Spaces:
Sleeping
Sleeping
File size: 22,058 Bytes
792c6a9 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 | # utils.py
from googlesearch import search
import requests
from bs4 import BeautifulSoup
import time
from huggingface_hub import InferenceClient
from langchain_groq import ChatGroq
from langchain_core.prompts import ChatPromptTemplate
from gtts import gTTS
import os
from google import genai
from translate import Translator
import re
from openai import OpenAI
import json
from config import (
llm,
client,
API_URL,
HEADERS,
openrouter_client,
translator,
genai_client,
genai_client2,
)
from googlesearch import search as google_search
def extract_news(company, num_articles=10):
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/123.0.0.0 Safari/537.36"
}
articles = []
seen_urls = set()
sentiment_count = {"Positive": 0, "Negative": 0, "Neutral": 0}
print(f"Fetching {num_articles} valid articles...")
while len(articles) < num_articles:
try:
for url in google_search(f"{company} news", num=num_articles * 2, stop=num_articles * 2):
if url in seen_urls or any(
blocked in url
for blocked in ["youtube.com", "youtu.be", "twitter.com", "x.com"]
):
continue
seen_urls.add(url)
print(f"\nFetching article {len(articles) + 1}: {url}")
try:
article_response = requests.get(url, headers=headers, timeout=10)
article_response.raise_for_status()
article_soup = BeautifulSoup(article_response.text, "html.parser")
title = (
article_soup.find("title").get_text()
if article_soup.find("title")
else "No Title Found"
)
if title == "No Title Found":
print(f"Skipping article {url} due to missing title")
continue
paragraphs = article_soup.find_all("p")
content = " ".join([para.get_text() for para in paragraphs])
if not content.strip():
print(f"No content found for {url}")
continue
# Summarize the article
summary, translated_summary = summarize_article(content)
# Sentiment analysis
sentiment, reason = analyze_sentiment(summary)
if sentiment in sentiment_count:
sentiment_count[sentiment] += 1
# Extract relevant topics using LLM
topics = extract_topics(content)
metadata = {
"URL": url,
"Title": title,
"Summary": summary,
"Summary(in Hindi)": translated_summary,
"Sentiment": sentiment,
"Reason(for Sentiment)": reason,
"Topics": topics,
}
articles.append(metadata)
if len(articles) >= num_articles:
break
time.sleep(1)
except requests.exceptions.Timeout:
print(f"Timeout error while fetching {url}")
except requests.exceptions.RequestException as e:
print(f"Failed to fetch article {url}: {e}")
except Exception as e:
print(f"Error while searching for articles: {e}")
final_sentiment = final_sentiment_analysis(sentiment_count)
# Create final JSON output
output = {
"Company": company,
"Articles": articles,
"Comparative Sentiment Score": {
"Sentiment Distribution": sentiment_count,
"Total Sentiments": sum(sentiment_count.values()),
},
"Final Sentiment Analysis": final_sentiment,
}
print(f"\nSuccessfully fetched {len(articles)} valid articles.")
# Pretty print JSON
print(json.dumps(output, indent=4))
return output
def clean_and_parse_json(response):
try:
# Extract JSON block using regex
cleaned_response = re.search(r"\{.*\}", response, re.DOTALL)
if cleaned_response:
cleaned_response = cleaned_response.group()
# Replace single quotes with double quotes
cleaned_response = cleaned_response.replace("'", '"')
# Remove trailing commas
cleaned_response = re.sub(r",\s*([\]}])", r"\1", cleaned_response)
# Parse JSON
return json.loads(cleaned_response)
else:
raise ValueError("No JSON found in response.")
except Exception as e:
print(f"Parsing error: {e}")
return {"Analysis": "Failed to parse structured output from the model."}
def final_sentiment_analysis(sentiment_count):
try:
sentiment_data = json.dumps(sentiment_count, indent=2)
print("Sentiment Data:\n", sentiment_data)
# Use Gemma for sentiment analysis
response = genai_client2.models.generate_content(
model="gemma-3-27b-it",
contents=(
f"Analyze the following sentiment distribution:\n\n"
f"{sentiment_data}\n\n"
f"Provide a FINAL sentiment analysis based on the distribution. "
f"Summarize the overall sentiment and any patterns or insights in EXACTLY THREE LINES. "
f"Return the result in a normal string format with summary about the sentiment and particular company current form."
