Commit
Β·
520de88
1
Parent(s):
ffe6715
pdf_generator function fixed
Browse files- reporter/report_generator.py +592 -256
reporter/report_generator.py
CHANGED
|
@@ -82,6 +82,18 @@ class ReportGenerator:
|
|
| 82 |
# Convert DetectionResult to dict for consistent access
|
| 83 |
detection_dict = detection_result.to_dict() if hasattr(detection_result, 'to_dict') else detection_result
|
| 84 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
# Generate detailed reasoning
|
| 86 |
reasoning = self.reasoning_generator.generate(ensemble_result = detection_result.ensemble_result,
|
| 87 |
metric_results = detection_result.metric_results,
|
|
@@ -91,7 +103,7 @@ class ReportGenerator:
|
|
| 91 |
)
|
| 92 |
|
| 93 |
# Extract detailed metrics from ACTUAL detection results
|
| 94 |
-
detailed_metrics = self._extract_detailed_metrics(
|
| 95 |
|
| 96 |
# Timestamp for filenames
|
| 97 |
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
|
@@ -100,7 +112,8 @@ class ReportGenerator:
|
|
| 100 |
|
| 101 |
# Generate requested formats
|
| 102 |
if ("json" in formats):
|
| 103 |
-
json_path = self._generate_json_report(
|
|
|
|
| 104 |
reasoning = reasoning,
|
| 105 |
detailed_metrics = detailed_metrics,
|
| 106 |
attribution_result = attribution_result,
|
|
@@ -111,7 +124,8 @@ class ReportGenerator:
|
|
| 111 |
|
| 112 |
if ("pdf" in formats):
|
| 113 |
try:
|
| 114 |
-
pdf_path = self._generate_pdf_report(
|
|
|
|
| 115 |
reasoning = reasoning,
|
| 116 |
detailed_metrics = detailed_metrics,
|
| 117 |
attribution_result = attribution_result,
|
|
@@ -129,44 +143,56 @@ class ReportGenerator:
|
|
| 129 |
return generated_files
|
| 130 |
|
| 131 |
|
| 132 |
-
def _extract_detailed_metrics(self,
|
| 133 |
"""
|
| 134 |
Extract detailed metrics with sub-metrics from ACTUAL detection result
|
| 135 |
"""
|
| 136 |
detailed_metrics = list()
|
| 137 |
-
metrics_data =
|
| 138 |
-
ensemble_data =
|
| 139 |
|
| 140 |
# Get actual metric weights from ensemble
|
| 141 |
metric_weights = ensemble_data.get("metric_contributions", {})
|
| 142 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 143 |
# Extract actual metric data
|
| 144 |
for metric_name, metric_result in metrics_data.items():
|
| 145 |
-
if not isinstance(metric_result, dict):
|
|
|
|
| 146 |
continue
|
| 147 |
|
| 148 |
-
if metric_result.get("error") is not None:
|
|
|
|
| 149 |
continue
|
| 150 |
|
| 151 |
# Get actual probabilities and confidence
|
| 152 |
-
ai_prob = metric_result.get("ai_probability", 0)
|
| 153 |
-
human_prob = metric_result.get("human_probability", 0)
|
| 154 |
-
confidence = metric_result.get("confidence", 0)
|
|
|
|
|
|
|
|
|
|
| 155 |
|
| 156 |
# Determine verdict based on actual probability
|
| 157 |
-
|
|
|
|
| 158 |
verdict = "AI"
|
| 159 |
-
|
| 160 |
-
|
|
|
|
| 161 |
verdict = "HUMAN"
|
| 162 |
|
| 163 |
else:
|
| 164 |
-
verdict = "MIXED
|
| 165 |
|
| 166 |
# Get actual weight or use default
|
| 167 |
weight = 0.0
|
|
|
|
| 168 |
if metric_name in metric_weights:
|
| 169 |
-
weight = metric_weights[metric_name].get("weight", 0.0)
|
| 170 |
|
| 171 |
# Extract actual detailed metrics from metric result
|
| 172 |
detailed_metrics_data = self._extract_metric_details(metric_name = metric_name,
|
|
@@ -177,16 +203,17 @@ class ReportGenerator:
|
|
| 177 |
description = self._get_metric_description(metric_name = metric_name)
|
| 178 |
|
| 179 |
detailed_metrics.append(DetailedMetric(name = metric_name,
|
| 180 |
-
ai_probability = ai_prob,
|
| 181 |
-
human_probability = human_prob,
|
| 182 |
-
confidence = confidence,
|
| 183 |
verdict = verdict,
|
| 184 |
description = description,
|
| 185 |
detailed_metrics = detailed_metrics_data,
|
| 186 |
-
weight = weight,
|
| 187 |
)
|
| 188 |
)
|
| 189 |
|
|
|
|
| 190 |
return detailed_metrics
|
| 191 |
|
| 192 |
|
|
@@ -226,7 +253,7 @@ class ReportGenerator:
|
|
| 226 |
return descriptions.get(metric_name, "Advanced text analysis metric.")
|
| 227 |
|
| 228 |
|
| 229 |
-
def _generate_json_report(self,
|
| 230 |
attribution_result: Optional[AttributionResult], highlighted_sentences: Optional[List] = None, filename: str = None) -> Path:
|
| 231 |
"""
|
| 232 |
Generate JSON format report with detailed metrics
|
|
@@ -273,54 +300,54 @@ class ReportGenerator:
|
|
| 273 |
}
|
| 274 |
|
| 275 |
# Use ACTUAL detection results from dictionary
|
| 276 |
-
ensemble_data
|
| 277 |
-
analysis_data
|
| 278 |
-
metrics_data_dict
|
| 279 |
-
performance_data
|
| 280 |
-
|
| 281 |
-
report_data
|
| 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 |
-
output_path
|
| 324 |
|
| 325 |
with open(output_path, 'w', encoding='utf-8') as f:
|
| 326 |
json.dump(obj = report_data,
|
|
@@ -333,271 +360,580 @@ class ReportGenerator:
|
|
| 333 |
return output_path
|
| 334 |
|
| 335 |
|
| 336 |
-
def _generate_pdf_report(self,
|
| 337 |
attribution_result: Optional[AttributionResult], highlighted_sentences: Optional[List] = None, filename: str = None) -> Path:
|
| 338 |
"""
|
| 339 |
Generate PDF format report with detailed metrics
|
| 340 |
"""
|
| 341 |
try:
|
| 342 |
from reportlab.lib import colors
|
| 343 |
-
from reportlab.lib.
|
| 344 |
-
from reportlab.
|
| 345 |
from reportlab.lib.units import inch
|
| 346 |
-
from reportlab.platypus import
|
| 347 |
-
from reportlab.lib.
