AI 教育视频生成：课件到视频的自动转换

输入：PPT 课件
    ↓
┌──────────────────────────────────────┐
│        AI 教育视频生成引擎             │
│                                      │
│  1. 课件解析                          │
│     ├── PPT → 图片 + 文本             │
│     ├── 知识点提取                    │
│     └── 教学大纲生成                  │
│                                      │
│  2. 脚本生成                          │
│     ├── 每页讲解文案                  │
│     ├── 动画效果规划                  │
│     └── 互动问题设计                  │
│                                      │
│  3. 视觉制作                          │
│     ├── 课件动画化                    │
│     ├── 知识图谱可视化                │
│     └── 板书效果                      │
│                                      │
│  4. 音频制作                          │
│     ├── 语音讲解（CosyVoice）         │
│     ├── 语速控制                      │
│     └── BGM                           │
│                                      │
│  5. 后期合成                          │
│     ├── 音画同步                      │
│     ├── 字幕生成                      │
│     └── 知识点标注                    │
└──────────────────────────────────────┘
    ↓
输出：教学视频（含字幕、互动）

from pptx import Presentation
import json

class PPTParser:
    """PPT 课件解析器"""
    
    def parse(self, ppt_path):
        """解析 PPT 为结构化数据"""
        prs = Presentation(ppt_path)
        slides = []
        
        for i, slide in enumerate(prs.slides):
            slide_data = {
                "slide_number": i + 1,
                "title": self._extract_title(slide),
                "text_content": self._extract_text(slide),
                "images": self._extract_images(slide),
                "tables": self._extract_tables(slide),
                "charts": self._extract_charts(slide),
                "notes": self._extract_notes(slide),
                "layout": self._detect_layout(slide),
            }
            slides.append(slide_data)
        
        return {
            "total_slides": len(slides),
            "slides": slides,
            "metadata": self._extract_metadata(prs)
        }
    
    def _extract_text(self, slide):
        """提取文本内容"""
        texts = []
        for shape in slide.shapes:
            if shape.has_text_frame:
                for para in shape.text_frame.paragraphs:
                    text = para.text.strip()
                    if text:
                        texts.append({
                            "text": text,
                            "level": para.level,
                            "position": (shape.left, shape.top)
                        })
        return texts

class KnowledgeExtractor:
    """从课件中提取知识点"""
    
    def __init__(self):
        self.llm = OpenAI()
    
    async def extract(self, slides_data):
        """提取知识点结构"""
        prompt = f"""
        基于以下课件内容，提取知识结构：
        
        课件内容：{json.dumps(slides_data, ensure_ascii=False)}
        
        返回 JSON 格式：
        {{
            "topics": [
                {{
                    "name": "主题名称",
                    "key_points": ["要点1", "要点2"],
                    "difficulty": "easy/medium/hard",
                    "prerequisites": ["前置知识"],
                    "slide_range": [start, end]
                }}
            ],
            "summary": "课程概要",
            "learning_objectives": ["学习目标1", "学习目标2"]
        }}
        """
        
        response = await self.llm.chat.completions.acreate(
            model="gpt-4o",
            messages=[{"role": "user", "content": prompt}],
            response_format={"type": "json_object"}
        )
        
        return json.loads(response.choices[0].message.content)

class TeachingScriptGenerator:
    """教学视频脚本生成"""
    
    def __init__(self):
        self.llm = OpenAI()
    
    async def generate(self, slide_data, knowledge):
        """为每页 PPT 生成讲解脚本"""
        
        prompt = f"""
        你是一位经验丰富的教师，正在制作教学视频。
        
        课件第 {slide_data['slide_number']} 页：
        标题：{slide_data['title']}
        内容：{slide_data['text_content']}
        
        知识点信息：
        主题：{knowledge['current_topic']}
        难度：{knowledge['difficulty']}
        
        请生成这一页的讲解脚本：
        1. 讲解文案（口语化，适合朗读，120-200字）
        2. 动画建议（哪些元素需要动画效果）
        3. 重点强调（哪些内容需要高亮/放大）
        4. 互动提问（每页1-2个问题）
        5. 预计讲解时长（秒）
        
        要求：
        - 语言生动，避免照念PPT
        - 有启发性，引导学生思考
        - 适当使用类比和举例
        - 中文，保留专业术语的英文
        """
        
        response = await self.llm.chat.completions.acreate(
            model="gpt-4o",
            messages=[{"role": "user", "content": prompt}],
            response_format={"type": "json_object"}
        )
        
        return json.loads(response.choices[0].message.content)