),
)
if response and response.text:
generated_text = response.text.strip()
print("Generated Text (Gemma):", generated_text)
else:
print("Error: No response from the model.")
generated_text = "Failed to generate response from Gemma."
print("Generated Text:", generated_text)
# Clean and parse JSON
# generated_text = clean_and_parse_json(generated_text)
return generated_text
except Exception as e:
print(f"Gemma sentiment analysis failed: {e}")
try:
print("Falling back to OpenRouter (qwen/qwq)...")
completion = openrouter_client.chat.completions.create(
model="qwen/qwq-32b:free",
messages=[
{
"role": "user",
"content": (
f"Analyze the following sentiment distribution:\n\n"
f"{sentiment_data}\n\n"
f"Provide a FINAL sentiment analysis based on the distribution. "
f"Summarize the overall sentiment and any patterns or insights in EXACTLY THREE LINES. "
f"Return the result in a normal string format with summary about the sentiment and particular company current form."
),
}
],
)
generated_text = completion.choices[0].message.content.strip()
print("OpenRouter Generated Text:", generated_text)
# Clean and parse JSON
# generated_text = clean_and_parse_json(generated_text)
return generated_text
except Exception as e:
print(f"OpenRouter sentiment analysis failed: {e}")
return {
"Summary": "Sentiment analysis failed using both models.",
"Dominant Sentiment": "Unknown",
"Insight": "No insights available.",
}
def summarize_article(content):
try:
# Use the Inference API for summarization
result = client.summarization(content, model="facebook/bart-large-cnn")
summary = result["summary_text"]
# Translate summary to Hindi
translated_summary = translator.translate(summary)
return summary, translated_summary
except Exception as e:
print(f"Hugging Face summarization failed: {e}")
print("Trying Gemini as fallback...")
try:
truncated_content = content[:1024]
# Use Gemini model as fallback
response = genai_client2.models.generate_content(
model="gemma-3-27b-it",
contents=f"Summarize the following article:\n\n{truncated_content}",
)
summary = response.text.strip()
translated_summary = translator.translate(summary)
return summary, translated_summary
except Exception as e:
print(f"Gemini summarization also failed: {e}")
return "Summary not available", "Summary not available"
def convert_text_to_speech(text, filename):
try:
# Try using gTTS for Hindi TTS
tts = gTTS(text=text, lang="hi")
tts.save(filename)
return filename
except Exception as e:
print(f"Error during TTS conversion with gTTS: {e}")
print("Switching to Hugging Face MMS-TTS...")
# Use Hugging Face MMS-TTS as a fallback
try:
payload = {"inputs": text}
response = requests.post(API_URL, headers=HEADERS, json=payload)
if response.status_code == 200:
with open(filename, "wb") as f:
f.write(response.content)
print("Audio file generated using Hugging Face MMS-TTS.")
return filename
else:
print(
f"Hugging Face API error: {response.status_code} - {response.text}"
)
return None
except Exception as e:
print(f"Error during TTS conversion with Hugging Face MMS-TTS: {e}")
return None
def analyze_sentiment(content):
try:
# Use Gemini LLM for sentiment analysis as the primary method
truncated_content = content
response = genai_client.models.generate_content(
model="gemma-3-27b-it",
contents=(
f"Analyze the sentiment of the following article:\n\n{truncated_content}\n\n"
"Return the response as simple string . It shoild have sentiment and reason both are seperated by comma"
"Sentiment should be Positive, Negative, or Neutral. No other extra text or contents."
