Fix tools bugs

requirements.txt

@@ -1,16 +1,21 @@
+-i https://pypi.org/simple
 aiofiles==24.1.0 ; python_version >= '3.8'
 annotated-types==0.7.0 ; python_version >= '3.8'
 anyio==4.9.0 ; python_version >= '3.9'
+beautifulsoup4==4.13.4 ; python_full_version >= '3.7.0'
 boto3==1.38.23
 botocore==1.38.23 ; python_version >= '3.9'
 certifi==2025.4.26 ; python_version >= '3.6'
 charset-normalizer==3.4.2 ; python_version >= '3.7'
 click==8.2.1 ; python_version >= '3.10'
+distro==1.9.0 ; python_version >= '3.6'
+duckduckgo-search==8.0.2
 et-xmlfile==2.0.0 ; python_version >= '3.8'
 fastapi==0.115.12 ; python_version >= '3.8'
 ffmpy==0.5.0 ; python_version >= '3.8' and python_version < '4.0'
 filelock==3.18.0 ; python_version >= '3.9'
 fsspec==2025.5.1 ; python_version >= '3.9'
+googlesearch-python==1.3.0
 gradio==5.31.0
 gradio-client==1.10.1 ; python_version >= '3.10'
 groovy==0.1.2 ; python_version >= '3.10'
@@ -21,6 +26,7 @@ httpx==0.28.1 ; python_version >= '3.8'
 huggingface-hub==0.32.1 ; python_full_version >= '3.8.0'
 idna==3.10 ; python_version >= '3.6'
 jinja2==3.1.6 ; python_version >= '3.7'
+jiter==0.10.0 ; python_version >= '3.9'
 jmespath==1.0.1 ; python_version >= '3.7'
 jsonpatch==1.33 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3, 3.4, 3.5, 3.6'
 jsonpointer==3.0.0 ; python_version >= '3.7'
@@ -30,16 +36,20 @@ langgraph-checkpoint==2.0.26 ; python_version >= '3.9'
 langgraph-prebuilt==0.2.1 ; python_version >= '3.9'
 langgraph-sdk==0.1.70 ; python_version >= '3.9'
 langsmith==0.3.42 ; python_version >= '3.9'
+lxml==5.4.0 ; python_version >= '3.6'
 markdown-it-py==3.0.0 ; python_version >= '3.8'
+markdownify==1.1.0
 markupsafe==3.0.2 ; python_version >= '3.9'
 mdurl==0.1.2 ; python_version >= '3.7'
 numpy==2.2.6 ; python_version >= '3.10'
+openai==1.82.0
 openpyxl==3.1.5
 orjson==3.10.18 ; python_version >= '3.9'
 ormsgpack==1.10.0 ; python_version >= '3.9'
 packaging==24.2 ; python_version >= '3.8'
 pandas==2.2.3 ; python_version >= '3.9'
 pillow==11.2.1 ; python_version >= '3.9'
+primp==0.15.0 ; python_version >= '3.8'
 pydantic==2.11.5 ; python_version >= '3.9'
 pydantic-core==2.33.2 ; python_version >= '3.9'
 pydub==0.25.1
@@ -61,6 +71,7 @@ shellingham==1.5.4 ; python_version >= '3.7'
 six==1.17.0 ; python_version >= '2.7' and python_version not in '3.0, 3.1, 3.2, 3.3'
 smolagents==1.16.1
 sniffio==1.3.1 ; python_version >= '3.7'
+soupsieve==2.7 ; python_version >= '3.8'
 starlette==0.46.2 ; sys_platform != 'emscripten'
 tenacity==9.1.2 ; python_version >= '3.9'
 tomlkit==0.13.2 ; python_version >= '3.8'

tools.py

@@ -8,8 +8,13 @@ from uuid import uuid4
 
 import boto3
 import fitz
+import requests
+from bs4 import BeautifulSoup
+from googlesearch import search
 from pandas import read_excel
 from smolagents import tool, Tool
+from requests.exceptions import HTTPError
+from urllib3.exceptions import ReadTimeoutError
 
 from definitions import TranscriptionJob
 from utils import get_file, s3_upload_file, s3_download_file, bedrock_runtime, BEDROCK_MODEL_ID
@@ -45,6 +50,7 @@ def excel_reader(task_id: str, file_name: str) -> str:
         return f"Error reading Excel file {file_name}: {e}"
 
