Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -10,10 +10,7 @@ import logging
|
|
| 10 |
from typing import List, Optional
|
| 11 |
|
| 12 |
# Configure logging
|
| 13 |
-
logging.basicConfig(
|
| 14 |
-
level=logging.INFO,
|
| 15 |
-
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
|
| 16 |
-
)
|
| 17 |
logger = logging.getLogger(__name__)
|
| 18 |
|
| 19 |
# Create FastAPI app
|
|
@@ -32,58 +29,21 @@ app.add_middleware(
|
|
| 32 |
allow_headers=["*"],
|
| 33 |
)
|
| 34 |
|
| 35 |
-
@app.on_event("startup")
|
| 36 |
-
async def startup_event():
|
| 37 |
-
"""Startup event - runs after server starts"""
|
| 38 |
-
logger.info("=" * 60)
|
| 39 |
-
logger.info("HTML to PDF Converter API Starting...")
|
| 40 |
-
logger.info("=" * 60)
|
| 41 |
-
logger.info("Checking dependencies...")
|
| 42 |
-
|
| 43 |
-
# Check if node is available
|
| 44 |
-
try:
|
| 45 |
-
result = subprocess.run(['node', '--version'], capture_output=True, text=True, timeout=5)
|
| 46 |
-
logger.info(f"Node.js version: {result.stdout.strip()}")
|
| 47 |
-
except Exception as e:
|
| 48 |
-
logger.error(f"Node.js not found: {e}")
|
| 49 |
-
|
| 50 |
-
# Check if chromium is available
|
| 51 |
-
chromium_path = os.environ.get('PUPPETEER_EXECUTABLE_PATH', '/usr/bin/chromium')
|
| 52 |
-
if os.path.exists(chromium_path):
|
| 53 |
-
logger.info(f"Chromium found at: {chromium_path}")
|
| 54 |
-
else:
|
| 55 |
-
logger.error(f"Chromium not found at: {chromium_path}")
|
| 56 |
-
|
| 57 |
-
# Check if puppeteer script exists
|
| 58 |
-
if os.path.exists('/app/puppeteer_pdf.js'):
|
| 59 |
-
logger.info("Puppeteer script found")
|
| 60 |
-
else:
|
| 61 |
-
logger.error("Puppeteer script not found!")
|
| 62 |
-
|
| 63 |
-
logger.info("=" * 60)
|
| 64 |
-
logger.info("API is ready!")
|
| 65 |
-
logger.info("=" * 60)
|
| 66 |
-
|
| 67 |
@app.get("/")
|
| 68 |
async def root():
|
| 69 |
-
"""Root endpoint -
|
| 70 |
return {
|
| 71 |
"status": "online",
|
| 72 |
"service": "HTML to PDF Converter API",
|
| 73 |
"version": "1.0.0",
|
| 74 |
"docs": "/docs",
|
| 75 |
-
"
|
| 76 |
}
|
| 77 |
|
| 78 |
@app.get("/health")
|
| 79 |
async def health():
|
| 80 |
"""Health check endpoint"""
|
| 81 |
-
return {
|
| 82 |
-
"status": "healthy",
|
| 83 |
-
"service": "HTML to PDF Converter API",
|
| 84 |
-
"version": "1.0.0",
|
| 85 |
-
"endpoints": ["/convert", "/convert-html", "/convert-base64"]
|
| 86 |
-
}
|
| 87 |
|
| 88 |
@app.get("/ui", response_class=HTMLResponse)
|
| 89 |
async def ui():
|
|
@@ -148,14 +108,6 @@ async def ui():
|
|
| 148 |
}
|
| 149 |
ul { margin-left: 20px; }
|
| 150 |
li { margin: 8px 0; }
|
| 151 |
-
a { color: #667eea; text-decoration: none; font-weight: 500; }
|
| 152 |
-
a:hover { text-decoration: underline; }
|
| 153 |
-
.links {
|
| 154 |
-
display: flex;
|
| 155 |
-
gap: 15px;
|
| 156 |
-
margin: 20px 0;
|
| 157 |
-
flex-wrap: wrap;
|
| 158 |
-
}
|
| 159 |
.btn {
|
| 160 |
display: inline-block;
|
| 161 |
background: #667eea;
|
|
@@ -163,6 +115,7 @@ async def ui():
|
|
| 163 |
padding: 12px 24px;
|
| 164 |
border-radius: 6px;
|
| 165 |
text-decoration: none;
|
|
|
|
| 166 |
transition: all 0.3s;
|
| 167 |
}
|
| 168 |
.btn:hover {
|
|
@@ -182,19 +135,13 @@ async def ui():
|
|
| 182 |
page breaks, image embedding, and multiple aspect ratios.
