Spaces:
Sleeping
Sleeping
Update json_parser.py
Browse files- json_parser.py +48 -7
json_parser.py
CHANGED
|
@@ -39,27 +39,65 @@ class SSEParser:
|
|
| 39 |
def _clean_mermaid_content(self, content: str) -> Optional[str]:
|
| 40 |
"""Clean and extract mermaid diagram content"""
|
| 41 |
try:
|
| 42 |
-
#
|
| 43 |
-
if
|
| 44 |
-
content = content
|
| 45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
# Parse JSON if present
|
| 47 |
try:
|
| 48 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
# Handle different mermaid output formats
|
| 50 |
if "mermaid_output" in data:
|
| 51 |
content = data["mermaid_output"]
|
| 52 |
elif "mermaid_diagram" in data:
|
| 53 |
content = data["mermaid_diagram"]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 54 |
except json.JSONDecodeError:
|
| 55 |
pass
|
| 56 |
|
| 57 |
# Clean up markdown formatting
|
| 58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 59 |
|
| 60 |
-
return content.strip()
|
| 61 |
except Exception as e:
|
| 62 |
self.logger.error(f"Error cleaning mermaid content: {str(e)}")
|
|
|
|
| 63 |
return None
|
| 64 |
|
| 65 |
def parse_sse_event(self, data: str) -> Optional[Dict]:
|
|
@@ -89,9 +127,10 @@ class SSEParser:
|
|
| 89 |
thought = parsed_data.get("thought", "")
|
| 90 |
observation = parsed_data.get("observation", "")
|
| 91 |
tool = parsed_data.get("tool", "")
|
|
|
|
| 92 |
|
| 93 |
# Handle tool-specific formatting
|
| 94 |
-
if tool == "mermaid_diagrams":
|
| 95 |
try:
|
| 96 |
cleaned_content = self._clean_mermaid_content(observation)
|
| 97 |
if cleaned_content:
|
|
@@ -103,6 +142,7 @@ class SSEParser:
|
|
| 103 |
}
|
| 104 |
except Exception as e:
|
| 105 |
self.logger.error(f"Failed to parse mermaid data: {str(e)}")
|
|
|
|
| 106 |
|
| 107 |
return {
|
| 108 |
"type": "thought",
|
|
@@ -125,6 +165,7 @@ class SSEParser:
|
|
| 125 |
|
| 126 |
except Exception as e:
|
| 127 |
self.logger.error(f"Parse error: {str(e)}")
|
|
|
|
| 128 |
return None
|
| 129 |
|
| 130 |
def _process_observation(self, data: Dict) -> Dict:
|
|
|
|
| 39 |
def _clean_mermaid_content(self, content: str) -> Optional[str]:
|
| 40 |
"""Clean and extract mermaid diagram content"""
|
| 41 |
try:
|
| 42 |
+
# Handle tool output format
|
| 43 |
+
if isinstance(content, dict) and "tool_output" in content:
|
| 44 |
+
content = content["tool_output"]
|
| 45 |
|
| 46 |
+
# Remove tool response prefix/suffix if present
|
| 47 |
+
if isinstance(content, str):
|
| 48 |
+
if "tool response:" in content:
|
| 49 |
+
content = content.split("tool response:", 1)[1].strip()
|
| 50 |
+
if content.endswith('.'):
|
| 51 |
+
content = content[:-1]
|
| 52 |
+
|
| 53 |
# Parse JSON if present
|
| 54 |
try:
|
| 55 |
+
if isinstance(content, str):
|
| 56 |
+
data = json.loads(content)
|
| 57 |
+
else:
|
| 58 |
+
data = content
|
| 59 |
+
|
| 60 |
# Handle different mermaid output formats
|
| 61 |
if "mermaid_output" in data:
|
| 62 |
content = data["mermaid_output"]
|
| 63 |
elif "mermaid_diagram" in data:
|
| 64 |
content = data["mermaid_diagram"]
|
| 65 |
+
|
| 66 |
+
# If content is still JSON string, parse again
|
| 67 |
+
if isinstance(content, str) and content.startswith('{'):
|
| 68 |
+
try:
|
| 69 |
+
data = json.loads(content)
|
| 70 |
+
if "mermaid_output" in data:
|
| 71 |
+
content = data["mermaid_output"]
|
| 72 |
+
elif "mermaid_diagram" in data:
|
| 73 |
+
content = data["mermaid_diagram"]
|
| 74 |
+
except:
|
| 75 |
+
pass
|
| 76 |
+
|
| 77 |
except json.JSONDecodeError:
|
| 78 |
pass
|
| 79 |
|
| 80 |
# Clean up markdown formatting
|
| 81 |
+
if isinstance(content, str):
|
| 82 |
+
content = content.replace("```mermaid\n", "").replace("\n```", "")
|
| 83 |
+
content = content.strip()
|
| 84 |
+
|
| 85 |
+
# Remove any remaining JSON artifacts
|
| 86 |
+
if content.startswith('{'):
|
| 87 |
+
try:
|
| 88 |
+
data = json.loads(content)
|
| 89 |
+
if isinstance(data, dict):
|
| 90 |
+
content = next(iter(data.values()))
|
| 91 |
+
except:
|
| 92 |
+
pass
|
| 93 |
+
|
| 94 |
+
return content
|
| 95 |
+
|
| 96 |
+
return None
|
| 97 |
|
|
|
|
| 98 |
except Exception as e:
|
| 99 |
self.logger.error(f"Error cleaning mermaid content: {str(e)}")
|
| 100 |
+
self.logger.debug(f"Original content: {content}")
|
| 101 |
return None
|
| 102 |
|
| 103 |
def parse_sse_event(self, data: str) -> Optional[Dict]:
|
|
|
|
| 127 |
thought = parsed_data.get("thought", "")
|
| 128 |
observation = parsed_data.get("observation", "")
|
| 129 |
tool = parsed_data.get("tool", "")
|
| 130 |
+
tool_input = parsed_data.get("tool_input", "")
|
| 131 |
|
| 132 |
# Handle tool-specific formatting
|
| 133 |
+
if tool == "mermaid_diagrams" or "mermaid" in str(tool_input).lower():
|
| 134 |
try:
|
| 135 |
cleaned_content = self._clean_mermaid_content(observation)
|
| 136 |
if cleaned_content:
|
|
|
|
| 142 |
}
|
| 143 |
except Exception as e:
|
| 144 |
self.logger.error(f"Failed to parse mermaid data: {str(e)}")
|
| 145 |
+
self.logger.debug(f"Raw observation: {observation}")
|
| 146 |
|
| 147 |
return {
|
| 148 |
"type": "thought",
|
|
|
|
| 165 |
|
| 166 |
except Exception as e:
|
| 167 |
self.logger.error(f"Parse error: {str(e)}")
|
| 168 |
+
self.logger.debug(f"Raw data: {data}")
|
| 169 |
return None
|
| 170 |
|
| 171 |
def _process_observation(self, data: Dict) -> Dict:
|