fix: standardize path handling to use Path objects consistently and fix critical bug in code_retrieval

Author: wangjichao

codebase_rag/graph_updater.py

@@ -262,7 +262,7 @@ def _is_dependency_file(self, file_name: str, filepath: Path) -> bool:
         )
 
     def run(self) -> None:
-        absolute_path = self.repo_path.absolute().as_posix()
+        absolute_path = str(self.repo_path.absolute())
 
         self.ingestor.ensure_node_batch(
             cs.NODE_PROJECT,

codebase_rag/parsers/class_ingest/cpp_modules.py

@@ -83,7 +83,7 @@ def _process_export_module(
         {
             cs.KEY_QUALIFIED_NAME: interface_qn,
             cs.KEY_NAME: module_name,
-            cs.KEY_PATH: str(file_path.relative_to(repo_path)),
+            cs.KEY_PATH: file_path.relative_to(repo_path).as_posix(),
             cs.KEY_MODULE_TYPE: cs.CPP_MODULE_TYPE_INTERFACE,
         },
     )
@@ -117,7 +117,7 @@ def _process_module_implementation(
         {
             cs.KEY_QUALIFIED_NAME: impl_qn,
             cs.KEY_NAME: f"{module_name}{cs.CPP_IMPL_SUFFIX}",
-            cs.KEY_PATH: str(file_path.relative_to(repo_path)),
+            cs.KEY_PATH: file_path.relative_to(repo_path).as_posix(),
             cs.KEY_IMPLEMENTS_MODULE: module_name,
             cs.KEY_MODULE_TYPE: cs.CPP_MODULE_TYPE_IMPLEMENTATION,
         },

codebase_rag/tools/code_retrieval.py

@@ -39,36 +39,36 @@ async def find_code_snippet(self, qualified_name: str) -> CodeSnippet:
                 )
 
             res = results[0]
-            absolute_path_str = res.get("absolute_path")
             project_name = res.get("project_name")
-            file_path_str = res.get("relative_path")
-
-            if not absolute_path_str:
-                logger.warning(ls.NO_ABSOLUTE_PATH_FALLBACK.format(qn=qualified_name))
-
             start_line = res.get("start")
             end_line = res.get("end")
 
-            file_path_to_read = absolute_path_str or (
-                str(self.project_root.as_posix() / file_path_str)
-                if file_path_str
-                else ""
-            )
+            absolute_path_str = res.get("absolute_path")
+            relative_path_str = res.get("relative_path")
+
+            if absolute_path_str:
+                file_path_obj = Path(absolute_path_str)
+            elif relative_path_str:
+                file_path_obj = self.project_root / relative_path_str
+                logger.warning(ls.NO_ABSOLUTE_PATH_FALLBACK.format(qn=qualified_name))
+            else:
+                file_path_obj = None
 
-            if not all([file_path_to_read, start_line, end_line]):
+            if not (file_path_obj and start_line and end_line):
                 return CodeSnippet(
                     qualified_name=qualified_name,
                     source_code="",
-                    file_path=file_path_to_read or "",
+                    file_path=str(file_path_obj) if file_path_obj else "",
                     project_name=project_name,
                     line_start=0,
                     line_end=0,
                     found=False,
                     error_message=te.CODE_MISSING_LOCATION,
                 )
 
-            full_path = Path(file_path_to_read)
-            with full_path.open("r", encoding=ENCODING_UTF8) as f:
+            assert file_path_obj is not None
+
+            with file_path_obj.open("r", encoding=ENCODING_UTF8) as f:
                 all_lines = f.readlines()
 
             snippet_lines = all_lines[start_line - 1 : end_line]
@@ -77,7 +77,7 @@ async def find_code_snippet(self, qualified_name: str) -> CodeSnippet:
             return CodeSnippet(
                 qualified_name=qualified_name,
                 source_code=source_code,
-                file_path=file_path_to_read,
+                file_path=str(file_path_obj),
                 project_name=project_name,
                 line_start=start_line,
                 line_end=end_line,