Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,11 +2,9 @@ import gradio as gr
|
|
| 2 |
import pdfplumber
|
| 3 |
import matplotlib.pyplot as plt
|
| 4 |
import numpy as np
|
| 5 |
-
import re
|
| 6 |
from word2number import w2n
|
| 7 |
-
|
| 8 |
from typing import Tuple, List, Dict
|
| 9 |
-
import os
|
| 10 |
|
| 11 |
# Custom CSS for styling
|
| 12 |
css = """
|
|
@@ -112,11 +110,6 @@ css = """
|
|
| 112 |
}
|
| 113 |
"""
|
| 114 |
|
| 115 |
-
# Salesforce Authentication (Make sure your credentials are stored securely)
|
| 116 |
-
sf = Salesforce(username='Kushalpavansekharm503@agentforce.com',
|
| 117 |
-
password='Kushal@123',
|
| 118 |
-
security_token='WwUIFWBVUjeKn9VPKyWJmawY0')
|
| 119 |
-
|
| 120 |
def extract_text_from_pdf(pdf_path: str) -> str:
|
| 121 |
"""Extract text from PDF using pdfplumber"""
|
| 122 |
text = ""
|
|
@@ -183,22 +176,90 @@ def calculate_risk_score(penalty_count: int, penalty_values: List[float], obliga
|
|
| 183 |
else:
|
| 184 |
return score, "High"
|
| 185 |
|
| 186 |
-
def
|
| 187 |
-
"""
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
|
| 200 |
-
|
| 201 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 202 |
|
| 203 |
def analyze_pdf(file_obj) -> List:
|
| 204 |
"""Main analysis function for Gradio interface"""
|
|
@@ -229,32 +290,85 @@ def analyze_pdf(file_obj) -> List:
|
|
| 229 |
total_penalties, penalty_values, total_obligations, total_delays
|
| 230 |
)
|
| 231 |
|
| 232 |
-
#
|
| 233 |
-
|
| 234 |
-
risk_score, risk_level, str(penalty_counts), str(penalty_values), str(penalty_values),
|
| 235 |
-
str(obligation_counts), str(delay_counts)
|
| 236 |
-
)
|
| 237 |
|
| 238 |
-
#
|
| 239 |
-
|
| 240 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 241 |
except Exception as e:
|
| 242 |
return [f"Error: {str(e)}"] * 6
|
| 243 |
|
| 244 |
-
# Create Gradio interface
|
| 245 |
with gr.Blocks(css=css, title="PDF Contract Risk Analyzer") as demo:
|
| 246 |
gr.Markdown("# 📄 PDF Contract Risk Analyzer")
|
| 247 |
gr.Markdown("Upload a contract PDF to analyze penalties, obligations, and delays.")
|
| 248 |
|
| 249 |
with gr.Row():
|
| 250 |
-
|
| 251 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 252 |
|
| 253 |
with gr.Row():
|
| 254 |
-
|
| 255 |
-
|
| 256 |
|
| 257 |
-
submit_btn.click(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 258 |
|
| 259 |
if __name__ == "__main__":
|
| 260 |
-
demo.launch()
|
|
|
|
| 2 |
import pdfplumber
|
| 3 |
import matplotlib.pyplot as plt
|
| 4 |
import numpy as np
|
|
|
|
| 5 |
from word2number import w2n
|
| 6 |
+
import re
|
| 7 |
from typing import Tuple, List, Dict
|
|
|
|
| 8 |
|
| 9 |
# Custom CSS for styling
|
| 10 |
css = """
|
|
|
|
| 110 |
}
|
| 111 |
"""
|
| 112 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 113 |
def extract_text_from_pdf(pdf_path: str) -> str:
|
| 114 |
"""Extract text from PDF using pdfplumber"""
|
| 115 |
text = ""
|
|
|
|
| 176 |
else:
|
| 177 |
return score, "High"
|
| 178 |
|
| 179 |
+
def generate_combined_risk_display(risk_score: float, risk_level: str) -> Tuple[str, plt.Figure]:
|
| 180 |
+
"""Generate a combined display with all three risk levels in one layout"""
|
| 181 |
+
fig, ax = plt.subplots(figsize=(10, 3))
|
| 182 |
+
ax.axis('off')
|
| 183 |
+
|
| 184 |
+
risk_levels = ["Low", "Medium", "High"]
|
| 185 |
+
colors = ['#28a745', '#ffc107', '#dc3545']
|
| 186 |
+
|
| 187 |
+
# Create HTML for the text display
|
| 188 |
+
html_parts = []
|
| 189 |
+
html_parts.append("<div class='combined-risk-container'>")
|
| 190 |
+
|
| 191 |
+
for i, level in enumerate(risk_levels):
