implement functions to fact check

Implement functions:
reliable_url: returns if an url is part of a selection of reliable sources
url_to_claim: gets the message of the news
support_decision: decides if a news is supporting a given claim
web_search: search for news related to the main claim
check: returns the final decision about some claim based on the web results

app.py

@@ -1,69 +0,0 @@
-from smolagents import CodeAgent,DuckDuckGoSearchTool, HfApiModel,load_tool,tool
-import datetime
-import requests
-import pytz
-import yaml
-from tools.final_answer import FinalAnswerTool
-
-from Gradio_UI import GradioUI
-
-# Below is an example of a tool that does nothing. Amaze us with your creativity !
-@tool
-def my_custom_tool(arg1:str, arg2:int)-> str: #it's import to specify the return type
-    #Keep this format for the description / args / args description but feel free to modify the tool
-    """A tool that does nothing yet 
-    Args:
-        arg1: the first argument
-        arg2: the second argument
-    """
-    return "What magic will you build ?"
-
-@tool
-def get_current_time_in_timezone(timezone: str) -> str:
-    """A tool that fetches the current local time in a specified timezone.
-    Args:
-        timezone: A string representing a valid timezone (e.g., 'America/New_York').
-    """
-    try:
-        # Create timezone object
-        tz = pytz.timezone(timezone)
-        # Get current time in that timezone
-        local_time = datetime.datetime.now(tz).strftime("%Y-%m-%d %H:%M:%S")
-        return f"The current local time in {timezone} is: {local_time}"
-    except Exception as e:
-        return f"Error fetching time for timezone '{timezone}': {str(e)}"
-
-
-final_answer = FinalAnswerTool()
-
-# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
-# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud' 
-
-model = HfApiModel(
-max_tokens=2096,
-temperature=0.5,
-model_id='Qwen/Qwen2.5-Coder-32B-Instruct',# it is possible that this model may be overloaded
-custom_role_conversions=None,
-)
-
-
-# Import tool from Hub
-image_generation_tool = load_tool("agents-course/text-to-image", trust_remote_code=True)
-
-with open("prompts.yaml", 'r') as stream:
-    prompt_templates = yaml.safe_load(stream)
-    
-agent = CodeAgent(
-    model=model,
-    tools=[final_answer], ## add your tools here (don't remove final answer)
-    max_steps=6,
-    verbosity_level=1,
-    grammar=None,
-    planning_interval=None,
-    name=None,
-    description=None,
-    prompt_templates=prompt_templates
-)
-
-
-GradioUI(agent).launch()

