Spaces:
Sleeping
Sleeping
File size: 16,010 Bytes
4565986 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 |
import os
import pandas as pd
# import wikipediaapi
from markdownify import markdownify as md
from smolagents import tool, LiteLLMModel
import whisper
from youtube_transcript_api import YouTubeTranscriptApi
from youtube_transcript_api.formatters import JSONFormatter
import base64
import mimetypes
import requests # Keep for consistency, though not used for fetching image in this version
import os # Added for os.path.join
import re
from bs4 import BeautifulSoup, Tag, Comment
# that could be better done via a managed agent, but this is a quick hack to get it working
@tool
def describe_image_file(local_image_path: str) -> str:
"""
Describe the contents of a local image file in detail and return the description as text.
Args:
local_image_path (str): The path to the local image file to be described.
Returns:
str: A detailed description of the image contents.
"""
model = LiteLLMModel(
model_id='ollama/gemma3:27b',
api_base="https://192.168.5.217:8000", # replace with remote open-ai compatible server if necessary
api_key=os.getenv("OLLAMA_REVPROXY_SRVML"),
num_ctx=16384, # ollama default is 2048 which will often fail horribly. 8192 works for easy tasks, more is better. Check https://huggingface.co/spaces/NyxKrage/LLM-Model-VRAM-Calculator to calculate how much VRAM this will need for the selected model
ssl_verify=False, # Explicitly disable SSL verification
extra_headers={
"Authorization": f"Bearer {os.getenv('OLLAMA_REVPROXY_SRVML')}", # Explicitly set auth header
},
flatten_messages_as_text = False
)
text_prompt = "What is in this image? Describe it in detail."
try:
if not os.path.exists(local_image_path):
raise FileNotFoundError(f"Image file not found at {local_image_path}. Please ensure it was downloaded correctly.")
# 1. Read the image content from the local file
with open(local_image_path, "rb") as image_file:
image_content_bytes = image_file.read()
# 2. Base64 encode the image content
base64_image_bytes = base64.b64encode(image_content_bytes)
base64_image_string = base64_image_bytes.decode('utf-8')
# 3. Set MIME type based on file extension
if local_image_path.lower().endswith('.png'):
content_type = 'image/png'
elif local_image_path.lower().endswith('.jpg') or local_image_path.lower().endswith('.jpeg'):
content_type = 'image/jpeg'
elif local_image_path.lower().endswith('.gif'):
content_type = 'image/gif'
elif local_image_path.lower().endswith('.bmp'):
content_type = 'image/bmp'
elif local_image_path.lower().endswith('.webp'):
content_type = 'image/webp'
else:
content_type = mimetypes.guess_type(local_image_path)[0] or 'application/octet-stream'
print(f"Using specified MIME type: {content_type}")
# 4. Construct the data URI
data_uri = f"data:{content_type};base64,{base64_image_string}"
# Construct the messages payload
messages = [
{
"role": "user",
"content": [
{"type": "text", "text": text_prompt},
{
"type": "image_url",
"image_url": {
"url": data_uri # Use the base64 data URI here
}
}
]
}
]
# Assuming 'model' is your LiteLLMModel instance initialized in a previous cell (e.g., cell 'dfc845ab')
if 'model' not in locals():
raise NameError("Variable 'model' is not defined. Please run the cell that initializes the LiteLLMModel.")
response = model.generate(messages)
return response
except FileNotFoundError as fnf_err:
print(f"File error: {fnf_err}")
except NameError as ne:
print(f"A required variable might not be defined (e.g., filename, model): {ne}")
print("Please ensure the cells defining these variables have been run.")
except Exception as e:
print(f"An error occurred: {e}")
@tool
def get_youtube_video_transcript(video_id: str) -> str:
"""
Fetches the transcript of a YouTube video by its ID and returns it in JSON format.
The video ID can be found in the YouTube video URL:
https://www.youtube.com/watch?v=VIDEO_ID, where VIDEO_ID is the part after "v=".
example: for the url https://www.youtube.com/watch?v=L1vXCYZAYYM the video_id is "L1vXCYZAYYM".
Args:
video_id (str): The YouTube video ID.
Returns:
str: The transcript in JSON format.
"""
ytt_api = YouTubeTranscriptApi()
transcript = ytt_api.fetch(video_id)
formatter = JSONFormatter()
# .format_transcript(transcript) turns the transcript into a JSON string.
json_formatted = formatter.format_transcript(transcript)
return json_formatted
@tool
def transcribe_mp3(mp3_path: str, model_size: str = "base") -> str:
"""
Transcribe an MP3 file to text using Whisper.