|
| 348 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 349 |
except ImportError:
|
| 350 |
raise ImportError("reportlab is required for PDF generation. Install: pip install reportlab")
|
| 351 |
|
| 352 |
output_path = self.output_dir / filename
|
| 353 |
|
| 354 |
-
# Create PDF
|
| 355 |
doc = SimpleDocTemplate(str(output_path),
|
| 356 |
-
pagesize =
|
| 357 |
-
rightMargin =
|
| 358 |
-
leftMargin =
|
| 359 |
-
topMargin =
|
| 360 |
-
bottomMargin =
|
| 361 |
)
|
| 362 |
|
| 363 |
# Container for PDF elements
|
| 364 |
-
elements
|
| 365 |
-
styles
|
| 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 |
-
# Use detection results from
|
| 392 |
-
ensemble_data
|
| 393 |
-
analysis_data
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 394 |
|
| 395 |
# Title and main sections
|
| 396 |
elements.append(Paragraph("AI Text Detection Analysis Report", title_style))
|
| 397 |
-
elements.append(Paragraph(f"Generated on {datetime.now().strftime('%B %d, %Y at %I:%M %p')}",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 398 |
elements.append(Spacer(1, 0.3*inch))
|
| 399 |
|
| 400 |
-
# Verdict
|
| 401 |
-
elements.append(Paragraph("
|
| 402 |
-
verdict_data = [['Final Verdict:', ensemble_data.get("final_verdict", "Unknown")],
|
| 403 |
-
['AI Probability:', f"{ensemble_data.get('ai_probability', 0):.1%}"],
|
| 404 |
-
['Human Probability:', f"{ensemble_data.get('human_probability', 0):.1%}"],
|
| 405 |
-
['Mixed Probability:', f"{ensemble_data.get('mixed_probability', 0):.1%}"],
|
| 406 |
-
['Overall Confidence:', f"{ensemble_data.get('overall_confidence', 0):.1%}"],
|
| 407 |
-
['Uncertainty Score:', f"{ensemble_data.get('uncertainty_score', 0):.1%}"],
|
| 408 |
-
['Consensus Level:', f"{ensemble_data.get('consensus_level', 0):.1%}"],
|
| 409 |
-
]
|
| 410 |
-
|
| 411 |
-
verdict_table = Table(verdict_data, colWidths=[2*inch, 3*inch])
|
| 412 |
-
verdict_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (0, -1), colors.HexColor('#f8fafc')),
|
| 413 |
-
('FONTNAME', (0, 0), (0, -1), 'Helvetica-Bold'),
|
| 414 |
-
('FONTSIZE', (0, 0), (-1, -1), 10),
|
| 415 |
-
('BOTTOMPADDING', (0, 0), (-1, -1), 6),
|
| 416 |
-
])
|
| 417 |
-
)
|
| 418 |
|
| 419 |
-
|
| 420 |
-
|
|
|
|
| 421 |
|
| 422 |
-
|
| 423 |
-
|
| 424 |
-
|
| 425 |
-
|
| 426 |
-
|
| 427 |
-
|
| 428 |
-
|
| 429 |
-
|
| 430 |
-
|
| 431 |
-
|
| 432 |
-
|
| 433 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 434 |
])
|
| 435 |
)
|
| 436 |
-
|
| 437 |
elements.append(content_table)
|
| 438 |
-
elements.append(Spacer(1, 0.
|
| 439 |
|
| 440 |
-
#
|
| 441 |
-
elements.append(Paragraph("
|
| 442 |
-
elements.append(Paragraph("Method: Confidence Calibrated Aggregation", styles['Normal']))
|
| 443 |
-
elements.append(Spacer(1, 0.1*inch))
|
| 444 |
|
| 445 |
-
# Metric weights table
|
| 446 |
metric_contributions = ensemble_data.get("metric_contributions", {})
|
| 447 |
-
|
| 448 |
-
|
| 449 |
-
|
| 450 |
-
|
| 451 |
-
|
| 452 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 453 |
|
| 454 |
-
weight_table = Table(weight_data, colWidths=[
|
| 455 |
-
weight_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0),
|
| 456 |
('TEXTCOLOR', (0, 0), (-1, 0), colors.white),
|
| 457 |
('ALIGN', (0, 0), (-1, -1), 'LEFT'),
|
| 458 |
('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
|
| 459 |
('FONTSIZE', (0, 0), (-1, -1), 9),
|
| 460 |
-
('BOTTOMPADDING', (0, 0), (-1, -1),
|
| 461 |
-
('
|
|
|
|
|
|
|
|
|
|
| 462 |
])
|
| 463 |
-
|
|
|
|
| 464 |
elements.append(weight_table)
|
| 465 |
-
elements.append(Spacer(1, 0.
|
| 466 |
|
| 467 |
-
# Detailed
|
| 468 |
-
elements.append(Paragraph("
|
| 469 |
-
|
| 470 |
-
|
| 471 |
-
|
| 472 |
-
|
| 473 |
-
|
| 474 |
-
|
| 475 |
-
|
| 476 |
-
|
| 477 |
-
|
| 478 |
-
|
| 479 |
-
|
| 480 |
-
|
| 481 |
-
|
| 482 |
-
|
| 483 |
-
|
| 484 |
-
elements.append(metric_table)
|
| 485 |
-
elements.append(Paragraph(metric.description, body_style))
|
| 486 |
-
|
| 487 |
-
# Add detailed sub-metrics if available
|
| 488 |
-
if metric.detailed_metrics:
|
| 489 |
-
elements.append(Paragraph("Detailed Metrics:", styles['Heading4']))
|
| 490 |
-
sub_metric_data = [['Metric', 'Value']]
|
| 491 |
-
for sub_name, sub_value in list(metric.detailed_metrics.items())[:6]: # Show top 6
|
| 492 |
-
sub_metric_data.append([sub_name.replace('_', ' ').title(), f"{sub_value:.2f}"])
|
| 493 |
|
| 494 |
-
|
| 495 |
-
|
| 496 |
-
|
| 497 |
-
|
| 498 |
-
|
| 499 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 500 |
|
| 501 |
-
|
| 502 |
-
|
| 503 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 504 |
|
| 505 |
# Detection Reasoning
|
| 506 |
-
elements.append(Paragraph("
|
| 507 |
-
|
| 508 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 509 |
|
| 510 |
# Key Indicators
|
| 511 |
-
|
| 512 |
-
|
| 513 |
-
|
| 514 |
-
|
| 515 |
-
|
| 516 |
-
|
| 517 |
-
|
| 518 |
-
|
| 519 |
-
|
| 520 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 521 |
|
| 522 |
-
#
|
| 523 |
-
elements.append(
|
| 524 |
-
elements.append(Paragraph(reasoning.uncertainty_analysis, body_style))
|
| 525 |
|
| 526 |
# Model Attribution Section
|
| 527 |
if attribution_result:
|
| 528 |
-
elements.append(
|
| 529 |
-
|
|
|
|
|
|
|
| 530 |
|
| 531 |
-
|
| 532 |
-
|
| 533 |
-
|
|
|
|
| 534 |
|
| 535 |
-
|
| 536 |
-
['Attribution Confidence:', f"{confidence:.1f}%"],
|
| 537 |
-
['Domain Used:', attribution_result.domain_used.value.title()],
|
| 538 |
-
]
|
| 539 |
|
| 540 |
-
attribution_table
|
| 541 |
-
attribution_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (0, -1), colors.HexColor('#f8fafc')),
|
| 542 |
('FONTNAME', (0, 0), (0, -1), 'Helvetica-Bold'),
|
| 543 |
-
('FONTSIZE', (0, 0), (-1, -1),
|
| 544 |
-
('BOTTOMPADDING', (0, 0), (-1, -1),
|
|
|
|
|
|
|
| 545 |
])
|
| 546 |
)
|
| 547 |
-
|
| 548 |
elements.append(attribution_table)
|
| 549 |
-
elements.append(Spacer(1, 0.