|
| 183 |
</p>
|
| 184 |
|
| 185 |
-
<div
|
| 186 |
<a href="/docs" class="btn">📚 API Documentation</a>
|
| 187 |
<a href="/health" class="btn">💚 Health Check</a>
|
| 188 |
</div>
|
| 189 |
|
| 190 |
<h2>📡 Endpoints</h2>
|
| 191 |
|
| 192 |
-
<div class="endpoint">
|
| 193 |
-
<h3>GET /health</h3>
|
| 194 |
-
<p>Check API health status</p>
|
| 195 |
-
<pre>curl https://YOUR-SPACE.hf.space/health</pre>
|
| 196 |
-
</div>
|
| 197 |
-
|
| 198 |
<div class="endpoint">
|
| 199 |
<h3>POST /convert</h3>
|
| 200 |
<p>Upload HTML file and convert to PDF</p>
|
|
@@ -202,56 +149,26 @@ async def ui():
|
|
| 202 |
<ul>
|
| 203 |
<li><code>html_file</code> - HTML file (required)</li>
|
| 204 |
<li><code>aspect_ratio</code> - "16:9", "1:1", or "9:16" (optional)</li>
|
| 205 |
-
<li><code>
|
| 206 |
-
<li><code>images</code> - Image files to embed (optional, multiple files)</li>
|
| 207 |
</ul>
|
| 208 |
-
<pre>curl -X POST https://YOUR-SPACE.hf.space/convert \\
|
| 209 |
-
-F "html_file=@document.html" \\
|
| 210 |
-
-F "images=@logo.png" \\
|
| 211 |
-
-o output.pdf</pre>
|
| 212 |
</div>
|
| 213 |
|
| 214 |
<div class="endpoint">
|
| 215 |
<h3>POST /convert-html</h3>
|
| 216 |
<p>Convert HTML string to PDF</p>
|
| 217 |
-
<p><strong>Parameters:</strong></p>
|
| 218 |
-
<ul>
|
| 219 |
-
<li><code>html_content</code> - HTML as string (required)</li>
|
| 220 |
-
<li><code>aspect_ratio</code> - "16:9", "1:1", or "9:16" (optional)</li>
|
| 221 |
-
</ul>
|
| 222 |
-
<pre>curl -X POST https://YOUR-SPACE.hf.space/convert-html \\
|
| 223 |
-
-F "html_content=<html><body><h1>Hello!</h1></body></html>" \\
|
| 224 |
-
-o output.pdf</pre>
|
| 225 |
</div>
|
| 226 |
|
| 227 |
<div class="endpoint">
|
| 228 |
<h3>POST /convert-base64</h3>
|
| 229 |
<p>Convert HTML to base64-encoded PDF (returns JSON)</p>
|
| 230 |
-
<p><strong>Response:</strong></p>
|
| 231 |
-
<pre>{
|
| 232 |
-
"success": true,
|
| 233 |
-
"pdf_base64": "JVBERi0xLjQK...",
|
| 234 |
-
"aspect_ratio": "9:16",
|
| 235 |
-
"size_bytes": 45678
|
| 236 |
-
}</pre>
|
| 237 |
</div>
|
| 238 |
|
| 239 |
<h2>📐 Aspect Ratios</h2>
|
| 240 |
<ul>
|
| 241 |
-
<li><code>16:9</code> - Landscape
|
| 242 |
-
<li><code>1:1</code> - Square
|
| 243 |
-
<li><code>9:16</code> - Portrait
|
| 244 |
</ul>
|
| 245 |
-
|
| 246 |
-
<h2>📄 HTML Structure for Page Breaks</h2>
|
| 247 |
-
