Update tools.py

tools.py

@@ -1,7 +1,12 @@
-# tools.py
 import os
+import re
+from pathlib import Path
+import tempfile
+
 import openai
-from smolagents.tools import Tool
+import pandas as pd
+from tabulate import tabulate
+from smolagents.tools import Tool, PipelineTool
 
 # Load your OpenAI API key
 openai.api_key = os.getenv("OPENAI_API_KEY")
@@ -13,17 +18,6 @@ YOUR FINAL ANSWER should be:
 - If you are asked for a number, don't use commas, symbols or units (e.g. %, $, km) unless explicitly asked.
 - If you are asked for a string, don't use articles ("a", "the"), abbreviations (e.g. "NYC"), or extra words; write digits in plain text unless specified otherwise.
 - If you are asked for a comma separated list, apply the above rules to each element.
-Examples:
-Q: How many legs does a spider have?
-Thought: It's basic biology.
-FINAL ANSWER: 8
-Q: What are the primary colors in light?
-Thought: Physics concept.
-FINAL ANSWER: red, green, blue
-Q: Who won the Best Picture Oscar in 1994?
-Thought: Film history.
-FINAL ANSWER: Forrest Gump
-Now it's your turn.
 """
 
 class AnswerTool(Tool):
@@ -47,3 +41,48 @@ class AnswerTool(Tool):
         if "FINAL ANSWER:" in text:
             return text.split("FINAL ANSWER:")[-1].strip()
         return text
+
+class SpeechToTextTool(PipelineTool):
+    """Transcribes audio files using OpenAI Whisper."""
+    name = "transcriber"
+    description = "Transcribes a local audio file to text using Whisper API."
+    inputs = {"audio": {"type": "string", "description": "Path to audio file."}}
+    output_type = "string"
+    default_checkpoint = "openai/whisper-1"
+
+    def __call__(self, audio: str) -> str:
+        return self._transcribe(audio)
+
+    @staticmethod
+    def _transcribe(audio_path: str) -> str:
+        path = Path(audio_path).expanduser().resolve()
+        if not path.is_file():
+            raise FileNotFoundError(f"Audio file not found: {path}")
+        with path.open("rb") as fp:
+            response = openai.audio.transcriptions.create(
+                file=fp,
+                model="whisper-1",
+                response_format="text"
+            )
+        return response  # already plain text
+
+class ExcelToTextTool(Tool):
+    """Renders an Excel worksheet as Markdown text."""
+    name = "excel_to_text"
+    description = "Read an Excel file and return a Markdown table."
+    inputs = {
+        "excel_path": {"type": "string", "description": "Path to Excel file."},
+        "sheet_name": {"type": "string", "description": "Sheet name or index as string.", "nullable": True},
+    }
+    output_type = "string"
+
+    def forward(self, excel_path: str, sheet_name: str = None) -> str:
+        path = Path(excel_path).expanduser().resolve()
+        if not path.exists():
+            return f"Error: Excel file not found at {path}"
+        # Determine sheet
+        sheet = 0 if not sheet_name or sheet_name == "" else (int(sheet_name) if sheet_name.isdigit() else sheet_name)
+        df = pd.read_excel(path, sheet_name=sheet)
+        if hasattr(df, "to_markdown"):
+            return df.to_markdown(index=False)
+        return tabulate(df, headers="keys", tablefmt="github", showindex=False)