Spaces:
Sleeping
Sleeping
File size: 38,442 Bytes
9e269fa 02a3004 f3950a2 02a3004 f3950a2 4cce6f9 f3950a2 02a3004 f3950a2 02a3004 f3950a2 02a3004 9e269fa c2d81e1 02a3004 c2d81e1 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa e616a02 02a3004 e616a02 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 9e269fa 02a3004 d64cc8e 02a3004 d64cc8e 02a3004 d64cc8e 02a3004 9e269fa 301c92f |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 |
import streamlit as st
import requests
import os
import base64
import json
import time
from io import BytesIO
from dotenv import load_dotenv
from bs4 import BeautifulSoup
from urllib.parse import urljoin, quote_plus
import fitz # PyMuPDF
import pycountry # for country list
from groq import Groq
import tempfile
# Load environment variables
load_dotenv()
# API Keys - set these in your .env file or in Hugging Face Secrets
SERPER_API_KEY = os.getenv("SERPER_API_KEY")
GROQ_API_KEY = os.getenv("GROQ_API_KEY") # Using Groq as free LLM API
# Initialize Groq client (free alternative to OpenAI)
groq_client = Groq(api_key=GROQ_API_KEY)
# Constants
COUNTRIES = sorted([country.name for country in pycountry.countries])
# Set up sessions state for persistence
if 'search_results' not in st.session_state:
st.session_state.search_results = []
if 'selected_pdf' not in st.session_state:
st.session_state.selected_pdf = None
if 'form_fields' not in st.session_state:
st.session_state.form_fields = []
if 'pdf_bytes' not in st.session_state:
st.session_state.pdf_bytes = None
if 'search_history' not in st.session_state:
st.session_state.search_history = []
if 'form_selected' not in st.session_state:
st.session_state.form_selected = False
if 'selected_form_url' not in st.session_state:
st.session_state.selected_form_url = ""
if 'selected_form_index' not in st.session_state:
st.session_state.selected_form_index = -1
# Search using SERPER API
def serper_search(query, country_code="pk"):
url = "https://google.serper.dev/search"
# Always use Pakistan-specific domains
country_domain = "site:.gov.pk OR site:.fbr.gov.pk"
# Build the search query
search_query = f"{query} tax form {country_domain} filetype:pdf"
data = {
"q": search_query,
"gl": "pk", # Force Pakistan search
"hl": "en"
}
try:
# Check if API key is available
if not SERPER_API_KEY:
# Fall back to a free method if no API key
st.warning("SERPER API key not found. Falling back to limited search.")
return fallback_search(query, "pk")
headers = {"X-API-KEY": SERPER_API_KEY}
response = requests.post(url, json=data, headers=headers)
if response.status_code == 200:
results = response.json().get("organic", [])
# Store in session state
st.session_state.search_results = results
return results
else:
st.error(f"Search API error: {response.status_code}")
return []
except Exception as e:
st.error(f"Search failed: {str(e)}")
return []
# Fallback search method (limited, but free)
def fallback_search(query, country_code=""):
try:
# Format country code for search
country_name = next((country.name for country in pycountry.countries if country.alpha_2.lower() == country_code.lower()), "")
# Use a different free API or direct scraping approach
search_query = quote_plus(f"{query} {country_name} tax form pdf")
url = f"https://ddg-api.herokuapp.com/search?query={search_query}&limit=5"
response = requests.get(url)
if response.status_code == 200:
results = response.json()
# Convert to a format similar to Serper
formatted_results = []
for result in results:
formatted_results.append({
"title": result.get("title", ""),
"link": result.get("link", ""),
"snippet": result.get("snippet", "")
})
return formatted_results
return []
except Exception as e:
st.error(f"Fallback search failed: {str(e)}")
return []
# Use LLM to extract relevant information from search results
def analyze_search_results(results, query, country):
if not GROQ_API_KEY:
return results # Return unprocessed results if no LLM available
try:
# Prepare results for LLM analysis
results_text = json.dumps(results[:5], indent=2)
prompt = f"""
I'm looking for tax forms for {country} related to "{query}".