|
| 550 |
|
| 551 |
# Model probabilities table
|
| 552 |
if attribution_result.model_probabilities:
|
| 553 |
-
elements.append(Paragraph("
|
| 554 |
|
| 555 |
-
prob_data = [['
|
| 556 |
|
| 557 |
-
# Show top
|
| 558 |
-
sorted_models = sorted(attribution_result.model_probabilities.items(),
|
| 559 |
-
key = lambda x: x[1],
|
| 560 |
-
reverse = True)[:5]
|
| 561 |
|
| 562 |
for model_name, probability in sorted_models:
|
| 563 |
display_name = model_name.replace("_", " ").replace("-", " ").title()
|
| 564 |
-
|
| 565 |
-
|
| 566 |
-
|
| 567 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 568 |
('TEXTCOLOR', (0, 0), (-1, 0), colors.white),
|
| 569 |
('ALIGN', (0, 0), (-1, -1), 'LEFT'),
|
|
|
|
| 570 |
('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
|
| 571 |
('FONTSIZE', (0, 0), (-1, -1), 9),
|
| 572 |
-
('BOTTOMPADDING', (0, 0), (-1, -1),
|
| 573 |
-
('
|
|
|
|
|
|
|
|
|
|
| 574 |
])
|
| 575 |
)
|
| 576 |
-
|
| 577 |
elements.append(prob_table)
|
| 578 |
-
elements.append(Spacer(1, 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 579 |
|
| 580 |
-
|
| 581 |
-
|
| 582 |
-
|
| 583 |
-
|
| 584 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 585 |
|
| 586 |
-
#
|
| 587 |
-
|
| 588 |
-
|
| 589 |
-
|
| 590 |
-
|
|
|
|
|
|
|
|
|
|
| 591 |
|
| 592 |
-
|
| 593 |
-
elements.append(
|
| 594 |
-
|
| 595 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 596 |
|
| 597 |
# Build PDF
|
| 598 |
doc.build(elements)
|
| 599 |
|
| 600 |
-
logger.info(f"PDF report saved: {output_path}")
|
| 601 |
return output_path
|
| 602 |
|
| 603 |
|
|
|
|
| 82 |
# Convert DetectionResult to dict for consistent access
|
| 83 |
detection_dict = detection_result.to_dict() if hasattr(detection_result, 'to_dict') else detection_result
|
| 84 |
|
| 85 |
+
# DEBUG: Check structure
|
| 86 |
+
logger.debug(f"detection_dict keys: {list(detection_dict.keys())}")
|
| 87 |
+
|
| 88 |
+
# Extract the actual detection data from the structure: The full response has 'detection_result' key, but we need the inner data
|
| 89 |
+
if ("detection_result" in detection_dict):
|
| 90 |
+
detection_data = detection_dict["detection_result"]
|
| 91 |
+
logger.debug("Extracted detection_result from outer dict")
|
| 92 |
+
|
| 93 |
+
else:
|
| 94 |
+
detection_data = detection_dict
|
| 95 |
+
logger.debug("Using detection_dict directly")
|
| 96 |
+
|
| 97 |
# Generate detailed reasoning
|
| 98 |
reasoning = self.reasoning_generator.generate(ensemble_result = detection_result.ensemble_result,
|
| 99 |
metric_results = detection_result.metric_results,
|
|
|
|
| 103 |
)
|
| 104 |
|
| 105 |
# Extract detailed metrics from ACTUAL detection results
|
| 106 |
+
detailed_metrics = self._extract_detailed_metrics(detection_data)
|
| 107 |
|
| 108 |
# Timestamp for filenames
|
| 109 |
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
|
|
|
|
| 112 |
|
| 113 |
# Generate requested formats
|
| 114 |
if ("json" in formats):
|
| 115 |
+
json_path = self._generate_json_report(detection_data = detection_data,
|
| 116 |
+
detection_dict_full = detection_dict,
|
| 117 |
reasoning = reasoning,
|
| 118 |
detailed_metrics = detailed_metrics,
|
| 119 |
attribution_result = attribution_result,
|
|
|
|
| 124 |
|
| 125 |
if ("pdf" in formats):
|
| 126 |
try:
|
| 127 |
+
pdf_path = self._generate_pdf_report(detection_data = detection_data,
|
| 128 |
+
detection_dict_full = detection_dict,
|
| 129 |
reasoning = reasoning,
|
| 130 |
detailed_metrics = detailed_metrics,
|
| 131 |
attribution_result = attribution_result,
|
|
|
|
| 143 |
return generated_files
|
| 144 |
|
| 145 |
|
| 146 |
+
def _extract_detailed_metrics(self, detection_data: Dict) -> List[DetailedMetric]:
|
| 147 |
"""
|
| 148 |
Extract detailed metrics with sub-metrics from ACTUAL detection result
|
| 149 |
"""
|
| 150 |
detailed_metrics = list()
|
| 151 |
+
metrics_data = detection_data.get("metrics", {})
|
| 152 |
+
ensemble_data = detection_data.get("ensemble", {})
|
| 153 |
|
| 154 |
# Get actual metric weights from ensemble
|
| 155 |
metric_weights = ensemble_data.get("metric_contributions", {})
|
| 156 |
|
| 157 |
+
# Log what we're working with
|
| 158 |
+
logger.debug(f"Extracting metrics from {len(metrics_data)} metrics")
|
| 159 |
+
logger.debug(f"Metric names: {list(metrics_data.keys())}")
|
| 160 |
+
|
| 161 |
# Extract actual metric data
|
| 162 |
for metric_name, metric_result in metrics_data.items():
|
| 163 |
+
if (not isinstance(metric_result, dict)):
|
| 164 |
+
logger.warning(f"Metric {metric_name} is not a dict: {type(metric_result)}")
|
| 165 |
continue
|
| 166 |
|
| 167 |
+
if (metric_result.get("error") is not None):
|
| 168 |
+
logger.warning(f"Metric {metric_name} has error: {metric_result.get('error')}")
|
| 169 |
continue
|
| 170 |
|
| 171 |
# Get actual probabilities and confidence
|
| 172 |
+
ai_prob = metric_result.get("ai_probability", 0)
|
| 173 |
+
human_prob = metric_result.get("human_probability", 0)
|
| 174 |
+
confidence = metric_result.get("confidence", 0)
|
| 175 |
+
|
| 176 |
+
# DEBUG: Log extracted values
|
| 177 |
+
logger.debug(f"Metric {metric_name}: AI={ai_prob}, Human={human_prob}, Confidence={confidence}")
|
| 178 |
|
| 179 |
# Determine verdict based on actual probability
|
| 180 |
+
# 60% threshold in decimal
|
| 181 |
+
if (ai_prob >= 0.6):
|
| 182 |
verdict = "AI"
|
| 183 |
+
|
| 184 |
+
# 40% threshold in decimal
|
| 185 |
+
elif (ai_prob <= 0.4):
|
| 186 |
verdict = "HUMAN"
|
| 187 |
|
| 188 |
else:
|
| 189 |
+
verdict = "MIXED"
|
| 190 |
|
| 191 |
# Get actual weight or use default
|
| 192 |
weight = 0.0
|
| 193 |
+
|
| 194 |
if metric_name in metric_weights:
|
| 195 |
+
weight = metric_weights[metric_name].get("weight", 0.0)
|
| 196 |
|
| 197 |
# Extract actual detailed metrics from metric result
|
| 198 |
detailed_metrics_data = self._extract_metric_details(metric_name = metric_name,
|
|
|
|
| 203 |
description = self._get_metric_description(metric_name = metric_name)
|
| 204 |
|
| 205 |
detailed_metrics.append(DetailedMetric(name = metric_name,
|
| 206 |
+
ai_probability = ai_prob * 100, # Convert to percentage
|
| 207 |
+
human_probability = human_prob * 100, # Convert to percentage
|
| 208 |
+
confidence = confidence * 100, # Convert to percentage
|
| 209 |
verdict = verdict,
|
| 210 |
description = description,
|
| 211 |
detailed_metrics = detailed_metrics_data,
|
| 212 |
+
weight = weight * 100, # Convert to percentage
|
| 213 |
)
|
| 214 |
)
|
| 215 |
|
| 216 |
+
logger.debug(f"Extracted {len(detailed_metrics)} detailed metrics")
|
| 217 |
return detailed_metrics
|
| 218 |
|
| 219 |
|
|
|
|
| 253 |
return descriptions.get(metric_name, "Advanced text analysis metric.")
