): 泛型类型参数 SHA1 规范化 + 自举链接修复
- 类型参数统一 "{定义模块SHA1}.{Class}"(ast.AST/AST → f467978339422f32.AST),
  消除跨模块特化名分裂导致的 undefined symbol(list[AST|t.CPtr].set 根因)
- TypeGenResolve 新增 _norm_type_arg_name/_is_sha1_prefixed/_concat_sha1_class;
  resolve_annotation_type 兼容 "{16hex}.Class" Attribute 形态(跳过
  imported_modules 检查,优先完整名命中定义侧 struct)
- CallGenCtor._extract_type_args_from_slice / extract_class_name_from_annotation
  新增 from_imports 参数,8 处调用方传 trans._from_imports 保持注册/查找一致
- 修复模块级函数链式调用推断(get_perm_pool() 联合返回 → .alloc() 裸名 alloc)
- 跨模块方法 declare self 改为类指针 %"sha1.Class"*,调用侧按形参 coerce bitcast
- 撤销泛型特化双 SHA1 前缀,恢复单前缀(模板 SHA1)
- 新增 --phase2command xxx.bat:不执行 llc/链接,命令写入 bat
2026-08-23 12:00:04 +08:00

1377 lines
75 KiB
Plaintext
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
import lib.core.Gen.GenApi as GenApi
import t, c
from stdint import *
import stdio
import string
import stdlib
import memhub
import viperlib
import w32.fileio as fileio
import w32.win32file
import w32.win32base
import sys
import ast
import llvmlite
import subprocess
import argparse
import lib.core.VLogger as VLogger
import lib.core.Handles.HandlesTranslator as HandlesTranslator
import lib.core.Handles.HandlesExprCall as HandlesExprCall
import lib.Phase.BuildPipeline as BuildPipeline
import lib.core.IncludesScanner as IncludesScanner
import lib.core.CompilerFactory as CompilerFactory
import lib.Stub.StubMerger as StubMerger
import lib.Projectrans.Utils as Utils
import lib.Projectrans.Config as Config
import lib.Stub.StubMerger.StubMergerBuild as StubMergerBuild
import lib.Stub.StubMerger.StubMergerCollect as StubMergerCollect
import lib.Stub.StubMerger.StubMergerIR as StubMergerIR
import lib.Stub.StubMerger.StubMergerStore as StubMergerStore
import lib.Stub.StubMerger.StubMergerTopo as StubMergerTopo
import lib.Stub.StubGen.Converter as StubConverter
# 全局 mbuddy 指针
_mbuddy: memhub.MemBuddy | t.CPtr
# 源代码缓冲区大小(1MB)
SRC_BUF_SIZE: t.CDefine = 1048576
# 最大源文件数(递归扫描子目录后文件数增多:App 现有 65 个 .vp,
# 2026-08-14 CompilerFactory.vp 加入后超出 64 被截断 → 链接 undefined symbol
# CompilerFactory.get_perm_pool。增大到 128 留余量)
MAX_SRC_FILES: t.CDefine = 128
# 全局栈金丝雀(避免局部变量改变栈布局)
_g_canary: t.CUInt32T
@t.NoVTable
class SrcFileEntry:
"""源文件条目"""
Path: str # 完整路径
Sha1: str # SHA1 前16字符
# ============================================================
# _ScanDirForPyFiles - 递归扫描目录,收集 .vp/.py 文件到 entries
#
# 替代旧的非递归 FindFirstFileA(dir/*.py) 扫描,支持子目录
# (如 App/lib/core/Handles/)。遇到目录递归扫描,遇到 .vp/.py
# 文件读取内容、计算 SHA1、加入 entries。
#
# Args:
# mb: 内存池
# dir_path: 当前扫描目录
# entries: SrcFileEntry 数组
# entry_size: 单个条目大小(SrcFileEntry.__sizeof__())
# file_count: 当前已收集的文件数
# max_files: 最大文件数
#
# Returns:
# 新的 file_count
# ============================================================
def _ScanDirForPyFiles(mb: memhub.MemBuddy | t.CPtr,
dir_path: str,
entries: SrcFileEntry | t.CPtr,
entry_size: t.CSizeT,
file_count: int, max_files: int) -> int:
"""递归扫描目录,收集 .vp/.py 文件到 entries,返回新的 file_count"""
if dir_path is None or entries is None:
return file_count
if file_count >= max_files:
return file_count
dir_len: t.CSizeT = string.strlen(dir_path)
pattern: bytes = stdlib.malloc(dir_len + 16)
if pattern is None:
return file_count
viperlib.snprintf(pattern, dir_len + 16, "%s/*", dir_path)
find_data_size: t.CSizeT = w32.win32file.WIN32_FIND_DATAA.__sizeof__()
find_data: w32.win32file.WIN32_FIND_DATAA | t.CPtr = stdlib.malloc(find_data_size + 16)
if find_data is None:
stdlib.free(pattern)
return file_count
string.memset(find_data, 0, find_data_size + 16)
handle: w32.win32base.HANDLE = w32.win32file.FindFirstFileA(pattern, find_data)
if handle == w32.win32base.INVALID_HANDLE_VALUE:
stdlib.free(pattern)
stdlib.free(find_data)
return file_count
while True:
fname: str = find_data.cFileName
if fname is None:
if w32.win32file.FindNextFileA(handle, find_data) == 0:
break
continue
fname_len: t.CSizeT = string.strlen(fname)
if fname_len <= 0:
if w32.win32file.FindNextFileA(handle, find_data) == 0:
break
continue
# 跳过 . 和 ..
is_dot: int = 0
if fname_len == 1 and fname[0] == '.':
is_dot = 1
elif fname_len == 2 and fname[0] == '.' and fname[1] == '.':
is_dot = 1
if is_dot != 0:
if w32.win32file.FindNextFileA(handle, find_data) == 0:
break
continue
# 检查是否是目录
attrs: ULONG = find_data.dwFileAttributes
is_dir: int = 0
if (attrs & w32.win32file.FILE_ATTRIBUTE_DIRECTORY) != 0:
is_dir = 1
if is_dir != 0:
# 递归扫描子目录
sub_dir: bytes = Utils.JoinPath(dir_path, fname)
if sub_dir is not None:
file_count = _ScanDirForPyFiles(mb, sub_dir, entries, entry_size, file_count, max_files)
stdlib.free(sub_dir)
if w32.win32file.FindNextFileA(handle, find_data) == 0:
break
continue
# 检查是否是 .vp 或 .py 文件(.vp 为 Viper 源,.py 向后兼容)
if fname_len > 3 and Utils.is_vp_or_py_at(fname, 0, fname_len) != 0 \
and file_count < max_files:
full_path: bytes = Utils.JoinPath(dir_path, fname)
if full_path is not None:
sbuf: bytes = Utils.read_file_to_buf(full_path, SRC_BUF_SIZE)
if sbuf is None:
if w32.win32file.FindNextFileA(handle, find_data) == 0:
break
continue
# 统一用 IncludesScanner.compute_file_sha1(CRLF→LF 归一化),
# 与 StubMerger.vp:1510 对同一批 App 源文件的口径一致;
# 此前用 Utils.compute_sha1(原始字节)会导致 stub/define
# 符号前缀不一致 → 链接报 undefined reference。
sha1_val: str = IncludesScanner.compute_file_sha1(mb, full_path)
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: str = 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
w32.win32file.FindClose(handle)
stdlib.free(pattern)
stdlib.free(find_data)
return file_count
# ============================================================
# _PreRegisterAppGenericTemplates - mark2 前预注册 App 源文件泛型类模板
#
# 与 Phase1 的 includes 全量预注册同理:泛型模板表(GenericClassNames 等
# 4 张 + 模板节点)是 pool 懒分配的跨文件持久表,若在 mark2 之后
# (Phase A-pre 翻译 App 文件时)才首次注册,reset_to(mark2) 会回收表与
# 节点 → 后续查重/特化遍历悬空内存崩溃。App 文件数量少,mark2 前预注册
# 产生的 AST 驻留内存可忽略。重复注册由 _register_generic_template 查重跳过。
# ============================================================
def _PreRegisterAppGenericTemplates(mb: memhub.MemBuddy | t.CPtr,
