Commit
·
7c8e9de
1
Parent(s):
9738813
Add the missing duckdukgo function
Browse files- agent.py +67 -43
- requirements.txt +2 -1
agent.py
CHANGED
@@ -1,5 +1,8 @@
|
|
1 |
from typing import TypedDict, Annotated
|
2 |
import os
|
|
|
|
|
|
|
3 |
from langchain_community.tools import DuckDuckGoSearchRun
|
4 |
from langchain_community.document_loaders import WikipediaLoader, YoutubeLoader
|
5 |
from langchain_community.document_loaders.youtube import TranscriptFormat
|
@@ -57,49 +60,6 @@ def analyze_youtube_video(video_url: str) -> str:
|
|
57 |
Returns:
|
58 |
str: The transcript content of the YouTube video
|
59 |
"""
|
60 |
-
# try:
|
61 |
-
# # Method 1: Try with basic YoutubeLoader first
|
62 |
-
# try:
|
63 |
-
# loader = YoutubeLoader.from_youtube_url(
|
64 |
-
# video_url,
|
65 |
-
# add_video_info=True,
|
66 |
-
# language=["en", "en-US", "en-GB"] # Try multiple English variants
|
67 |
-
# )
|
68 |
-
# docs = loader.load()
|
69 |
-
|
70 |
-
# if docs:
|
71 |
-
# content = ""
|
72 |
-
# for doc in docs:
|
73 |
-
# title = doc.metadata.get('title', 'Unknown Video')
|
74 |
-
# author = doc.metadata.get('author', 'Unknown Author')
|
75 |
-
# length = doc.metadata.get('length', 'Unknown')
|
76 |
-
|
77 |
-
# content += f"Video Title: {title}\n"
|
78 |
-
# content += f"Author: {author}\n"
|
79 |
-
# content += f"Length: {length} seconds\n"
|
80 |
-
# content += f"Transcript:\n{doc.page_content}\n\n"
|
81 |
-
|
82 |
-
# return content
|
83 |
-
# except Exception as e1:
|
84 |
-
# print(f"Method 1 failed: {e1}")
|
85 |
-
|
86 |
-
# Method 2: Try without video info
|
87 |
-
# try:
|
88 |
-
# loader = YoutubeLoader.from_youtube_url(
|
89 |
-
# video_url,
|
90 |
-
# add_video_info=False,
|
91 |
-
# language=["en"]
|
92 |
-
# )
|
93 |
-
# docs = loader.load()
|
94 |
-
|
95 |
-
# if docs:
|
96 |
-
# content = f"Video URL: {video_url}\n"
|
97 |
-
# content += f"Transcript:\n{docs[0].page_content}\n\n"
|
98 |
-
# return content
|
99 |
-
# except Exception as e2:
|
100 |
-
# print(f"Method 2 failed: {e2}")
|
101 |
-
|
102 |
-
# # Method 3: Try with chunked format
|
103 |
try:
|
104 |
loader = YoutubeLoader.from_youtube_url(
|
105 |
video_url,
|
@@ -118,6 +78,70 @@ def analyze_youtube_video(video_url: str) -> str:
|
|
118 |
except Exception as e:
|
119 |
print(f"Analyze video failed: {e}")
|
120 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
121 |
# Initialize Langfuse CallbackHandler globally
|
122 |
def get_langfuse_handler():
|
123 |
"""Get configured Langfuse handler"""
|
|
|
1 |
from typing import TypedDict, Annotated
|
2 |
import os
|
3 |
+
import requests
|
4 |
+
from urllib.parse import urlparse
|
5 |
+
from pathlib import Path
|
6 |
from langchain_community.tools import DuckDuckGoSearchRun
|
7 |
from langchain_community.document_loaders import WikipediaLoader, YoutubeLoader
|
8 |
from langchain_community.document_loaders.youtube import TranscriptFormat
|
|
|
60 |
Returns:
|
61 |
str: The transcript content of the YouTube video
|
62 |
"""
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
63 |
try:
|
64 |
loader = YoutubeLoader.from_youtube_url(
|
65 |
video_url,
|
|
|
78 |
except Exception as e:
|
79 |
print(f"Analyze video failed: {e}")
|
80 |
|
81 |
+
# Create file download tool
|
82 |
+
@tool
|
83 |
+
def download_file(url: str, filename: str = None, download_dir: str = "downloaded_files") -> str:
|
84 |
+
"""Download a file from a URL and save it locally.
