Spaces:
Running
Running
download results
Browse files- app.py +39 -8
- validation.py +52 -1
app.py
CHANGED
|
@@ -7,30 +7,34 @@ import requests
|
|
| 7 |
|
| 8 |
def process_file(file):
|
| 9 |
results = []
|
|
|
|
| 10 |
|
| 11 |
# Check 1: JSON validation
|
| 12 |
json_valid, json_message, json_data = validate_json(file.name)
|
| 13 |
results.append(("JSON Format Validation", json_valid, json_message))
|
| 14 |
|
| 15 |
if not json_valid:
|
| 16 |
-
return results
|
| 17 |
|
| 18 |
# Check 2: Croissant validation
|
| 19 |
croissant_valid, croissant_message = validate_croissant(json_data)
|
| 20 |
results.append(("Croissant Schema Validation", croissant_valid, croissant_message))
|
| 21 |
|
| 22 |
if not croissant_valid:
|
| 23 |
-
return results
|
| 24 |
|
| 25 |
# Check 3: Records validation
|
| 26 |
records_valid, records_message = validate_records(json_data)
|
| 27 |
results.append(("Records Generation Test", records_valid, records_message))
|
| 28 |
|
| 29 |
-
|
|
|
|
|
|
|
|
|
|
| 30 |
|
| 31 |
def create_ui():
|
| 32 |
with gr.Blocks(theme=gr.themes.Soft()) as app:
|
| 33 |
-
gr.Markdown("# ππ₯ Croissant
|
| 34 |
gr.Markdown("""
|
| 35 |
Upload your Croissant JSON-LD file or enter a URL to validate if it meets the requirements for NeurIPS submission.
|
| 36 |
The validator will check:
|
|
@@ -66,6 +70,18 @@ def create_ui():
|
|
| 66 |
with gr.Group():
|
| 67 |
# Validation results
|
| 68 |
validation_results = gr.HTML(visible=False)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 69 |
|
| 70 |
# Define CSS for the validation UI
|
| 71 |
gr.HTML("""
|
|
@@ -256,16 +272,31 @@ def create_ui():
|
|
| 256 |
|
| 257 |
def on_validate(file):
|
| 258 |
if file is None:
|
| 259 |
-
return gr.update(visible=False)
|
| 260 |
|
| 261 |
# Process the file and get results
|
| 262 |
-
results = process_file(file)
|
| 263 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 264 |
|
| 265 |
# Connect UI events to functions
|
| 266 |
tabs.select(on_tab_change, None, [active_tab, upload_progress, validation_results])
|
| 267 |
file_input.change(on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results])
|
| 268 |
-
validate_btn.click(
|
|
|
|
|
|
|
|
|
|
|
|
|
| 269 |
fetch_btn.click(fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results])
|
| 270 |
|
| 271 |
# Footer
|
|
|
|
| 7 |
|
| 8 |
def process_file(file):
|
| 9 |
results = []
|
| 10 |
+
json_data = None
|
| 11 |
|
| 12 |
# Check 1: JSON validation
|
| 13 |
json_valid, json_message, json_data = validate_json(file.name)
|
| 14 |
results.append(("JSON Format Validation", json_valid, json_message))
|
| 15 |
|
| 16 |
if not json_valid:
|
| 17 |
+
return results, None
|
| 18 |
|
| 19 |
# Check 2: Croissant validation
|
| 20 |
croissant_valid, croissant_message = validate_croissant(json_data)
|
| 21 |
results.append(("Croissant Schema Validation", croissant_valid, croissant_message))
|
| 22 |
|
| 23 |
if not croissant_valid:
|
| 24 |
+
return results, None
|
| 25 |
|
| 26 |
# Check 3: Records validation
|
| 27 |
records_valid, records_message = validate_records(json_data)
|
| 28 |
results.append(("Records Generation Test", records_valid, records_message))
|
| 29 |
|
| 30 |
+
# Generate detailed report
|
| 31 |
+
report = generate_validation_report(file.name, json_data, results)
|
| 32 |
+
|
| 33 |
+
return results, report
|
| 34 |
|
| 35 |
def create_ui():
|
| 36 |
with gr.Blocks(theme=gr.themes.Soft()) as app:
|
| 37 |
+
gr.Markdown("# ππ₯ Croissant Validator for NeurIPS D&B")
|
| 38 |
gr.Markdown("""
|
| 39 |
Upload your Croissant JSON-LD file or enter a URL to validate if it meets the requirements for NeurIPS submission.