TEACHING_STYLES = {
    "academic": {
        "description": "学术风格，严谨专业",
        "tone": "正式",
        "examples": "学术案例",
        "speed": 0.9  # 语速稍慢
    },
    "engaging": {
        "description": "生动有趣，适合青少年",
        "tone": "活泼",
        "examples": "生活类比",
        "speed": 1.0
    },
    "concise": {
        "description": "精炼高效，适合成人教育",
        "tone": "简洁",
        "examples": "实际应用",
        "speed": 1.15
    },
    "storytelling": {
        "description": "故事化叙事，适合科普",
        "tone": "叙事",
        "examples": "历史故事",
        "speed": 0.95
    }
}

from moviepy.editor import *
from PIL import Image, ImageDraw, ImageFont

class SlideAnimator:
    """课件动画化"""
    
    def animate_slide(self, slide_image, animation_plan):
        """为静态 PPT 页面添加动画效果"""
        clips = []
        
        for anim in animation_plan:
            if anim["type"] == "fade_in":
                clip = self._fade_in(slide_image, anim)
            elif anim["type"] == "zoom":
                clip = self._zoom_effect(slide_image, anim)
            elif anim["type"] == "highlight":
                clip = self._highlight(slide_image, anim)
            elif anim["type"] == "draw_arrow":
                clip = self._draw_arrow(slide_image, anim)
            elif anim["type"] == "typewriter":
                clip = self._typewriter(slide_image, anim)
            
            clips.append(clip)
        
        return concatenate_videoclips(clips)
    
    def _fade_in(self, image, anim):
        """淡入效果"""
        clip = ImageClip(image).set_duration(anim["duration"])
        clip = clip.fadein(0.5)
        return clip
    
    def _zoom_effect(self, image, anim):
        """放大效果"""
        clip = ImageClip(image).set_duration(anim["duration"])
        clip = clip.resize(lambda t: 1 + 0.1 * t / anim["duration"])
        return clip
    
    def _highlight(self, image, anim):
        """高亮效果"""
        # 在指定区域添加高亮框
        img = Image.open(image)
        draw = ImageDraw.Draw(img)
        x1, y1, x2, y2 = anim["region"]
        draw.rectangle([x1, y1, x2, y2], outline="yellow", width=3)
        return ImageClip(img).set_duration(anim["duration"])

class BlackboardEffect:
    """模拟黑板板书效果"""
    
    def create_writing_animation(self, text, duration=3):
        """创建手写板书动画"""
        # 1. 生成板书图片
        board = self._create_board()
        
        # 2. 逐字显示
        frames = []
        chars = list(text)
        for i in range(len(chars) + 1):
            frame = self._render_partial(board, chars[:i])
            frames.append(frame)
        
        # 3. 转为视频
        clip = ImageSequenceClip(frames, fps=len(chars) / duration)
        return clip

class NarrationGenerator:
    """教学旁白生成"""
    
    def __init__(self):
        self.tts = CosyVoice2("pretrained_model")
    
    async def generate_narration(self, script, style="engaging"):
        """生成教学旁白"""
        style_config = TEACHING_STYLES[style]
        
        audio_segments = []
        for slide_script in script:
            audio = self.tts.synthesize(
                text=slide_script["narration"],
                voice_id="teacher_friendly",
                emotion="neutral",
                speed=style_config["speed"]
            )
            audio_segments.append({
                "audio": audio,
                "duration": slide_script["estimated_duration"],
                "slide_number": slide_script["slide_number"]
            })
        
        return audio_segments

class EducationVideoGenerator:
    """完整的教育视频生成器"""
    
    async def generate(self, ppt_path, style="engaging", 
                       languages=["zh"], output_dir="./output"):
        
        # 1. 解析 PPT
        slides_data = self.parser.parse(ppt_path)
        
        # 2. 提取知识点
        knowledge = await self.extractor.extract(slides_data)
        
        # 3. 生成脚本
        scripts = []
        for slide in slides_data["slides"]:
            script = await self.script_gen.generate(slide, knowledge)
            scripts.append(script)
        
        # 4. 生成视觉
        video_segments = []
        for slide, script in zip(slides_data["slides"], scripts):
            # 截取 PPT 页面为图片
            slide_image = self._render_slide(slide)
            
            # 添加动画
            animated = self.animator.animate_slide(slide_image, script["animations"])
            video_segments.append(animated)
        
        # 5. 生成旁白
        narration = await self.narrator.generate_narration(scripts, style)
        
        # 6. 合成
        final_video = self._compose(video_segments, narration)
        
        # 7. 字幕
        subtitles = self._generate_subtitles(narration, languages)
        
        # 8. 导出
        final = self._add_subtitles(final_video, subtitles)
        final.write_videofile(f"{output_dir}/lesson.mp4")
        
        return final

class InteractiveVideoGenerator:
    """互动教学视频"""
    
    async def add_interactions(self, video, scripts):
        """添加互动元素"""
        interactions = []
        
        for script in scripts:
            if "quiz" in script:
                interactions.append({
                    "timestamp": script["end_time"],
                    "type": "quiz",
                    "question": script["quiz"]["question"],
                    "options": script["quiz"]["options"],
                    "answer": script["quiz"]["answer"],
                    "explanation": script["quiz"]["explanation"]
                })
            
            if "pause_and_think" in script:
                interactions.append({
                    "timestamp": script["end_time"],
                    "type": "pause",
                    "duration": 5,
                    "prompt": script["pause_and_think"]
                })
        
        return interactions