Merge branch 'dev' into code_intepreter

metagpt/utils/project_repo.py

@@ -99,6 +99,13 @@ class ProjectRepo(FileRepository):
         self.tests = self._git_repo.new_file_repository(relative_path=TEST_CODES_FILE_REPO)
         self.test_outputs = self._git_repo.new_file_repository(relative_path=TEST_OUTPUTS_FILE_REPO)
         self._srcs_path = None
+        self.code_files_exists()
+
+    def __str__(self):
+        repo_str = f"ProjectRepo({self._git_repo.workdir})"
+        docs_str = f"Docs({self.docs.all_files})"
+        srcs_str = f"Srcs({self.srcs.all_files})"
+        return f"{repo_str}\n{docs_str}\n{srcs_str}"
 
     @property
     async def requirement(self):

metagpt/utils/repair_llm_raw_output.py

@@ -119,15 +119,22 @@ def repair_json_format(output: str) -> str:
         logger.info(f"repair_json_format: {'}]'}")
     elif output.startswith("{") and output.endswith("]"):
         output = output[:-1] + "}"
-
-    # remove `#` in output json str, usually appeared in `glm-4`
+    # remove comments in output json string, after json value content, maybe start with #, maybe start with //
     arr = output.split("\n")
     new_arr = []
-    for line in arr:
-        idx = line.find("#")
-        if idx >= 0:
-            line = line[:idx]
-        new_arr.append(line)
+    for json_line in arr:
+        # look for # or // comments and make sure they are not inside the string value
+        comment_index = -1
+        for match in re.finditer(r"(\".*?\"|\'.*?\')|(#|//)", json_line):
+            if match.group(1):  # if the string value
+                continue
+            if match.group(2):  # if comments
+                comment_index = match.start(2)
+                break
+        # if comments, then delete them
+        if comment_index != -1:
+            json_line = json_line[:comment_index].rstrip()
+        new_arr.append(json_line)
     output = "\n".join(new_arr)
     return output