<p>Use class <code>page</code> or <code>slide</code> for automatic page breaks:</p>
|
| 248 |
-
<pre><div class="page">
|
| 249 |
-
<h1>Page 1</h1>
|
| 250 |
-
</div>
|
| 251 |
-
|
| 252 |
-
<div class="page">
|
| 253 |
-
<h1>Page 2</h1>
|
| 254 |
-
</div></pre>
|
| 255 |
</div>
|
| 256 |
</body>
|
| 257 |
</html>
|
|
@@ -319,8 +236,6 @@ def convert_to_pdf(html: str, ratio: str) -> bytes:
|
|
| 319 |
logger.error(f"Conversion failed: {result.stderr}")
|
| 320 |
raise Exception(f"PDF conversion failed: {result.stderr}")
|
| 321 |
|
| 322 |
-
logger.info("Conversion successful")
|
| 323 |
-
|
| 324 |
pdf_file = html_file.replace('.html', '.pdf')
|
| 325 |
|
| 326 |
if not os.path.exists(pdf_file):
|
|
@@ -329,7 +244,7 @@ def convert_to_pdf(html: str, ratio: str) -> bytes:
|
|
| 329 |
with open(pdf_file, 'rb') as f:
|
| 330 |
pdf_bytes = f.read()
|
| 331 |
|
| 332 |
-
logger.info(f"PDF
|
| 333 |
return pdf_bytes
|
| 334 |
|
| 335 |
finally:
|
|
@@ -344,13 +259,10 @@ async def convert(
|
|
| 344 |
):
|
| 345 |
"""Convert uploaded HTML file to PDF"""
|
| 346 |
try:
|
| 347 |
-
logger.info(f"Received conversion request: {html_file.filename}")
|
| 348 |
-
|
| 349 |
html = (await html_file.read()).decode('utf-8')
|
| 350 |
|
| 351 |
if auto_detect:
|
| 352 |
aspect_ratio = detect_aspect_ratio(html)
|
| 353 |
-
logger.info(f"Auto-detected aspect ratio: {aspect_ratio}")
|
| 354 |
elif not aspect_ratio:
|
| 355 |
aspect_ratio = "9:16"
|
| 356 |
|
|
@@ -359,7 +271,6 @@ async def convert(
|
|
| 359 |
|
| 360 |
# Embed images if provided
|
| 361 |
if images:
|
| 362 |
-
logger.info(f"Embedding {len(images)} image(s)")
|
| 363 |
for img in images:
|
| 364 |
img_bytes = await img.read()
|
| 365 |
b64 = base64.b64encode(img_bytes).decode()
|
|
@@ -370,7 +281,6 @@ async def convert(
|
|
| 370 |
|
| 371 |
pdf = convert_to_pdf(html, aspect_ratio)
|
| 372 |
|
| 373 |
-
logger.info("Returning PDF response")
|
| 374 |
return Response(
|
| 375 |
content=pdf,
|
| 376 |
media_type="application/pdf",
|
|
@@ -381,7 +291,7 @@ async def convert(
|
|
| 381 |
)
|
| 382 |
|
| 383 |
except Exception as e:
|
| 384 |
-
logger.error(f"Error
|
| 385 |
raise HTTPException(500, str(e))
|
| 386 |
|
| 387 |
@app.post("/convert-html")
|
|
@@ -392,8 +302,6 @@ async def convert_html(
|
|
| 392 |
):
|
| 393 |
"""Convert HTML string to PDF"""
|
| 394 |
try:
|
| 395 |
-