|
| 192 |
+
active = level == risk_level
|
| 193 |
+
score = risk_score if active else 0
|
| 194 |
+
|
| 195 |
+
# Add to HTML
|
| 196 |
+
html_parts.append(f"""
|
| 197 |
+
<div class='risk-row'>
|
| 198 |
+
<div class='risk-label risk-{level.lower()}'>{level} Risk</div>
|
| 199 |
+
<div class='risk-score risk-{level.lower()}'>{score:.1f}%</div>
|
| 200 |
+
<div class='heatmap-wrapper'>
|
| 201 |
+
<img src='data:image/png;base64,{create_mini_heatmap(score, colors[i])}' style='width:100%'>
|
| 202 |
+
</div>
|
| 203 |
+
</div>
|
| 204 |
+
""")
|
| 205 |
+
|
| 206 |
+
html_parts.append("</div>")
|
| 207 |
+
|
| 208 |
+
return "\n".join(html_parts), fig
|
| 209 |
+
|
| 210 |
+
def create_mini_heatmap(score: float, color: str) -> str:
|
| 211 |
+
"""Create a small heatmap for one risk level"""
|
| 212 |
+
fig, ax = plt.subplots(figsize=(8, 0.5))
|
| 213 |
+
|
| 214 |
+
if score > 0:
|
| 215 |
+
gradient = np.linspace(0, score/100, 256).reshape(1, -1)
|
| 216 |
+
else:
|
| 217 |
+
gradient = np.zeros((1, 256))
|
| 218 |
+
|
| 219 |
+
gradient = np.vstack((gradient, gradient))
|
| 220 |
+
|
| 221 |
+
ax.imshow(gradient, aspect='auto', cmap=plt.cm.colors.LinearSegmentedColormap.from_list('custom', ['white', color]))
|
| 222 |
+
ax.set_axis_off()
|
| 223 |
+
plt.tight_layout()
|
| 224 |
+
|
| 225 |
+
# Save to base64 string
|
| 226 |
+
from io import BytesIO
|
| 227 |
+
buf = BytesIO()
|
| 228 |
+
plt.savefig(buf, format='png', bbox_inches='tight', pad_inches=0)
|
| 229 |
+
plt.close(fig)
|
| 230 |
+
import base64
|
| 231 |
+
return base64.b64encode(buf.getvalue()).decode('utf-8')
|
| 232 |
+
|
| 233 |
+
def format_warning_message(count: int, items: str, item_type: str) -> str:
|
| 234 |
+
"""Format warning message based on count"""
|
| 235 |
+
if count == 0:
|
| 236 |
+
return f"""<div class="success-box">✓ No {item_type} clauses detected - Good!</div>"""
|
| 237 |
+
elif count < 5:
|
| 238 |
+
return f"""<div class="warning-box">⚠️ {count} {item_type} clauses detected</div>"""
|
| 239 |
+
else:
|
| 240 |
+
return f"""<div class="danger-box">⚠️⚠️ {count} {item_type} clauses detected - High Risk!</div>"""
|
| 241 |
+
|
| 242 |
+
def create_vertical_count_display(counts: Dict[str, int], total: int, section_type: str) -> str:
|
| 243 |
+
"""Create a vertical display for counts with section styling"""
|
| 244 |
+
items_html = []
|
| 245 |
+
for kw, count in counts.items():
|
| 246 |
+
items_html.append(f"""
|
| 247 |
+
<div class="count-item">
|
| 248 |
+
<span class="count-label">{kw.title()}:</span>
|
| 249 |
+
<span class="count-value">{count}</span>
|
| 250 |
+
</div>
|
| 251 |
+
""")
|
| 252 |
+
|
| 253 |
+
return f"""
|
| 254 |
+
<div class="section-container {section_type}-box">
|
| 255 |
+
<div class="section-title">{section_type.title()} Analysis</div>
|
| 256 |
+
<div style="margin-bottom: 10px;">
|
| 257 |
+
<span style="font-weight: bold;">Total:</span>
|
| 258 |
+
<span style="margin-left: 10px;">{total}</span>
|
| 259 |
+
</div>
|
| 260 |
+
{"".join(items_html)}
|
| 261 |
+
</div>
|
| 262 |
+
"""
|
| 263 |
|
| 264 |
def analyze_pdf(file_obj) -> List:
|
| 265 |
"""Main analysis function for Gradio interface"""
|
|
|
|
| 290 |
total_penalties, penalty_values, total_obligations, total_delays
|
| 291 |
)
|
| 292 |
|
| 293 |
+
# Generate combined risk display
|
| 294 |
+
risk_display, _ = generate_combined_risk_display(risk_score, risk_level)
|
|
|
|
|
|
|
|
|
|
| 295 |
|
| 296 |
+
# Prepare warning/success messages
|
| 297 |
+
penalty_warning = format_warning_message(total_penalties, "penalty", "penalty")
