Spaces:
Sleeping
Sleeping
File size: 6,094 Bytes
396f5a0 100ac43 396f5a0 100ac43 396f5a0 100ac43 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 |
from smolagents import Tool, tool
from youtube_transcript_api import YouTubeTranscriptApi
from transformers import AutoModelForSpeechSeq2Seq, AutoProcessor, pipeline
import torch
@tool
def fetch_webpage(url: str, convert_to_markdown: bool = True) -> str:
"""
Visit a website / url and fetch the content of the webpage.
if markdown conversion is enabled, it will remove script and style and return the text content as markdown else return raw unfiltered HTML
Args:
url (str): The URL to fetch.
convert_to_markdown (bool): If True, convert the HTML content to Markdown format. else return the raw HTML.
Returns:
str: The HTML content of the URL.
"""
import requests
from bs4 import BeautifulSoup
from markdownify import markdownify as md
content = None
headers = {
'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36'
}
response = requests.get(url, timeout=30, headers=headers)
if (convert_to_markdown):
soup = BeautifulSoup(response.text, "html.parser")
# remove script and style tags
for script in soup(["script", "style"]):
script.extract()
# for wikipedia only keep the main content
if "wikipedia.org" in url:
elements_to_remove = [
# Navigation and reference elements
{'class': 'navbox'},
{'class': 'navbox-group'},
{'class': 'reflist'},
{'class': 'navigation-box'},
{'class': 'sister-project'},
{'class': 'metadata'},
{'class': 'interlanguage-link'},
{'class': 'catlinks'},
{'id': 'References'},
{'id': 'External_links'},
{'id': 'Further_reading'},
{'id': 'See_also'},
{'id': 'Notes'},
]
for selector in elements_to_remove:
elements = soup.find_all(attrs=selector)
for element in elements:
# For ID-based elements, remove the parent section
if 'id' in selector:
parent = element.parent
if parent and parent.name in ['h2', 'h3', 'h4']:
# Remove heading and all content until next heading
current = parent
while current and current.next_sibling:
next_elem = current.next_sibling
if (hasattr(next_elem, 'name') and
next_elem.name in ['h2', 'h3', 'h4']):
break
if hasattr(next_elem, 'decompose'):
next_elem.decompose()
else:
current = next_elem
parent.decompose()
else:
element.decompose()
main_content = soup.find("main",{"id":"content"})
if main_content:
content = md(str(main_content),strip=['script', 'style'], heading_style="ATX").strip()
else:
content = md(response.text,strip=['script', 'style'], heading_style="ATX").strip()
else:
# Fallback for all other sites - from chatgpt - not tested
content = md(str(soup), strip=['script', 'style'], heading_style="ATX").strip()
else:
content = response.text
return content
@tool
def read_file_tool(file_path: str) -> str:
"""
Tool to read a file and return its content.
Args:
file_path (str): Path to the file to read.
Returns:
str: Content of the file or error message.
"""
try:
with open(file_path, "r") as file:
return file.read()
except Exception as e:
return f"Error reading file: {str(e)}"
@tool
def get_youtube_transcript(video_id: str) -> str:
"""
Fetches the transcript of a YouTube video given its video ID.
Args:
video_id (str): The ID of the YouTube video. Pass in the video ID, NOT the video URL. For a video with the URL https://www.youtube.com/watch?v=12345 the ID is 12345.
Returns:
str: The transcript of the YouTube video. as a single string with each line separated by a newline character.
"""
# Initialize the YouTubeTranscriptApi
ytt_api = YouTubeTranscriptApi()
fetched_transcript = ytt_api.fetch(video_id)
raw_data = fetched_transcript.to_raw_data()
# raw data is in the form of [{ 'text': 'Hey there', 'start': 0.0, 'duration': 1.54 }, { 'text': 'how are you',, 'start': 1.54, 'duration': 4.16 }, ... ] we will return ony the text element as lines
transcript = "\n".join([item['text'] for item in raw_data])
return transcript
@tool
def transcribe_audio(audio_path: str) -> str:
"""
Speech to Text - transcribes audio file and returns the text
Args:
audio_path (str): Local file path to the audio
Returns:
str: The transcript of the audio file
"""
device = "cuda:0" if torch.cuda.is_available() else "cpu"
torch_dtype = torch.float16 if torch.cuda.is_available() else torch.float32
model_id = "openai/whisper-small"
model = AutoModelForSpeechSeq2Seq.from_pretrained(
model_id, torch_dtype=torch_dtype, low_cpu_mem_usage=True, use_safetensors=True
)
model.to(device)
processor = AutoProcessor.from_pretrained(model_id)
pipe = pipeline(
"automatic-speech-recognition",
model=model,
tokenizer=processor.tokenizer,
feature_extractor=processor.feature_extractor,
torch_dtype=torch_dtype,
device=device,
chunk_length_s=30,
)
result = pipe(audio_path)
return result
|