logger.info("Received HTML string conversion request")
|
| 396 |
-
|
| 397 |
if auto_detect:
|
| 398 |
aspect_ratio = detect_aspect_ratio(html_content)
|
| 399 |
elif not aspect_ratio:
|
|
@@ -414,7 +322,7 @@ async def convert_html(
|
|
| 414 |
)
|
| 415 |
|
| 416 |
except Exception as e:
|
| 417 |
-
logger.error(f"Error
|
| 418 |
raise HTTPException(500, str(e))
|
| 419 |
|
| 420 |
@app.post("/convert-base64")
|
|
@@ -425,8 +333,6 @@ async def convert_base64(
|
|
| 425 |
):
|
| 426 |
"""Convert HTML to base64-encoded PDF"""
|
| 427 |
try:
|
| 428 |
-
logger.info("Received base64 conversion request")
|
| 429 |
-
|
| 430 |
if auto_detect:
|
| 431 |
aspect_ratio = detect_aspect_ratio(html_content)
|
| 432 |
elif not aspect_ratio:
|
|
@@ -446,7 +352,7 @@ async def convert_base64(
|
|
| 446 |
})
|
| 447 |
|
| 448 |
except Exception as e:
|
| 449 |
-
logger.error(f"Error
|
| 450 |
raise HTTPException(500, str(e))
|
| 451 |
|
| 452 |
if __name__ == "__main__":
|
|
|
|
| 10 |
from typing import List, Optional
|
| 11 |
|
| 12 |
# Configure logging
|
| 13 |
+
logging.basicConfig(level=logging.INFO)
|
|
|
|
|
|
|
|
|
|
| 14 |
logger = logging.getLogger(__name__)
|
| 15 |
|
| 16 |
# Create FastAPI app
|
|
|
|
| 29 |
allow_headers=["*"],
|
| 30 |
)
|
| 31 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
@app.get("/")
|
| 33 |
async def root():
|
| 34 |
+
"""Root endpoint - returns immediately for health check"""
|
| 35 |
return {
|
| 36 |
"status": "online",
|
| 37 |
"service": "HTML to PDF Converter API",
|
| 38 |
"version": "1.0.0",
|
| 39 |
"docs": "/docs",
|
| 40 |
+
"ui": "/ui"
|
| 41 |
}
|
| 42 |
|
| 43 |
@app.get("/health")
|
| 44 |
async def health():
|
| 45 |
"""Health check endpoint"""
|
| 46 |
+
return {"status": "healthy"}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 47 |
|
| 48 |
@app.get("/ui", response_class=HTMLResponse)
|
| 49 |
async def ui():
|
|
|
|
| 108 |
}
|
| 109 |
ul { margin-left: 20px; }
|
| 110 |
li { margin: 8px 0; }
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
.btn {
|
| 112 |
display: inline-block;
|
| 113 |
background: #667eea;
|
|
|
|
| 115 |
padding: 12px 24px;
|
| 116 |
border-radius: 6px;
|
| 117 |
text-decoration: none;
|
| 118 |
+
margin: 5px;
|
| 119 |
transition: all 0.3s;
|
| 120 |
}
|
| 121 |
.btn:hover {
|
|
|
|
| 135 |
page breaks, image embedding, and multiple aspect ratios.