|
| 40 |
The validator will check:
|
|
|
|
| 70 |
with gr.Group():
|
| 71 |
# Validation results
|
| 72 |
validation_results = gr.HTML(visible=False)
|
| 73 |
+
with gr.Row():
|
| 74 |
+
report_text = gr.Textbox(
|
| 75 |
+
label="Detailed Report",
|
| 76 |
+
visible=False,
|
| 77 |
+
show_copy_button=True,
|
| 78 |
+
lines=10
|
| 79 |
+
)
|
| 80 |
+
report_md = gr.File(
|
| 81 |
+
label="Download Report",
|
| 82 |
+
visible=False,
|
| 83 |
+
file_types=[".md"]
|
| 84 |
+
)
|
| 85 |
|
| 86 |
# Define CSS for the validation UI
|
| 87 |
gr.HTML("""
|
|
|
|
| 272 |
|
| 273 |
def on_validate(file):
|
| 274 |
if file is None:
|
| 275 |
+
return [gr.update(visible=False)] * 3
|
| 276 |
|
| 277 |
# Process the file and get results
|
| 278 |
+
results, report = process_file(file)
|
| 279 |
+
|
| 280 |
+
# Save report to file
|
| 281 |
+
if report:
|
| 282 |
+
report_path = "validation_report.md"
|
| 283 |
+
with open(report_path, "w") as f:
|
| 284 |
+
f.write(report)
|
| 285 |
+
|
| 286 |
+
return [
|
| 287 |
+
build_results_html(results),
|
| 288 |
+
gr.update(visible=True, value=report) if report else gr.update(visible=False),
|
| 289 |
+
gr.update(visible=True, value="validation_report.md") if report else gr.update(visible=False)
|
| 290 |
+
]
|
| 291 |
|
| 292 |
# Connect UI events to functions
|
| 293 |
tabs.select(on_tab_change, None, [active_tab, upload_progress, validation_results])
|
| 294 |
file_input.change(on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results])
|
| 295 |
+
validate_btn.click(
|
| 296 |
+
on_validate,
|
| 297 |
+
inputs=file_input,
|
| 298 |
+
outputs=[validation_results, report_text, report_md]
|
| 299 |
+
)
|
| 300 |
fetch_btn.click(fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results])
|
| 301 |
|
| 302 |
# Footer
|
validation.py
CHANGED
|
@@ -61,4 +61,55 @@ def validate_records(json_data):
|
|
| 61 |
except Exception as e:
|
| 62 |
error_details = traceback.format_exc()
|
| 63 |
error_message = f"β Unexpected error during records validation: {str(e)}\n\n{error_details}"
|
| 64 |
-
return False, error_message
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 61 |
except Exception as e:
|
| 62 |
error_details = traceback.format_exc()
|
| 63 |
error_message = f"β Unexpected error during records validation: {str(e)}\n\n{error_details}"
|
| 64 |
+
return False, error_message
|
| 65 |
+
|
| 66 |
+
def generate_validation_report(file_path, json_data, validation_results):
|
| 67 |
+
"""Generate a detailed markdown report of the validation results."""
|
| 68 |
+
report = []
|
| 69 |
+
|
| 70 |
+
# Header
|
| 71 |
+
report.append("# CROISSANT VALIDATION REPORT")
|
| 72 |
+
report.append("=" * 80)
|
| 73 |
+
report.append("")
|
| 74 |
+
|
| 75 |
+
# Validation Results Section
|
| 76 |
+
report.append("## VALIDATION RESULTS")
|
| 77 |
+
report.append("-" * 80)
|
| 78 |
+
report.append(f"Starting validation for file: {file_path}")
|
| 79 |
+
|
| 80 |
+
# Process each validation step
|
| 81 |
+
for step_name, passed, message in validation_results:
|
| 82 |
+
report.append("")
|
| 83 |
+
report.append(f"### {step_name}")
|
| 84 |
+
report.append("β" if passed else "β")
|
| 85 |
+
report.append(message.replace("β
", "β").replace("β", "β"))
|
| 86 |
+
|
| 87 |
+
# Add extra details for record sets if available
|
| 88 |
+
if step_name == "Records Generation Test" and json_data:
|
| 89 |
+
try:
|
| 90 |
+
dataset = mlc.Dataset(jsonld=json_data)
|
| 91 |
+
for record_set in dataset.metadata.record_sets:
|
| 92 |
+
report.append("")
|
| 93 |
+
report.append(f"#### Record Set: {record_set.uuid}")
|
| 94 |
+
report.append(f"Description: {record_set.description}")
|
| 95 |
+
if record_set.data_types:
|
| 96 |
+
report.append(f"Data Types: {record_set.data_types}")
|
| 97 |
+
report.append("")
|
| 98 |
+
report.append("Fields:")
|
| 99 |
+
for field in record_set.fields:
|
| 100 |
+
report.append(f"- {field.name} ({field.data_type})")
|
| 101 |
+
if field.description:
|
| 102 |
+
report.append(f" Description: {field.description}")
|
| 103 |
+
except Exception as e:
|
| 104 |
+
report.append(f"Error getting record set details: {str(e)}")
|
| 105 |
+
|
| 106 |
+
# JSON-LD Reference Section
|
| 107 |
+
report.append("")
|
| 108 |
+
report.append("## JSON-LD REFERENCE")
|
| 109 |
+
report.append("=" * 80)
|
| 110 |
+
report.append("")
|
| 111 |
+
report.append("```json")
|
| 112 |
+
report.append(json.dumps(json_data, indent=2))
|
| 113 |
+
report.append("```")
|
| 114 |
+
|
| 115 |
+
return "\n".join(report)
|