Args:
mp3_path (str): Path to the MP3 file.
model_size (str): Whisper model size (tiny, base, small, medium, large).
Returns:
str: Transcribed text.
"""
transcription_path = mp3_path.replace(".mp3", "_transcript.txt")
# Check if transcription already exists
if os.path.exists(transcription_path):
with open(transcription_path, 'r', encoding='utf-8') as f:
return f.read()
# Load model
model = whisper.load_model(model_size)
# Transcribe
result = model.transcribe(mp3_path)
transcription = result["text"]
# Save transcription to file
with open(transcription_path, 'w', encoding='utf-8') as f:
f.write(transcription)
# Return the text
return transcription
@tool
def get_text_from_ascii_file(filepath: str) -> str:
"""
Reads the content of an ASCII text file and returns it as a string.
Args:
filepath (str): The path to the ASCII text file.
Returns:
str: The content of the file as a string.
"""
if not os.path.exists(filepath):
raise FileNotFoundError(f"The file at {filepath} does not exist.")
with open(filepath, "r") as f:
return f.read()
# @tool
# def get_wikipedia_page_content(page_title: str, lang: str='en') -> str:
# """
# This function uses the `wikipediaapi` library to retrieve the content of a specified Wikipedia page in a given language.
# For example: for the url 'https://en.wikipedia.org/wiki/Python_(programming_language)' the page_title would be 'Python_(programming_language)' and the lang would be 'en'.
# It returns the content of the page as a Markdown-formatted string.
# Args:
# page_title (str): The title of the Wikipedia page to fetch.
# lang (str): The language of the Wikipedia page (default is 'en' for English).
# Returns:
# str: The content of the Wikipedia page.
# """
# MY_EMAIL = os.getenv("MY_EMAIL", None)
# if MY_EMAIL is None:
# raise ValueError("MY_EMAIL environment variable is not set. Please set it to your email address.")
# wiki_wiki = wikipediaapi.Wikipedia(user_agent=f'Wiki Agent ({MY_EMAIL})', language=lang)
# page = wiki_wiki.page(page_title)
# if not page.exists():
# raise ValueError(f"The Wikipedia page '{page_title}' does not exist.")
# return md(page.text)
@tool
def get_wikipedia_markdown(
title: str,
lang: str = 'en',
ignore_references: bool = True,
ignore_links: bool = True
) -> str:
"""
Fetches the main content of a Wikipedia page and returns it as Markdown,
excluding infoboxes, navigation templates, images, and—if requested—the
References, Further reading, and External links sections. It's recommended
to start with ignore_references=True and ignore_links=True
to reduce the amount of output to the pure infomation.
Args:
title (str): Wikipedia page title (e.g., "Mercedes_Sosa").
lang (str): Language code (default 'en').
ignore_references (bool): If True, drop "References", "Further reading",
and "External links" sections entirely.
ignore_links (bool): If True, strip out all <a> tags entirely.
Returns:
str: Markdown-formatted content of the main article body.
"""
# 1. Fetch raw HTML
url = f"https://{lang}.wikipedia.org/wiki/{title}"
try:
response = requests.get(url)
response.raise_for_status()
except requests.exceptions.HTTPError as e:
# use wikipedia's API to check if the page exists
api_url = f"https://{lang}.wikipedia.org/w/api.php"
search_params = {
'list': 'search',
'srprop': '',
'srlimit': 10,
'limit': 10,
'srsearch': title.replace("_", " "),
'srinfo': 'suggestion',
'format': 'json',
'action': 'query'
}
headers = {
'User-Agent': "mozilla /5.0 (Windows NT 10.0; Win64; x64)"
}
r = requests.get(api_url, params=search_params, headers=headers)
raw_results = r.json()
search_results = [d['title'].replace(" ", "_") for d in raw_results['query']['search']]
if ('searchinfo' in raw_results['query']) and ('suggestion' in raw_results['query']['searchinfo']):
search_results.insert(0, raw_results['query']['searchinfo']['suggestion'].replace(" ", "_"))
errorMsg = f"Could not fetch page '{title}' for language '{lang}' (HTTP {response.status_code})."