entries: SrcFileEntry | t.CPtr,
file_count: int,
entry_size: t.CSizeT) -> int:
"""mark2 前预注册 App 源文件的泛型类模板,返回处理到的泛型类数量"""
if mb is None or entries is None or file_count <= 0:
return 0
handled: int = 0
for i in range(file_count):
ea_pa: t.CUInt64T = t.CUInt64T(entries) + i * entry_size
ent_pa: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_pa, t.CPtr))
if ent_pa is None or ent_pa.Path is None or ent_pa.Sha1 is None:
continue
# 读取文件
f_pa: fileio.File | t.CPtr = fileio.File(ent_pa.Path, fileio.MODE.R)
if f_pa.closed:
continue
src_buf_pa: bytes = stdlib.malloc(SRC_BUF_SIZE)
if src_buf_pa is None:
f_pa.close()
continue
bytes_read_pa: LONG = f_pa.read_all(src_buf_pa, SRC_BUF_SIZE)
f_pa.close()
if bytes_read_pa <= 0:
stdlib.free(src_buf_pa)
continue
if bytes_read_pa < SRC_BUF_SIZE:
src_buf_pa[bytes_read_pa] = 0
else:
src_buf_pa[SRC_BUF_SIZE - 1] = 0
# 解析 AST(只解析,不完整翻译;用永久池使模板节点跨 mark2 reset 存活)
perm_pa: memhub.MemBuddy | t.CPtr = CompilerFactory.get_perm_pool()
if perm_pa is None:
perm_pa = mb
lx_pa: ast.Lexer | t.CPtr = ast.new_lexer(perm_pa)
if lx_pa is None:
stdlib.free(src_buf_pa)
continue
ast._lexer_init(lx_pa, src_buf_pa, perm_pa)
tokens_pa: ast.Token | t.CPtr = ast.tokenize(lx_pa)
tree_pa: ast.AST | t.CPtr = ast.parse_tokens(perm_pa, tokens_pa)
if tree_pa is None:
stdlib.free(src_buf_pa)
continue
# 遍历顶层 ClassDef,注册泛型模板(type_params 非 None)
ch_pa: list[ast.AST | t.CPtr] | t.CPtr = tree_pa.children
if ch_pa is not None:
cn_pa: t.CSizeT = len(ch_pa)
for ci_pa in range(cn_pa):
child_pa: ast.AST | t.CPtr = ch_pa.get(ci_pa)
if child_pa is None or child_pa.kind() != ast.ASTKind.ClassDef:
continue
cd_pa: ast.ClassDef | t.CPtr = (ast.ClassDef | t.CPtr)(child_pa)
if cd_pa is None or cd_pa.name is None:
continue
tp_pa: list[str] | t.CPtr = cd_pa.type_params
if tp_pa is None:
continue
GenApi.set_current_file_type(ent_pa.Path)
GenApi._register_generic_template(perm_pa, cd_pa, ent_pa.Sha1)
handled += 1
stdlib.free(src_buf_pa)
if handled > 0:
fb_pg3: str = VLogger.fmt_buf()
if fb_pg3 is not None:
viperlib.snprintf(fb_pg3, 1024, "mark2 前预注册 %d 个 App 泛型类模板", handled)
VLogger.info(fb_pg3, "PhaseA")
return handled
def RunMultiFileProject(mb: memhub.MemBuddy | t.CPtr,
source_dir: str, temp_dir: str, output_dir: str,
cc_cmd: str, cc_flags: str,
linker_cmd: str, linker_flags: str, linker_output: str,
includes_binary_dir: str,
includes_dir: str,
do_phase1: int, do_phase2: int,
log: VLogger.Logger | t.CPtr,
args: argparse.ParsedArgs | t.CPtr) -> int:
"""多文件项目编译
Returns: 0 成功,非 0 失败
"""
if source_dir is None:
VLogger.error("source_dir 为空", "project")
return 1
# === 栈金丝雀检查(全局变量,不影响栈布局)===
global _g_canary
_g_canary = 305419896
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "多文件项目编译: %s", source_dir)
VLogger.info(fb, "project")
# === 1. 递归扫描 source_dir 下的 .vp/.py 文件(包括子目录)===
entry_size: t.CSizeT = SrcFileEntry.__sizeof__()
entries: SrcFileEntry | t.CPtr = stdlib.malloc(MAX_SRC_FILES * entry_size)
if entries is None:
return 1
string.memset(entries, 0, MAX_SRC_FILES * entry_size)
file_count: int = _ScanDirForPyFiles(mb, source_dir, entries, entry_size, 0, MAX_SRC_FILES)
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "共 %d 个源文件", file_count)
VLogger.info(fb, "project")
if file_count == 0:
stdlib.free(entries)
return 1
# 初始化 AST 表
ast._init_tables(mb)
# mark2 之前预注册 App 源文件的泛型类模板:泛型模板表/节点是 pool 懒分配,
# 必须在 mark2 之前注册才能跨 reset_to(mark2) 存活(与 Phase1 的 includes
# 全量预注册同理),否则 Phase A-pre 首次注册后 reset 回收 → 后续查重/特化
# 遍历悬空内存崩溃(1.log FuncTest ASAN access-violation 同类根因)。
_PreRegisterAppGenericTemplates(mb, entries, file_count, entry_size)
# OOM 清理(对齐 Phase1 1b 的 mark1/reset_to 模式):
# Phase2 的 includes 预注册(Phase A-pre-inc)、App 源文件预注册(Phase A-pre)、
# App 全量翻译(Phase A)、缺失 includes 翻译(Phase B+)都会产生大量瞬时数据
# (AST/tokens/lexer/IR/Handler/dump 临时)。无回收时 98 个 includes 的 AST/IR
# 全量累积 pool → 1GB OOM(NewAstTest 依赖 ast 模块场景)。
# 此处打 mark2,各处翻译完成后 reset_to(mark2) 回收该文件瞬时数据;
# 跨文件持久表(struct/enum/docstring/CDefine/函数注册表等)已 stdlib 化
# 或落在 mark2 之前,reset 不受影响(与 Phase1 1b 同翻译路径,已验证安全)。
mark2: t.CSizeT = mb.mark()
# mark2 之后新注册的 struct/enum 条目改 stdlib 分配(跨 reset 存活,对齐 Phase1)
GenApi.set_stdlib_alloc(1)
# 确保 temp/output 目录存在
BuildPipeline.ensure_dir(temp_dir)
BuildPipeline.ensure_dir(output_dir)
# === 0.5 扫描 includes 目录填充 Sha1Store(消除 Phase1 依赖)===
# Phase2 不依赖 Phase1 的副作用:如果 Sha1Store 为空(Phase1 未运行或提前返回),
# 自行扫描 includes 目录并填充 Sha1Store,确保后续"编译缺失 includes"逻辑能正常工作
if includes_dir is not None:
if StubMergerStore.GetSha1StoreCount() == 0:
inc_scan_result: IncludesScanner.ScanResult | t.CPtr = IncludesScanner.scan_includes(mb, includes_dir)
if inc_scan_result is None:
VLogger.error("includes 扫描失败,无法填充 Sha1Store", "project")
if inc_scan_result is not None:
inc_filled: int = StubMergerStore.PopulateSha1MapStore(inc_scan_result)
fb_inc: str = VLogger.fmt_buf()
if fb_inc is not None:
viperlib.snprintf(fb_inc, 1024, "Phase2 自主填充 includes Sha1Store: %d 个", inc_filled)
VLogger.info(fb_inc, "project")
# 追加 App 源文件 SHA1 到 _sha1_map.txt(人类可读输出,程序内部不读取此文件)
# 格式: {sha1}:{rel_path}\n (保留目录结构,如 lib/core/VLogger.vp)
# 同步追加到内存存储器(AppendToSha1MapStore,供跨模块 CDefine 查找)
td_len_am: t.CSizeT = string.strlen(temp_dir)
src_dir_len_am: t.CSizeT = string.strlen(source_dir)
map_path_am: bytes = stdlib.malloc(td_len_am + 32)
if map_path_am is not None:
viperlib.snprintf(map_path_am, td_len_am + 32, "%s/_sha1_map.txt", temp_dir)
mf: fileio.File | t.CPtr = fileio.File(map_path_am, fileio.MODE.A)
line_am: bytes = stdlib.malloc(512)
# 先遍历一次填充内存存储器(不依赖文件 I/O),再写入 _sha1_map.txt
app_filled: int = 0
for i in range(file_count):
ea_am: t.CUInt64T = t.CUInt64T(entries) + i * entry_size
ent_am: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_am, t.CPtr))
if ent_am is None or ent_am.Path is None or ent_am.Sha1 is None:
continue
# 计算相对路径(去除 source_dir 前缀,保留目录结构)
p_str: str = ent_am.Path
p_len: t.CSizeT = string.strlen(p_str)
rel_path_am: str = p_str
if p_len > src_dir_len_am + 1:
if string.strncmp(p_str, source_dir, src_dir_len_am) == 0:
rel_path_am = p_str + src_dir_len_am + 1
# 同步追加到内存存储器(供跨模块 CDefine 查找)
# rel_path_am 格式如 "lib/core/StubMerger.vp",与 _sha1_map.txt 一致
if StubMergerStore.AppendToSha1MapStore(ent_am.Sha1, rel_path_am) != 0:
fb_apf: str = VLogger.fmt_buf()
if fb_apf is not None:
viperlib.snprintf(fb_apf, 1024, "AppendToSha1MapStore 失败: %s", rel_path_am)
VLogger.error(fb_apf, "project")
app_filled += 1
# 写入 _sha1_map.txt(人类可读输出)
if not mf.closed and line_am is not None:
viperlib.snprintf(line_am, 512, "%s:%s\n", ent_am.Sha1, rel_path_am)
ll_am: t.CSizeT = string.strlen(line_am)
mf.write(line_am, ll_am)
if line_am is not None:
stdlib.free(line_am)
if not mf.closed:
mf.close()
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "已追加 %d 个 App 文件到 _sha1_map.txt 和内存存储器", app_filled)
VLogger.info(fb, "project")
stdlib.free(map_path_am)
# 模块 SHA1 映射已由 PopulateSha1MapStore + AppendToSha1MapStore 统一填充到
# StubMerger 全局存储器,HandlesExprCall._lookup_module_sha1 直接查全局存储器,
# 不再需要构建独立的 SHA1 映射表副本。
# === Phase A-pre-vpi: 预生成所有 App 源文件的 .vpi stub ===
#
# 与 Phase1 1b-pre 同理:依赖模块的 .vpi 必须在依赖方翻译前可用。
# Phase A 主循环按扫描顺序翻译(非拓扑),若 StubMergerCollect/IR/Store
# 先于 __init__.vp 翻译,__init__.vpi 尚未生成 → LoadDepVpiIntoModule
# 找不到 vpi → 跨模块全局变量 declare 未注入 → ExprGenAttr fallback
# 以 i8* 创建带前缀 global(@b0b568._sha1_store_count = external global i8*)
# → llc "icmp slt 类型不匹配"(22fc/2e23/daad 根因,见 vpi 时间戳顺序)。
# 此处先为所有 App 源文件预生成 vpi,保证 Phase A 翻译任意模块时其依赖
# vpi 已存在(App 源文件自身的 vpi 随后由 TranslateFileGetTrans 覆盖重写,
# 内容一致,此处仅保证"先于依赖方")。
td_len_vpa: t.CSizeT = string.strlen(temp_dir)
PYI_BUF_SIZE_P2: t.CSizeT = 262144
vpi_failed_p2: int = 0
for i_vp2 in range(file_count):
ea_vp2: t.CUInt64T = t.CUInt64T(entries) + i_vp2 * entry_size
ent_vp2: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_vp2, t.CPtr))
if ent_vp2 is None or ent_vp2.Path is None or ent_vp2.Sha1 is None:
vpi_failed_p2 += 1
continue
# vpi 已存在则跳过(sha1 基于内容,同名 vpi 必然有效;跨次编译缓存)
vpi_path_vp2: str = StubMergerCollect._sliced_path(temp_dir, td_len_vpa, ent_vp2.Sha1, "vpi")
if vpi_path_vp2 is None:
vpi_failed_p2 += 1
continue
vf_vp2: fileio.File | t.CPtr = fileio.File(vpi_path_vp2, fileio.MODE.R)
if not vf_vp2.closed:
vf_vp2.close()
stdlib.free(vpi_path_vp2)
continue
stdlib.free(vpi_path_vp2)
# 读取源文件 → 解析 AST → 生成 vpi
# (rel_path 传 None:与 TranslateFileGetTrans 后续覆盖写的内容一致)
f_vp2: fileio.File | t.CPtr = fileio.File(ent_vp2.Path, fileio.MODE.R)
if f_vp2.closed:
vpi_failed_p2 += 1
continue
src_buf_vp2: bytes = stdlib.malloc(SRC_BUF_SIZE)
if src_buf_vp2 is None:
f_vp2.close()
vpi_failed_p2 += 1
continue
bytes_read_vp2: LONG = f_vp2.read_all(src_buf_vp2, SRC_BUF_SIZE)
f_vp2.close()
if bytes_read_vp2 <= 0:
stdlib.free(src_buf_vp2)
vpi_failed_p2 += 1
continue
if bytes_read_vp2 < SRC_BUF_SIZE:
src_buf_vp2[bytes_read_vp2] = 0
else:
src_buf_vp2[SRC_BUF_SIZE - 1] = 0
lx_vp2: ast.Lexer | t.CPtr = ast.new_lexer(mb)
if lx_vp2 is None:
stdlib.free(src_buf_vp2)
vpi_failed_p2 += 1
continue
ast._lexer_init(lx_vp2, src_buf_vp2, mb)
tokens_vp2: ast.Token | t.CPtr = ast.tokenize(lx_vp2)
tree_vp2: ast.AST | t.CPtr = ast.parse_tokens(mb, tokens_vp2)
if tree_vp2 is None:
stdlib.free(src_buf_vp2)
vpi_failed_p2 += 1
continue
pyi_buf_vp2: bytes = stdlib.malloc(PYI_BUF_SIZE_P2)
if pyi_buf_vp2 is not None:
pyi_pos_vp2: t.CSizeT = StubConverter._GeneratePyiFromAst(
mb, tree_vp2, None, pyi_buf_vp2, PYI_BUF_SIZE_P2)
if pyi_pos_vp2 > 0:
pyi_path_vp2: str = StubMergerCollect._sliced_path(temp_dir, td_len_vpa, ent_vp2.Sha1, "vpi")
if pyi_path_vp2 is not None:
pf_vp2: fileio.File | t.CPtr = fileio.File(pyi_path_vp2, fileio.MODE.W)
if not pf_vp2.closed:
pf_vp2.write(pyi_buf_vp2, pyi_pos_vp2)
pf_vp2.close()
stdlib.free(pyi_path_vp2)
stdlib.free(pyi_buf_vp2)
else:
vpi_failed_p2 += 1
stdlib.free(src_buf_vp2)
# OOM 清理:回收本文件 AST(vpi 已写盘;mark2 后持久数据已 stdlib 化)
mb.reset_to(mark2)
if vpi_failed_p2 > 0:
fb_vp2: str = VLogger.fmt_buf()
if fb_vp2 is not None:
viperlib.snprintf(fb_vp2, 1024, "Phase A-pre-vpi 有 %d 个文件 vpi 预生成失败", vpi_failed_p2)
VLogger.error(fb_vp2, "PhaseA")
return 1
# === 2. Phase A: 为每个文件生成 stub + text ===
# app_deps_buf: 收集所有 App 源文件的直接依赖模块名(空格分隔)
# 供 Phase B+ 的 _BuildReachableSha1Set 作为初始 worklist,实现按需编译
# 声明在 if 块外,确保 Phase B+ 和清理代码能访问
# 65536 与 StubMerger.REACH_WORKLIST_BUF 对齐:全量 App imports 拼接
# 超 8192(追加82 同源溢出),8192 会静默丢依赖致 Phase B+ 可达集合不全
APP_DEPS_BUF_SIZE: t.CSizeT = 65536
app_deps_buf: bytes = None
app_deps_len: t.CSizeT = 0
if do_phase1 != 0:
if log is not None:
log.banner("Phase A: 生成 stub + text")
# === Phase A-pre-inc: includes 预注册 struct + 依赖填充 ===
# 必须在 App 源文件预注册之前执行:App 源文件(strict_mode=1)可能继承
# 或引用 includes 中的 struct(如 ast.AST)。若 includes struct 未注册,
# App 源文件翻译会失败 → .obj 缺失 → undefined reference。
# 同时填充 PopulateIncludesDeps,供 Phase B+ 的 _BuildReachableSha1Set 使用。
# 多遍扫描:解决跨文件继承问题(如 Constant(AST) 在 AST 未注册时被跳过)。
if includes_dir is not None and includes_binary_dir is not None:
store_count_inc_pre: int = StubMergerStore.GetSha1StoreCount()
if store_count_inc_pre > 0:
store_sha1_inc_pre: bytes = StubMergerStore.GetSha1StoreArrPtr()
store_rel_inc_pre: bytes = StubMergerStore.GetSha1StoreRelArrPtr()
store_incdir_inc_pre: bytes = StubMergerStore.GetSha1StoreIncdirArrPtr()
deps_filled_pre: int = 0
PHASE_A_INC_MAX_PASSES: t.CInt = 3
# [修复] 记录预注册失败的 includes:多遍后仍失败 → 报错终止编译。
# 不允许静默降级——静默失败会让 App 源文件的跨模块类构造退化为
# 普通函数调用(返回类型默认推断成 i32),调用点报误导性的
# "跨模块调用 'parser.add_argument' 找不到模块 SHA1"
# (如 argparse.ArgumentParser 类未注册导致 parser 变量类型崩坏)。
# declare_only=1 时 import 的 SHA1 检查跳过(容忍依赖未注册),
# 因此预注册失败只可能是文件自身无法翻译(语法/解析/编译错误)。
# 注意:Sha1Store 因 AppendToSha1MapStore 追加了 App 源文件 SHA1
# (供跨模块 CDefine 查找),此处必须跳过 App 文件——App 文件的
# 预注册由 Phase A-pre(遍历 entries,declare_only=1)负责。
# App 文件是严格模式(仅本地+import 可见)且预注册阶段依赖未就绪,
# 若在此预注册必然失败,误报"预注册失败: main.vp"。
inc_fail_flags: bytes = stdlib.malloc(store_count_inc_pre)
if inc_fail_flags is not None:
string.memset(inc_fail_flags, 1, store_count_inc_pre)
for pass_inc_i in range(PHASE_A_INC_MAX_PASSES):
struct_count_before_inc: int = GenApi.get_struct_count()