Here are search results:
{results_text}
Please analyze these results and tell me:
1. Which result is most likely the official tax form I need?
2. Is this result from an official government source?
3. What specific form number or name should I be looking for?
4. Any additional forms I might need based on this search intent?
Format your response as JSON with the following keys:
{{
"best_result_index": 0-4 (index of the best result, or -1 if none are good),
"is_official": true/false,
"form_name": "string",
"form_description": "string",
"additional_forms": ["form1", "form2"]
}}
"""
# Call Groq API with mixed model approach (prefer cheaper model)
completion = groq_client.chat.completions.create(
model="llama3-8b-8192", # Free/cheaper model
messages=[{"role": "user", "content": prompt}],
temperature=0.0,
max_tokens=800
)
try:
# Parse the response as JSON
analysis = json.loads(completion.choices[0].message.content)
return results, analysis
except json.JSONDecodeError:
# If parsing fails, return the original results
return results, None
except Exception as e:
st.error(f"LLM analysis failed: {str(e)}")
return results, None
# Try to download PDF
def fetch_pdf(url):
try:
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"
}
st.write(f"Attempting to download PDF from: {url}")
r = requests.get(url, headers=headers, timeout=15)
st.write(f"Response status code: {r.status_code}")
st.write(f"Content-Type: {r.headers.get('Content-Type', 'Not specified')}")
if r.status_code == 200:
if 'application/pdf' in r.headers.get('Content-Type', ''):
st.success("Successfully retrieved PDF!")
return BytesIO(r.content)
else:
st.info("URL doesn't point directly to a PDF. Searching for PDF links on the page...")
# Try to find PDF links if this is an HTML page
pdf_url = find_pdf_in_html_page(url, r.text)
if pdf_url:
st.info(f"Found PDF link: {pdf_url}")
return fetch_pdf(pdf_url)
else:
st.warning("No PDF links found on the page")
else:
st.error(f"Failed to retrieve URL: {r.status_code}")
return None
except Exception as e:
st.error(f"Error fetching PDF: {str(e)}")
return None
# Scrape .pdf links from HTML page
def find_pdf_in_html_page(url, html_content=None):
try:
if not html_content:
r = requests.get(url, timeout=10)
html_content = r.text
soup = BeautifulSoup(html_content, "html.parser")
pdf_links = []
# Look for PDF links
st.write("Scanning page for PDF links...")
for link in soup.find_all('a', href=True):
href = link['href']
if href.lower().endswith('.pdf'):
full_url = href if href.startswith("http") else urljoin(url, href)
pdf_links.append((full_url, link.text.strip()))
st.write(f"Found PDF link: {full_url} - {link.text.strip()}")
st.write(f"Total PDF links found: {len(pdf_links)}")
# First, look for links with "tax", "form", or "return" in them
for link_url, link_text in pdf_links:
combined_text = (link_url + " " + link_text).lower()
if any(keyword in combined_text for keyword in ["tax", "form", "return", "income"]):
st.success(f"Selected most relevant PDF: {link_url}")
return link_url
# If no specific tax links, return the first PDF link
if pdf_links:
st.info(f"No tax-specific PDFs found. Using first PDF: {pdf_links[0][0]}")
return pdf_links[0][0]
st.warning("No PDF links found on the page")
return None
except Exception as e:
st.error(f"Error finding PDF links: {str(e)}")
return None
# Display PDF safely with error handling
def display_pdf(file_bytesio):
try:
file_bytesio.seek(0)
base64_pdf = base64.b64encode(file_bytesio.read()).decode('utf-8')
pdf_display = f'<iframe src="data:application/pdf;base64,{base64_pdf}" width="100%" height="800" type="application/pdf"></iframe>'
st.markdown(pdf_display, unsafe_allow_html=True)
st.success("PDF loaded successfully!")
# Add a direct download option for better user experience
file_bytesio.seek(0)
except Exception as e:
st.error(f"Error displaying PDF: {str(e)}")
st.info("If the PDF isn't displaying, you can try using the direct link.")
# Extract interactive fields from PDF
def extract_form_fields(file_bytesio):
try:
file_bytesio.seek(0)
doc = fitz.open(stream=file_bytesio, filetype="pdf")
fields = []
widget_types = {
fitz.PDF_WIDGET_TYPE_TEXT: "Text Field",
fitz.PDF_WIDGET_TYPE_CHECKBOX: "Checkbox",
fitz.PDF_WIDGET_TYPE_RADIOBUTTON: "Radio Button",
fitz.PDF_WIDGET_TYPE_COMBOBOX: "Dropdown",
fitz.PDF_WIDGET_TYPE_LISTBOX: "List Box"
}
for page_num, page in enumerate(doc):
widgets = page.widgets()
for widget in widgets:
field_type = widget_types.get(widget.field_type, "Unknown")
field_info = {
"name": widget.field_name or f"Field_{page_num}_{len(fields)}",
"type": field_type,
"value": widget.field_value,
"options": widget.choice_values if hasattr(widget, "choice_values") else None,
"page": page_num + 1
}
fields.append(field_info)
return fields
except Exception as e:
st.error(f"Error extracting form fields: {str(e)}")
return []
# Use LLM to explain form fields
def explain_form_fields(fields, country, form_name):
if not GROQ_API_KEY or not fields:
return {}
try:
fields_json = json.dumps(fields, indent=2)
prompt = f"""
These are form fields from a tax form ({form_name}) from {country}:
{fields_json}
Please analyze these fields and:
1. Group them into logical sections (personal info, income, deductions, etc.)
2. Explain any technical tax terms in simple language
3. Identify which fields are mandatory vs. optional if possible
Format your response as JSON with the following structure:
{{
"sections": [
{{
"name": "section name",
"fields": ["field1", "field2"],
"explanation": "explanation of this section"
}}
],
"key_terms": {{
"term1": "simple explanation",
"term2": "simple explanation"
}},
"mandatory_fields": ["field1", "field2"]
}}
"""
completion = groq_client.chat.completions.create(
model="llama3-8b-8192",
messages=[{"role": "user", "content": prompt}],
temperature=0.1,
max_tokens=1000
)
try:
explanation = json.loads(completion.choices[0].message.content)
return explanation
except json.JSONDecodeError:
return {}
except Exception as e:
st.error(f"Error explaining form fields: {str(e)}")
return {}
# Fill PDF form with user data
def fill_pdf_form(file_bytesio, field_values):
try:
file_bytesio.seek(0)
doc = fitz.open(stream=file_bytesio, filetype="pdf")
# Fill in the form fields
for page in doc:
widgets = page.widgets()
for widget in widgets:
field_name = widget.field_name
if field_name in field_values:
widget.field_value = field_values[field_name]
widget.update()
# Save to a temporary file
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
doc.save(temp_file.name)
doc.close()
# Read the saved file back
with open(temp_file.name, "rb") as f:
filled_pdf = BytesIO(f.read())
# Clean up
os.unlink(temp_file.name)
return filled_pdf
except Exception as e:
st.error(f"Error filling form: {str(e)}")
return None
# Add to search history
def add_to_history(country, query, pdf_url=None):
timestamp = time.strftime("%Y-%m-%d %H:%M:%S")
st.session_state.search_history.append({
"timestamp": timestamp,
"country": country,
"query": query,
"pdf_url": pdf_url
})
# Keep only the last 10 searches
if len(st.session_state.search_history) > 10:
st.session_state.search_history = st.session_state.search_history[-10:]
# Get country code from name
def get_country_code(country_name):
try:
country = pycountry.countries.get(name=country_name)
return country.alpha_2 if country else ""
except:
return ""
# Function to suggest other forms
def suggest_other_forms():
"""Suggest other forms from search results if current form isn't fillable"""
if ('search_results' in st.session_state and
st.session_state.search_results and
'selected_pdf' in st.session_state):
selected_idx = st.session_state.selected_pdf
st.markdown("### π Try These Other Forms")
# Display up to 5 alternative forms
other_forms = [r for i, r in enumerate(st.session_state.search_results[:5])
if i != selected_idx]
for idx, result in enumerate(other_forms):
title = result.get('title', 'Untitled Form')
link = result.get('link', '')
st.markdown(f"**{idx+1}. {title}**")
if st.button(f"Try Form #{idx+1}", key=f"try_form_{idx}"):
with st.spinner(f"Fetching alternative form #{idx+1}..."):
pdf_bytes = fetch_pdf(link)
if pdf_bytes:
st.session_state.pdf_bytes = pdf_bytes
st.session_state.selected_pdf = idx
st.session_state.form_fields = extract_form_fields(pdf_bytes)
# Add this new function
def tax_agent_response(user_query, tax_form_type=None, form_fields=None):
"""Generate an agent-like response to user tax questions using LLM"""
if not GROQ_API_KEY:
return "I need an LLM API key to provide detailed assistance. Please upload a PDF or search for forms directly."
try:
# Create context from available information
context = f"The user is asking about Pakistani tax: '{user_query}'\n"
if tax_form_type:
context += f"They previously selected tax form type: {tax_form_type}\n"
if form_fields and len(form_fields) > 0:
fields_sample = ", ".join([f["name"] for f in form_fields[:5]])
context += f"They are looking at a form with fields including: {fields_sample}\n"
# Pakistan-specific tax information to help ground the response
context += """
Pakistan tax information:
- FBR (Federal Board of Revenue) is the main tax authority
- Common tax forms include income tax returns, sales tax returns, and withholding tax statements
- The tax year in Pakistan typically runs from July to June
- NTN (National Tax Number) is required for filing taxes in Pakistan
"""
prompt = f"""
{context}
As a LifePilot Tax Agent specialized in Pakistani taxation, provide a helpful response to their query.
