diff --git a/domain/task.py b/domain/task.py index 7cac821..fc06f40 100644 --- a/domain/task.py +++ b/domain/task.py @@ -9,6 +9,12 @@ from enum import Enum from dataclasses import dataclass, field from typing import Dict, Any, Optional, List from datetime import datetime +from urllib.parse import urlparse, unquote +import os + + +# 支持的图片扩展名 +IMAGE_EXTENSIONS = {'.jpg', '.jpeg', '.png', '.webp', '.bmp', '.gif'} class TaskType(Enum): @@ -369,6 +375,37 @@ class Task: """获取绑定的素材 URL（实际可下载的 HTTP URL）""" return self.payload.get('boundMaterialUrl') + def get_material_type(self) -> str: + """ + 获取素材类型 + + 优先使用服务端下发的 materialType 字段， + 如果不存在则根据 URL 后缀自动推断。 + + Returns: + 素材类型："video" 或 "image" + """ + # 优先使用服务端下发的类型 + material_type = self.payload.get('materialType') + if material_type in ('video', 'image'): + return material_type + + # 降级：根据 URL 后缀推断 + material_url = self.get_material_url() + if material_url: + parsed = urlparse(material_url) + path = unquote(parsed.path) + _, ext = os.path.splitext(path) + if ext.lower() in IMAGE_EXTENSIONS: + return 'image' + + # 默认视频类型 + return 'video' + + def is_image_material(self) -> bool: + """判断素材是否为图片类型""" + return self.get_material_type() == 'image' + def get_render_spec(self) -> RenderSpec: """获取渲染规格""" return RenderSpec.from_dict(self.payload.get('renderSpec')) diff --git a/handlers/render_video.py b/handlers/render_video.py index eb70e98..9f9982c 100644 --- a/handlers/render_video.py +++ b/handlers/render_video.py @@ -9,14 +9,23 @@ import os import logging from typing import List, Optional, Tuple +from urllib.parse import urlparse, unquote from handlers.base import BaseHandler -from domain.task import Task, TaskType, RenderSpec, OutputSpec, Effect +from domain.task import Task, TaskType, RenderSpec, OutputSpec, Effect, IMAGE_EXTENSIONS from domain.result import TaskResult, ErrorCode logger = logging.getLogger(__name__) +def _get_extension_from_url(url: str) -> str: + """从 URL 提取文件扩展名""" + parsed = urlparse(url) + path = unquote(parsed.path) + _, ext = os.path.splitext(path) + return ext.lower() if ext else '' + + class RenderSegmentVideoHandler(BaseHandler): """ 视频片段渲染处理器 @@ -46,18 +55,18 @@ class RenderSegmentVideoHandler(BaseHandler): "Missing material URL (boundMaterialUrl or sourceRef)" ) - # 检查 URL 格式：必须是 HTTP/HTTPS 协议 + # 检查 URL 格式：必须是 HTTP 或 HTTPS 协议 if not material_url.startswith(('http://', 'https://')): source_ref = task.get_source_ref() bound_url = task.get_bound_material_url() logger.error( f"[task:{task.task_id}] Invalid material URL format: '{material_url}'. " f"boundMaterialUrl={bound_url}, sourceRef={source_ref}. " - f"Server should provide boundMaterialUrl with HTTP URL." + f"Server should provide boundMaterialUrl with HTTP/HTTPS URL." ) return TaskResult.fail( ErrorCode.E_SPEC_INVALID, - f"Invalid material URL: '{material_url}' is not a valid HTTP URL. " + f"Invalid material URL: '{material_url}' is not a valid HTTP/HTTPS URL. " f"Server must provide boundMaterialUrl." ) @@ -65,15 +74,44 @@ class RenderSegmentVideoHandler(BaseHandler): output_spec = task.get_output_spec() duration_ms = task.get_duration_ms() - # 1. 下载素材 - input_file = os.path.join(work_dir, 'input.mp4') + # 1. 检测素材类型并确定输入文件扩展名 + is_image = task.is_image_material() + if is_image: + # 图片素材：根据 URL 确定扩展名 + ext = _get_extension_from_url(material_url) + if not ext or ext not in IMAGE_EXTENSIONS: + ext = '.jpg' # 默认扩展名 + input_file = os.path.join(work_dir, f'input{ext}') + else: + input_file = os.path.join(work_dir, 'input.mp4') + + # 2. 下载素材 if not self.download_file(material_url, input_file): return TaskResult.fail( ErrorCode.E_INPUT_UNAVAILABLE, f"Failed to download material: {material_url}" ) - # 2. 下载 LUT（如有） + # 3. 图片素材转换为视频 + if is_image: + video_input_file = os.path.join(work_dir, 'input_video.mp4') + if not self._convert_image_to_video( + image_file=input_file, + output_file=video_input_file, + duration_ms=duration_ms, + output_spec=output_spec, + render_spec=render_spec, + task_id=task.task_id + ): + return TaskResult.fail( + ErrorCode.E_FFMPEG_FAILED, + "Failed to convert image to video" + ) + # 使用转换后的视频作为输入 + input_file = video_input_file + logger.info(f"[task:{task.task_id}] Image converted to video successfully") + + # 4. 