for si_inc in range(store_count_inc_pre):
inc_sha1_pre: str = store_sha1_inc_pre + t.CSizeT(si_inc) * 17
inc_rel_pre: str = store_rel_inc_pre + t.CSizeT(si_inc) * StubMerger.MAX_REL_PATH_LEN
if inc_rel_pre is None or inc_rel_pre[0] == '\0':
continue
# 跳过 App 源文件(main.vp 等):由 Phase A-pre 预注册,不算失败
is_app_file_inc: int = 0
for i_app_inc in range(file_count):
ea_app_inc: t.CUInt64T = t.CUInt64T(entries) + i_app_inc * entry_size
ent_app_inc: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_app_inc, t.CPtr))
if ent_app_inc is not None and ent_app_inc.Sha1 is not None \
and string.strcmp(ent_app_inc.Sha1, inc_sha1_pre) == 0:
is_app_file_inc = 1
break
if is_app_file_inc != 0 and inc_fail_flags is not None:
inc_fail_flags[si_inc] = 0
continue
rel_len_pre: t.CSizeT = string.strlen(inc_rel_pre)
# 所属 includes 根目录(多目录逐条取自 IncDir 数组,空串回退 includes_dir)
incdir_pre: str = store_incdir_inc_pre + t.CSizeT(si_inc) * StubMerger.MAX_INC_DIR_LEN
if incdir_pre is None or incdir_pre[0] == '\0':
incdir_pre = includes_dir
incdir_len_pre: t.CSizeT = string.strlen(incdir_pre)
full_path_pre: bytes = stdlib.malloc(incdir_len_pre + rel_len_pre + 2)
if full_path_pre is None:
continue
viperlib.snprintf(full_path_pre, incdir_len_pre + rel_len_pre + 2,
"%s/%s", incdir_pre, inc_rel_pre)
pkg_inc_pre: str = GenApi.compute_package_from_relpath(mb, inc_rel_pre)
tr_inc_pre: HandlesTranslator.Translator | t.CPtr = BuildPipeline.TranslateFileGetTrans(
mb, full_path_pre, inc_sha1_pre, pkg_inc_pre, 1)
stdlib.free(full_path_pre)
if tr_inc_pre is not None:
# 预注册成功:清除失败标记
if inc_fail_flags is not None:
inc_fail_flags[si_inc] = 0
# 依赖填充只在第一遍执行(PopulateIncludesDeps 内部有去重)
# 注意:必须在 reset_to(mark2) 之前调用——内部深拷贝到 stdlib
if pass_inc_i == 0 and tr_inc_pre._imported_modules is not None \
and StubMergerStore.PopulateIncludesDeps(
inc_sha1_pre, tr_inc_pre._imported_modules) == 0:
deps_filled_pre = deps_filled_pre + 1
# OOM 清理:回收本 includes 文件预注册翻译的瞬时数据
# (AST/tokens/lexer/IR/Handler);struct 注册表条目 stdlib 分配不受影响
mb.reset_to(mark2)
struct_count_after_inc: int = GenApi.get_struct_count()
# 收敛检查:本遍没有新结构体注册 → 所有类已注册
if struct_count_after_inc == struct_count_before_inc:
break
# 多遍后仍预注册失败的 includes → 报错终止(不允许降级)
if inc_fail_flags is not None:
for si_inc_f in range(store_count_inc_pre):
if inc_fail_flags[si_inc_f] != 0:
rel_f: str = store_rel_inc_pre + t.CSizeT(si_inc_f) * StubMerger.MAX_REL_PATH_LEN
fb_inc_f: str = VLogger.fmt_buf()
if fb_inc_f is not None:
viperlib.snprintf(fb_inc_f, 1024,
"includes 预注册失败: %s(文件无法翻译,可能是语法/解析/编译错误),终止编译",
rel_f)
VLogger.error(fb_inc_f, "prePhaseA")
sys.exit(1)
stdlib.free(inc_fail_flags)
fb_inc_pre: str = VLogger.fmt_buf()
if fb_inc_pre is not None:
viperlib.snprintf(fb_inc_pre, 1024, "includes 预注册 + 依赖填充: %d/%d", deps_filled_pre, store_count_inc_pre)
VLogger.info(fb_inc_pre, "prePhaseA")
# === Phase A-pre: 预注册所有源文件的 struct/enum/union ===
# 解决循环引用问题:circ_a 翻译时需要知道 circ_b.ClassB 的 struct 定义
# 仅注册 struct/enum/union(declare_only=1),不翻译方法体
# 多遍扫描:解决跨文件继承的字母序问题(如 HandlesAnnAssign.vp/.py 在 HandlesBase.vp/.py 之前,
# AnnAssignHandle 继承 Mixin 时 Mixin 未注册)。每遍注册新类后,下一遍子类可继承。
VLogger.info("预注册 struct/enum/union...", "PhaseA")
src_dir_len_pre: t.CSizeT = string.strlen(source_dir)
PHASE_A_PRE_MAX_PASSES: t.CInt = 3
for pass_i in range(PHASE_A_PRE_MAX_PASSES):
struct_count_before: int = GenApi.get_struct_count()
for i in range(file_count):
ea_pre: t.CUInt64T = t.CUInt64T(entries) + i * entry_size
ent_pre: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_pre, t.CPtr))
if ent_pre is None or ent_pre.Path is None:
continue
pkg_pre: str = None
if string.strlen(ent_pre.Path) > src_dir_len_pre + 1:
rel_path_pre: str = ent_pre.Path + src_dir_len_pre + 1
pkg_pre = GenApi.compute_package_from_relpath(mb, rel_path_pre)
tr_pre: HandlesTranslator.Translator | t.CPtr = BuildPipeline.TranslateFileGetTrans(mb, ent_pre.Path, ent_pre.Sha1, pkg_pre, 1)
if tr_pre is None:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "预注册失败: %s", ent_pre.Path)
VLogger.error(fb, "PhaseA")
# OOM 清理:回收本 App 文件预注册翻译的瞬时数据
mb.reset_to(mark2)
struct_count_after: int = GenApi.get_struct_count()
# 收敛检查:本遍没有新结构体注册 → 所有类已注册
if struct_count_after == struct_count_before:
break
VLogger.info("预注册完成", "PhaseA")
PHASE_A_IR_SIZE: t.CSizeT = 1048576
td_len_pa: t.CSizeT = string.strlen(temp_dir)
# 设置 temp 目录路径,使 HandlesExprCall 能读取依赖模块的 text.ll
# 用于跨模块方法返回类型查找(避免指针返回值被截断为 i32)
GenApi.SetTempDir(temp_dir)
# app_deps_buf: 收集所有 App 源文件的直接依赖模块名(空格分隔)
# 供 Phase B+ 的 _BuildReachableSha1Set 作为初始 worklist 使用
# 声明已在 if 块外,此处仅赋值
app_deps_buf = stdlib.malloc(APP_DEPS_BUF_SIZE)
app_deps_len = 0
if app_deps_buf is not None:
app_deps_buf[0] = '\0'
# Phase A 多遍重试:同项目多源文件时,main.vp 可能先于其 import 的
# mbuddytest.vp 翻译(文件系统顺序),跨模块函数返回类型精确解析会因
# 依赖 text.ll 未生成而标记 _deps_pending=1(见 CallGenMixin 跨模块
# 返回类型解析段)。下遍重译即可命中(依赖已生成 text.ll)。
# 每遍至少完成一个文件,最多 file_count 遍收敛。
done_flags: bytes = stdlib.malloc(file_count)
if done_flags is None:
VLogger.error("Phase A done_flags 分配失败", "PhaseA")
stdlib.free(entries)
return 1
string.memset(done_flags, 0, file_count)
PHASE_A_MAX_PASSES: t.CInt = 32
for pass_a in range(PHASE_A_MAX_PASSES):
completed_this_pass: int = 0
for i in range(file_count):
if done_flags[i] != 0:
continue
ea: t.CUInt64T = t.CUInt64T(entries) + i * entry_size
ent: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea, t.CPtr))
if ent is None or ent.Path is None:
done_flags[i] = 1
completed_this_pass += 1
continue
# 计算源文件的包名(相对 source_dir 的目录部分)
src_dir_len_pa: t.CSizeT = string.strlen(source_dir)
pkg_pa: str = None
if string.strlen(ent.Path) > src_dir_len_pa + 1:
rel_path_pa: str = ent.Path + src_dir_len_pa + 1
pkg_pa = GenApi.compute_package_from_relpath(mb, rel_path_pa)
tr_a: HandlesTranslator.Translator | t.CPtr = BuildPipeline.TranslateFileGetTrans(mb, ent.Path, ent.Sha1, pkg_pa, 0, temp_dir)
if tr_a is None:
# 翻译失败(语法/解析错误):记录待下遍重试,收敛检查统一报错
# OOM 清理:失败翻译也已产生 AST/tokens 瞬时数据,立即回收
mb.reset_to(mark2)
continue