If they are asking about which tax form they need, explain the options and help them decide.
If they are asking about how to fill a specific field, provide guidance.
If they need information about tax filing deadlines or procedures, provide accurate information.
Your response should be:
1. Conversational and helpful
2. Specific to Pakistan's tax system
3. Brief but informative
"""
# Call Groq API
completion = groq_client.chat.completions.create(
model="llama3-8b-8192", # Or "mixtral-8x7b-32768" if available
messages=[{"role": "user", "content": prompt}],
temperature=0.3,
max_tokens=800
)
return completion.choices[0].message.content
except Exception as e:
return f"I encountered an error while processing your question: {str(e)}"
# Add this function to recommend tax form types
def recommend_tax_form_type(user_query):
"""Recommend appropriate tax form type based on user's situation"""
if not GROQ_API_KEY:
return ["Income Tax Return", "Sales Tax Return", "Withholding Tax Statement"]
try:
prompt = f"""
The user is asking about Pakistani taxes: "{user_query}"
Based on their query, which of these Pakistani tax form types would be most relevant?
- Income Tax Return
- Sales Tax Return
- Withholding Tax Statement
- Property Tax
- Customs Duty
- Advance Tax
- Wealth Statement
Return only the names of the top 3 most relevant form types as a JSON array:
["Form Type 1", "Form Type 2", "Form Type 3"]
"""
completion = groq_client.chat.completions.create(
model="llama3-8b-8192",
messages=[{"role": "user", "content": prompt}],
temperature=0.1,
max_tokens=100
)
try:
# Parse the response as JSON
return json.loads(completion.choices[0].message.content)
except json.JSONDecodeError:
# Fallback to default options
return ["Income Tax Return", "Sales Tax Return", "Withholding Tax Statement"]
except Exception as e:
return ["Income Tax Return", "Sales Tax Return", "Withholding Tax Statement"]
# Main Streamlit app
def main():
st.set_page_config(
page_title="LifePilot - Pakistan Tax Form Finder",
page_icon="π",
layout="wide",
initial_sidebar_state="expanded"
)
# Sidebar
with st.sidebar:
st.title("πLifePilot")
st.caption("Pakistan Tax Assistant")
st.subheader("π History")
if st.session_state.search_history:
for idx, item in enumerate(reversed(st.session_state.search_history)):
with st.expander(f"{item['query']}"):
st.write(f"π
{item['timestamp']}")
if item['pdf_url']:
st.write(f"[Open PDF]({item['pdf_url']})")
else:
st.info("Your search history will appear hereπ")
st.divider()
st.markdown("πLifePilot| Made with Streamlit")
# Main content area
st.title("Pakistan Tax Form Finder")
st.markdown("Search, preview, and get assistance with official Pakistani tax forms.")
# Two main tabs: Search Forms and Upload Forms
tab1, tab2, tab3 = st.tabs(["π€ Tax Assistant", "π Search Forms", "π€ Upload Form"])
# Tab 1: Tax Assistant
with tab1:
st.header("Pakistan Tax Assistant")
st.markdown("Ask any questions about Pakistani taxes or which forms you need.")
# User query input
user_query = st.text_input("Ask your tax question:", placeholder="Which tax form do I need as a salaried employee?")
