Spaces:
Sleeping
Sleeping
url ingest
Browse files- app.py +82 -18
- requirements.txt +2 -1
app.py
CHANGED
|
@@ -3,6 +3,7 @@ import json
|
|
| 3 |
import time
|
| 4 |
import traceback
|
| 5 |
from validation import validate_json, validate_croissant, validate_records
|
|
|
|
| 6 |
|
| 7 |
def process_file(file):
|
| 8 |
results = []
|
|
@@ -31,24 +32,32 @@ def create_ui():
|
|
| 31 |
with gr.Blocks(theme=gr.themes.Soft()) as app:
|
| 32 |
gr.Markdown("# Croissant JSON-LD Validator for NeurIPS")
|
| 33 |
gr.Markdown("""
|
| 34 |
-
Upload your Croissant JSON-LD file to validate if it meets the requirements for NeurIPS submission.
|
| 35 |
The validator will check:
|
| 36 |
1. If the file is valid JSON
|
| 37 |
2. If it passes Croissant schema validation
|
| 38 |
3. If records can be generated within a reasonable time
|
| 39 |
""")
|
| 40 |
|
| 41 |
-
with gr.
|
| 42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 43 |
|
| 44 |
upload_progress = gr.HTML(
|
| 45 |
"""<div class="progress-container">
|
| 46 |
-
<div class="progress-status">Ready for
|
| 47 |
</div>""", visible=True)
|
| 48 |
|
| 49 |
-
validate_btn = gr.Button("Validate", variant="primary")
|
| 50 |
|
| 51 |
-
# Create containers for
|
| 52 |
validation_results = gr.HTML(visible=False)
|
| 53 |
|
| 54 |
# Define CSS for the validation UI
|
|
@@ -119,22 +128,68 @@ def create_ui():
|
|
| 119 |
<div class="progress-status">β
File uploaded successfully</div>
|
| 120 |
</div>""", gr.update(visible=False)
|
| 121 |
|
| 122 |
-
def
|
| 123 |
-
if
|
| 124 |
-
return
|
| 125 |
-
|
| 126 |
-
|
| 127 |
-
results = process_file(file)
|
| 128 |
|
| 129 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
html = '<div class="validation-results">'
|
| 131 |
|
| 132 |
for i, (test_name, passed, message) in enumerate(results):
|
| 133 |
-
# Determine status class
|
| 134 |
status_class = "status-success" if passed else "status-error"
|
| 135 |
status_icon = "β" if passed else "β"
|
| 136 |
|
| 137 |
-
# Use direct inline JavaScript manipulation instead of a function call
|
| 138 |
html += f'''
|
| 139 |
<div class="validation-step" id="step-{i}">
|
| 140 |
<div class="step-header" onclick="
|
|
@@ -160,19 +215,28 @@ def create_ui():
|
|
| 160 |
'''
|
| 161 |
|
| 162 |
html += '</div>'
|
| 163 |
-
|
| 164 |
-
# No separate JavaScript function needed
|
| 165 |
return gr.update(value=html, visible=True)
|
| 166 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 167 |
file_input.change(on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results])
|
| 168 |
validate_btn.click(on_validate, inputs=file_input, outputs=validation_results)
|
|
|
|
| 169 |
|
|
|
|
| 170 |
gr.HTML("""
|
| 171 |
<div style="text-align: center; margin-top: 20px;">
|
| 172 |
<p>Based on the <a href="https://github.com/mlcommons/croissant" target="_blank">Croissant format</a> from MLCommons.</p>
|
| 173 |
</div>
|
| 174 |
""")
|
| 175 |
-
|
| 176 |
return app
|
| 177 |
|
| 178 |
if __name__ == "__main__":
|
|
|
|
| 3 |
import time
|
| 4 |
import traceback
|
| 5 |
from validation import validate_json, validate_croissant, validate_records
|
| 6 |
+
import requests
|
| 7 |
|
| 8 |
def process_file(file):
|
| 9 |
results = []
|
|
|
|
| 32 |
with gr.Blocks(theme=gr.themes.Soft()) as app:
|
| 33 |
gr.Markdown("# Croissant JSON-LD Validator for NeurIPS")
|
| 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:
|
| 37 |
1. If the file is valid JSON
|
| 38 |
2. If it passes Croissant schema validation
|
| 39 |
3. If records can be generated within a reasonable time
|
| 40 |
""")
|
| 41 |
|
| 42 |
+
with gr.Tabs() as tabs:
|
| 43 |
+
with gr.TabItem("Upload File"):
|
| 44 |
+
file_input = gr.File(label="Upload Croissant JSON-LD File", file_types=[".json", ".jsonld"])
|
| 45 |
+
|
| 46 |
+
with gr.TabItem("URL Input"):
|
| 47 |
+
url_input = gr.Textbox(
|
| 48 |
+
label="Enter Croissant JSON-LD URL",
|
| 49 |
+
placeholder="https://huggingface.co/api/datasets/facebook/natural_reasoning/croissant"
|
| 50 |
+
)
|
| 51 |
+
fetch_btn = gr.Button("Fetch and Validate", variant="primary")
|
| 52 |
|
| 53 |
upload_progress = gr.HTML(
|
| 54 |
"""<div class="progress-container">
|
| 55 |
+
<div class="progress-status">Ready for validation</div>
|
| 56 |
</div>""", visible=True)
|
| 57 |
|
| 58 |
+
validate_btn = gr.Button("Validate Uploaded File", variant="primary")
|
| 59 |
|
| 60 |
+
# Create containers for validation results
|
| 61 |
validation_results = gr.HTML(visible=False)
|
| 62 |
|
| 63 |
# Define CSS for the validation UI
|
|
|
|
| 128 |
<div class="progress-status">β
File uploaded successfully</div>
|
| 129 |
</div>""", gr.update(visible=False)
|
| 130 |
|
| 131 |
+
def fetch_from_url(url):
|
| 132 |
+
if not url:
|
| 133 |
+
return """<div class="progress-container">
|
| 134 |
+
<div class="progress-status">Please enter a URL</div>
|
| 135 |
+
</div>""", gr.update(visible=False)
|
|
|
|
| 136 |
|
| 137 |
+
try:
|
| 138 |
+
# Fetch JSON from URL
|
| 139 |
+
response = requests.get(url, timeout=10)
|
| 140 |
+
response.raise_for_status() # Raise exception for 4XX/5XX status codes
|
| 141 |
+
|
| 142 |
+
# Try to parse as JSON
|
| 143 |
+
json_data = response.json()
|
| 144 |
+
|
| 145 |
+
# Show success message
|
| 146 |
+
progress_html = """<div class="progress-container">
|
| 147 |
+
<div class="progress-status">β
JSON fetched successfully from URL</div>
|
| 148 |
+
</div>"""
|
| 149 |
+
|
| 150 |
+
# Validate the fetched JSON
|
| 151 |
+
results = []
|
| 152 |
+
|
| 153 |
+
# Check 1: JSON validation (already done by parsing)
|
| 154 |
+
results.append(("JSON Format Validation", True, "β
The URL returned valid JSON."))