# 收集 _imported_modules 到 app_deps_buf(供 Phase B+ 按需编译)。
# 依赖在翻译早期已填充,即使 _deps_pending 也收集。
if tr_a._imported_modules is not None and app_deps_buf is not None:
# 逐模块去重拼接:多个 App 文件 import 同一模块(builtins/
# _list/_dict/stdlib/t/c 等公共依赖)会让 app_deps 无去重
# 膨胀到接近 64KB,_BuildReachableSha1Set 的初始 worklist
# 即满 → BFS 展开必报"依赖 worklist 溢出"。
im_len_a: t.CSizeT = string.strlen(tr_a._imported_modules)
im_pos_a: t.CSizeT = 0
while im_pos_a < im_len_a:
im_start_a: t.CSizeT = im_pos_a
while im_pos_a < im_len_a and tr_a._imported_modules[im_pos_a] != ' ':
im_pos_a += 1
im_mod_len: t.CSizeT = im_pos_a - im_start_a
if im_pos_a < im_len_a:
im_pos_a += 1
if im_mod_len == 0:
continue
im_mod_buf: bytes = stdlib.malloc(im_mod_len + 1)
if im_mod_buf is None:
continue
string.strncpy(im_mod_buf, tr_a._imported_modules + im_start_a, im_mod_len)
im_mod_buf[im_mod_len] = '\0'
# 已在 app_deps 中则跳过(词边界精确匹配)
if app_deps_len > 0 and GenApi.is_module_imported(app_deps_buf, im_mod_buf) != 0:
stdlib.free(im_mod_buf)
continue
if app_deps_len + im_mod_len + 2 < APP_DEPS_BUF_SIZE:
if app_deps_len > 0:
app_deps_buf[app_deps_len] = ' '
app_deps_len = app_deps_len + 1
string.strcpy(app_deps_buf + app_deps_len, im_mod_buf)
app_deps_len = app_deps_len + im_mod_len
app_deps_buf[app_deps_len] = '\0'
stdlib.free(im_mod_buf)
if tr_a._deps_pending != 0:
# 依赖模块 text.ll 未生成(同项目文件顺序),不写盘,下遍重试
fb_dp: str = VLogger.fmt_buf()
if fb_dp is not None:
viperlib.snprintf(fb_dp, 1024,
"Phase A 依赖未就绪,下遍重试: %s", ent.Path)
# --nowarn-register 亦抑制此类 Phase A 依赖重试 WARN(刷屏收敛)
if VLogger.nowarn_register() == 0:
VLogger.warning(fb_dp, "PhaseA-DBG")
# 回滚本文件已注册类的 MethodsEmitted:第 1 遍 _translate_oop_methods
# 发射方法体前已置 MethodsEmitted=1(防重入),但方法体内跨模块调用
# 依赖 text.ll 未生成触发 _deps_pending → 本遍作废;若不回滚,
# 下遍重译 existing 分支被 MethodsEmitted==1 防重复检查拦截 → 类方法
# 永不发射 → 链接 undefined symbol(追加87 根因)。
# 必须先于 mb.reset_to(mark2)(MethodsEmitted 在 perm_pool 上跨 reset 存活)。
if ent.Sha1 is not None:
GenApi.reset_module_methods_emitted(ent.Sha1)
# OOM 清理:本文件翻译已产生瞬时数据(下遍会重新翻译),立即回收
mb.reset_to(mark2)
continue
# dump stub IR (declarations only)
stub_buf_a: bytes = stdlib.malloc(PHASE_A_IR_SIZE)
if stub_buf_a is None:
VLogger.error("stub 缓冲区分配失败", "PhaseA")
tr_a.dump_ir(stub_buf_a, PHASE_A_IR_SIZE, llvmlite.OUTPUT_STUB)
stub_len_a: t.CSizeT = string.strlen(stub_buf_a)
# save stub.ll (切片路径: temp_dir/{sha1前缀}/{sha1}.stub.ll)
stub_path_a: str = StubMergerCollect._sliced_path(temp_dir, td_len_pa, ent.Sha1, "stub.ll")
if stub_path_a is not None:
sf_a: fileio.File | t.CPtr = fileio.File(stub_path_a, fileio.MODE.W)
if not sf_a.closed:
sf_a.write(stub_buf_a, stub_len_a)
sf_a.close()
stdlib.free(stub_path_a)
stdlib.free(stub_buf_a)
# dump text IR (definitions only)
text_buf_a: bytes = stdlib.malloc(PHASE_A_IR_SIZE)
if text_buf_a is None:
VLogger.error("text 缓冲区分配失败", "PhaseA")
tr_a.dump_ir(text_buf_a, PHASE_A_IR_SIZE, llvmlite.OUTPUT_TEXT)
text_len_a: t.CSizeT = string.strlen(text_buf_a)
# save text.ll (切片路径)
text_path_a: str = StubMergerCollect._sliced_path(temp_dir, td_len_pa, ent.Sha1, "text.ll")
if text_path_a is not None:
tf_a: fileio.File | t.CPtr = fileio.File(text_path_a, fileio.MODE.W)
if not tf_a.closed:
tf_a.write(text_buf_a, text_len_a)
tf_a.close()
stdlib.free(text_path_a)
stdlib.free(text_buf_a)
# save dependencies (_imported_modules) for Phase B (切片路径)
deps_path_a: str = StubMergerCollect._sliced_path(temp_dir, td_len_pa, ent.Sha1, "deps.txt")
if deps_path_a is not None:
df_a: fileio.File | t.CPtr = fileio.File(deps_path_a, fileio.MODE.W)
if not df_a.closed:
if tr_a._imported_modules is not None:
dl_a: t.CSizeT = string.strlen(tr_a._imported_modules)
df_a.write(tr_a._imported_modules, dl_a)
df_a.close()
stdlib.free(deps_path_a)
done_flags[i] = 1
completed_this_pass += 1
# OOM 清理:stub/text/deps 已写盘,回收本文件翻译瞬时数据
# (AST/tokens/lexer/IR/Handler);app_deps_buf 已深拷贝依赖,不受影响
mb.reset_to(mark2)
# 收敛检查:所有文件完成则结束
all_done: int = 1
for i_ck in range(file_count):
if done_flags[i_ck] == 0:
all_done = 0
break
if all_done != 0:
break
if completed_this_pass == 0:
# 无进展:剩余文件无法完成(真实编译错误),终止并列出未完成文件
for i_err in range(file_count):
if done_flags[i_err] == 0:
ea_err: t.CUInt64T = t.CUInt64T(entries) + i_err * entry_size
ent_err: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea_err, t.CPtr))
if ent_err is not None and ent_err.Path is not None:
fb_err: str = VLogger.fmt_buf()
if fb_err is not None:
viperlib.snprintf(fb_err, 1024,
"Phase A 无法完成翻译: %s(请检查依赖模块是否已编译/注册)", ent_err.Path)
VLogger.error(fb_err, "PhaseA")
stdlib.free(done_flags)
VLogger.error("Phase A 多遍收敛失败,终止编译", "PhaseA")
stdlib.free(entries)
return 1
stdlib.free(done_flags)
if do_phase2 == 0:
VLogger.info("Phase A 完成(仅 stub 生成)", "project")
if app_deps_buf is not None:
stdlib.free(app_deps_buf)
stdlib.free(entries)
return 0
# === 3. Phase B: 编译每个文件为 .obj ===
if do_phase2 != 0:
if log is not None:
log.banner("Phase B: 编译 .obj")
# 收集 .obj 路径(增大到 64KB,避免 includes .obj 路径溢出导致链接丢失符号)
# reachable_set/reachable_count: Phase B+ 填充,Phase C 用于按需链接
reachable_set: bytes = None
reachable_count: int = 0
OBJ_PATHS_SIZE: t.CSizeT = 65536
obj_paths: bytes = stdlib.malloc(OBJ_PATHS_SIZE)
if obj_paths is None:
return 1
obj_paths[0] = '\0'
obj_pos: t.CSizeT = 0
# main_obj_path: 存放定义了用户 main 的 .obj 路径(链接时放在最前面,
# 避免 --allow-multiple-definition 选择了其他模块的 wrapper main)
main_obj_path: bytes = stdlib.malloc(512)
if main_obj_path is None:
return 1
main_obj_path[0] = '\0'
compiled_count: int = 0
# 组合 IR 缓冲区大小(4MB,足够容纳本地 stub + 所有依赖 stub + 本地 text)
COMBINED_IR_SIZE: t.CSizeT = 4194304
# llc 任务表(stdlib.malloc,跨线程存活):连续 LlcTask 记录数组,
# 循环内只写 .ll + 构造任务,循环后 run_llc_parallel 分批并行跑 llc
llc_tasks: bytes = stdlib.malloc(t.CSizeT(file_count) * BuildPipeline.LlcTask.__sizeof__())
if llc_tasks is None:
VLogger.error("llc 任务表分配失败", "PhaseB")
return 1
llc_task_count: int = 0
for i in range(file_count):
ea: t.CUInt64T = t.CUInt64T(entries) + i * entry_size
ent: SrcFileEntry | t.CPtr = (SrcFileEntry | t.CPtr)(t.CVoid(ea, t.CPtr))
if ent is None or ent.Path is None or ent.Sha1 is None:
continue
