Merge pull request #799 from HuZixia/huzixia

Solve JSONDecodeError

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
 

tests/metagpt/utils/test_repair_llm_raw_output.py

@@ -141,6 +141,32 @@ def test_repair_json_format():
     output = repair_llm_raw_output(output=raw_output, req_keys=[None], repair_type=RepairType.JSON)
     assert output == target_output
 
+    raw_output = """
+{
+    "Language": "en_us",  // define language
+    "Programming Language": "Python" # define code language
+}
+"""
+    target_output = """{
+    "Language": "en_us",  
+    "Programming Language": "Python"
+}"""
+    output = repair_llm_raw_output(output=raw_output, req_keys=[None], repair_type=RepairType.JSON)
+    assert output == target_output
+
+    raw_output = """
+    {
+        "Language": "#en_us#",  // define language
+        "Programming Language": "//Python # Code // Language//" # define code language
+    }
+    """
+    target_output = """{
+        "Language": "#en_us#",  
+        "Programming Language": "//Python # Code // Language//"
+    }"""
+    output = repair_llm_raw_output(output=raw_output, req_keys=[None], repair_type=RepairType.JSON)
+    assert output == target_output
+
 
 def test_repair_invalid_json():
     from metagpt.utils.repair_llm_raw_output import repair_invalid_json