|
| 155 |
+
|
| 156 |
+
# Check 2: Croissant validation
|
| 157 |
+
croissant_valid, croissant_message = validate_croissant(json_data)
|
| 158 |
+
results.append(("Croissant Schema Validation", croissant_valid, croissant_message))
|
| 159 |
+
|
| 160 |
+
if not croissant_valid:
|
| 161 |
+
return progress_html, build_results_html(results)
|
| 162 |
+
|
| 163 |
+
# Check 3: Records validation
|
| 164 |
+
records_valid, records_message = validate_records(json_data)
|
| 165 |
+
results.append(("Records Generation Test", records_valid, records_message))
|
| 166 |
+
|
| 167 |
+
return progress_html, build_results_html(results)
|
| 168 |
+
|
| 169 |
+
except requests.exceptions.RequestException as e:
|
| 170 |
+
error_message = f"β Error fetching URL: {str(e)}"
|
| 171 |
+
return f"""<div class="progress-container">
|
| 172 |
+
<div class="progress-status">{error_message}</div>
|
| 173 |
+
</div>""", gr.update(visible=False)
|
| 174 |
+
except json.JSONDecodeError as e:
|
| 175 |
+
error_message = f"β URL did not return valid JSON: {str(e)}"
|
| 176 |
+
return f"""<div class="progress-container">
|
| 177 |
+
<div class="progress-status">{error_message}</div>
|
| 178 |
+
</div>""", gr.update(visible=False)
|
| 179 |
+
except Exception as e:
|
| 180 |
+
error_message = f"β Unexpected error: {str(e)}"
|
| 181 |
+
return f"""<div class="progress-container">
|
| 182 |
+
<div class="progress-status">{error_message}</div>
|
| 183 |
+
</div>""", gr.update(visible=False)
|
| 184 |
+
|
| 185 |
+
def build_results_html(results):
|
| 186 |
+
# Extract the HTML generation logic to a reusable function
|
| 187 |
html = '<div class="validation-results">'
|
| 188 |
|
| 189 |
for i, (test_name, passed, message) in enumerate(results):
|
|
|
|
| 190 |
status_class = "status-success" if passed else "status-error"
|
| 191 |
status_icon = "β" if passed else "β"
|
| 192 |
|
|
|
|
| 193 |
html += f'''
|
| 194 |
<div class="validation-step" id="step-{i}">
|
| 195 |
<div class="step-header" onclick="
|
|
|
|
| 215 |
'''
|
| 216 |
|
| 217 |
html += '</div>'
|
|
|
|
|
|
|
| 218 |
return gr.update(value=html, visible=True)
|
| 219 |
|
| 220 |
+
def on_validate(file):
|
| 221 |
+
if file is None:
|
| 222 |
+
return gr.update(visible=False)
|
| 223 |
+
|
| 224 |
+
# Process the file and get results
|
| 225 |
+
results = process_file(file)
|
| 226 |
+
return build_results_html(results)
|
| 227 |
+
|
| 228 |
+
# Connect UI events to functions
|
| 229 |
file_input.change(on_file_upload, inputs=file_input, outputs=[upload_progress, validation_results])
|
| 230 |
validate_btn.click(on_validate, inputs=file_input, outputs=validation_results)
|
| 231 |
+
fetch_btn.click(fetch_from_url, inputs=url_input, outputs=[upload_progress, validation_results])
|
| 232 |
|
| 233 |
+
# Footer
|
| 234 |
gr.HTML("""
|
| 235 |
<div style="text-align: center; margin-top: 20px;">
|
| 236 |
<p>Based on the <a href="https://github.com/mlcommons/croissant" target="_blank">Croissant format</a> from MLCommons.</p>
|
| 237 |
</div>
|
| 238 |
""")
|
| 239 |
+
|
| 240 |
return app
|
| 241 |
|
| 242 |
if __name__ == "__main__":
|
requirements.txt
CHANGED
|
@@ -1,3 +1,4 @@
|
|
| 1 |
gradio>=3.50.2
|
| 2 |
mlcroissant
|
| 3 |
-
func_timeout
|
|
|
|
|
|
| 1 |
gradio>=3.50.2
|
| 2 |
mlcroissant
|
| 3 |
+
func_timeout
|
| 4 |
+
requests
|