# 组合本地 stub + 所有依赖 stub + 本地 text → 完整 IR
combined_ir: bytes = stdlib.malloc(COMBINED_IR_SIZE)
if combined_ir is None:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "combined_ir 分配失败: %s", ent.Path)
VLogger.error(fb, "PhaseB")
continue
combined_len: t.CSizeT = StubMergerBuild.BuildCombinedIR(temp_dir, ent.Sha1, combined_ir, COMBINED_IR_SIZE)
if combined_len == 0:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "BuildCombinedIR 失败: %s", ent.Path)
VLogger.error(fb, "PhaseB")
stdlib.free(combined_ir)
continue
# 写 .ll(llc 线程只读此文件);llc 本身改为循环后批量并行
ret_w: int = BuildPipeline.write_ir_to_file(combined_ir, combined_len, temp_dir, ent.Sha1)
stdlib.free(combined_ir)
if ret_w != 0:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "写 .ll 失败: %s", ent.Path)
VLogger.error(fb, "PhaseB")
sys.exit(1)
# 构造 llc 任务(命令串 stdlib.malloc,线程运行期不被 reset 回收)
llc_ir_path: str = BuildPipeline.build_sliced_path(temp_dir, ent.Sha1, "ll")
if llc_ir_path is None:
VLogger.error("llc 输入路径构造失败", "PhaseB")
sys.exit(1)
tslot: BuildPipeline.LlcTask | t.CPtr = (BuildPipeline.LlcTask | t.CPtr)(t.CVoid(
t.CUInt64T(llc_tasks) + t.CSizeT(llc_task_count) * BuildPipeline.LlcTask.__sizeof__(), t.CPtr))
if BuildPipeline.init_llc_task(tslot, llc_ir_path, output_dir, ent.Sha1,
cc_cmd, cc_flags) != 0:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "llc 任务构造失败: %s", ent.Path)
VLogger.error(fb, "PhaseB")
sys.exit(1)
llc_task_count += 1
compiled_count += 1
# 显示已提交 llc 的文件
fb_comp: str = VLogger.fmt_buf()
if fb_comp is not None:
viperlib.snprintf(fb_comp, 1024, "[%d/%d] llc 已提交: %s", i + 1, file_count, ent.Path)
VLogger.info(fb_comp, "PhaseB")
# 构造 .obj 路径,检测是否是 main 模块(test_main.vp/.py 或 main.vp/.py)
od_len: t.CSizeT = string.strlen(output_dir)
is_main_mod: int = 0
if string.strstr(ent.Path, "test_main.vp") is not None or string.strstr(ent.Path, "test_main.py") is not None:
is_main_mod = 1
elif string.strstr(ent.Path, "main.vp") is not None or string.strstr(ent.Path, "main.py") is not None:
is_main_mod = 1
# 切片路径: output_dir/{sha1前缀}/{sha1}.obj
obj_path_sliced: str = StubMergerCollect._sliced_path(output_dir, od_len, ent.Sha1, "obj")
if obj_path_sliced is not None:
op_sliced_len: t.CSizeT = string.strlen(obj_path_sliced)
if is_main_mod != 0:
# main 模块: 复制到 main_obj_path(确保链接时 main 在最前)
if op_sliced_len < 512:
string.strcpy(main_obj_path, obj_path_sliced)
else:
VLogger.error("main_obj_path 缓冲区不足,终止编译", "PhaseB")
else:
# 其他模块: 追加到 obj_paths
if obj_pos + op_sliced_len + 2 < OBJ_PATHS_SIZE:
if obj_pos > 0:
obj_paths[obj_pos] = ' '
obj_pos += 1
string.strcpy(obj_paths + obj_pos, obj_path_sliced)
obj_pos += op_sliced_len
obj_paths[obj_pos] = '\0'
else:
VLogger.error(".obj 路径缓冲区不足", "PhaseB")
stdlib.free(obj_path_sliced)
# 并行执行全部 llc(分批 LLC_MAX_THREADS 线程)
ret_p: int = BuildPipeline.run_llc_parallel(llc_tasks, llc_task_count,
BuildPipeline.LLC_MAX_THREADS)
if ret_p != 0:
VLogger.error("llc 编译失败,终止编译", "PhaseB")
sys.exit(1)
# 释放 llc 任务(命令串 + 记录)
ti: int = 0
while ti < llc_task_count:
tfree: BuildPipeline.LlcTask | t.CPtr = (BuildPipeline.LlcTask | t.CPtr)(t.CVoid(
t.CUInt64T(llc_tasks) + t.CSizeT(ti) * BuildPipeline.LlcTask.__sizeof__(), t.CPtr))
BuildPipeline.free_llc_task(tfree)
ti += 1
stdlib.free(llc_tasks)
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "编译完成: %d/%d", compiled_count, file_count)
VLogger.success(fb, "PhaseB")
if compiled_count == 0:
VLogger.error("无成功编译的文件", "PhaseB")
stdlib.free(entries)
return 1
# === 3.5 编译缺失的 includes 文件(按依赖图按需编译)===
# 只编译被 App 源文件直接/间接引用的 includes 文件(可达集合),
# 而非遍历整个 includes 目录。通过 _BuildReachableSha1Set 构建依赖图。
if includes_dir is not None and includes_binary_dir is not None:
inc_compiled: int = 0
td_len_mi: t.CSizeT = string.strlen(temp_dir)
# 诊断日志:确认 Phase B+ 入口和 Sha1Store 状态
fb_pbp_enter: str = VLogger.fmt_buf()
if fb_pbp_enter is not None:
viperlib.snprintf(fb_pbp_enter, 1024,
"PhaseB+ 入口: store_count=%d includes_binary_dir=%s",
StubMergerStore.GetSha1StoreCount(), includes_binary_dir)
VLogger.info(fb_pbp_enter, "PhaseB+")
# 创建 includes_binary_dir 目录(确保 .obj 能写入和 collect_obj_files 能扫描)
BuildPipeline.ensure_dir(includes_binary_dir)
# declare_only 预处理已移至 Phase A 之前(Phase A-pre-inc),
# 确保 Phase A 翻译 App 源文件时 includes struct 已注册。
# PopulateIncludesDeps 已在 Phase A-pre-inc 中填充。
# 构建可达 SHA1 集合(依赖图遍历)
# app_deps_buf 为 None 时回退到扫描 source_dir(非递归)
REACHABLE_SET_SIZE: t.CSizeT = t.CSizeT(StubMerger.MAX_INCLUDES_SHA1) * 17
reachable_set: bytes = stdlib.malloc(REACHABLE_SET_SIZE)
reachable_count: int = 0
if reachable_set is not None:
string.memset(reachable_set, 0, REACHABLE_SET_SIZE)
reachable_count = StubMergerBuild._BuildReachableSha1Set(
mb, source_dir, temp_dir, reachable_set, app_deps_buf)
fb_reach: str = VLogger.fmt_buf()
if fb_reach is not None:
viperlib.snprintf(fb_reach, 1024, "可达 includes SHA1: %d 个", reachable_count)
VLogger.info(fb_reach, "PhaseB+")
# fast-fail: 可达集合构建失败(依赖未找到),永不回退,直接终止
if reachable_count < 0:
VLogger.error("可达集合构建失败(fast-fail),终止编译", "PhaseB+")
return 1
else:
VLogger.error("reachable_set 分配失败,终止编译", "PhaseB+")
return 1
# 释放 app_deps_buf(已构建完 reachable_set,不再需要)
if app_deps_buf is not None:
stdlib.free(app_deps_buf)
app_deps_buf = None
# 从全局存储器获取 SHA1/模块名 数组:复制局部副本(bytes),
# 避免 store 联合返回值(bytes|t.CPtr)跨函数传参/索引的地址异常
# (与 _BuildReachableSha1Set 的 BFS 一致)。
# rel/incdir 仅本函数内按索引读取(不跨函数传参),保留 store 指针。
td_len_pb: t.CSizeT = string.strlen(temp_dir)
store_sha1_arr_mi: bytes = stdlib.malloc(StubMerger.MAX_INCLUDES * 17)
store_mod_arr_mi: bytes = stdlib.malloc(StubMerger.MAX_INCLUDES * 64)
store_count_mi: int = 0
if store_sha1_arr_mi is not None and store_mod_arr_mi is not None:
store_count_mi = StubMergerIR._BuildIncludesSha1Map(
temp_dir, td_len_pb, store_sha1_arr_mi, store_mod_arr_mi)
store_rel_arr_mi: bytes = StubMergerStore.GetSha1StoreRelArrPtr()
store_incdir_arr_mi: bytes = StubMergerStore.GetSha1StoreIncdirArrPtr()
if store_count_mi > 0:
# 拓扑排序可达 includes:被依赖模块先完整翻译。跨模块 vtable 生成
# 需要依赖模块的函数 TypeStr 已注册到全局表(如 astaux 继承 base 的 AST)
topo_idx: t.CSizeT | t.CPtr = (t.CSizeT | t.CPtr)(stdlib.malloc(8 * (store_count_mi + 1)))