),
)
# print("RESponse", response)
# Parse JSON response
# generated_text = response.text.strip()
# print("Generated Response:", generated_text)
# Split the response into sentiment and reason
parts = response.text.split(",", 1)
if len(parts) == 2:
sentiment = parts[0].strip()
reason = parts[1].strip()
print(f"Sentiment: {sentiment}")
print(f"Reason: {reason}")
else:
print("Invalid response format")
if sentiment.lower() not in ["positive", "negative", "neutral"]:
sentiment = "Sentiment analysis not available"
return sentiment, reason
except Exception as e:
print(f"Error during Gemini LLM sentiment analysis: {e}")
print("Trying HF Inference as fallback for sentiment analysis...")
try:
# Fallback to HF Inference for sentiment analysis
truncated_content = content
result = client.text_classification(
truncated_content,
model="finiteautomata/bertweet-base-sentiment-analysis",
)
hf_sentiment = result[0]["label"]
# Map HF inference sentiment labels to required format
sentiment_mapping = {"POS": "Positive", "NEG": "Negative", "NEU": "Neutral"}
sentiment = sentiment_mapping.get(
hf_sentiment, "Sentiment analysis not available"
)
return sentiment, ""
except Exception as e:
print(f"HF Inference sentiment analysis also failed: {e}")
return "Sentiment analysis not available", ""
def extract_topics(content):
try:
prompt = ChatPromptTemplate.from_messages(
[
(
"system",
"You are a helpful assistant that extracts key topics from articles.",
),
(
"human",
f"Extract the **three most relevant and suitable two words topics** from the following article:\n\n{content}\n\n"
"Provide exactly 5 topics, separated by commas. Avoid numbering or extra details.",
),
]
)
chain = prompt | llm
response = chain.invoke({})
# Clean and process response
if isinstance(response, str):
topics = response.split(",")
elif hasattr(response, "content") and isinstance(response.content, str):
topics = response.content.split(",")
else:
topics = ["Topics not available"]
# Remove unwanted characters and extra spaces
topics = [re.sub(r"^\d+[\.\)\-]?\s*", "", topic.strip()) for topic in topics]
return topics
except Exception as e:
print(f"Groq topic extraction failed: {e}")
print("Trying Gemini as fallback...")
try:
# Use Gemini for topic extraction
response = genai_client.models.generate_content(
model="gemma-3-27b-it",
contents=f"Extract exactly 5 key topics (two or one words) from the following article:\n\n{content}\n\n"
"Return them as a comma-separated list without numbering or extra details.",
)
topics = response.text.strip().split(",")
# Clean up and format the topics
topics = [
re.sub(r"^\d+[\.\)\-]?\s*", "", topic.strip()) for topic in topics
]
return topics
except Exception as e:
print(f"Gemini topic extraction also failed: {e}")
return ["Topics not available"]
def clean_response(response):
cleaned_response = re.search(r"\{.*\}", response, re.DOTALL)
if cleaned_response:
return cleaned_response.group()
return response
def generate_comparative_analysis(articles):
num_articles = min(len(articles["Articles"]), 5)
all_contents = "\n\n".join(
[
f"Article {i + 1}: {article['Summary']}"
for i, article in enumerate(articles["Articles"][:num_articles])
]
)
print(all_contents)
try:
# Use Gemma for comparative analysis
response = genai_client.models.generate_content(
model="gemma-3-27b-it",
contents=(
f"Perform a detailed comparative analysis of the following {num_articles} articles:\n\n"
f"{all_contents}\n\n"
f"Provide differences in coverage and potential impacts between each pair of articles.\n"
f"Return the result in the following structured JSON format:\n"
f"{{\n"
f' "Coverage Differences": [\n'
f" {{\n"
f' "Comparison": "<summary of difference between article X and article Y>",\n'
f' "Impact": "<impact of these differences>"\n'
f" }}\n"
f" ]\n"
f"}}"
),
)
generated_text = response.text.strip()
print("Generated Text (Gemma):", generated_text)
# Clean and parse the response
cleaned_response = clean_response(generated_text)
structured_response = json.loads(cleaned_response)
except Exception as e:
print(f"Gemma comparative analysis failed: {e}")
try:
print("Falling back to OpenRouter (qwq)...")
