update sentiment_tool.py

tools/sentiment_tool.py

@@ -1,5 +1,3 @@
-# tools/sentiment_tool.py
-
 import os
 import requests
 from crewai.tools import BaseTool
@@ -7,47 +5,41 @@ from openai import OpenAI
 from typing import Type
 from pydantic import BaseModel, Field
 
-# -------------------------
-# Environment Variables
-# -------------------------
 SERPER_API_KEY = os.getenv("SERPER_API_KEY")
 OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
 
 client = OpenAI(api_key=OPENAI_API_KEY)
 
-# -------------------------
-# Input Schema
-# -------------------------
+# ------------------------
+# INPUT SCHEMA
+# ------------------------
 class SentimentInput(BaseModel):
-    query: str = Field(default="bitcoin", description="Cryptocurrency name to analyze.")
+    query: str = Field(default="bitcoin", description="Cryptocurrency name to evaluate sentiment for.")
 
-# -------------------------
-# Sentiment Tool
-# -------------------------
+# ------------------------
+# SENTIMENT TOOL
+# ------------------------
 class SentimentTool(BaseTool):
     name: str = "get_crypto_sentiment"
     description: str = (
-        "Fetches cryptocurrency sentiment using Serper (news + Reddit), "
-        "then uses GPT-4.1 to classify sentiment as bullish, bearish, or neutral."
+        "Fetches recent cryptocurrency news and Reddit discussions using Serper.dev, "
+        "then performs sentiment analysis using OpenAI GPT. Returns structured JSON."
     )
-    args_schema: Type[BaseModel] = SentimentInput
-
-    def _run(self, query: str = "bitcoin") -> dict:
-        headers = {
-            "X-API-KEY": SERPER_API_KEY,
-            "Content-Type": "application/json"
-        }
+    arg_schema: Type[BaseModel] = SentimentInput
 
-        # ----------------------------------------
-        # 1) FETCH NEWS USING SERPER
-        # ----------------------------------------
+    def _run(self, query: str = "bitcoin") -> str:
+        # ============================
+        # 1) FETCH NEWS VIA SERPER
+        # ============================
         news_headlines = []
         news_error = None
+
         try:
             news_payload = {
-                "q": f"{query} cryptocurrency",
-                "num": 5
+                "q": f"{query} crypto news",
+                "num": 10
             }
+            headers = {"X-API-KEY": SERPER_API_KEY, "Content-Type": "application/json"}
 
             news_res = requests.post(
                 "https://google.serper.dev/news",
@@ -57,24 +49,28 @@ class SentimentTool(BaseTool):
             )
             news_res.raise_for_status()
 
-            news_json = news_res.json()
-            news_items = news_json.get("news", [])
-            news_headlines = [item.get("title") for item in news_items if "title" in item]
+            news_json = news_res.json().get("news", [])
+            news_headlines = [n.get("title") for n in news_json if n.get("title")]
+            news_headlines = news_headlines[:10]
 
         except Exception as e:
             news_error = str(e)
 
-        # ----------------------------------------
-        # 2) FETCH REDDIT POSTS USING SERPER (CORRECT ENDPOINT)
-        # ----------------------------------------
+        # ============================
+        # 2) FORCED REDDIT SCRAPING (RELIABLE)
+        # ============================
         reddit_titles = []
         reddit_error = None
 
         try:
+            # Serper search that *forces* Reddit results
             reddit_payload = {
-                "q": query,
-                "num": 5,
-                "reddit": True  # <-- THIS IS THE CORRECT WAY TO FILTER REDDIT RESULTS
+                "q": (
+                    f"site:reddit.com/r/cryptocurrency OR "
+                    f"site:reddit.com/r/{query} "
+                    f"{query} discussion latest"
+                ),
+                "num": 10
             }
 
             reddit_res = requests.post(
@@ -86,45 +82,45 @@ class SentimentTool(BaseTool):
             reddit_res.raise_for_status()
 
             reddit_json = reddit_res.json()
-            reddit_items = reddit_json.get("reddit", [])
+            organic_results = reddit_json.get("organic", [])
 
             reddit_titles = [
                 item.get("title")
-                for item in reddit_items
-                if "title" in item
+                for item in organic_results
+                if "reddit.com" in item.get("link", "")
             ]
 
+            reddit_titles = reddit_titles[:5]
+
         except Exception as e:
             reddit_error = str(e)
 
-        # ----------------------------------------
-        # 3) COMBINE TEXT FOR SENTIMENT CLASSIFICATION
-        # ----------------------------------------
+        # ============================
+        # 3) SENTIMENT ANALYSIS
+        # ============================
         combined_text = (
             "News Headlines:\n" + "\n".join(news_headlines) +
             "\n\nReddit Posts:\n" + "\n".join(reddit_titles)
         )
 
-        # ----------------------------------------
-        # 4) GPT-4.1 SENTIMENT INTERPRETATION
-        # ----------------------------------------
         sentiment_prompt = f"""
 You are a cryptocurrency sentiment analyst.
 
-Classify sentiment toward **{query}** based on NEWS and REDDIT content.
+Based on the following combined news headlines and Reddit discussions, classify the overall sentiment toward "{query}" as **bullish**, **bearish**, or **neutral**.
 
-Return ONLY valid JSON using this format:
+Return only valid JSON in this format:
 
 {{
-  "sentiment": "bullish" | "bearish" | "neutral",
+  "sentiment": "bullish/bearish/neutral",
   "reasoning": "short explanation",
   "news_headlines": [...],
   "reddit_titles": [...],
-  "news_error": "... or null",
-  "reddit_error": "... or null"
+  "news_error": null or string,
+  "reddit_error": null or string
 }}
 
-Content to analyze:
+CONTENT TO ANALYSE:
+-------------------
 {combined_text}
 """
 
@@ -132,29 +128,22 @@ Content to analyze:
             completion = client.chat.completions.create(
                 model="gpt-4.1",
                 messages=[
-                    {"role": "system", "content": "You extract structured sentiment from crypto news and Reddit posts."},
+                    {"role": "system", "content": "You are a precise sentiment classifier. Respond only with JSON."},
                     {"role": "user", "content": sentiment_prompt}
                 ],
                 temperature=0.2
             )
-
-            # GPT returns JSON string → CrewAI needs dict
-            result_text = completion.choices[0].message.content
-            import json
-            parsed = json.loads(result_text)
-
-            # Ensure errors included even if GPT forgets them
-            parsed["news_error"] = news_error
-            parsed["reddit_error"] = reddit_error
-
-            return parsed
+            sentiment_json = completion.choices[0].message.content
+            return sentiment_json
 
         except Exception as e:
+            # Return structured failure JSON for debugging
             return {
                 "sentiment": "unknown",
-                "reasoning": "GPT sentiment classification failed.",
+                "reasoning": "LLM sentiment analysis failed.",
                 "news_headlines": news_headlines,
                 "reddit_titles": reddit_titles,
                 "news_error": news_error,
-                "reddit_error": str(e)
+                "reddit_error": reddit_error,
+                "llm_error": str(e)
             }