Spaces:
Running
Running
Update src/aibom_generator/utils.py
Browse files- src/aibom_generator/utils.py +84 -166
src/aibom_generator/utils.py
CHANGED
@@ -1,21 +1,17 @@
|
|
1 |
"""
|
2 |
Utility functions for the AIBOM Generator.
|
3 |
"""
|
|
|
4 |
import json
|
5 |
import logging
|
6 |
import os
|
7 |
import re
|
8 |
import uuid
|
9 |
-
import requests
|
10 |
-
import jsonschema
|
11 |
from typing import Dict, List, Optional, Any, Union, Tuple
|
12 |
from enum import Enum
|
13 |
|
14 |
logger = logging.getLogger(__name__)
|
15 |
|
16 |
-
# CycloneDX schema URL for version 1.6
|
17 |
-
CYCLONEDX_SCHEMA_URL = "https://raw.githubusercontent.com/CycloneDX/specification/master/schema/bom-1.6.schema.json"
|
18 |
-
|
19 |
# Validation severity levels
|
20 |
class ValidationSeverity(Enum):
|
21 |
ERROR = "error"
|
@@ -107,73 +103,6 @@ def validate_spdx(license_entry):
|
|
107 |
return license_entry in spdx_licenses
|
108 |
|
109 |
|
110 |
-
def _load_cyclonedx_schema():
|
111 |
-
"""
|
112 |
-
Load the CycloneDX JSON schema from the specified URL.
|
113 |
-
|
114 |
-
Returns:
|
115 |
-
dict: The loaded schema or a minimal schema if loading fails
|
116 |
-
"""
|
117 |
-
try:
|
118 |
-
response = requests.get(CYCLONEDX_SCHEMA_URL)
|
119 |
-
response.raise_for_status()
|
120 |
-
schema = response.json()
|
121 |
-
logger.info(f"Successfully loaded CycloneDX schema from {CYCLONEDX_SCHEMA_URL}")
|
122 |
-
return schema
|
123 |
-
except Exception as e:
|
124 |
-
logger.error(f"Failed to load CycloneDX schema: {e}")
|
125 |
-
# Fallback to a minimal schema validation
|
126 |
-
return {
|
127 |
-
"type": "object",
|
128 |
-
"required": ["bomFormat", "specVersion", "serialNumber", "version"]
|
129 |
-
}
|
130 |
-
|
131 |
-
|
132 |
-
def _validate_schema(aibom: Dict[str, Any], schema: Dict[str, Any]) -> List[Dict[str, Any]]:
|
133 |
-
"""
|
134 |
-
Validate an AIBOM against the CycloneDX JSON schema.
|
135 |
-
|
136 |
-
Args:
|
137 |
-
aibom: The AIBOM to validate
|
138 |
-
schema: The CycloneDX schema
|
139 |
-
|
140 |
-
Returns:
|
141 |
-
List of validation issues
|
142 |
-
"""
|
143 |
-
issues = []
|
144 |
-
|
145 |
-
if not schema:
|
146 |
-
issues.append({
|
147 |
-
"severity": ValidationSeverity.WARNING.value,
|
148 |
-
"code": "SCHEMA_UNAVAILABLE",
|
149 |
-
"message": "CycloneDX schema unavailable, performing minimal validation",
|
150 |
-
"path": "$"
|
151 |
-
})
|
152 |
-
return issues
|
153 |
-
|
154 |
-
try:
|
155 |
-
jsonschema.validate(instance=aibom, schema=schema)
|
156 |
-
except jsonschema.exceptions.ValidationError as e:
|
157 |
-
# Extract the JSON path where the validation error occurred
|
158 |
-
path = ".".join(str(p) for p in e.path) if e.path else "$"
|
159 |
-
|
160 |
-
issues.append({
|
161 |
-
"severity": ValidationSeverity.ERROR.value,
|
162 |
-
"code": "SCHEMA_VALIDATION_ERROR",
|
163 |
-
"message": str(e),
|
164 |
-
"path": path
|
165 |
-
})
|
166 |
-
except Exception as e:
|
167 |
-
issues.append({
|
168 |
-
"severity": ValidationSeverity.ERROR.value,
|
169 |
-
"code": "SCHEMA_VALIDATION_EXCEPTION",
|
170 |
-
"message": f"Unexpected error during schema validation: {str(e)}",
|
171 |
-
"path": "$"
|
172 |
-
})
|
173 |
-
|
174 |
-
return issues
|
175 |
-
|
176 |
-
|
177 |
def _validate_ai_requirements(aibom: Dict[str, Any]) -> List[Dict[str, Any]]:
|
178 |
"""
|
179 |
Validate AI-specific requirements for an AIBOM.