|
| 254 |
|
| 255 |
|
| 256 |
+
def _generate_json_report(self, detection_data: Dict, detection_dict_full: Dict, reasoning: DetailedReasoning, detailed_metrics: List[DetailedMetric],
|
| 257 |
attribution_result: Optional[AttributionResult], highlighted_sentences: Optional[List] = None, filename: str = None) -> Path:
|
| 258 |
"""
|
| 259 |
Generate JSON format report with detailed metrics
|
|
|
|
| 300 |
}
|
| 301 |
|
| 302 |
# Use ACTUAL detection results from dictionary
|
| 303 |
+
ensemble_data = detection_data.get("ensemble", {})
|
| 304 |
+
analysis_data = detection_data.get("analysis", {})
|
| 305 |
+
metrics_data_dict = detection_data.get("metrics", {})
|
| 306 |
+
performance_data = detection_data.get("performance", {})
|
| 307 |
+
|
| 308 |
+
report_data = {"report_metadata" : {"generated_at" : datetime.now().isoformat(),
|
| 309 |
+
"version" : "1.0.0",
|
| 310 |
+
"format" : "json",
|
| 311 |
+
"report_id" : filename.replace('.json', ''),
|
| 312 |
+
},
|
| 313 |
+
"overall_results" : {"final_verdict" : ensemble_data.get("final_verdict", "Unknown"),
|
| 314 |
+
"ai_probability" : ensemble_data.get("ai_probability", 0),
|
| 315 |
+
"human_probability" : ensemble_data.get("human_probability", 0),
|
| 316 |
+
"mixed_probability" : ensemble_data.get("mixed_probability", 0),
|
| 317 |
+
"overall_confidence" : ensemble_data.get("overall_confidence", 0),
|
| 318 |
+
"uncertainty_score" : ensemble_data.get("uncertainty_score", 0),
|
| 319 |
+
"consensus_level" : ensemble_data.get("consensus_level", 0),
|
| 320 |
+
"domain" : analysis_data.get("domain", "general"),
|
| 321 |
+
"domain_confidence" : analysis_data.get("domain_confidence", 0),
|
| 322 |
+
"text_length" : analysis_data.get("text_length", 0),
|
| 323 |
+
"sentence_count" : analysis_data.get("sentence_count", 0),
|
| 324 |
+
},
|
| 325 |
+
"ensemble_analysis" : {"method_used" : "confidence_calibrated",
|
| 326 |
+
"metric_weights" : ensemble_data.get("metric_contributions", {}),
|
| 327 |
+
"reasoning" : ensemble_data.get("reasoning", []),
|
| 328 |
+
},
|
| 329 |
+
"detailed_metrics" : metrics_data,
|
| 330 |
+
"detection_reasoning" : {"summary" : reasoning.summary,
|
| 331 |
+
"key_indicators" : reasoning.key_indicators,
|
| 332 |
+
"metric_explanations" : reasoning.metric_explanations,
|
| 333 |
+
"supporting_evidence" : reasoning.supporting_evidence,
|
| 334 |
+
"contradicting_evidence" : reasoning.contradicting_evidence,
|
| 335 |
+
"confidence_explanation" : reasoning.confidence_explanation,
|
| 336 |
+
"domain_analysis" : reasoning.domain_analysis,
|
| 337 |
+
"ensemble_analysis" : reasoning.ensemble_analysis,
|
| 338 |
+
"uncertainty_analysis" : reasoning.uncertainty_analysis,
|
| 339 |
+
"recommendations" : reasoning.recommendations,
|
| 340 |
+
},
|
| 341 |
+
"highlighted_text" : highlighted_data,
|
| 342 |
+
"model_attribution" : attribution_data,
|
| 343 |
+
"performance_metrics" : {"total_processing_time" : performance_data.get("total_time", 0),
|
| 344 |
+
"metrics_execution_time" : performance_data.get("metrics_time", {}),
|
| 345 |
+
"warnings" : detection_data.get("warnings", []),
|
| 346 |
+
"errors" : detection_data.get("errors", []),
|
| 347 |
+
}
|
| 348 |
+
}
|
| 349 |
+
|
| 350 |
+
output_path = self.output_dir / filename
|
| 351 |
|
| 352 |
with open(output_path, 'w', encoding='utf-8') as f:
|
| 353 |
json.dump(obj = report_data,
|
|
|
|
| 360 |
return output_path
|
| 361 |
|
| 362 |
|
| 363 |
+
def _generate_pdf_report(self, detection_data: Dict, detection_dict_full: Dict, reasoning: DetailedReasoning, detailed_metrics: List[DetailedMetric],
|
| 364 |
attribution_result: Optional[AttributionResult], highlighted_sentences: Optional[List] = None, filename: str = None) -> Path:
|
| 365 |
"""
|
| 366 |
Generate PDF format report with detailed metrics
|
| 367 |
"""
|
| 368 |
try:
|
| 369 |
from reportlab.lib import colors
|
| 370 |
+
from reportlab.lib.units import cm
|
| 371 |
+
from reportlab.platypus import Table
|
| 372 |
from reportlab.lib.units import inch
|
| 373 |
+
from reportlab.platypus import Spacer
|
| 374 |
+
from reportlab.lib.pagesizes import A4
|
| 375 |
+
from reportlab.lib.enums import TA_LEFT
|
| 376 |
+
from reportlab.platypus import PageBreak
|
| 377 |
+
from reportlab.platypus import Paragraph
|
| 378 |
+
from reportlab.lib.enums import TA_RIGHT
|
| 379 |
+
from reportlab.graphics import renderPDF
|
| 380 |
+
from reportlab.lib.enums import TA_CENTER
|
| 381 |
+
from reportlab.platypus import TableStyle
|
| 382 |
+
from reportlab.pdfgen.canvas import Canvas
|
| 383 |
+
from reportlab.lib.enums import TA_JUSTIFY
|
| 384 |
+
from reportlab.lib.pagesizes import letter
|
| 385 |
+
from reportlab.graphics.shapes import Line
|
| 386 |
+
from reportlab.graphics.shapes import Rect
|
| 387 |
+
from reportlab.platypus import KeepTogether
|
| 388 |
+
from reportlab.graphics.shapes import Circle
|
| 389 |
+
from reportlab.graphics.shapes import Drawing
|
| 390 |
+
from reportlab.lib.styles import ParagraphStyle
|
| 391 |
+
from reportlab.platypus import SimpleDocTemplate
|
| 392 |
+
from reportlab.graphics.charts.piecharts import Pie
|
| 393 |
+
from reportlab.platypus.flowables import HRFlowable
|
| 394 |
+
from reportlab.lib.styles import getSampleStyleSheet
|
| 395 |
+
from reportlab.graphics.charts.textlabels import Label
|
| 396 |
+
from reportlab.graphics.widgets.markers import makeMarker
|
| 397 |
+
|
| 398 |
except ImportError:
|
| 399 |
raise ImportError("reportlab is required for PDF generation. Install: pip install reportlab")
|
| 400 |
|
| 401 |
output_path = self.output_dir / filename
|
| 402 |
|
| 403 |
+
# Create PDF with premium settings
|
| 404 |
doc = SimpleDocTemplate(str(output_path),
|
| 405 |
+
pagesize = A4,
|
| 406 |
+
rightMargin = 0.75*inch,
|
| 407 |
+
leftMargin = 0.75*inch,
|
| 408 |
+
topMargin = 0.75*inch,
