jarvis-ai-assistant 0.7.8__py3-none-any.whl → 1.0.2__py3-none-any.whl

jarvis/jarvis_c2rust/collector.py

@@ -20,21 +20,19 @@
 - 本模块注重正确性/鲁棒性而非性能
 - 不尝试发现传递包含；仅通过添加-I <file.parent>来辅助解析器
 """
+
 from __future__ import annotations
 
 from pathlib import Path
-from typing import List, Optional, Dict
-
-
-from jarvis.jarvis_c2rust.scanner import (
-    _try_import_libclang,
-    find_compile_commands,
-    load_compile_commands,
-    scan_file,
-)
-
+from typing import Dict
+from typing import List
+from typing import Optional
 
-HEADER_EXTS = {".h", ".hh", ".hpp", ".hxx"}
+from jarvis.jarvis_c2rust.constants import HEADER_EXTS
+from jarvis.jarvis_c2rust.scanner import _try_import_libclang
+from jarvis.jarvis_c2rust.scanner import find_compile_commands
+from jarvis.jarvis_c2rust.scanner import load_compile_commands
+from jarvis.jarvis_c2rust.scanner import scan_file
 
 
 def _guess_lang_header_flag(file: Path) -> List[str]:
@@ -47,7 +45,10 @@ def _guess_lang_header_flag(file: Path) -> List[str]:
     # 对于.h文件默认使用C头文件(保守选择)
     return ["-x", "c-header"]
 
-def _ensure_parse_args_for_header(file: Path, base_args: Optional[List[str]]) -> List[str]:
+
+def _ensure_parse_args_for_header(
+    file: Path, base_args: Optional[List[str]]
+) -> List[str]:
     """
     确保头文件解析所需的最小参数集:
     - 通过-I包含file.parent目录
@@ -92,6 +93,7 @@ def _ensure_parse_args_for_header(file: Path, base_args: Optional[List[str]]) ->
 
     return args
 
+
 def _collect_decl_function_names(cindex, file: Path, args: List[str]) -> List[str]:
     """
     对于没有内联定义的头文件的回退方案:
@@ -110,10 +112,19 @@ def _collect_decl_function_names(cindex, file: Path, args: List[str]) -> List[st
             kind = node.kind.name
         except Exception:
             kind = ""
-        if kind in {"FUNCTION_DECL", "CXX_METHOD", "FUNCTION_TEMPLATE", "CONSTRUCTOR", "DESTRUCTOR"}:
+        if kind in {
+            "FUNCTION_DECL",
+            "CXX_METHOD",
+            "FUNCTION_TEMPLATE",
+            "CONSTRUCTOR",
+            "DESTRUCTOR",
+        }:
             loc_file = getattr(getattr(node, "location", None), "file", None)
             try:
-                same_file = loc_file is not None and Path(loc_file.name).resolve() == file.resolve()
+                same_file = (
+                    loc_file is not None
+                    and Path(loc_file.name).resolve() == file.resolve()
+                )
             except Exception:
                 same_file = False
             if same_file:
@@ -133,6 +144,7 @@ def _collect_decl_function_names(cindex, file: Path, args: List[str]) -> List[st
         pass
     return names
 
+
 def collect_function_names(
     files: List[Path],
     out_path: Path,
@@ -166,7 +178,8 @@ def collect_function_names(
     # 准备libclang
     cindex = _try_import_libclang()
     if cindex is None:
-        from clang import cindex as _ci  # type: ignore
+        from clang import cindex as _ci
+
         cindex = _ci
 
     # 准备compile_commands参数映射(如果提供了根目录则全局一次，否则每个文件单独处理)
@@ -205,7 +218,11 @@ def collect_function_names(
             funcs = scan_file(cindex, hf, args)
         except Exception:
             try:
-                funcs = scan_file(cindex, hf, _ensure_parse_args_for_header(hf, ["-I", str(hf.parent)]))
+                funcs = scan_file(
+                    cindex,
+                    hf,
+                    _ensure_parse_args_for_header(hf, ["-I", str(hf.parent)]),
+                )
             except Exception:
                 funcs = []
 
