| 1 | # Copyright (C) 2025 AIDC-AI |
| 2 | # |
| 3 | # Licensed under the Apache License, Version 2.0 (the "License"); |
| 4 | # you may not use this file except in compliance with the License. |
| 5 | # You may obtain a copy of the License at |
| 6 | # http://www.apache.org/licenses/LICENSE-2.0 |
| 7 | # Unless required by applicable law or agreed to in writing, software |
| 8 | # distributed under the License is distributed on an "AS IS" BASIS, |
| 9 | # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 10 | # See the License for the specific language governing permissions and |
| 11 | # limitations under the License. |
| 12 | |
| 13 | """ |
| 14 | Video generation endpoints |
| 15 | |
| 16 | Supports both synchronous and asynchronous video generation. |
| 17 | """ |
| 18 | |
| 19 | import os |
| 20 | from fastapi import APIRouter, HTTPException, Request |
| 21 | from loguru import logger |
| 22 | |
| 23 | from api.dependencies import PixelleVideoDep |
| 24 | from api.schemas.video import ( |
| 25 | VideoGenerateRequest, |
| 26 | VideoGenerateResponse, |
| 27 | VideoGenerateAsyncResponse, |
| 28 | ) |
| 29 | from api.tasks import task_manager, TaskType |
| 30 | |
| 31 | router = APIRouter(prefix="/video", tags=["Video Generation"]) |
| 32 | |
| 33 | |
| 34 | def path_to_url(request: Request, file_path: str) -> str: |
| 35 | """ |
| 36 | Convert file path to accessible URL |
| 37 | |
| 38 | Handles both absolute and relative paths, extracting the path relative |
| 39 | to the output directory for URL construction. |
| 40 | |
| 41 | Args: |
| 42 | request: FastAPI Request object (provides base_url from actual request) |
| 43 | file_path: Absolute or relative file path |
| 44 | |
| 45 | Returns: |
| 46 | Full URL to access the file |
| 47 | |
| 48 | Examples: |
| 49 | Windows: G:\\...\\output\\20251205_233630_c939\\final.mp4 |
| 50 | -> http://localhost:8000/api/files/20251205_233630_c939/final.mp4 |
| 51 | |
| 52 | Linux: /home/user/.../output/20251205_233630_c939/final.mp4 |
| 53 | -> http://localhost:8000/api/files/20251205_233630_c939/final.mp4 |
| 54 | |
| 55 | Domain: With domain request -> https://your-domain.com/api/files/... |
| 56 | """ |
| 57 | from pathlib import Path |
| 58 | import os |
| 59 | |
| 60 | # Normalize path separators to forward slashes first (for cross-platform compatibility) |
| 61 | file_path = file_path.replace("\\", "/") |
| 62 | |
| 63 | # Check if it's an absolute path (works for both Windows and Linux) |
| 64 | is_absolute = os.path.isabs(file_path) or Path(file_path).is_absolute() |
| 65 | |
| 66 | if is_absolute: |
| 67 | # Find "output" in the path and get everything after it |
| 68 | # Split by / to work with normalized paths |
| 69 | parts = file_path.split("/") |
| 70 | try: |
| 71 | output_idx = parts.index("output") |
| 72 | # Get all parts after "output" and join them |
| 73 | relative_parts = parts[output_idx + 1:] |
| 74 | file_path = "/".join(relative_parts) |
| 75 | except ValueError: |
| 76 | # If "output" not in path, use the filename only |
| 77 | file_path = Path(file_path).name |
| 78 | else: |
| 79 | # If relative path starting with "output/", remove it |
| 80 | if file_path.startswith("output/"): |
| 81 | file_path = file_path[7:] # Remove "output/" |
| 82 | |
| 83 | # Build URL using request's base_url (automatically matches the request host) |
| 84 | base_url = str(request.base_url).rstrip('/') |
| 85 | return f"{base_url}/api/files/{file_path}" |
| 86 | |
| 87 | |
| 88 | @router.post("/generate/sync", response_model=VideoGenerateResponse) |
| 89 | async def generate_video_sync( |
| 90 | request_body: VideoGenerateRequest, |
| 91 | pixelle_video: PixelleVideoDep, |
| 92 | request: Request |
| 93 | ): |
| 94 | """ |
| 95 | Generate video synchronously |
| 96 | |
| 97 | This endpoint blocks until video generation is complete. |
| 98 | Suitable for small videos (< 30 seconds). |
| 99 | |
| 100 | **Note**: May timeout for large videos. Use `/generate/async` instead. |
| 101 | |
