返回 VideoClaw
act_extract_sequel_en.txt
根目录 / video-claw / video-claw / backend / prompts / script / act_extract_sequel_en.txt
1 You are a professional script analyst. You need to accurately split the continued (sequel) script content into a JSON array based on episodic labels (such as "Episode {start_episode_num}", etc.).
2
3 ---
4
5 [Script Input]
6
7 {sequel_script}
8
9 ---
10
11 [Splitting Requirements]
12
13 1. Group the input script content into the corresponding `episode_number`. This continuation starts from Episode {start_episode_num}.
14 2. For `act_title`, summarize a concise title of no more than 20 words.
15 3. For `content`, preserve the full original script text of that episode (including scene descriptions, action directions, and character dialogue).
16
17 **Note (Strictly Adhere to):**
18 1. **Output Format**: You must output ONLY a pure JSON array. No Markdown formatting (such as ```json code blocks), explanatory text, or closing remarks are allowed.
19 2. Ensure the JSON structure is complete and valid.
20
21 ---
22
23 [Output Format Example (ONLY return this JSON array)]
24 [
25 {{
26 "episode_number": {start_episode_num},
27 "act_title": "Episode Title",
28 "content": "Detailed script content..."
29 }}
30 ]
31
31 lines Plain Text