@@ -243,8 +260,10 @@ def collect_function_names(
     out_path = Path(out_path)
     out_path.parent.mkdir(parents=True, exist_ok=True)
     try:
-        out_path.write_text("\n".join(ordered_names) + ("\n" if ordered_names else ""), encoding="utf-8")
+        out_path.write_text(
+            "\n".join(ordered_names) + ("\n" if ordered_names else ""), encoding="utf-8"
+        )
     except Exception as e:
         raise RuntimeError(f"写入输出文件失败: {out_path}: {e}")
 
-    return out_path
+    return out_path

jarvis/jarvis_c2rust/constants.py

@@ -3,6 +3,8 @@
 C2Rust 转译器常量定义
 """
 
+from typing import Set
+
 # 数据文件常量
 C2RUST_DIRNAME = ".jarvis/c2rust"
 
@@ -24,3 +26,61 @@ CONSECUTIVE_FIX_FAILURE_THRESHOLD = 10  # 连续修复失败次数阈值，达
 MAX_FUNCTION_RETRIES = 10  # 函数重新开始处理的最大次数
 DEFAULT_PLAN_MAX_RETRIES_ENTRY = 5  # run_transpile 入口函数的 plan_max_retries 默认值
 
+# 文件扩展名常量
+SOURCE_EXTS: Set[str] = {
+    ".c",
+    ".cc",
+    ".cpp",
+    ".cxx",
+    ".C",
+    ".h",
+    ".hh",
+    ".hpp",
+    ".hxx",
+}
+HEADER_EXTS = {".h", ".hh", ".hpp", ".hxx"}
+
+# AST节点类型常量
+TYPE_KINDS: Set[str] = {
+    "STRUCT_DECL",
+    "UNION_DECL",
+    "ENUM_DECL",
+    "CXX_RECORD_DECL",  # C++ class/struct/union
+    "TYPEDEF_DECL",
+    "TYPE_ALIAS_DECL",
+}
+
+# 运行状态文件
+RUN_STATE_JSON = "run_state.json"
+
+# Library Replacer 相关常量
+# LLM评估重试配置
+MAX_LLM_RETRIES = 3  # LLM评估最大重试次数
+
+# 源码片段读取配置
+DEFAULT_SOURCE_SNIPPET_MAX_LINES = 200  # 默认源码片段最大行数
+SUBTREE_SOURCE_SNIPPET_MAX_LINES = 120  # 子树提示词中源码片段最大行数
+
+# 子树提示词构建配置
+MAX_SUBTREE_NODES_META = 200  # 子树节点元数据列表最大长度
+MAX_SUBTREE_EDGES = 400  # 子树边列表最大长度
+MAX_DOT_EDGES = 200  # DOT图边数阈值（超过此值不生成DOT）
+MAX_CHILD_SAMPLES = 2  # 子节点采样数量
+MAX_SOURCE_SAMPLES = 3  # 代表性源码样本最大数量（注释说明）
+
+# 显示配置
+MAX_NOTES_DISPLAY_LENGTH = 200  # 备注显示最大长度
+
+# 输出文件路径配置
+DEFAULT_SYMBOLS_OUTPUT = "symbols_library_pruned.jsonl"  # 默认符号表输出文件名
+DEFAULT_MAPPING_OUTPUT = "library_replacements.jsonl"  # 默认替代映射输出文件名
+SYMBOLS_PRUNE_OUTPUT = "symbols_prune.jsonl"  # 兼容符号表输出文件名
+ORDER_PRUNE_OUTPUT = "translation_order_prune.jsonl"  # 剪枝阶段转译顺序输出文件名
+ORDER_ALIAS_OUTPUT = "translation_order.jsonl"  # 通用转译顺序输出文件名
+DEFAULT_CHECKPOINT_FILE = "library_replacer_checkpoint.json"  # 默认检查点文件名
+
+# Checkpoint配置
+DEFAULT_CHECKPOINT_INTERVAL = 1  # 默认检查点保存间隔（每评估N个节点保存一次）
+
+# JSON格式化配置
+JSON_INDENT = 2  # JSON格式化缩进空格数