| 102 | Request body includes all video generation parameters. |
| 103 | See VideoGenerateRequest schema for details. |
| 104 | |
| 105 | Returns path to generated video, duration, and file size. |
| 106 | """ |
| 107 | try: |
| 108 | logger.info(f"Sync video generation: {request_body.text[:50]}...") |
| 109 | |
| 110 | # Auto-determine media_width and media_height from template meta tags (required) |
| 111 | if not request_body.frame_template: |
| 112 | raise ValueError("frame_template is required to determine media size") |
| 113 | |
| 114 | from pixelle_video.services.frame_html import HTMLFrameGenerator |
| 115 | from pixelle_video.utils.template_util import resolve_template_path |
| 116 | template_path = resolve_template_path(request_body.frame_template) |
| 117 | generator = HTMLFrameGenerator(template_path) |
| 118 | media_width, media_height = generator.get_media_size() |
| 119 | logger.debug(f"Auto-determined media size from template: {media_width}x{media_height}") |
| 120 | |
| 121 | # Build video generation parameters |
| 122 | video_params = { |
| 123 | "text": request_body.text, |
| 124 | "mode": request_body.mode, |
| 125 | "title": request_body.title, |
| 126 | "n_scenes": request_body.n_scenes, |
| 127 | "min_narration_words": request_body.min_narration_words, |
| 128 | "max_narration_words": request_body.max_narration_words, |
| 129 | "min_image_prompt_words": request_body.min_image_prompt_words, |
| 130 | "max_image_prompt_words": request_body.max_image_prompt_words, |
| 131 | "media_width": media_width, |
| 132 | "media_height": media_height, |
| 133 | "media_workflow": request_body.media_workflow, |
| 134 | "video_fps": request_body.video_fps, |
| 135 | "frame_template": request_body.frame_template, |
| 136 | "prompt_prefix": request_body.prompt_prefix, |
| 137 | "bgm_path": request_body.bgm_path, |
| 138 | "bgm_volume": request_body.bgm_volume, |
| 139 | } |
| 140 | |
| 141 | # Add TTS workflow if specified |
| 142 | if request_body.tts_workflow: |
| 143 | video_params["tts_workflow"] = request_body.tts_workflow |
| 144 | |
| 145 | # Add ref_audio if specified |
| 146 | if request_body.ref_audio: |
| 147 | video_params["ref_audio"] = request_body.ref_audio |
| 148 | |
| 149 | # Legacy voice_id support (deprecated) |
| 150 | if request_body.voice_id: |
| 151 | logger.warning("voice_id parameter is deprecated, please use tts_workflow instead") |
| 152 | video_params["voice_id"] = request_body.voice_id |
| 153 | |
| 154 | # Add custom template parameters if specified |
| 155 | if request_body.template_params: |
| 156 | video_params["template_params"] = request_body.template_params |
| 157 | |
| 158 | # Call video generator service |
| 159 | result = await pixelle_video.generate_video(**video_params) |
| 160 | |
| 161 | # Get file size |
| 162 | file_size = os.path.getsize(result.video_path) if os.path.exists(result.video_path) else 0 |
| 163 | |
| 164 | # Convert path to URL |
| 165 | video_url = path_to_url(request, result.video_path) |
| 166 | |
| 167 | return VideoGenerateResponse( |
| 168 | video_url=video_url, |
| 169 | duration=result.duration, |
| 170 | file_size=file_size |
| 171 | ) |
| 172 | |
| 173 | except Exception as e: |
| 174 | logger.error(f"Sync video generation error: {e}") |
| 175 | raise HTTPException(status_code=500, detail=str(e)) |
| 176 | |
| 177 | |
| 178 | @router.post("/generate/async", response_model=VideoGenerateAsyncResponse) |
| 179 | async def generate_video_async( |
| 180 | request_body: VideoGenerateRequest, |
| 181 | pixelle_video: PixelleVideoDep, |
| 182 | request: Request |
| 183 | ): |
| 184 | """ |
| 185 | Generate video asynchronously |
| 186 | |
| 187 | Creates a background task for video generation. |
| 188 | Returns immediately with a task_id for tracking progress. |
| 189 | |
| 190 | **Workflow:** |
| 191 | 1. Submit video generation request |
| 192 | 2. Receive task_id in response |
| 193 | 3. Poll `/api/tasks/{task_id}` to check status |
| 194 | 4. When status is "completed", retrieve video from result |