if "all_contents" not in locals():
all_contents = "\n\n".join(
[
f"Article {i+1}: {article['Summary']}"
for i, article in enumerate(articles["Articles"])
]
)
completion = openrouter_client.chat.completions.create(
model="qwen/qwq-32b:free",
messages=[
{
"role": "user",
"content": (
f"Perform a detailed comparative analysis of the following {num_articles} articles:\n\n"
f"{all_contents}\n\n"
f"Provide differences in coverage and potential impacts between each pair of articles.\n"
f"Return the result in the following structured JSON format:\n"
f"{{\n"
f' "Coverage Differences": [\n'
f" {{\n"
f' "Comparison": "<summary of difference between article X and article Y>",\n'
f' "Impact": "<impact of these differences>"\n'
f" }}\n"
f" ]\n"
f"}}"
),
}
],
)
# Extract response
generated_text = completion.choices[0].message.content.strip()
print("Generated Text (OpenRouter):", generated_text)
# Clean and parse the response
cleaned_response = clean_response(generated_text)
structured_response = json.loads(cleaned_response)
except Exception as e:
print(f"OpenRouter fallback failed: {e}")
structured_response = {
"Coverage Differences": [
{
"Comparison": "Failed to generate comparative analysis from both Gemma and OpenRouter.",
"Impact": "Check the model's output format or prompt.",
}
]
}
# Merge the result back into the input
articles["Coverage Differences"] = structured_response["Coverage Differences"]
return articles
def add_topic_overlap_to_coverage_differences(news_articles):
try:
# Get the list of topics from each article
article_topics = [
set(article["Topics"]) for article in news_articles["Articles"]
]
if len(article_topics) < 2:
raise ValueError(
"At least two articles are required for topic overlap analysis."
)
# Step 1: Find topics that are common in at least two articles
topic_occurrences = {}
for i, topics in enumerate(article_topics):
for topic in topics:
if topic not in topic_occurrences:
topic_occurrences[topic] = []
topic_occurrences[topic].append(
i + 1
) # Storing article index (1-based)
# Step 2: Filter out topics appearing in at least two articles
common_topics = [
f"{topic} (Article {', '.join(map(str, indexes))})"
for topic, indexes in topic_occurrences.items()
if len(indexes) >= 2
]
# Step 3: Find unique topics for each article
unique_topics = []
for i, topics in enumerate(article_topics):
other_topics = set.union(*(article_topics[:i] + article_topics[i + 1 :]))
unique = topics - other_topics
unique_topics.append(list(unique))
# Step 4: Build the output structure for Coverage Differences
topic_overlap = {"Common Topics": common_topics, "Unique Topics": {}}
for i, topics in enumerate(unique_topics):
topic_overlap["Unique Topics"][f"Article {i + 1}"] = topics
# Step 5: Add topic_overlap under "Coverage Differences"
if "Coverage Differences" not in news_articles:
news_articles["Coverage Differences"] = []
news_articles["Coverage Differences"].append({"Topic Overlap": topic_overlap})
return news_articles
except Exception as e:
print(f"Error: {e}")
return None
def generate_company_analysis_in_hindi(company_name, articles):
try:
# Extract summaries and sentiment from updated articles
summaries = "\n\n".join(
[
f"Article {i+1}: {article['Summary']}\nSentiment: {article.get('Sentiment', 'Unknown')}"
for i, article in enumerate(articles["Articles"])
]
)
# Create the prompt in English, request output in Hindi
prompt = (
f"Please prepare a detailed analysis of the company '{company_name}' based on the following articles. "
f"Include the sentiment and key points from each article.\n\n"
f"{summaries}\n\n"
f"The analysis should cover the overall performance of the company, the sentiment reflected in the articles, "
f"and a comparison of the articles' coverage.\n\n"
f"Output should be in **Hindi** and in a single detailed paragraph based on these articles.Don't provide in english"
)
# Generate content using Gemini Gemma model
response = genai_client2.models.generate_content(
model="gemma-3-27b-it",
contents=prompt,
)
if response and response.text:
analysis_text = response.text.strip()
print("\n🔎 Generated Company Analysis (Hindi):\n")
print(analysis_text)
return analysis_text
else:
raise ValueError("Gemini returned an empty response")
except Exception as e:
print(f"Gemini analysis generation failed: {e}")
return None
|