|
85 |
+
|
86 |
+
Args:
|
87 |
+
url: The URL of the file to download
|
88 |
+
filename: Optional custom filename. If not provided, will extract from URL
|
89 |
+
download_dir: Directory to save the file (default: 'downloaded_files')
|
90 |
+
|
91 |
+
Returns:
|
92 |
+
str: Success message with file path or error message
|
93 |
+
"""
|
94 |
+
try:
|
95 |
+
# Create download directory if it doesn't exist
|
96 |
+
Path(download_dir).mkdir(exist_ok=True)
|
97 |
+
|
98 |
+
# Parse URL to get filename if not provided
|
99 |
+
if not filename:
|
100 |
+
parsed_url = urlparse(url)
|
101 |
+
filename = os.path.basename(parsed_url.path)
|
102 |
+
if not filename:
|
103 |
+
filename = "downloaded_file"
|
104 |
+
|
105 |
+
# Full file path
|
106 |
+
file_path = os.path.join(download_dir, filename)
|
107 |
+
|
108 |
+
# Download the file
|
109 |
+
print(f"Downloading file from: {url}")
|
110 |
+
response = requests.get(url, stream=True, timeout=30)
|
111 |
+
response.raise_for_status()
|
112 |
+
|
113 |
+
# Get file size for progress tracking
|
114 |
+
total_size = int(response.headers.get('content-length', 0))
|
115 |
+
|
116 |
+
# Write file in chunks
|
117 |
+
with open(file_path, 'wb') as file:
|
118 |
+
downloaded = 0
|
119 |
+
for chunk in response.iter_content(chunk_size=8192):
|
120 |
+
if chunk:
|
121 |
+
file.write(chunk)
|
122 |
+
downloaded += len(chunk)
|
123 |
+
|
124 |
+
# Get file info
|
125 |
+
file_size = os.path.getsize(file_path)
|
126 |
+
file_size_mb = file_size / (1024 * 1024)
|
127 |
+
|
128 |
+
return f"""File downloaded successfully!
|
129 |
+
|
130 |
+
File Details:
|
131 |
+
- URL: {url}
|
132 |
+
- Saved as: {file_path}
|
133 |
+
- File size: {file_size_mb:.2f} MB
|
134 |
+
- Content type: {response.headers.get('content-type', 'Unknown')}
|
135 |
+
|
136 |
+
You can now access the file at: {os.path.abspath(file_path)}"""
|
137 |
+
|
138 |
+
except requests.exceptions.RequestException as e:
|
139 |
+
return f"Error downloading file from {url}: Network error - {str(e)}"
|
140 |
+
except OSError as e:
|
141 |
+
return f"Error saving file: File system error - {str(e)}"
|
142 |
+
except Exception as e:
|
143 |
+
return f"Error downloading file from {url}: {str(e)}"
|
144 |
+
|
145 |
# Initialize Langfuse CallbackHandler globally
|
146 |
def get_langfuse_handler():
|
147 |
"""Get configured Langfuse handler"""
|
requirements.txt
CHANGED
@@ -6,4 +6,5 @@ langgraph
|
|
6 |
langfuse
|
7 |
langchain-openai
|
8 |
youtube-transcript-api
|
9 |
-
pytube
|
|
|
|
6 |
langfuse
|
7 |
langchain-openai
|
8 |
youtube-transcript-api
|
9 |
+
pytube
|
10 |
+
duckduckgo-search
|