|
| 298 |
+
obligation_warning = format_warning_message(total_obligations, "obligation", "obligation")
|
| 299 |
+
delay_warning = format_warning_message(total_delays, "delay", "delay")
|
| 300 |
+
|
| 301 |
+
# Create vertical displays
|
| 302 |
+
penalty_display = create_vertical_count_display(penalty_counts, total_penalties, "penalty")
|
| 303 |
+
obligation_display = create_vertical_count_display(obligation_counts, total_obligations, "obligation")
|
| 304 |
+
delay_display = create_vertical_count_display(delay_counts, total_delays, "delay")
|
| 305 |
+
|
| 306 |
+
# Combine warnings with displays
|
| 307 |
+
penalty_output = f"{penalty_warning}\n{penalty_display}"
|
| 308 |
+
obligation_output = f"{obligation_warning}\n{obligation_display}"
|
| 309 |
+
delay_output = f"{delay_warning}\n{delay_display}"
|
| 310 |
+
|
| 311 |
+
penalty_amounts = "\n".join([f"- ${amt:,.2f}" for amt in penalty_values[:5]]) if penalty_values else "No specific penalty amounts found"
|
| 312 |
+
|
| 313 |
+
# Find example sentences with penalties
|
| 314 |
+
penalty_sentences = []
|
| 315 |
+
for sentence in re.split(r'(?<=[.!?])\s+', text):
|
| 316 |
+
if any(kw.lower() in sentence.lower() for kw in penalty_keywords):
|
| 317 |
+
penalty_sentences.append(sentence.strip())
|
| 318 |
+
|
| 319 |
+
penalty_examples = "\n\n".join([f"{i+1}. {sent}" for i, sent in enumerate(penalty_sentences[:3])]) if penalty_sentences else "No penalty clauses found"
|
| 320 |
+
|
| 321 |
+
# Return all results
|
| 322 |
+
return [
|
| 323 |
+
risk_display,
|
| 324 |
+
penalty_output,
|
| 325 |
+
f"{len(penalty_values)} amounts found\n\n{penalty_amounts}",
|
| 326 |
+
obligation_output,
|
| 327 |
+
delay_output,
|
| 328 |
+
penalty_examples
|
| 329 |
+
]
|
| 330 |
except Exception as e:
|
| 331 |
return [f"Error: {str(e)}"] * 6
|
| 332 |
|
| 333 |
+
# Create Gradio interface
|
| 334 |
with gr.Blocks(css=css, title="PDF Contract Risk Analyzer") as demo:
|
| 335 |
gr.Markdown("# 📄 PDF Contract Risk Analyzer")
|
| 336 |
gr.Markdown("Upload a contract PDF to analyze penalties, obligations, and delays.")
|
| 337 |
|
| 338 |
with gr.Row():
|
| 339 |
+
with gr.Column():
|
| 340 |
+
file_input = gr.File(label="Upload PDF", file_types=[".pdf"])
|
| 341 |
+
submit_btn = gr.Button("Analyze PDF", variant="primary")
|
| 342 |
+
|
| 343 |
+
with gr.Column():
|
| 344 |
+
gr.Markdown("### 🔍 Overall Risk Assessment")
|
| 345 |
+
risk_display = gr.HTML(label="Risk Analysis")
|
| 346 |
+
|
| 347 |
+
with gr.Row():
|
| 348 |
+
with gr.Column():
|
| 349 |
+
gr.Markdown("### 📊 Penalties Analysis")
|
| 350 |
+
penalty_count = gr.HTML(label="Penalty Clauses")
|
| 351 |
+
gr.Markdown("### Penalty Amounts")
|
| 352 |
+
penalty_amounts = gr.Textbox(label="", lines=5)
|
| 353 |
+
|
| 354 |
+
with gr.Column():
|
| 355 |
+
gr.Markdown("### ⚖️ Obligations Analysis")
|
| 356 |
+
obligation_count = gr.HTML(label="Obligation Clauses")
|
| 357 |
+
|
| 358 |
+
with gr.Column():
|
| 359 |
+
gr.Markdown("### ⏱️ Delays Analysis")
|
| 360 |
+
delay_count = gr.HTML(label="Delay Clauses")
|
| 361 |
|
| 362 |
with gr.Row():
|
| 363 |
+
gr.Markdown("### 🔎 Extracted Penalty Clauses")
|
| 364 |
+
penalty_examples = gr.Textbox(label="Example Penalty Clauses", lines=5)
|
| 365 |
|
| 366 |
+
submit_btn.click(
|
| 367 |
+
fn=analyze_pdf,
|
| 368 |
+
inputs=file_input,
|
| 369 |
+
outputs=[risk_display, penalty_count, penalty_amounts,
|
| 370 |
+
obligation_count, delay_count, penalty_examples]
|
| 371 |
+
)
|
| 372 |
|
| 373 |
if __name__ == "__main__":
|
| 374 |
+
demo.launch()
|