| 195 | |
| 196 | Request body includes all video generation parameters. |
| 197 | See VideoGenerateRequest schema for details. |
| 198 | |
| 199 | Returns task_id for tracking progress. |
| 200 | """ |
| 201 | try: |
| 202 | logger.info(f"Async video generation: {request_body.text[:50]}...") |
| 203 | |
| 204 | # Create task |
| 205 | task = task_manager.create_task( |
| 206 | task_type=TaskType.VIDEO_GENERATION, |
| 207 | request_params=request_body.model_dump() |
| 208 | ) |
| 209 | |
| 210 | # Define async execution function |
| 211 | async def execute_video_generation(): |
| 212 | """Execute video generation in background""" |
| 213 | # Auto-determine media_width and media_height from template meta tags (required) |
| 214 | if not request_body.frame_template: |
| 215 | raise ValueError("frame_template is required to determine media size") |
| 216 | |
| 217 | from pixelle_video.services.frame_html import HTMLFrameGenerator |
| 218 | from pixelle_video.utils.template_util import resolve_template_path |
| 219 | template_path = resolve_template_path(request_body.frame_template) |
| 220 | generator = HTMLFrameGenerator(template_path) |
| 221 | media_width, media_height = generator.get_media_size() |
| 222 | logger.debug(f"Auto-determined media size from template: {media_width}x{media_height}") |
| 223 | |
| 224 | # Build video generation parameters |
| 225 | video_params = { |
| 226 | "text": request_body.text, |
| 227 | "mode": request_body.mode, |
| 228 | "title": request_body.title, |
| 229 | "n_scenes": request_body.n_scenes, |
| 230 | "min_narration_words": request_body.min_narration_words, |
| 231 | "max_narration_words": request_body.max_narration_words, |
| 232 | "min_image_prompt_words": request_body.min_image_prompt_words, |
| 233 | "max_image_prompt_words": request_body.max_image_prompt_words, |
| 234 | "media_width": media_width, |
| 235 | "media_height": media_height, |
| 236 | "media_workflow": request_body.media_workflow, |
| 237 | "video_fps": request_body.video_fps, |
| 238 | "frame_template": request_body.frame_template, |
| 239 | "prompt_prefix": request_body.prompt_prefix, |
| 240 | "bgm_path": request_body.bgm_path, |
| 241 | "bgm_volume": request_body.bgm_volume, |
| 242 | # Progress callback can be added here if needed |
| 243 | # "progress_callback": lambda event: task_manager.update_progress(...) |
| 244 | } |
| 245 | |
| 246 | # Add TTS workflow if specified |
| 247 | if request_body.tts_workflow: |
| 248 | video_params["tts_workflow"] = request_body.tts_workflow |
| 249 | |
| 250 | # Add ref_audio if specified |
| 251 | if request_body.ref_audio: |
| 252 | video_params["ref_audio"] = request_body.ref_audio |
| 253 | |
| 254 | # Legacy voice_id support (deprecated) |
| 255 | if request_body.voice_id: |
| 256 | logger.warning("voice_id parameter is deprecated, please use tts_workflow instead") |
| 257 | video_params["voice_id"] = request_body.voice_id |
| 258 | |
| 259 | # Add custom template parameters if specified |
| 260 | if request_body.template_params: |
| 261 | video_params["template_params"] = request_body.template_params |
| 262 | |
| 263 | result = await pixelle_video.generate_video(**video_params) |
| 264 | |
| 265 | # Get file size |
| 266 | file_size = os.path.getsize(result.video_path) if os.path.exists(result.video_path) else 0 |
| 267 | |
| 268 | # Convert path to URL |
| 269 | video_url = path_to_url(request, result.video_path) |
| 270 | |
| 271 | return { |
| 272 | "video_url": video_url, |
| 273 | "duration": result.duration, |
| 274 | "file_size": file_size |
| 275 | } |
| 276 | |
| 277 | # Start execution |
| 278 | await task_manager.execute_task( |
| 279 | task_id=task.task_id, |
| 280 | coro_func=execute_video_generation |
| 281 | ) |
| 282 | |
| 283 | return VideoGenerateAsyncResponse( |
| 284 | task_id=task.task_id |
| 285 | ) |
| 286 | |
| 287 | except Exception as e: |
| 288 | logger.error(f"Async video generation error: {e}") |
| 289 | raise HTTPException(status_code=500, detail=str(e)) |
| 290 | |
| 291 |