topo_count: int = 0
if topo_idx is None:
VLogger.error("拓扑排序索引分配失败,终止编译", "PhaseB+")
topo_count = 0
else:
topo_count = StubMergerTopo.BuildTopoSortedIndices(
store_count_mi, store_sha1_arr_mi, store_mod_arr_mi,
reachable_set, reachable_count, topo_idx)
for si_mi in range(topo_count):
# 当前条目(拓扑序索引,仅含可达模块)
ti: int = topo_idx[si_mi]
# 获取当前条目的 SHA1 和 rel_path
inc_sha1_mi: str = store_sha1_arr_mi + t.CSizeT(ti) * 17
inc_rel_mi: str = store_rel_arr_mi + t.CSizeT(ti) * StubMerger.MAX_REL_PATH_LEN
# 检查 .obj 是否已存在于 includes.binary(切片子目录)
ibd_len_mi: t.CSizeT = string.strlen(includes_binary_dir)
check_pat_mi: str = StubMergerCollect._sliced_path(includes_binary_dir, ibd_len_mi, inc_sha1_mi, "obj")
check_fd_mi: w32.win32file.WIN32_FIND_DATAA | t.CPtr = stdlib.malloc(w32.win32file.WIN32_FIND_DATAA.__sizeof__())
obj_exists_mi: int = 0
if check_pat_mi is not None and check_fd_mi is not None:
string.memset(check_fd_mi, 0, w32.win32file.WIN32_FIND_DATAA.__sizeof__())
check_h_mi: w32.win32base.HANDLE = w32.win32file.FindFirstFileA(check_pat_mi, check_fd_mi)
if check_h_mi != w32.win32base.INVALID_HANDLE_VALUE:
w32.win32file.FindClose(check_h_mi)
obj_exists_mi = 1
if obj_exists_mi != 0:
if check_pat_mi is not None:
stdlib.free(check_pat_mi)
stdlib.free(check_fd_mi)
continue
# .obj 不存在,需要编译
# 构造源文件路径: {所属 includes 根目录}/{rel_path}(多目录逐条取 IncDir,空串回退 includes_dir)
rel_path_len_mi: t.CSizeT = string.strlen(inc_rel_mi)
incdir_mi: str = store_incdir_arr_mi + t.CSizeT(ti) * StubMerger.MAX_INC_DIR_LEN
if incdir_mi is None or incdir_mi[0] == '\0':
incdir_mi = includes_dir
inc_dir_len_mi: t.CSizeT = string.strlen(incdir_mi)
src_fp_mi: bytes = stdlib.malloc(inc_dir_len_mi + 1 + rel_path_len_mi + 1)
if src_fp_mi is None:
stdlib.free(check_pat_mi)
if check_fd_mi is not None:
stdlib.free(check_fd_mi)
VLogger.error("includes 源路径分配失败,终止编译", "PhaseB+")
viperlib.snprintf(src_fp_mi, inc_dir_len_mi + 1 + rel_path_len_mi + 1,
"%s/%s", incdir_mi, inc_rel_mi)
# 检查是否为声明文件(只有 declare 没有实质 define)
# 判断方法: text.ll 中若有混淆函数 define(含 @")则为实现文件
is_decl_mi: int = -1
tpath_mi: str = StubMergerCollect._sliced_path(temp_dir, td_len_mi, inc_sha1_mi, "text.ll")
if tpath_mi is not None:
tf_mi: fileio.File | t.CPtr = fileio.File(tpath_mi, fileio.MODE.R)
if not tf_mi.closed:
is_decl_mi = 1
tbuf_mi: bytes = stdlib.malloc(StubMerger.STUB_READ_BUF_SIZE)
tbr_mi: t.CInt64T = 0
if tbuf_mi is not None:
tbr_mi = tf_mi.read_all(tbuf_mi, StubMerger.STUB_READ_BUF_SIZE)
if tbuf_mi is not None and tbr_mi > 0:
if tbr_mi < StubMerger.STUB_READ_BUF_SIZE:
tbuf_mi[tbr_mi] = '\0'
else:
tbuf_mi[StubMerger.STUB_READ_BUF_SIZE - 1] = '\0'
# 逐行扫描: 找 define 行即为实现文件
# (不再要求行中含 @",因为 memhub 等模块的函数名
# 如 @ca40915f11b8b6ad._align_up 不带引号但仍是实现文件)
tpos_mi: t.CSizeT = 0
while tpos_mi < tbr_mi:
tls_mi: t.CSizeT = tpos_mi
while tpos_mi < tbr_mi:
if tbuf_mi[tpos_mi] == '\n':
break
tpos_mi += 1
tll_mi: t.CSizeT = tpos_mi - tls_mi
if tpos_mi < tbr_mi:
tpos_mi += 1
if tll_mi >= 7 and string.strncmp(tbuf_mi + tls_mi, "define ", 7) == 0:
is_decl_mi = 0
break
stdlib.free(tbuf_mi)
tf_mi.close()
stdlib.free(tpath_mi)
# is_decl_mi == -1: text.ll 不存在(Phase1 未翻译此文件)
# is_decl_mi == 1: 声明文件(仅 declare 无 define)
# is_decl_mi == 0: 实现文件(有 define)
# 只跳过声明文件(is_decl_mi == 1);text.ll 不存在时继续,让后续翻译逻辑处理
if is_decl_mi == 1:
stdlib.free(check_pat_mi)
if check_fd_mi is not None:
stdlib.free(check_fd_mi)
stdlib.free(src_fp_mi)
continue
# 尝试 BuildCombinedIR(stub/text 应已由 Phase1 生成)
inc_combined_mi: bytes = stdlib.malloc(COMBINED_IR_SIZE)
inc_combined_len_mi: t.CSizeT = 0
if inc_combined_mi is not None:
inc_combined_len_mi = StubMergerBuild.BuildCombinedIR(temp_dir, inc_sha1_mi, inc_combined_mi, COMBINED_IR_SIZE)
# 如果 stub/text 不存在,翻译源文件并保存 stub + text,然后重试
if inc_combined_len_mi == 0 and inc_combined_mi is not None:
# 计算 includes 文件的包名(相对 includes_dir 的目录部分)
inc_pkg_mi: str = None
if string.strlen(src_fp_mi) > inc_dir_len_mi + 1:
inc_rel_mi2: str = src_fp_mi + inc_dir_len_mi + 1
inc_pkg_mi = GenApi.compute_package_from_relpath(mb, inc_rel_mi2)
tr_mi: HandlesTranslator.Translator | t.CPtr = BuildPipeline.TranslateFileGetTrans(mb, src_fp_mi, inc_sha1_mi, inc_pkg_mi)
if tr_mi is not None:
PBP_IR_SIZE: t.CSizeT = 1048576
# 保存 stub.ll (切片路径)
inc_stub_buf: bytes = stdlib.malloc(PBP_IR_SIZE)
if inc_stub_buf is not None:
tr_mi.dump_ir(inc_stub_buf, PBP_IR_SIZE, llvmlite.OUTPUT_STUB)
inc_stub_len: t.CSizeT = string.strlen(inc_stub_buf)
inc_stub_path: str = StubMergerCollect._sliced_path(temp_dir, td_len_mi, inc_sha1_mi, "stub.ll")
if inc_stub_path is not None:
isf: fileio.File | t.CPtr = fileio.File(inc_stub_path, fileio.MODE.W)
if not isf.closed:
isf.write(inc_stub_buf, inc_stub_len)
isf.close()
stdlib.free(inc_stub_path)
stdlib.free(inc_stub_buf)
# 保存 text.ll (切片路径)
inc_text_buf: bytes = stdlib.malloc(PBP_IR_SIZE)
if inc_text_buf is not None:
tr_mi.dump_ir(inc_text_buf, PBP_IR_SIZE, llvmlite.OUTPUT_TEXT)
inc_text_len: t.CSizeT = string.strlen(inc_text_buf)
inc_text_path: str = StubMergerCollect._sliced_path(temp_dir, td_len_mi, inc_sha1_mi, "text.ll")
if inc_text_path is not None:
itf: fileio.File | t.CPtr = fileio.File(inc_text_path, fileio.MODE.W)
if not itf.closed:
itf.write(inc_text_buf, inc_text_len)
itf.close()
stdlib.free(inc_text_path)
stdlib.free(inc_text_buf)
# 重试 BuildCombinedIR
inc_combined_len_mi = StubMergerBuild.BuildCombinedIR(temp_dir, inc_sha1_mi, inc_combined_mi, COMBINED_IR_SIZE)
# OOM 清理:stub/text 已写盘,回收本 includes 文件翻译瞬时数据
# (AST/tokens/lexer/IR/Handler);BuildCombinedIR 只读文件,不受影响
mb.reset_to(mark2)
if inc_combined_len_mi == 0:
# 诊断:检查 stub.ll/text.ll 是否存在
diag_stub: str = StubMergerCollect._sliced_path(temp_dir, td_len_mi, inc_sha1_mi, "stub.ll")
diag_text: str = StubMergerCollect._sliced_path(temp_dir, td_len_mi, inc_sha1_mi, "text.ll")
diag_stub_ex: int = 0
diag_text_ex: int = 0
if diag_stub is not None:
df_diag_s: fileio.File | t.CPtr = fileio.File(diag_stub, fileio.MODE.R)
if not df_diag_s.closed:
diag_stub_ex = 1
df_diag_s.close()