下载 LUT（如有） lut_file = None if render_spec.lut_url: lut_file = os.path.join(work_dir, 'lut.cube') @@ -81,7 +119,7 @@ class RenderSegmentVideoHandler(BaseHandler): logger.warning(f"[task:{task.task_id}] Failed to download LUT, continuing without it") lut_file = None - # 3. 下载叠加层（如有） + # 5. 下载叠加层（如有） overlay_file = None if render_spec.overlay_url: # 根据 URL 后缀确定文件扩展名 @@ -93,13 +131,13 @@ class RenderSegmentVideoHandler(BaseHandler): logger.warning(f"[task:{task.task_id}] Failed to download overlay, continuing without it") overlay_file = None - # 4. 计算 overlap 时长（用于转场帧冻结） + # 6. 计算 overlap 时长（用于转场帧冻结） # 头部 overlap: 来自前一片段的出场转场 overlap_head_ms = task.get_overlap_head_ms() # 尾部 overlap: 当前片段的出场转场 overlap_tail_ms = task.get_overlap_tail_ms_v2() - # 5. 构建 FFmpeg 命令 + # 7. 构建 FFmpeg 命令 output_file = os.path.join(work_dir, 'output.mp4') cmd = self._build_command( input_file=input_file, @@ -113,25 +151,25 @@ class RenderSegmentVideoHandler(BaseHandler): overlap_tail_ms=overlap_tail_ms ) - # 6. 执行 FFmpeg + # 8. 执行 FFmpeg if not self.run_ffmpeg(cmd, task.task_id): return TaskResult.fail( ErrorCode.E_FFMPEG_FAILED, "FFmpeg rendering failed" ) - # 7. 验证输出文件 + # 9. 验证输出文件 if not self.ensure_file_exists(output_file, min_size=4096): return TaskResult.fail( ErrorCode.E_FFMPEG_FAILED, "Output file is missing or too small" ) - # 8. 获取实际时长 + # 10. 获取实际时长 actual_duration = self.probe_duration(output_file) actual_duration_ms = int(actual_duration * 1000) if actual_duration else duration_ms - # 9. 上传产物 + # 11. 上传产物 video_url = self.upload_file(task.task_id, 'video', output_file) if not video_url: return TaskResult.fail( @@ -139,7 +177,7 @@ class RenderSegmentVideoHandler(BaseHandler): "Failed to upload video" ) - # 10. 构建结果（包含 overlap 信息） + # 12. 构建结果（包含 overlap 信息） result_data = { 'videoUrl': video_url, 'actualDurationMs': actual_duration_ms, @@ -156,6 +194,96 @@ class RenderSegmentVideoHandler(BaseHandler): finally: self.cleanup_work_dir(work_dir) + def _convert_image_to_video( + self, + image_file: str, + output_file: str, + duration_ms: int, + output_spec: OutputSpec, + render_spec: RenderSpec, + task_id: str + ) -> bool: + """ + 将图片转换为视频 + + 使用 FFmpeg 将静态图片转换为指定时长的视频， + 同时应用缩放填充和变速处理。 + + Args: + image_file: 输入图片文件路径 + output_file: 输出视频文件路径 + duration_ms: 目标时长（毫秒） + output_spec: 输出规格 + render_spec: 渲染规格 + task_id: 任务 ID（用于日志） + + Returns: + 是否成功 + """ + width = output_spec.width + height = output_spec.height + fps = output_spec.fps + + # 计算实际时长（考虑变速） + speed = float(render_spec.speed) if render_spec.speed else 1.0 + if speed <= 0: + speed = 1.0 + # 变速后的实际播放时长 + actual_duration_sec = (duration_ms / 1000.0) / speed + + # 构建 FFmpeg 命令 + cmd = [ + 'ffmpeg', '-y', '-hide_banner', + '-loop', '1', # 循环输入图片 + '-i', image_file, + '-t', str(actual_duration_sec), # 输出时长 + ] + + # 构建滤镜：缩放填充到目标尺寸 + filters = [] + + # 裁切处理（与视频相同逻辑） + if render_spec.crop_enable and render_spec.face_pos: + try: + fx, fy = map(float, render_spec.face_pos.split(',')) + target_ratio = width / height + filters.append( + f"crop='min(iw,ih*{target_ratio})':'min(ih,iw/{target_ratio})':" + f"'(iw-min(iw,ih*{target_ratio}))*{fx}':" + f"'(ih-min(ih,iw/{target_ratio}))*{fy}'" + ) + except (ValueError, ZeroDivisionError): + logger.warning(f"[task:{task_id}] Invalid face position: {render_spec.face_pos}") + elif render_spec.zoom_cut: + target_ratio = width / height + filters.append( + f"crop='min(iw,ih*{target_ratio})':'min(ih,iw/{target_ratio})'" + ) + + # 缩放填充 + filters.append( + f"scale={width}:{height}:force_original_aspect_ratio=decrease," + f"pad={width}:{height}:(ow-iw)/2:(oh-ih)/2:black" + ) + + # 格式转换（确保兼容性） + filters.append("format=yuv420p") + + cmd.extend(['-vf', ','.join(filters)]) + + # 编码参数 + cmd.extend([ + '-c:v', 'libx264', + '-preset', 'fast', + '-crf', '18', + '-r', str(fps), + '-an', # 无音频 + output_file + ]) + + logger.info(f"[task:{task_id}] Converting image to video: {actual_duration_sec:.2f}s at {fps}fps") + return self.run_ffmpeg(cmd, task_id) + def _build_command( self, input_file: str,