# Current form context
current_form_type = None
if 'form_fields' in st.session_state and st.session_state.form_fields:
current_form_type = "Tax Form with fillable fields"
# Process query when submitted
if user_query:
with st.spinner("Processing your question..."):
agent_response = tax_agent_response(
user_query,
tax_form_type=current_form_type,
form_fields=st.session_state.form_fields if 'form_fields' in st.session_state else None
)
st.markdown("### Response:")
st.markdown(agent_response)
# Initialize recommended_types to empty list by default
recommended_types = []
# Recommend form types if needed
if "which form" in user_query.lower() or "what form" in user_query.lower() or "do i need" in user_query.lower():
st.markdown("### Recommended Form Types:")
recommended_types = recommend_tax_form_type(user_query)
# Auto-fetch the first recommended form without requiring a button click
if recommended_types:
form_type = recommended_types[0]
with st.spinner(f"Automatically fetching {form_type} forms..."):
results = serper_search(form_type, "pk")
if results:
st.session_state.search_results = results
st.success(f"Found {len(results)} {form_type} forms")
# Display first result and fetch PDF automatically
if results[0].get('link', ''):
with st.spinner("Fetching the most relevant form..."):
pdf_bytes = fetch_pdf(results[0].get('link', ''))
if pdf_bytes:
st.session_state.pdf_bytes = pdf_bytes
st.session_state.form_fields = extract_form_fields(pdf_bytes)
add_to_history("Pakistan", form_type, results[0].get('link', ''))
st.success("Form fetched successfully!")
# Show form preview immediately
st.markdown("### π Form Preview")
display_pdf(pdf_bytes)
# If form has fields, show them
if st.session_state.form_fields:
st.markdown("### π Form Fields")
# Get field explanations if LLM is available
field_explanations = {}
if GROQ_API_KEY:
with st.spinner("Analyzing form fields..."):
explanations = explain_form_fields(
st.session_state.form_fields,
"Pakistan",
f"{form_type} form"
)
if explanations:
field_explanations = explanations
# Display fields with explanations if available
if field_explanations and "sections" in field_explanations:
for section in field_explanations["sections"]:
with st.expander(f"π {section['name']}"):
st.write(section["explanation"])
for field_name in section["fields"]:
matching_fields = [f for f in st.session_state.form_fields
if f["name"] == field_name]
if matching_fields:
field = matching_fields[0]
st.write(f"**{field['name']}** ({field['type']})")
else:
# Simple field display without explanations
for field in st.session_state.form_fields:
st.write(f"**{field['name']}** ({field['type']})")
else:
st.error("Unable to fetch this form automatically. Please try another search.")
# Only show buttons for all recommended types if we have any
if recommended_types:
for idx, form_type in enumerate(recommended_types):
if st.button(f"Find {form_type} Forms", key=f"find_{idx}"):
# Set up search for this form type
with st.spinner(f"Searching for {form_type} forms..."):
results = serper_search(form_type, "pk")
if results:
st.session_state.search_results = results
st.success(f"Found {len(results)} {form_type} forms")
# Display first result
if results[0].get('link', ''):
with st.spinner("Fetching the most relevant form..."):
pdf_bytes = fetch_pdf(results[0].get('link', ''))
if pdf_bytes:
st.session_state.pdf_bytes = pdf_bytes
st.session_state.form_fields = extract_form_fields(pdf_bytes)
add_to_history("Pakistan", form_type, results[0].get('link', ''))
st.success("Form fetched successfully!")
# Show form preview immediately
st.markdown("### π Form Preview")
display_pdf(pdf_bytes)
else:
st.error("Unable to fetch this form. Please try another search.")
# Still show buttons for all recommended types
for idx, form_type in enumerate(recommended_types):
if st.button(f"Find {form_type} Forms", key=f"find_{idx}"):
# Set up search for this form type
with st.spinner(f"Searching for {form_type} forms..."):
results = serper_search(form_type, "pk")
if results:
st.session_state.search_results = results
st.success(f"Found {len(results)} {form_type} forms")
# Display first result
if results[0].get('link', ''):
with st.spinner("Fetching the most relevant form..."):
pdf_bytes = fetch_pdf(results[0].get('link', ''))
if pdf_bytes:
st.session_state.pdf_bytes = pdf_bytes
st.session_state.form_fields = extract_form_fields(pdf_bytes)
add_to_history("Pakistan", form_type, results[0].get('link', ''))
st.success("Form fetched successfully!")
# Show form preview immediately
st.markdown("### π Form Preview")
display_pdf(pdf_bytes)
else:
st.error("Unable to fetch this form. Please try another search.")
# Tab 2: Search Forms (modified version of original tab1)
with tab2:
# Set Pakistan as default country
country = "Pakistan"
st.info("π΅π° This application is focused on Pakistani tax forms.")
# Form type selection
form_type = st.selectbox(
"π What tax form are you looking for?",
["Income Tax Return", "Sales Tax Return", "Withholding Tax Statement",
"Property Tax", "Customs Duty", "Advance Tax", "Wealth Statement"]
)
# Additional form details for search refinement
custom_query = st.text_input(
"βοΈ Specific form or additional details:",
placeholder="e.g., Salaried individuals, business income, etc."