|
| 136 |
</p>
|
| 137 |
|
| 138 |
+
<div>
|
| 139 |
<a href="/docs" class="btn">📚 API Documentation</a>
|
| 140 |
<a href="/health" class="btn">💚 Health Check</a>
|
| 141 |
</div>
|
| 142 |
|
| 143 |
<h2>📡 Endpoints</h2>
|
| 144 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 145 |
<div class="endpoint">
|
| 146 |
<h3>POST /convert</h3>
|
| 147 |
<p>Upload HTML file and convert to PDF</p>
|
|
|
|
| 149 |
<ul>
|
| 150 |
<li><code>html_file</code> - HTML file (required)</li>
|
| 151 |
<li><code>aspect_ratio</code> - "16:9", "1:1", or "9:16" (optional)</li>
|
| 152 |
+
<li><code>images</code> - Image files to embed (optional)</li>
|
|
|
|
| 153 |
</ul>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 154 |
</div>
|
| 155 |
|
| 156 |
<div class="endpoint">
|
| 157 |
<h3>POST /convert-html</h3>
|
| 158 |
<p>Convert HTML string to PDF</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
</div>
|
| 160 |
|
| 161 |
<div class="endpoint">
|
| 162 |
<h3>POST /convert-base64</h3>
|
| 163 |
<p>Convert HTML to base64-encoded PDF (returns JSON)</p>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 164 |
</div>
|
| 165 |
|
| 166 |
<h2>📐 Aspect Ratios</h2>
|
| 167 |
<ul>
|
| 168 |
+
<li><code>16:9</code> - Landscape - Presentations</li>
|
| 169 |
+
<li><code>1:1</code> - Square - Social media</li>
|
| 170 |
+
<li><code>9:16</code> - Portrait - Documents (default)</li>
|
| 171 |
</ul>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 172 |
</div>
|
| 173 |
</body>
|
| 174 |
</html>
|
|
|
|
| 236 |
logger.error(f"Conversion failed: {result.stderr}")
|
| 237 |
raise Exception(f"PDF conversion failed: {result.stderr}")
|
| 238 |
|
|
|
|
|
|
|
| 239 |
pdf_file = html_file.replace('.html', '.pdf')
|
| 240 |
|
| 241 |
if not os.path.exists(pdf_file):
|
|
|
|
| 244 |
with open(pdf_file, 'rb') as f:
|
| 245 |
pdf_bytes = f.read()
|
| 246 |
|
| 247 |
+
logger.info(f"PDF generated: {len(pdf_bytes)} bytes")
|
| 248 |
return pdf_bytes
|
| 249 |
|
| 250 |
finally:
|
|
|
|
| 259 |
):
|
| 260 |
"""Convert uploaded HTML file to PDF"""
|
| 261 |
try:
|
|
|
|
|
|
|
| 262 |
html = (await html_file.read()).decode('utf-8')
|
| 263 |
|
| 264 |
if auto_detect:
|
| 265 |
aspect_ratio = detect_aspect_ratio(html)
|
|
|
|
| 266 |
elif not aspect_ratio:
|
| 267 |
aspect_ratio = "9:16"
|
| 268 |
|
|
|
|
| 271 |
|
| 272 |
# Embed images if provided
|
| 273 |
if images:
|
|
|
|
| 274 |
for img in images:
|
| 275 |
img_bytes = await img.read()
|
| 276 |
b64 = base64.b64encode(img_bytes).decode()
|
|
|
|
| 281 |
|
| 282 |
pdf = convert_to_pdf(html, aspect_ratio)
|
| 283 |
|
|
|
|
| 284 |
return Response(
|
| 285 |
content=pdf,
|
| 286 |
media_type="application/pdf",
|
|
|
|
| 291 |
)
|
| 292 |
|
| 293 |
except Exception as e:
|
| 294 |
+
logger.error(f"Error: {str(e)}")
|
| 295 |
raise HTTPException(500, str(e))
|
| 296 |
|
| 297 |
@app.post("/convert-html")
|
|
|
|
| 302 |
):
|
| 303 |
"""Convert HTML string to PDF"""
|
| 304 |
try:
|
|
|
|
|
|
|
| 305 |
if auto_detect:
|
| 306 |
aspect_ratio = detect_aspect_ratio(html_content)
|
| 307 |
elif not aspect_ratio:
|
|
|
|
| 322 |
)
|
| 323 |
|
| 324 |
except Exception as e:
|
| 325 |
+
logger.error(f"Error: {str(e)}")
|
| 326 |
raise HTTPException(500, str(e))
|
| 327 |
|
| 328 |
@app.post("/convert-base64")
|
|
|
|
| 333 |
):
|
| 334 |
"""Convert HTML to base64-encoded PDF"""
|
| 335 |
try:
|
|
|
|
|
|
|
| 336 |
if auto_detect:
|
| 337 |
aspect_ratio = detect_aspect_ratio(html_content)
|
| 338 |
elif not aspect_ratio:
|
|
|
|
| 352 |
})
|
| 353 |
|
| 354 |
except Exception as e:
|
| 355 |
+
logger.error(f"Error: {str(e)}")
|
| 356 |
raise HTTPException(500, str(e))
|
| 357 |
|
| 358 |
if __name__ == "__main__":
|