|
| 409 |
+
bottomMargin = 0.75*inch,
|
| 410 |
)
|
| 411 |
|
| 412 |
# Container for PDF elements
|
| 413 |
+
elements = list()
|
| 414 |
+
styles = getSampleStyleSheet()
|
| 415 |
+
|
| 416 |
+
# Premium Color Scheme
|
| 417 |
+
PRIMARY_COLOR = colors.HexColor('#3b82f6') # Blue-600
|
| 418 |
+
SUCCESS_COLOR = colors.HexColor('#10b981') # Emerald-500
|
| 419 |
+
WARNING_COLOR = colors.HexColor('#f59e0b') # Amber-500
|
| 420 |
+
DANGER_COLOR = colors.HexColor('#ef4444') # Red-500
|
| 421 |
+
INFO_COLOR = colors.HexColor('#8b5cf6') # Violet-500
|
| 422 |
+
GRAY_LIGHT = colors.HexColor('#f8fafc') # Gray-50
|
| 423 |
+
GRAY_MEDIUM = colors.HexColor('#e2e8f0') # Gray-200
|
| 424 |
+
GRAY_DARK = colors.HexColor('#334155') # Gray-700
|
| 425 |
+
TEXT_COLOR = colors.HexColor('#1e293b') # Gray-800
|
| 426 |
+
|
| 427 |
+
# Premium Custom Styles
|
| 428 |
+
title_style = ParagraphStyle('PremiumTitle',
|
| 429 |
+
parent = styles['Heading1'],
|
| 430 |
+
fontName = 'Helvetica-Bold',
|
| 431 |
+
fontSize = 28,
|
| 432 |
+
textColor = PRIMARY_COLOR,
|
| 433 |
+
spaceAfter = 20,
|
| 434 |
+
alignment = TA_CENTER,
|
| 435 |
+
)
|
| 436 |
+
|
| 437 |
+
subtitle_style = ParagraphStyle('PremiumSubtitle',
|
| 438 |
+
parent = styles['Normal'],
|
| 439 |
+
fontName = 'Helvetica',
|
| 440 |
+
fontSize = 12,
|
| 441 |
+
textColor = GRAY_DARK,
|
| 442 |
+
spaceAfter = 30,
|
| 443 |
+
alignment = TA_CENTER,
|
| 444 |
+
)
|
| 445 |
|
| 446 |
+
section_style = ParagraphStyle('PremiumSection',
|
| 447 |
+
parent = styles['Heading2'],
|
| 448 |
+
fontName = 'Helvetica-Bold',
|
| 449 |
+
fontSize = 18,
|
| 450 |
+
textColor = TEXT_COLOR,
|
| 451 |
+
spaceAfter = 12,
|
| 452 |
+
spaceBefore = 20,
|
| 453 |
+
underlineWidth = 1,
|
| 454 |
+
underlineColor = PRIMARY_COLOR,
|
| 455 |
+
)
|
| 456 |
+
|
| 457 |
+
subsection_style = ParagraphStyle('PremiumSubSection',
|
| 458 |
+
parent = styles['Heading3'],
|
| 459 |
+
fontName = 'Helvetica-Bold',
|
| 460 |
+
fontSize = 14,
|
| 461 |
+
textColor = GRAY_DARK,
|
| 462 |
+
spaceAfter = 8,
|
| 463 |
+
spaceBefore = 16,
|
| 464 |
+
)
|
| 465 |
+
|
| 466 |
+
body_style = ParagraphStyle('PremiumBody',
|
| 467 |
+
parent = styles['BodyText'],
|
| 468 |
+
fontName = 'Helvetica',
|
| 469 |
+
fontSize = 11,
|
| 470 |
+
textColor = TEXT_COLOR,
|
| 471 |
+
alignment = TA_JUSTIFY,
|
| 472 |
+
spaceAfter = 8,
|
| 473 |
+
)
|
| 474 |
+
|
| 475 |
+
verdict_style = ParagraphStyle('VerdictStyle',
|
| 476 |
+
parent = styles['Heading2'],
|
| 477 |
+
fontName = 'Helvetica-Bold',
|
| 478 |
+
fontSize = 22,
|
| 479 |
+
spaceAfter = 5,
|
| 480 |
+
)
|
| 481 |
+
|
| 482 |
+
metric_name_style = ParagraphStyle('MetricNameStyle',
|
| 483 |
+
parent = styles['Heading3'],
|
| 484 |
+
fontName = 'Helvetica-Bold',
|
| 485 |
+
fontSize = 13,
|
| 486 |
+
textColor = GRAY_DARK,
|
| 487 |
+
spaceAfter = 4,
|
| 488 |
+
)
|
| 489 |
|
| 490 |
+
# Use detection results from detection_data
|
| 491 |
+
ensemble_data = detection_data.get("ensemble", {})
|
| 492 |
+
analysis_data = detection_data.get("analysis", {})
|
| 493 |
+
performance_data = detection_data.get("performance", {})
|
| 494 |
+
|
| 495 |
+
# Extract values
|
| 496 |
+
ai_prob = ensemble_data.get("ai_probability", 0)
|
| 497 |
+
human_prob = ensemble_data.get("human_probability", 0)
|
| 498 |
+
mixed_prob = ensemble_data.get("mixed_probability", 0)
|
| 499 |
+
confidence = ensemble_data.get("overall_confidence", 0)
|
| 500 |
+
uncertainty = ensemble_data.get("uncertainty_score", 0)
|
| 501 |
+
consensus = ensemble_data.get("consensus_level", 0)
|
| 502 |
+
final_verdict = ensemble_data.get("final_verdict", "Unknown")
|
| 503 |
+
|
| 504 |
+
# Determine colors based on verdict
|
| 505 |
+
if ("Human".lower() in final_verdict.lower()):
|
| 506 |
+
verdict_color = SUCCESS_COLOR
|
| 507 |
+
|
| 508 |
+
elif ("AI".lower() in final_verdict.lower()):
|
| 509 |
+
verdict_color = DANGER_COLOR
|
| 510 |
+
|
| 511 |
+
elif ("Mixed".lower() in final_verdict.lower()):
|
| 512 |
+
verdict_color = WARNING_COLOR
|
| 513 |
+
|
| 514 |
+
else:
|
| 515 |
+
verdict_color = PRIMARY_COLOR
|
| 516 |
+
|
| 517 |
+
# Create header with logo/company name
|
| 518 |
+
header_style = ParagraphStyle('HeaderStyle',
|
| 519 |
+
parent = styles['Normal'],
|
| 520 |
+
fontName = 'Helvetica-Bold',
|
| 521 |
+
fontSize = 10,
|
| 522 |
+
textColor = GRAY_DARK,
|
| 523 |
+
alignment = TA_RIGHT,
|
| 524 |
+
)
|
| 525 |
+
|
| 526 |
+
# Header
|
| 527 |
+
elements.append(Paragraph("AI DETECTION ANALYTICS", header_style))
|
| 528 |
+
elements.append(HRFlowable(width = "100%",
|
| 529 |
+
thickness = 1,
|
| 530 |
+
color = PRIMARY_COLOR,
|
| 531 |
+
spaceAfter = 20,
|
| 532 |
+
)
|
| 533 |
+
)
|
| 534 |
|
| 535 |
# Title and main sections
|
| 536 |
elements.append(Paragraph("AI Text Detection Analysis Report", title_style))
|
| 537 |
+
elements.append(Paragraph(f"Generated on {datetime.now().strftime('%B %d, %Y at %I:%M %p')}", subtitle_style))
|
| 538 |
+
|
| 539 |
+
# Add decorative line
|
| 540 |
+
elements.append(HRFlowable(width = "80%",
|
| 541 |
+
thickness = 2,
|
| 542 |
+
color = PRIMARY_COLOR,
|
| 543 |
+
spaceBefore = 10,
|
| 544 |
+
spaceAfter = 30,
|
| 545 |
+
hAlign = 'CENTER',
|
| 546 |
+
)
|
| 547 |
+
)
|
| 548 |
+
|
| 549 |
+
# Quick Stats Banner
|
| 550 |
+
stats_data = [['', 'AI', 'HUMAN', 'MIXED'],
|
| 551 |
+
['Probability', f"{ai_prob:.1%}", f"{human_prob:.1%}", f"{mixed_prob:.1%}"]
|
| 552 |
+
]
|
| 553 |
+
|
| 554 |
+
stats_table = Table(stats_data, colWidths = [1.5*inch, 1*inch, 1*inch, 1*inch])
|
| 555 |
+
stats_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0), PRIMARY_COLOR),
|
| 556 |
+
('TEXTCOLOR', (0, 0), (-1, 0), colors.white),
|
| 557 |
+
('BACKGROUND', (1, 1), (1, 1), DANGER_COLOR),
|
| 558 |
+
('BACKGROUND', (2, 1), (2, 1), SUCCESS_COLOR),
|
| 559 |
+
('BACKGROUND', (3, 1), (3, 1), WARNING_COLOR),