|
@@ -379,13 +308,94 @@ def _generate_validation_recommendations(issues: List[Dict[str, Any]]) -> List[s
|
|
379 |
return recommendations
|
380 |
|
381 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
382 |
def calculate_completeness_score(aibom: Dict[str, Any], validate: bool = True) -> Dict[str, Any]:
|
383 |
"""
|
384 |
-
Calculate completeness score for an AIBOM and optionally validate against
|
385 |
|
386 |
Args:
|
387 |
aibom: The AIBOM to score and validate
|
388 |
-
validate: Whether to perform validation
|
389 |
|
390 |
Returns:
|
391 |
Dictionary containing score and validation results
|
@@ -517,98 +527,6 @@ def calculate_completeness_score(aibom: Dict[str, Any], validate: bool = True) -
|
|
517 |
return result
|
518 |
|
519 |
|
520 |
-
def validate_aibom(aibom: Dict[str, Any]) -> Dict[str, Any]:
|
521 |
-
"""
|
522 |
-
Validate an AIBOM against the CycloneDX schema and AI-specific requirements.
|
523 |
-
|
524 |
-
Args:
|
525 |
-
aibom: The AIBOM to validate
|
526 |
-
|
527 |
-
Returns:
|
528 |
-
Validation report with issues and recommendations
|
529 |
-
"""
|
530 |
-
# Initialize validation report
|
531 |
-
report = {
|
532 |
-
"valid": True,
|
533 |
-
"schema_valid": True,
|
534 |
-
"ai_valid": True,
|
535 |
-
"issues": [],
|
536 |
-
"recommendations": [],
|
537 |
-
"summary": {
|
538 |
-
"error_count": 0,
|
539 |
-
"warning_count": 0,
|
540 |
-
"info_count": 0
|
541 |
-
}
|
542 |
-
}
|
543 |
-
|
544 |
-
# Load schema
|
545 |
-
schema = _load_cyclonedx_schema()
|
546 |
-
|
547 |
-
# Validate against CycloneDX schema
|
548 |
-
schema_issues = _validate_schema(aibom, schema)
|
549 |
-
if schema_issues:
|
550 |
-
report["schema_valid"] = False
|
551 |
-
report["valid"] = False
|
552 |
-
report["issues"].extend(schema_issues)
|
553 |
-
|
554 |
-
# Validate AI-specific requirements
|
555 |
-
ai_issues = _validate_ai_requirements(aibom)
|
556 |
-
if ai_issues:
|
557 |
-
report["ai_valid"] = False
|
558 |
-
report["valid"] = False
|
559 |
-
report["issues"].extend(ai_issues)
|
560 |
-
|
561 |
-
# Generate recommendations
|
562 |
-
report["recommendations"] = _generate_validation_recommendations(report["issues"])
|
563 |
-
|
564 |
-
# Update summary counts
|
565 |
-
for issue in report["issues"]:
|
566 |
-
if issue["severity"] == ValidationSeverity.ERROR.value:
|
567 |
-
report["summary"]["error_count"] += 1
|
568 |
-
elif issue["severity"] == ValidationSeverity.WARNING.value:
|
569 |
-
report["summary"]["warning_count"] += 1
|
570 |
-
elif issue["severity"] == ValidationSeverity.INFO.value:
|
571 |
-
report["summary"]["info_count"] += 1
|
572 |
-
|
573 |
-
return report
|
574 |
-
|
575 |
-
|
576 |
-
def get_validation_summary(report: Dict[str, Any]) -> str:
|
577 |
-
"""
|
578 |
-
Get a human-readable summary of the validation report.