stdlib.free(diag_stub)
if diag_text is not None:
df_diag_t: fileio.File | t.CPtr = fileio.File(diag_text, fileio.MODE.R)
if not df_diag_t.closed:
diag_text_ex = 1
df_diag_t.close()
stdlib.free(diag_text)
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024,
"BuildCombinedIR 失败: %s (stub=%d text=%d is_decl=%d)",
src_fp_mi, diag_stub_ex, diag_text_ex, is_decl_mi)
VLogger.error(fb, "PhaseB+")
if inc_combined_mi is not None:
stdlib.free(inc_combined_mi)
stdlib.free(check_pat_mi)
if check_fd_mi is not None:
stdlib.free(check_fd_mi)
stdlib.free(src_fp_mi)
continue
# 编译为 .obj(输出到 includes_binary_dir,与存在性检查和链接收集一致)
inc_cret_mi: int = BuildPipeline.compile_module_to_obj(
inc_combined_mi, inc_combined_len_mi, temp_dir, includes_binary_dir, inc_sha1_mi,
cc_cmd, cc_flags)
stdlib.free(inc_combined_mi)
if inc_cret_mi != 0:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "llc 编译失败,终止编译: %s", src_fp_mi)
VLogger.error(fb, "PhaseB+")
sys.exit(1)
inc_compiled += 1
# 显示编译完成的 includes 文件
fb_inc_comp: str = VLogger.fmt_buf()
if fb_inc_comp is not None:
viperlib.snprintf(fb_inc_comp, 1024, "[includes %d] 编译完成: %s", inc_compiled, src_fp_mi)
VLogger.info(fb_inc_comp, "PhaseB+")
# 添加到 obj_paths (切片路径,从 includes_binary_dir 取)
ibd_len_mi2: t.CSizeT = string.strlen(includes_binary_dir)
inc_obj_sliced: str = StubMergerCollect._sliced_path(includes_binary_dir, ibd_len_mi2, inc_sha1_mi, "obj")
if inc_obj_sliced is not None:
inc_obj_len: t.CSizeT = string.strlen(inc_obj_sliced)
if obj_pos + inc_obj_len + 2 < OBJ_PATHS_SIZE:
if obj_pos > 0:
obj_paths[obj_pos] = ' '
obj_pos += 1
string.strcpy(obj_paths + obj_pos, inc_obj_sliced)
obj_pos += inc_obj_len
obj_paths[obj_pos] = '\0'
else:
VLogger.error("includes .obj 路径缓冲区溢出,终止编译", "PhaseB+")
stdlib.free(inc_obj_sliced)
# 释放本次迭代的临时内存
if check_pat_mi is not None:
stdlib.free(check_pat_mi)
if check_fd_mi is not None:
stdlib.free(check_fd_mi)
stdlib.free(src_fp_mi)
# 释放拓扑排序索引与局部映射副本
if topo_idx is not None:
stdlib.free(topo_idx)
if store_sha1_arr_mi is not None:
stdlib.free(store_sha1_arr_mi)
if store_mod_arr_mi is not None:
stdlib.free(store_mod_arr_mi)
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "编译缺失 includes: %d 个", inc_compiled)
VLogger.info(fb, "PhaseB+")
# 释放 app_deps_buf(Phase B+ 块未执行时此处兜底释放)
if app_deps_buf is not None:
stdlib.free(app_deps_buf)
app_deps_buf = None
# === 4. Phase C: 链接所有 .obj → .exe ===
if log is not None:
log.banner("Phase C: 链接")
od_len2: t.CSizeT = string.strlen(output_dir)
lo_len: t.CSizeT = string.strlen(linker_output)
exe_path: bytes = stdlib.malloc(od_len2 + lo_len + 2)
if exe_path is not None:
viperlib.snprintf(exe_path, od_len2 + lo_len + 2, "%s/%s", output_dir, linker_output)
else:
exe_path = linker_output
VLogger.error("exe_path 分配失败,终止编译", "PhaseC")
# 构造最终 .obj 路径列表:main_obj_path 在前,reachable .obj 在后
final_obj_paths: bytes = stdlib.malloc(OBJ_PATHS_SIZE + 512)
if final_obj_paths is None:
stdlib.free(entries)
return 1
final_obj_paths[0] = '\0'
fop_pos: t.CSizeT = 0
# main_obj_path 放在最前面(确保 --allow-multiple-definition 选择用户 main)
if main_obj_path[0] != '\0':
mlen: t.CSizeT = string.strlen(main_obj_path)
string.strcpy(final_obj_paths, main_obj_path)
fop_pos = mlen
# 合并 Phase B/B+ 编译的 App 非 main 模块 .obj(obj_paths,空格分隔)。
# 此前 final_obj_paths 只含 main_obj + reachable includes,App 依赖文件
# (main import 的 syslibtest_impl/os_backend 等)的 .obj 从未入链——追加28
# 前靠 reachable 误含 App 时从 includes.binary 的重复 obj 兜底(undefined
# symbol: ae2038ced7b4634f.run 根因)。obj_paths 同时含 Phase B+ 本次
# 新编译的 includes obj(与下方 reachable 收集重复,--allow-multiple-
# definition 容忍,仅本次编译的少量文件)。
if obj_pos > 0:
if fop_pos + obj_pos + 1 < OBJ_PATHS_SIZE + 512:
if fop_pos > 0:
final_obj_paths[fop_pos] = ' '
fop_pos += 1
string.strcpy(final_obj_paths + fop_pos, obj_paths)
fop_pos += obj_pos
final_obj_paths[fop_pos] = '\0'
# 按图链接:遍历 reachable_set,只收集可达 includes 的 .obj
# 不再全量收集 includes.binary 下的所有 .obj
if reachable_set is not None and reachable_count > 0:
ibd_len_pc: t.CSizeT = string.strlen(includes_binary_dir)
td_len_pc: t.CSizeT = string.strlen(temp_dir)
ri_pc: int
for ri_pc in range(reachable_count):
rsha1_pc: str = reachable_set + t.CSizeT(ri_pc) * 17
if rsha1_pc[0] == '\0':
continue
obj_path_pc: str = StubMergerCollect._sliced_path(includes_binary_dir, ibd_len_pc, rsha1_pc, "obj")
if obj_path_pc is not None:
# 检查 .obj 文件是否存在
cf_pc: fileio.File | t.CPtr = fileio.File(obj_path_pc, fileio.MODE.R)
if not cf_pc.closed:
cf_pc.close()
# .obj 存在,追加到 final_obj_paths
op_len_pc: t.CSizeT = string.strlen(obj_path_pc)
if fop_pos + op_len_pc + 2 < OBJ_PATHS_SIZE + 512:
if fop_pos > 0:
final_obj_paths[fop_pos] = ' '
fop_pos += 1
string.strcpy(final_obj_paths + fop_pos, obj_path_pc)
fop_pos += op_len_pc
final_obj_paths[fop_pos] = '\0'
else:
# .obj 缺失:纯声明模块(text.ll 无 define 行,如 stdio.vp 全 pass)
# 无需 .obj,静默跳过;实现文件缺失才是编译产物异常(fail-fast)
if StubMergerIR.IsDeclOnlyModule(temp_dir, td_len_pc, rsha1_pc) != 1:
fb_pc: str = VLogger.fmt_buf()
if fb_pc is not None:
viperlib.snprintf(fb_pc, 1024, "可达模块 .obj 缺失: %s", obj_path_pc)
VLogger.error(fb_pc, "PhaseC")
stdlib.free(obj_path_pc)
final_obj_paths[fop_pos] = '\0'
obj_paths_len: t.CSizeT = fop_pos
# 释放 reachable_set(Phase C 已使用完毕)
if reachable_set is not None:
stdlib.free(reachable_set)
reachable_set = None
# 按图链接:不传 includes_binary_dir(传 None),避免全量收集
lret: int = BuildPipeline.link_objs_to_exe(
final_obj_paths, obj_paths_len,
linker_cmd, linker_flags, exe_path,
None)
if lret == 0:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "输出: %s", exe_path)
VLogger.success(fb, "PhaseC")
if args.get_bool("run"):
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "执行: %s", exe_path)
VLogger.info(fb, "run")
rp: subprocess.CompletedProcess | t.CPtr = subprocess.run(exe_path, False, False)
if rp is not None:
fb: str = VLogger.fmt_buf()
if fb is not None:
viperlib.snprintf(fb, 1024, "退出码: %d", rp.returncode)
VLogger.info(fb, "run")
else:
VLogger.error("链接失败", "PhaseC")
stdlib.free(entries)
return 1
stdlib.free(entries)
return 0