Add JSON preprocessing for common model generated json errors

Author: dwash96

cecli/coders/agent_coder.py

@@ -21,6 +21,7 @@
 # All conversation functions are now available via ConversationChunks class
 from cecli.helpers.conversation.manager import ConversationManager
 from cecli.helpers.conversation.tags import MessageTag
+from cecli.helpers.responses import preprocess_json
 from cecli.helpers.similarity import (
     cosine_similarity,
     create_bigram_vector,
@@ -58,15 +59,14 @@ def __init__(self, *args, **kwargs):
             "viewfileswithsymbol",
             "grep",
             "listchanges",
-            "extractlines",
             "shownumberedcontext",
         }
         self.write_tools = {
             "command",
             "commandinteractive",
-            "insertblock",
-            "replaceblock",
-            "replaceall",
+            "deletetext",
+            "indenttext",
+            "inserttext",
             "replacetext",
             "undochange",
         }
@@ -244,11 +244,20 @@ async def _execute_local_tool_calls(self, tool_calls_list):
                     json_chunks = utils.split_concatenated_json(args_string)
                     for chunk in json_chunks:
                         try:
-                            parsed_args_list.append(json.loads(chunk))
-                        except json.JSONDecodeError:
+                            parsed_args_list.append(json.loads(preprocess_json(chunk)))
+                        except json.JSONDecodeError as e:
                             self.io.tool_warning(
                                 f"Could not parse JSON chunk for tool {tool_name}: {chunk}"
                             )
+                            tool_responses.append(
+                                {
+                                    "role": "tool",
+                                    "tool_call_id": tool_call.id,
+                                    "content": (
+                                        f"Could not parse JSON chunk for tool {tool_name}: {str(e)}"
+                                    ),
+                                }
+                            )
                             continue
                 if not parsed_args_list and not args_string:
                     parsed_args_list.append({})

cecli/helpers/responses.py

@@ -0,0 +1,19 @@
+import re
+
+
+def preprocess_json(response: str) -> str:
+    response = fix_json_backslashes(response)
+    return response
+
+
+def fix_json_backslashes(raw_str: str) -> str:
+    """
+    Finds invalid JSON escape sequences and escapes the backslash
+    so it is treated as a literal character.
+    """
+    # Look for a backslash NOT followed by valid escape characters
+    # We use a capturing group for the backslash to replace it
+    invalid_escape_pattern = r'\\(?!(["\\\/bfnrt]|u[0-9a-fA-F]{4}))'
+
+    # Replace the single backslash with a double backslash
+    return re.sub(invalid_escape_pattern, r"\\\\", raw_str)

cecli/tools/utils/helpers.py

@@ -3,6 +3,8 @@
 import re
 import traceback
 
+from cecli.helpers.responses import preprocess_json
+
 
 class ToolError(Exception):
     """Custom exception for tool-specific errors that should be reported to the LLM."""
@@ -258,7 +260,7 @@ def parse_arg_as_list(arg):
         import json
 
         try:
-            parsed = json.loads(arg)
+            parsed = json.loads(preprocess_json(arg))
             if isinstance(parsed, list):
                 return parsed
             else:

cecli/tools/utils/output.py

@@ -1,6 +1,8 @@
 import json
 import re
 
+from cecli.helpers.responses import preprocess_json
+
 
 def print_tool_response(coder, mcp_server, tool_response):
     """
@@ -65,7 +67,7 @@ def tool_body_unwrapped(coder, tool_response):
     color_start, color_end = color_markers(coder)
 
     try:
-        args_dict = json.loads(tool_response.function.arguments)
+        args_dict = json.loads(preprocess_json(tool_response.function.arguments))
         first_key = True
         for key, value in args_dict.items():
             # Convert explicit \\n sequences to actual newlines using regex