|
579 |
-
|
580 |
-
Args:
|
581 |
-
report: Validation report
|
582 |
-
|
583 |
-
Returns:
|
584 |
-
Human-readable summary
|
585 |
-
"""
|
586 |
-
if report["valid"]:
|
587 |
-
summary = "✅ AIBOM is valid and complies with CycloneDX schema and AI requirements.\n"
|
588 |
-
else:
|
589 |
-
summary = "❌ AIBOM validation failed.\n"
|
590 |
-
|
591 |
-
summary += f"\nSummary:\n"
|
592 |
-
summary += f"- Errors: {report['summary']['error_count']}\n"
|
593 |
-
summary += f"- Warnings: {report['summary']['warning_count']}\n"
|
594 |
-
summary += f"- Info: {report['summary']['info_count']}\n"
|
595 |
-
|
596 |
-
if not report["valid"]:
|
597 |
-
summary += "\nIssues:\n"
|
598 |
-
for issue in report["issues"]:
|
599 |
-
severity = issue["severity"].upper()
|
600 |
-
code = issue["code"]
|
601 |
-
message = issue["message"]
|
602 |
-
path = issue["path"]
|
603 |
-
summary += f"- [{severity}] {code}: {message} (at {path})\n"
|
604 |
-
|
605 |
-
summary += "\nRecommendations:\n"
|
606 |
-
for i, recommendation in enumerate(report["recommendations"], 1):
|
607 |
-
summary += f"{i}. {recommendation}\n"
|
608 |
-
|
609 |
-
return summary
|
610 |
-
|
611 |
-
|
612 |
def merge_metadata(primary: Dict[str, Any], secondary: Dict[str, Any]) -> Dict[str, Any]:
|
613 |
result = secondary.copy()
|
614 |
for key, value in primary.items():
|
|
|
1 |
"""
|
2 |
Utility functions for the AIBOM Generator.
|
3 |
"""
|
4 |
+
|
5 |
import json
|
6 |
import logging
|
7 |
import os
|
8 |
import re
|
9 |
import uuid
|
|
|
|
|
10 |
from typing import Dict, List, Optional, Any, Union, Tuple
|
11 |
from enum import Enum
|
12 |
|
13 |
logger = logging.getLogger(__name__)
|
14 |
|
|
|
|
|
|
|
15 |
# Validation severity levels
|
16 |
class ValidationSeverity(Enum):
|
17 |
ERROR = "error"
|
|
|
103 |
return license_entry in spdx_licenses
|
104 |
|
105 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
106 |
def _validate_ai_requirements(aibom: Dict[str, Any]) -> List[Dict[str, Any]]:
|
107 |
"""
|
108 |
Validate AI-specific requirements for an AIBOM.
|
|
|
308 |
return recommendations
|
309 |
|
310 |
|
311 |
+
def validate_aibom(aibom: Dict[str, Any]) -> Dict[str, Any]:
|
312 |
+
"""
|
313 |
+
Validate an AIBOM against AI-specific requirements.