)
# Build the search query
search_query = custom_query if custom_query else form_type
# Search button
search_button = st.button("π Search for Pakistani Tax Forms", use_container_width=True)
# Only show search results when search button is clicked
if search_button:
with st.spinner("Searching for Pakistani tax forms..."):
# Perform search
results = serper_search(search_query, "pk")
if results:
# Try to analyze results with LLM if available
if GROQ_API_KEY:
results, analysis = analyze_search_results(results, search_query, "Pakistan")
# Show LLM analysis if available
if analysis and isinstance(analysis, dict):
best_idx = analysis.get("best_result_index", -1)
if best_idx >= 0 and best_idx < len(results):
st.success(f"β
Found: {analysis.get('form_name', 'Tax Form')}")
st.info(analysis.get('form_description', ''))
# If additional forms are suggested
additional = analysis.get('additional_forms', [])
if additional:
st.markdown("**You might also need:**")
for form in additional:
st.markdown(f"- {form}")
# Display results in a cleaner format
st.markdown("### π Found Forms")
for idx, result in enumerate(results[:5]):
title = result.get('title', 'Untitled Form')
link = result.get('link', '')
snippet = result.get('snippet', '')
with st.container():
st.subheader(f"{idx+1}. {title}")
st.write(snippet)
# Get button and Download PDF button side by side
col1, col2 = st.columns(2)
with col1:
# Replace Select Form button with Get Form button
pdf_bytes = fetch_pdf(link)
if pdf_bytes:
st.session_state.pdf_bytes = pdf_bytes
st.session_state.form_fields = extract_form_fields(pdf_bytes)
add_to_history("Pakistan", search_query, link)
st.success("Form fetched successfully!")
else:
st.error("Unable to fetch this form. Please try another.")
with col2:
# Direct link to open in new tab
st.markdown(
f"""<a href="{link}" target="_blank">
<button style="background-color:#4CAF50;color:white;padding:6px 12px;
border:none;border-radius:4px;cursor:pointer;width:100%;">
ποΈ View Original
</button>
</a>""",
unsafe_allow_html=True
)
st.divider()
else:
st.warning("No results found. Try different search terms.")
# Display the PDF directly if it exists in session state
if 'pdf_bytes' in st.session_state and st.session_state.pdf_bytes:
st.markdown("### π Form Preview")
display_pdf(st.session_state.pdf_bytes)
# If form has fields, show them
if 'form_fields' in st.session_state and st.session_state.form_fields:
st.markdown("### π Form Fields")
# Get field explanations if LLM is available
field_explanations = {}
if GROQ_API_KEY:
with st.spinner("Analyzing form fields..."):
explanations = explain_form_fields(
st.session_state.form_fields,
"Pakistan",
f"{form_type} form"
)
if explanations:
field_explanations = explanations
# Display fields with explanations if available
if field_explanations and "sections" in field_explanations:
for section in field_explanations["sections"]:
with st.expander(f"π {section['name']}"):
st.write(section["explanation"])
for field_name in section["fields"]:
matching_fields = [f for f in st.session_state.form_fields
if f["name"] == field_name]
if matching_fields:
field = matching_fields[0]
st.write(f"**{field['name']}** ({field['type']})")
else:
# Simple field display without explanations
for field in st.session_state.form_fields:
st.write(f"**{field['name']}** ({field['type']})")
# Option to fill form
st.markdown("### βοΈ Fill This Form")
st.info("This feature will help you fill in the form fields.")
if st.button("Start Filling Form"):
st.session_state.is_filling = True
else:
st.warning("β οΈ This form doesn't have fillable fields. It may be a scanned document or not an interactive form.")
st.info("You can still view and download the form, but automatic filling isn't available.")
# Option to download the non-fillable form
st.download_button(
label="π₯ Download Form",
data=st.session_state.pdf_bytes,
file_name="pakistan_tax_form.pdf",
mime="application/pdf"
)
# Add a question and answer section for this form
st.markdown("### β Questions about this form?")
form_question = st.text_input("Ask a question about this form:", key="form_question_input")
if form_question:
with st.spinner("Getting answer..."):
form_response = tax_agent_response(form_question, tax_form_type=form_type)
st.markdown(form_response)
if __name__ == "__main__":
main() |