Update scripts/frontmatter_tool.py

scripts/frontmatter_tool.py

@@ -3,10 +3,13 @@ Frontmatter Generator Tool for Smolagents
 
 This tool helps generate consistent YAML frontmatter for documents,
 useful for RAG systems, static site generators, and document organization.
+Integrates with TextInspectorTool and MarkdownConverter for a complete
+document processing pipeline.
 """
 
 import re
 import yaml
+import json
 from datetime import datetime
 from typing import Dict, List, Optional, Any, Union
 from smolagents import Tool
@@ -20,6 +23,7 @@ class FrontmatterGeneratorTool(Tool):
     Generates or extracts YAML frontmatter for documents. Frontmatter provides structured 
     metadata for documents including title, author, date, description, and tags.
     Useful for document organization, RAG systems, and static site generators.
+    Works with content from the inspect_file_as_text tool to add metadata to documents.
     """
 
     inputs = {
@@ -27,37 +31,37 @@ class FrontmatterGeneratorTool(Tool):
             "type": "string",
             "description": "Document content (with or without existing frontmatter)",
         },
-        "title": {"type": "string", "description": "Document title", "optional": True},
+        "title": {"type": "string", "description": "Document title", "nullable": True},
         "author": {
             "type": "string",
             "description": "Document author(s)",
-            "optional": True,
+            "nullable": True,
         },
         "date": {
             "type": "string",
             "description": "Document date in YYYY-MM-DD format (defaults to today if not provided)",
-            "optional": True,
+            "nullable": True,
         },
         "date_format": {
             "type": "string",
             "description": "Format string for the document date (e.g., '%Y-%m-%d', '%d/%m/%Y'). Defaults to '%Y-%m-%d'",
-            "optional": True,
+            "nullable": True,
             "default": "%Y-%m-%d",
         },
         "description": {
             "type": "string",
             "description": "Brief description of the document",
-            "optional": True,
+            "nullable": True,
         },
         "tags": {
             "type": "string",
             "description": "Comma-separated list of tags",
-            "optional": True,
+            "nullable": True,
         },
         "additional_fields": {
             "type": "string",
             "description": "JSON string with additional frontmatter fields",
-            "optional": True,
+            "nullable": True,
         },
         "mode": {
             "type": "string",
@@ -130,6 +134,10 @@ class FrontmatterGeneratorTool(Tool):
             else:
                 return "Error: Empty content provided"
 
+        # Special handling for TextInspectorTool output
+        if content.startswith("Document content:"):
+            content = content[len("Document content:") :].strip()
+
         # Process based on mode
         try:
             if mode == "extract":
@@ -191,8 +199,6 @@ class FrontmatterGeneratorTool(Tool):
             return {}
 
         try:
-            import json
-
             return json.loads(additional_fields)
         except json.JSONDecodeError:
             raise ValueError("additional_fields must be a valid JSON string")
@@ -393,56 +399,3 @@ class FrontmatterGeneratorTool(Tool):
 
         # Combine frontmatter with content
         return frontmatter + clean_content
-
-
-# Example usage
-def example_usage():
-    """Simple example of how to use the FrontmatterGeneratorTool"""
-    tool = FrontmatterGeneratorTool()
-
-    # Example document
-    document = """# The Role of the Seat in International Arbitration
-
-This paper examines the significance of the seat in international arbitration proceedings,
-with a focus on recent developments in Australian law."""
-
-    # Generate frontmatter
-    result = tool.forward(
-        content=document,
-        title="The Role of the Seat in International Arbitration",
-        author="Matthew Barry",
-        date="13 March 2013",  # Now handles this format correctly
-        description="A legal analysis of the role of the seat in international arbitration.",
-        tags="international arbitration, enforcement, australian courts",
-        mode="generate",
-    )
-
-    print(result)
-    return result
-
-
-# Define what gets imported with "from frontmatter_generator import *"
-__all__ = ["FrontmatterGeneratorTool", "example_usage"]
-
-# Example usage
-if __name__ == "__main__":
-    tool = FrontmatterGeneratorTool()
-
-#     # Example document
-#     document = """# The Role of the Seat in International Arbitration
-
-# This paper examines the significance of the seat in international arbitration proceedings,
-# with a focus on recent developments in Australian law."""
-
-#     # Generate frontmatter
-#     result = tool.forward(
-#         content=document,
-#         title="The Role of the Seat in International Arbitration",
-#         author="Matthew Barry",
-#         date="2013-12-13",
-#         description="A legal analysis of the role of the seat in international arbitration.",
-#         tags="international arbitration, enforcement, australian courts",
-#         mode="generate"
-#     )
-
-#     print(result)