|
| 560 |
+
('TEXTCOLOR', (1, 1), (-1, 1), colors.white),
|
| 561 |
+
('ALIGN', (0, 0), (-1, -1), 'CENTER'),
|
| 562 |
+
('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
|
| 563 |
+
('FONTSIZE', (0, 0), (-1, -1), 11),
|
| 564 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 8),
|
| 565 |
+
('TOPPADDING', (0, 0), (-1, -1), 8),
|
| 566 |
+
('GRID', (0, 0), (-1, -1), 0.5, colors.white),
|
| 567 |
+
('BOX', (0, 0), (-1, -1), 1, PRIMARY_COLOR),
|
| 568 |
+
])
|
| 569 |
+
)
|
| 570 |
+
|
| 571 |
+
elements.append(stats_table)
|
| 572 |
elements.append(Spacer(1, 0.3*inch))
|
| 573 |
|
| 574 |
+
# Main Verdict Section with colored badge
|
| 575 |
+
elements.append(Paragraph("DETECTION VERDICT", section_style))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 576 |
|
| 577 |
+
verdict_box_data = [[Paragraph(f"<font size=18 color='{colors.toHex(verdict_color)}'><b>{final_verdict.upper()}</b></font>", ParagraphStyle('VerdictText', alignment=TA_CENTER)),
|
| 578 |
+
Paragraph(f"<font size=12>Confidence: <b>{confidence:.1%}</b></font><br/>" f"<font size=10>Uncertainty: {uncertainty:.1%} | Consensus: {consensus:.1%}</font>", ParagraphStyle('VerdictDetails', alignment=TA_CENTER))
|
| 579 |
+
]]
|
| 580 |
|
| 581 |
+
verdict_box = Table(verdict_box_data, colWidths=[2.5*inch, 3*inch])
|
| 582 |
+
|
| 583 |
+
verdict_box.setStyle(TableStyle([('BACKGROUND', (0, 0), (0, 0), GRAY_LIGHT),
|
| 584 |
+
('BACKGROUND', (1, 0), (1, 0), GRAY_LIGHT),
|
| 585 |
+
('BOX', (0, 0), (-1, -1), 1, verdict_color),
|
| 586 |
+
('ROUNDEDCORNERS', [10, 10, 10, 10]),
|
| 587 |
+
('ALIGN', (0, 0), (-1, -1), 'CENTER'),
|
| 588 |
+
('VALIGN', (0, 0), (-1, -1), 'MIDDLE'),
|
| 589 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 15),
|
| 590 |
+
('TOPPADDING', (0, 0), (-1, -1), 15),
|
| 591 |
+
])
|
| 592 |
+
)
|
| 593 |
+
|
| 594 |
+
elements.append(verdict_box)
|
| 595 |
+
elements.append(Spacer(1, 0.3*inch))
|
| 596 |
+
|
| 597 |
+
# Content Analysis in a sleek table
|
| 598 |
+
elements.append(Paragraph("CONTENT ANALYSIS", section_style))
|
| 599 |
+
|
| 600 |
+
domain = analysis_data.get("domain", "general").title().replace('_', ' ')
|
| 601 |
+
domain_confidence = analysis_data.get("domain_confidence", 0)
|
| 602 |
+
text_length = analysis_data.get("text_length", 0)
|
| 603 |
+
sentence_count = analysis_data.get("sentence_count", 0)
|
| 604 |
+
total_time = performance_data.get("total_time", 0)
|
| 605 |
+
|
| 606 |
+
# Create two-column layout for content analysis
|
| 607 |
+
content_data = [[Paragraph("<b>Content Domain</b>", body_style), Paragraph(f"<font color='{colors.toHex(INFO_COLOR)}'><b>{domain}</b></font> ({domain_confidence:.1%} confidence)", body_style)],
|
| 608 |
+
[Paragraph("<b>Text Statistics</b>", body_style), Paragraph(f"{text_length:,} words | {sentence_count:,} sentences", body_style)],
|
| 609 |
+
[Paragraph("<b>Processing Time</b>", body_style), Paragraph(f"{total_time:.2f} seconds", body_style)],
|
| 610 |
+
[Paragraph("<b>Analysis Method</b>", body_style), Paragraph("Confidence-Weighted Ensemble Aggregation", body_style)],
|
| 611 |
+
]
|
| 612 |
+
|
| 613 |
+
content_table = Table(content_data, colWidths = [2*inch, 4*inch])
|
| 614 |
+
|
| 615 |
+
content_table.setStyle(TableStyle([('FONTNAME', (0, 0), (0, -1), 'Helvetica-Bold'),
|
| 616 |
+
('FONTNAME', (1, 0), (1, -1), 'Helvetica'),
|
| 617 |
+
('FONTSIZE', (0, 0), (-1, -1), 10),
|
| 618 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 6),
|
| 619 |
+
('TOPPADDING', (0, 0), (-1, -1), 6),
|
| 620 |
+
('GRID', (0, 0), (-1, -1), 0.25, GRAY_MEDIUM),
|
| 621 |
+
('BACKGROUND', (0, 0), (0, -1), GRAY_LIGHT),
|
| 622 |
])
|
| 623 |
)
|
| 624 |
+
|
| 625 |
elements.append(content_table)
|
| 626 |
+
elements.append(Spacer(1, 0.3*inch))
|
| 627 |
|
| 628 |
+
# Metric Weights Visualization
|
| 629 |
+
elements.append(Paragraph("METRIC CONTRIBUTIONS", section_style))
|
|
|
|
|
|
|
| 630 |
|
|
|
|
| 631 |
metric_contributions = ensemble_data.get("metric_contributions", {})
|
| 632 |
+
|
| 633 |
+
if metric_contributions and len(metric_contributions) > 0:
|
| 634 |
+
# Create horizontal bar chart effect with table
|
| 635 |
+
weight_data = [['METRIC', 'WEIGHT', '']]
|
| 636 |
+
|
| 637 |
+
for metric_name, contribution in metric_contributions.items():
|
| 638 |
+
weight = contribution.get("weight", 0)
|
| 639 |
+
display_name = metric_name.title().replace('_', ' ')
|
| 640 |
+
|
| 641 |
+
# Create visual bar representation
|
| 642 |
+
bar_width = int(weight * 100)
|
| 643 |
+
bar_cell = f"[{'β' * bar_width}{'β' * (100-bar_width)}] {weight:.1%}"
|
| 644 |
+
|
| 645 |
+
weight_data.append([display_name, f"{weight:.1%}", bar_cell])
|
| 646 |
|
| 647 |
+
weight_table = Table(weight_data, colWidths=[2*inch, 1*inch, 3*inch])
|
| 648 |
+
weight_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0), PRIMARY_COLOR),
|
| 649 |
('TEXTCOLOR', (0, 0), (-1, 0), colors.white),
|
| 650 |
('ALIGN', (0, 0), (-1, -1), 'LEFT'),
|
| 651 |
('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
|
| 652 |
('FONTSIZE', (0, 0), (-1, -1), 9),
|
| 653 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 6),
|
| 654 |
+
('TOPPADDING', (0, 0), (-1, -1), 6),
|
| 655 |
+
('GRID', (0, 0), (-1, -1), 0.5, GRAY_MEDIUM),
|
| 656 |
+
('TEXTCOLOR', (2, 1), (2, -1), PRIMARY_COLOR),
|
| 657 |
+
('FONTNAME', (2, 1), (2, -1), 'Courier'),
|
| 658 |
])
|
| 659 |
+
)
|
| 660 |
+
|
| 661 |
elements.append(weight_table)
|
| 662 |
+
elements.append(Spacer(1, 0.3*inch))
|
| 663 |
|
| 664 |
+
# Detailed Metric Analysis with colored cards
|
| 665 |
+
elements.append(Paragraph("DETAILED METRIC ANALYSIS", section_style))
|
| 666 |
+
|
| 667 |
+
if detailed_metrics:
|
| 668 |
+
for metric in detailed_metrics:
|
| 669 |
+
# Determine metric color based on verdict
|
| 670 |
+
if (metric.verdict == "HUMAN"):
|
| 671 |