fact_checker.py

@@ -0,0 +1,179 @@
+from smolagents import ToolCallingAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
+import requests
+from typing import Literal
+from bs4 import BeautifulSoup
+import pytz
+import yaml
+from tools.final_answer import FinalAnswerTool
+
+from Gradio_UI import GradioUI
+
+def reliable_url(url)-> bool:
+    """A tool that decides if a given url is a reliable source
+    Args:
+        url: the address of the webpage we want to know whether it is realible or not
+    """
+    known_fact_checkers = ["reuters.com/fact-check",
+                              "g1.globo.com/fato-ou-fake",
+                              "estadao.com.br/estadao-verifica",
+                              "snopes.com",
+                              "factcheck.org"]
+    known_academic_sources = ['.edu', '.ac.']
+    known_news = ["globo.com",
+                  "bbc.com",
+                  "midianinja.org",
+                  "estadao.com",
+                  "brasil247.com",
+                  "espn.com.br"
+                 ]
+
+    reliable = known_academic_sources + known_fact_checkers + known_news
+    
+    for pattern in reliable:
+        if pattern in url:
+            return True
+    return False
+
+def url_to_claim(url:str)-> dict:
+    """A tool that receives a news url and returns the main claim of the webpage content and some aditional context
+    Args:
+        url: the address of the webpage we want to summarize into main claim and context
+    """
+    try:
+        response = requests.get(url, timeout=10)
+        response.raise_for_status()
+    except:
+        requests.exceptions.RequestException
+        return f"An error occurred: {requests.exceptions.RequestExceptions}"
+
+    soup = BeautifulSoup(response.content, "html.parser")
+    title = soup.title.text
+    context = soup.get_text()[:2000]
+
+    return {"claim": title, "context": context}
+
+def support_decision(claim:str, news:dict)-> Literal["supports", "contradicts", "unclear"]:
+    """A tool that decides if a given news article supports, contradicts or have an unclear opinion about a given claim
+    Args:
+        claim: the claim we want to decide whether it is supported or contradicted
+        news: the news that may be supporting or contradicting the claim
+    """
+    model = HfApiModel()
+
+    prompt = f"""
+    Analyze the sentiment of the provided news article in contrast with the given claim and respond with only one of the following words 
+    'supports', 'contradicts', or 'unclear'.
+
+    Claim: "{claim}"
+    
+    News: "{news}"
+    
+    Response:
+    """
+
+    response = model.run(prompt)
+    response = response.strip().lower()
+
+    if response in ["supports", "contradicts", "unclear"]:
+        return response
+    else:
+        return "An error ocurred: the model was not able to analyze the news article"
+
+def web_search(claim:str)-> list:
+    """A tool that receives a claim and gather news about it by making a web search
+    Args:
+        claim: the claim we are searching about
+    """
+    search_tool = DuckDuckGoSearchTool()
+    url_pattern = r'https?://[^\s\)\]\}]+'
+
+    query1 = f"{claim}"
+    results1 = search_tool(query1)
+    urls = re.findall(url_pattern, results1)
+
+    query2 = f"{claim} fact check"
+    results2 = search_tool(query2)
+    urls.append(re.findall(url_pattern, results2))
+    
+    return urls
+
+@tool
+def check(claim:str)-> str:
+    """A tool that receives a claim and answers if that information has support from reliable sources or not. This is the first tool you should use.
+    Args:
+        claim: the claim we want to check for support
+    """
+    support_reliable = []
+    contradict_reliable = []
+    support_others = []
+    contradict_others = []
+    
+    news_articles = web_search(claim)
+    for article in news_articles:
+        reliable_source = reliable_url(article)
+        news = url_to_claim(article)
+        decision = support_decision(claim, news)
+        if reliable_source:
+            if decision == "support":
+                support_reliable.append(article)
+            elif decision == "contradict":
+                contradict_reliable.append(article)
+        else:
+            if decision == "support":
+                support_others.append(article)
+            elif decision == "contradict":
+                contradict_others.append(article)
+
+    if len(support_reliable) == 0:
+        if len(contradict_reliable) > 0:
+            return f"There are reliable sources that contradict this information: {contradict_reliable}"
+        else:
+            if len(support_others) == 0:
+                if len(contradict_others) == 0:
+                    return "There are no sources available that talk about this topic"
+                else:
+                    return f"There are unverified sources that contradict this claim: {contradict_others}"
+            else:
+                if len(contradict_others) == 0:
+                    return f"There are unverified sources that support this information: {support_others}"
+                else:
+                    return f"There are unverified sources that contradict this claim and other unverified sources that support it: {contradict_others + support_others}"
+    else:
+        if len(contradict_reliable) == 0:
+            return f"There are reliable sources that support this information: {support_reliable}"
+        else:
+            return f"The information about this claim is inconsistent. More search is advised: {support_reliable, contradict_reliable}"
+final_answer = FinalAnswerTool()
+
+# If the agent does not answer, the model is overloaded, please use another model or the following Hugging Face Endpoint that also contains qwen2.5 coder:
+# model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud' 
+
+model = HfApiModel(
+max_tokens=2096,
+temperature=0.5,
+model_id='Qwen/Qwen2.5-Coder-32B-Instruct',
+custom_role_conversions=None,
+)
+
+
+with open("prompts.yaml", 'r') as stream:
+    prompt_templates = yaml.safe_load(stream)
+    
+agent = ToolCallingAgent(
+    model=model,
+    tools=[final_answer,
+           reliable_url,
+           url_to_claim,
+           claim_check,
+           DuckDuckGoSearchTool],
+    max_steps=6,
+    verbosity_level=1,
+    grammar=None,
+    planning_interval=None,
+    name=None,
+    description=None,
+    prompt_templates=prompt_templates
+)
+
+
+GradioUI(agent).launch()