 
+@tool
 def txt_reader(task_id: str, file_name: str) -> str:
     """Reads a text file and returns its content as a string.
 
@@ -59,6 +65,7 @@ def txt_reader(task_id: str, file_name: str) -> str:
         return f"Error reading file {file_name}: {e}"
 
 
+@tool
 def pdf_reader(task_id: str, file_name: str) -> str:
     """Reads a PDF file and returns its content as a string.
 
@@ -91,6 +98,7 @@ class AudioTranscriber(Tool):  # pylint: disable=C0115
             "description": "The name of the audio file to transcribe.",
         },
     }
+    output_type = "string"
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
@@ -151,7 +159,7 @@ def image_transcriber(text_prompt: str, task_id: str, file_name: str) -> str:
     """
     try:
         file_content = get_file(task_id)
-        base64_image = base64.b64encode(file_content).decode("utf-8")
+        base64_image = base64.b64encode(file_content.getvalue()).decode("utf-8")
         response = bedrock_runtime.invoke_model(
             modelId=BEDROCK_MODEL_ID,
             body=json.dumps(
@@ -176,7 +184,52 @@ def image_transcriber(text_prompt: str, task_id: str, file_name: str) -> str:
                     ],
                 }
             ),
-        ).read()
-        return json.loads(response)["message"]["content"][0]["text"]
+        )["body"].read()
+        return json.loads(response)["content"][0]["text"]
     except Exception as e:
         return f"Error processing image file {file_name}: {e}"
+
+
+def _get_content(url: str, timeout: int = 5) -> bytes:
+    resp = requests.get(url=url, timeout=timeout)
+    resp.raise_for_status()
+    return resp.content
+
+
+def _js_disable_message(text: str) -> bool:
+    return "JavaScript is disabled in this browser" in text
+
+
+@tool
+def search_engine(search_term: str) -> str:
+    """Search for the provided search term in Google Search
+
+    Args:
+        search_term (str): The term to search for on the web.
+    """
+    results = search(search_term, num_results=5)
+    for idx, url in enumerate(results, 1):
+        error_ocurred = False
+        try:
+            html_content = BeautifulSoup(_get_content(url), "html.parser")
+            # Remove headers and footers
+            for tag in html_content.find_all(["header", "footer", "nav", "aside"]):
+                tag.decompose()
+        except (ReadTimeoutError, HTTPError) as ex:
+            print("Got HTTP error when requesting %s. Error %s", url, ex)
+            error_ocurred = True
+
+        html_text = html_content.text
+        if _js_disable_message(html_text):
+            error_ocurred = True
+
+        if error_ocurred:
+            # if the last URL is not successfully requested, return a direct response as if it was the final answer
+            if len(results) == idx:
+                return "Sorry, got an HTTP error when requesting the internet"
+            # if there are more URLs to request, continue
+            continue
+
+        return html_text.replace("\n", "")
+
+    return "Could not retrieve any content from the search results."

utils.py

@@ -3,12 +3,9 @@ from io import BytesIO
 
 import boto3
 import requests
-from dotenv import load_dotenv
 
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
-BEDROCK_MODEL_ID = "anthropic.claude-3-5-sonnet-20241022-v2:0"
-
-load_dotenv()
+BEDROCK_MODEL_ID = "anthropic.claude-3-5-sonnet-20240620-v1:0"
 
 bedrock_runtime = boto3.client("bedrock-runtime", region_name=os.getenv("AWS_REGION"))