|
314 |
+
|
315 |
+
Args:
|
316 |
+
aibom: The AIBOM to validate
|
317 |
+
|
318 |
+
Returns:
|
319 |
+
Validation report with issues and recommendations
|
320 |
+
"""
|
321 |
+
# Initialize validation report
|
322 |
+
report = {
|
323 |
+
"valid": True,
|
324 |
+
"ai_valid": True,
|
325 |
+
"issues": [],
|
326 |
+
"recommendations": [],
|
327 |
+
"summary": {
|
328 |
+
"error_count": 0,
|
329 |
+
"warning_count": 0,
|
330 |
+
"info_count": 0
|
331 |
+
}
|
332 |
+
}
|
333 |
+
|
334 |
+
# Validate AI-specific requirements
|
335 |
+
ai_issues = _validate_ai_requirements(aibom)
|
336 |
+
if ai_issues:
|
337 |
+
report["ai_valid"] = False
|
338 |
+
report["valid"] = False
|
339 |
+
report["issues"].extend(ai_issues)
|
340 |
+
|
341 |
+
# Generate recommendations
|
342 |
+
report["recommendations"] = _generate_validation_recommendations(report["issues"])
|
343 |
+
|
344 |
+
# Update summary counts
|
345 |
+
for issue in report["issues"]:
|
346 |
+
if issue["severity"] == ValidationSeverity.ERROR.value:
|
347 |
+
report["summary"]["error_count"] += 1
|
348 |
+
elif issue["severity"] == ValidationSeverity.WARNING.value:
|
349 |
+
report["summary"]["warning_count"] += 1
|
350 |
+
elif issue["severity"] == ValidationSeverity.INFO.value:
|
351 |
+
report["summary"]["info_count"] += 1
|
352 |
+
|
353 |
+
return report
|
354 |
+
|
355 |
+
|
356 |
+
def get_validation_summary(report: Dict[str, Any]) -> str:
|
357 |
+
"""
|
358 |
+
Get a human-readable summary of the validation report.
|
359 |
+
|
360 |
+
Args:
|
361 |
+
report: Validation report
|
362 |
+
|
363 |
+
Returns:
|
364 |
+
Human-readable summary
|
365 |
+
"""
|
366 |
+
if report["valid"]:
|
367 |
+
summary = "✅ AIBOM is valid and complies with AI requirements.\n"
|
368 |
+
else:
|
369 |
+
summary = "❌ AIBOM validation failed.\n"
|
370 |
+
|
371 |
+
summary += f"\nSummary:\n"
|
372 |
+
summary += f"- Errors: {report['summary']['error_count']}\n"
|
373 |
+
summary += f"- Warnings: {report['summary']['warning_count']}\n"
|
374 |
+
summary += f"- Info: {report['summary']['info_count']}\n"
|
375 |
+
|
376 |
+
if not report["valid"]:
|
377 |
+
summary += "\nIssues:\n"
|
378 |
+
for issue in report["issues"]:
|
379 |
+
severity = issue["severity"].upper()
|
380 |
+
code = issue["code"]
|
381 |
+
message = issue["message"]
|
382 |
+
path = issue["path"]
|
383 |
+
summary += f"- [{severity}] {code}: {message} (at {path})\n"
|
384 |
+
|
385 |
+
summary += "\nRecommendations:\n"
|
386 |
+
for i, recommendation in enumerate(report["recommendations"], 1):
|
387 |
+
summary += f"{i}. {recommendation}\n"
|
388 |
+
|
389 |
+
return summary
|
390 |
+
|
391 |
+
|
392 |
def calculate_completeness_score(aibom: Dict[str, Any], validate: bool = True) -> Dict[str, Any]:
|
393 |
"""
|
394 |
+
Calculate completeness score for an AIBOM and optionally validate against AI requirements.
|
395 |
|
396 |
Args:
|
397 |
aibom: The AIBOM to score and validate
|
398 |
+
validate: Whether to perform validation
|
399 |
|
400 |
Returns:
|
401 |
Dictionary containing score and validation results
|
|
|
527 |
return result
|
528 |
|
529 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
530 |
def merge_metadata(primary: Dict[str, Any], secondary: Dict[str, Any]) -> Dict[str, Any]:
|
531 |
result = secondary.copy()
|
532 |
for key, value in primary.items():
|