+
metric_color = SUCCESS_COLOR
|
| 672 |
+
prob_color = SUCCESS_COLOR
|
| 673 |
+
|
| 674 |
+
elif( metric.verdict == "AI"):
|
| 675 |
+
metric_color = DANGER_COLOR
|
| 676 |
+
prob_color = DANGER_COLOR
|
| 677 |
+
|
| 678 |
+
else:
|
| 679 |
+
metric_color = WARNING_COLOR
|
| 680 |
+
prob_color = WARNING_COLOR
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 681 |
|
| 682 |
+
# Create metric card
|
| 683 |
+
metric_card_data = [[Paragraph(f"<font color='{colors.toHex(metric_color)}' size=12><b>{metric.name.upper().replace('_', ' ')}</b></font><br/>"
|
| 684 |
+
f"<font size=9>{metric.description}</font>",
|
| 685 |
+
ParagraphStyle('MetricTitle', alignment=TA_LEFT)),
|
| 686 |
+
|
| 687 |
+
Paragraph(f"<font size=11><b>VERDICT</b></font><br/>"
|
| 688 |
+
f"<font color='{colors.toHex(metric_color)}' size=12><b>{metric.verdict}</b></font>",
|
| 689 |
+
ParagraphStyle('MetricVerdict', alignment=TA_CENTER)),
|
| 690 |
+
|
| 691 |
+
Paragraph(f"<font size=11><b>AI PROBABILITY</b></font><br/>"
|
| 692 |
+
f"<font color='{colors.toHex(prob_color)}' size=12><b>{metric.ai_probability:.1f}%</b></font>",
|
| 693 |
+
ParagraphStyle('MetricProbability', alignment=TA_CENTER)),
|
| 694 |
+
|
| 695 |
+
Paragraph(f"<font size=11><b>WEIGHT</b></font><br/>"
|
| 696 |
+
f"<font size=12><b>{metric.weight:.1f}%</b></font>",
|
| 697 |
+
ParagraphStyle('MetricWeight', alignment=TA_CENTER)),
|
| 698 |
+
|
| 699 |
+
Paragraph(f"<font size=11><b>CONFIDENCE</b></font><br/>"
|
| 700 |
+
f"<font size=12><b>{metric.confidence:.1f}%</b></font>",
|
| 701 |
+
ParagraphStyle('MetricConfidence', alignment=TA_CENTER)),
|
| 702 |
+
]]
|
| 703 |
+
|
| 704 |
+
metric_table = Table(metric_card_data, colWidths = [2.5*inch, 1*inch, 1*inch, 0.8*inch, 0.8*inch])
|
| 705 |
+
|
| 706 |
+
metric_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0), GRAY_LIGHT),
|
| 707 |
+
('BOX', (0, 0), (-1, 0), 1, metric_color),
|
| 708 |
+
('LINEABOVE', (0, 0), (-1, 0), 2, metric_color),
|
| 709 |
+
('ALIGN', (0, 0), (-1, 0), 'CENTER'),
|
| 710 |
+
('VALIGN', (0, 0), (-1, 0), 'MIDDLE'),
|
| 711 |
+
('BOTTOMPADDING', (0, 0), (-1, 0), 10),
|
| 712 |
+
('TOPPADDING', (0, 0), (-1, 0), 10),
|
| 713 |
+
])
|
| 714 |
+
)
|
| 715 |
+
|
| 716 |
+
elements.append(metric_table)
|
| 717 |
+
|
| 718 |
+
# Add detailed sub-metrics if available
|
| 719 |
+
if metric.detailed_metrics:
|
| 720 |
+
elements.append(Spacer(1, 0.1*inch))
|
| 721 |
+
|
| 722 |
+
# Create a grid of sub-metrics
|
| 723 |
+
sub_items = list(metric.detailed_metrics.items())[:6]
|
| 724 |
+
sub_data = list()
|
| 725 |
+
|
| 726 |
+
for i in range(0, len(sub_items), 3):
|
| 727 |
+
row = list()
|
| 728 |
+
for j in range(3):
|
| 729 |
+
if (i + j < len(sub_items)):
|
| 730 |
+
sub_name, sub_value = sub_items[i + j]
|
| 731 |
+
|
| 732 |
+
# Format the value
|
| 733 |
+
if isinstance(sub_value, (int, float)):
|
| 734 |
+
if (sub_name.endswith('_score') or sub_name.endswith('_probability')):
|
| 735 |
+
formatted_value = f"{sub_value:.1f}%"
|
| 736 |
|
| 737 |
+
elif (sub_name.endswith('_ratio') or sub_name.endswith('_frequency')):
|
| 738 |
+
formatted_value = f"{sub_value:.3f}"
|
| 739 |
+
|
| 740 |
+
elif (sub_name.endswith('_entropy') or sub_name.endswith('_perplexity')):
|
| 741 |
+
formatted_value = f"{sub_value:.2f}"
|
| 742 |
+
|
| 743 |
+
else:
|
| 744 |
+
formatted_value = f"{sub_value:.2f}"
|
| 745 |
+
|
| 746 |
+
else:
|
| 747 |
+
formatted_value = str(sub_value)
|
| 748 |
+
|
| 749 |
+
row.append(f"<b>{sub_name.replace('_', ' ').title()}:</b> {formatted_value}")
|
| 750 |
+
|
| 751 |
+
else:
|
| 752 |
+
row.append("")
|
| 753 |
+
|
| 754 |
+
sub_data.append(row)
|
| 755 |
+
|
| 756 |
+
if sub_data:
|
| 757 |
+
sub_table = Table(sub_data, colWidths = [1.8*inch, 1.8*inch, 1.8*inch])
|
| 758 |
+
|
| 759 |
+
sub_table.setStyle(TableStyle([('FONTSIZE', (0, 0), (-1, -1), 8),
|
| 760 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 4),
|
| 761 |
+
('TOPPADDING', (0, 0), (-1, -1), 4),
|
| 762 |
+
('FONTNAME', (0, 0), (-1, -1), 'Helvetica'),
|
| 763 |
+
])
|
| 764 |
+
)
|
| 765 |
+
elements.append(sub_table)
|
| 766 |
+
|
| 767 |
+
elements.append(Spacer(1, 0.2*inch))
|
| 768 |
|
| 769 |
# Detection Reasoning
|
| 770 |
+
elements.append(Paragraph("DETECTION REASONING", section_style))
|
| 771 |
+
|
| 772 |
+
# Summary in a colored box
|
| 773 |
+
summary_box = Table([[Paragraph(f"<font size=11>{reasoning.summary}</font>", body_style)]], colWidths = [6.5*inch])
|
| 774 |
+
summary_box.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, -1), GRAY_LIGHT),
|
| 775 |
+
('BOX', (0, 0), (-1, -1), 1, PRIMARY_COLOR),
|
| 776 |
+
('PADDING', (0, 0), (-1, -1), 10),
|
| 777 |
+
])
|
| 778 |
+
)
|
| 779 |
+
|
| 780 |
+
elements.append(summary_box)
|
| 781 |
+
elements.append(Spacer(1, 0.2*inch))
|
| 782 |
|
| 783 |
# Key Indicators
|
| 784 |
+
if reasoning.key_indicators:
|
| 785 |
+
elements.append(Paragraph("KEY INDICATORS", subsection_style))
|
| 786 |
+
|
| 787 |
+
indicators_data = list()
|
| 788 |
+
|
| 789 |
+
for i in range(0, len(reasoning.key_indicators), 2):
|
| 790 |
+
row = list()
|
| 791 |
+
|
| 792 |
+
for j in range(2):
|
| 793 |
+
if (i + j < len(reasoning.key_indicators)):
|
| 794 |
+
indicator = reasoning.key_indicators[i + j]
|
| 795 |
+
# Add checkmark for positive indicators
|
| 796 |
+
if (indicator.startswith("β
") or indicator.startswith("β")):
|
| 797 |
+
icon_color = SUCCESS_COLOR
|
| 798 |
+
|
| 799 |
+
elif (indicator.startswith("β οΈ") or indicator.startswith("β")):
|
| 800 |
+
icon_color = WARNING_COLOR
|
| 801 |
+
|
| 802 |
+
else:
|
| 803 |
+
icon_color = PRIMARY_COLOR
|
| 804 |
+
|
| 805 |
+
row.append(Paragraph(f"<font color='{colors.toHex(icon_color)}'>β’</font> {indicator}", body_style))
|
| 806 |
+
|
| 807 |
+
else:
|
| 808 |
+
row.append("")
|
| 809 |
+
indicators_data.append(row)
|