if search_results:
errorMsg += f" Did you mean one of these pages? {', '.join(search_results)}"
raise ValueError(errorMsg) from e
html = response.text
# 2. Parse with BeautifulSoup and isolate the article’s main <div>
soup = BeautifulSoup(html, "lxml")
content_div = soup.find("div", class_="mw-parser-output") #
if content_div is None:
raise ValueError(f"Could not find main content for page '{title}'")
# 2a. Remove all “[edit]” links (<span class="mw-editsection">…)
for edit_span in content_div.find_all("span", class_="mw-editsection"):
edit_span.decompose() #
# 2b. Remove any superscript footnote markers (<sup class="reference">…)
for sup in content_div.find_all("sup", class_="reference"):
sup.decompose() #
# 2c. Remove any parser‐debug comments (e.g., “NewPP limit report…”, “Transclusion expansion time report…”)
for comment in content_div.find_all(string=lambda text: isinstance(text, Comment)):
comment_text = str(comment)
# If the comment contains debug keywords, extract it
if (
"NewPP limit report" in comment_text
or "Transclusion expansion time report" in comment_text
or "Saved in parser cache" in comment_text
):
comment.extract() #
# 3. Remove unwanted “boilerplate” elements:
# a) Infoboxes (sidebars)
for infobox in content_div.find_all("table", class_=re.compile(r"infobox")):
infobox.decompose() #
# b) Table of Contents
toc = content_div.find("div", id="toc")
if toc:
toc.decompose() #
# c) Navigation templates (navbox/vertical-navbox/metadata)
for nav in content_div.find_all(
["div", "table"],
class_=re.compile(r"navbox|vertical-navbox|metadata")
):
nav.decompose() #
# d) Thumbnails / image wrappers
for thumb in content_div.find_all("div", class_=re.compile(r"thumb")):
thumb.decompose() #
# e) Raw <img> tags
for img in content_div.find_all("img"):
img.decompose() #
# 4. Convert any remaining <table> into a Markdown table **in-place**
def table_to_markdown(table_tag: Tag) -> str:
"""
Converts a <table> into a Markdown-formatted table, preserving <th> headers.
"""
headers = []
header_row = table_tag.find("tr")
if header_row:
for th in header_row.find_all("th"):
headers.append(th.get_text(strip=True))
md_table = ""
if headers:
md_table += "| " + " | ".join(headers) + " |\n"
md_table += "| " + " | ".join("---" for _ in headers) + " |\n"
# Now process data rows (skip the first <tr> if it was header row)
for row in table_tag.find_all("tr")[1:]:
cells = row.find_all(["td", "th"])
if not cells:
continue
row_texts = [cell.get_text(strip=True) for cell in cells]
md_table += "| " + " | ".join(row_texts) + " |\n"
return md_table.rstrip()
for table in content_div.find_all("table"):
# Skip infobox/navigation tables (already removed above)
if "infobox" in table.get("class", []) or table.get("role") == "navigation":
continue
markdown_table = table_to_markdown(table) #
new_node = soup.new_string("\n\n" + markdown_table + "\n\n")
table.replace_with(new_node)
# 5. Remove “References”, “Further reading” & “External links” sections if requested
if ignore_references:
section_ids = {"references", "further_reading", "external_links"}
# We look for wrapper <div class="mw-heading mw-heading2"> or mw-heading3
for wrapper in content_div.find_all("div", class_=re.compile(r"mw-heading mw-heading[23]")):
heading_tag = wrapper.find(re.compile(r"^h[2-3]$"))
if heading_tag and heading_tag.get("id", "").strip().lower() in section_ids:
# Collect every sibling until the next wrapper of the same form
siblings_to_remove = []
for sib in wrapper.find_next_siblings():
if (
sib.name == "div"
and "mw-heading" in (sib.get("class") or [])
and re.match(r"mw-heading mw-heading[23]", " ".join(sib.get("class") or []))
):
break
siblings_to_remove.append(sib)
# First delete those siblings
for node in siblings_to_remove:
node.decompose() #
# Finally delete the wrapper itself
wrapper.decompose() #
# 6. Convert the cleaned HTML into Markdown
markdown_options = {}
if ignore_links:
markdown_options["strip"] = ["a"] # strip all <a> tags (keep only their text)
raw_html = "".join(str(child) for child in content_div.children)
markdown_text = md(raw_html, **markdown_options) #
# 7. Collapse 3+ blank lines into exactly two
markdown_text = re.sub(r"\n{3,}", "\n\n", markdown_text).strip()
return markdown_text
@tool
def read_xls_File(file_path: str) -> object:
"""This tool loads xls file into pandas and returns it.
Args:
file_path (str): File path to the xls file.
Returns:
object: The loaded xls file as a pandas DataFrame.
"""
return pd.read_excel(file_path) |