| 810 |
+
|
| 811 |
+
indicators_table = Table(indicators_data, colWidths=[3*inch, 3*inch])
|
| 812 |
+
indicators_table.setStyle(TableStyle([('VALIGN', (0, 0), (-1, -1), 'TOP'),
|
| 813 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 4),
|
| 814 |
+
])
|
| 815 |
+
)
|
| 816 |
+
|
| 817 |
+
elements.append(indicators_table)
|
| 818 |
+
elements.append(Spacer(1, 0.2*inch))
|
| 819 |
|
| 820 |
+
# Page break for attribution section
|
| 821 |
+
elements.append(PageBreak())
|
|
|
|
| 822 |
|
| 823 |
# Model Attribution Section
|
| 824 |
if attribution_result:
|
| 825 |
+
elements.append(Paragraph("AI MODEL ATTRIBUTION", section_style))
|
| 826 |
+
|
| 827 |
+
predicted_model = attribution_result.predicted_model.value.replace("_", " ").title()
|
| 828 |
+
attribution_confidence = attribution_result.confidence * 100
|
| 829 |
|
| 830 |
+
attribution_card_data = [[Paragraph("<b>PREDICTED MODEL</b>", subsection_style), Paragraph(f"<font size=14 color='{colors.toHex(INFO_COLOR)}'><b>{predicted_model}</b></font>", subsection_style)],
|
| 831 |
+
[Paragraph("<b>ATTRIBUTION CONFIDENCE</b>", subsection_style), Paragraph(f"<font size=14><b>{attribution_confidence:.1f}%</b></font>", subsection_style)],
|
| 832 |
+
[Paragraph("<b>DOMAIN USED</b>", subsection_style), Paragraph(f"<b>{attribution_result.domain_used.value.title()}</b>", subsection_style)],
|
| 833 |
+
]
|
| 834 |
|
| 835 |
+
attribution_table = Table(attribution_card_data, colWidths = [2.5*inch, 3.5*inch])
|
|
|
|
|
|
|
|
|
|
| 836 |
|
| 837 |
+
attribution_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (0, -1), GRAY_LIGHT),
|
|
|
|
| 838 |
('FONTNAME', (0, 0), (0, -1), 'Helvetica-Bold'),
|
| 839 |
+
('FONTSIZE', (0, 0), (-1, -1), 11),
|
| 840 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 8),
|
| 841 |
+
('TOPPADDING', (0, 0), (-1, -1), 8),
|
| 842 |
+
('GRID', (0, 0), (-1, -1), 0.5, GRAY_MEDIUM),
|
| 843 |
])
|
| 844 |
)
|
| 845 |
+
|
| 846 |
elements.append(attribution_table)
|
| 847 |
+
elements.append(Spacer(1, 0.3*inch))
|
| 848 |
|
| 849 |
# Model probabilities table
|
| 850 |
if attribution_result.model_probabilities:
|
| 851 |
+
elements.append(Paragraph("MODEL PROBABILITY DISTRIBUTION", subsection_style))
|
| 852 |
|
| 853 |
+
prob_data = [['MODEL', 'PROBABILITY', '']]
|
| 854 |
|
| 855 |
+
# Show top 8 models
|
| 856 |
+
sorted_models = sorted(attribution_result.model_probabilities.items(), key = lambda x: x[1], reverse=True)[:8]
|
|
|
|
|
|
|
| 857 |
|
| 858 |
for model_name, probability in sorted_models:
|
| 859 |
display_name = model_name.replace("_", " ").replace("-", " ").title()
|
| 860 |
+
bar_width = int(probability * 100)
|
| 861 |
+
|
| 862 |
+
prob_data.append([display_name,
|
| 863 |
+
f"{probability:.1%}",
|
| 864 |
+
f"[{'β' * bar_width}{'β' * (100-bar_width)}]"
|
| 865 |
+
])
|
| 866 |
+
|
| 867 |
+
prob_table = Table(prob_data, colWidths = [2.5*inch, 1*inch, 2.5*inch])
|
| 868 |
+
|
| 869 |
+
prob_table.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, 0), INFO_COLOR),
|
| 870 |
('TEXTCOLOR', (0, 0), (-1, 0), colors.white),
|
| 871 |
('ALIGN', (0, 0), (-1, -1), 'LEFT'),
|
| 872 |
+
('ALIGN', (1, 1), (1, -1), 'RIGHT'),
|
| 873 |
('FONTNAME', (0, 0), (-1, 0), 'Helvetica-Bold'),
|
| 874 |
('FONTSIZE', (0, 0), (-1, -1), 9),
|
| 875 |
+
('BOTTOMPADDING', (0, 0), (-1, -1), 6),
|
| 876 |
+
('TOPPADDING', (0, 0), (-1, -1), 6),
|
| 877 |
+
('GRID', (0, 0), (-1, -1), 0.5, GRAY_MEDIUM),
|
| 878 |
+
('FONTNAME', (2, 1), (2, -1), 'Courier'),
|
| 879 |
+
('TEXTCOLOR', (2, 1), (2, -1), INFO_COLOR),
|
| 880 |
])
|
| 881 |
)
|
| 882 |
+
|
| 883 |
elements.append(prob_table)
|
| 884 |
+
elements.append(Spacer(1, 0.3*inch))
|
| 885 |
+
|
| 886 |
+
# Recommendations in colored boxes
|
| 887 |
+
if reasoning.recommendations:
|
| 888 |
+
elements.append(Paragraph("RECOMMENDATIONS", section_style))
|
| 889 |
|
| 890 |
+
for i, recommendation in enumerate(reasoning.recommendations):
|
| 891 |
+
# Alternate colors for visual interest
|
| 892 |
+
if (i % 3 == 0):
|
| 893 |
+
rec_color = SUCCESS_COLOR
|
| 894 |
+
|
| 895 |
+
elif (i % 3 == 1):
|
| 896 |
+
rec_color = INFO_COLOR
|
| 897 |
+
|
| 898 |
+
else:
|
| 899 |
+
rec_color = WARNING_COLOR
|
| 900 |
+
|
| 901 |
+
rec_box = Table([[Paragraph(f"<font color='{colors.toHex(rec_color)}'>β</font> {recommendation}", body_style)]], colWidths=[6.5*inch])
|
| 902 |
+
|
| 903 |
+
rec_box.setStyle(TableStyle([('BACKGROUND', (0, 0), (-1, -1), GRAY_LIGHT),
|
| 904 |
+
('BOX', (0, 0), (-1, -1), 1, rec_color),
|
| 905 |
+
('PADDING', (0, 0), (-1, -1), 8),
|
| 906 |
+
('BOTTOMMARGIN', (0, 0), (-1, -1), 5),
|
| 907 |
+
])
|
| 908 |
+
)
|
| 909 |
+
|
| 910 |
+
elements.append(rec_box)
|
| 911 |
+
elements.append(Spacer(1, 0.1*inch))
|
| 912 |
|
| 913 |
+
# Footer with watermark
|
| 914 |
+
footer_style = ParagraphStyle('FooterStyle',
|
| 915 |
+
parent = styles['Normal'],
|
| 916 |
+
fontName = 'Helvetica',
|
| 917 |
+
fontSize = 9,
|
| 918 |
+
textColor = GRAY_DARK,
|
| 919 |
+
alignment = TA_CENTER,
|
| 920 |
+
)
|
| 921 |
|
| 922 |
+
elements.append(Spacer(1, 0.5*inch))
|
| 923 |
+
elements.append(HRFlowable(width="100%", thickness=0.5, color=GRAY_MEDIUM, spaceAfter=10))
|
| 924 |
+
|
| 925 |
+
footer_text = (f"Generated by AI Text Detector v2.0 | "
|
| 926 |
+
f"Processing Time: {total_time:.2f}s | "
|
| 927 |
+
f"Report ID: {filename.replace('.pdf', '')}")
|
| 928 |
+
|
| 929 |
+
elements.append(Paragraph(footer_text, footer_style))
|
| 930 |
+
elements.append(Paragraph("Confidential Analysis Report β’ Β© 2025 AI Detection Analytics",
|
| 931 |
+
ParagraphStyle('Copyright', parent=footer_style, fontSize=8, textColor=GRAY_MEDIUM)))
|
| 932 |
|
| 933 |
# Build PDF
|
| 934 |
doc.build(elements)
|
| 935 |
|
| 936 |
+
logger.info(f"Premium PDF report saved: {output_path}")
|
| 937 |
return output_path
|
| 938 |
|
| 939 |
|