import json import random import requests # Set API keys QWEN_API_KEY = "your qwen api key" PEXELS_API_KEY = "your pexels api key" CAPCUT_API_KEY = "your capcut api key" LICENSE_KEY="your capcut license key", def llm(query = ""): """ Call the Tongyi Qianwen large language model Parameters: Returns: dict: Dictionary containing title and list of sentences """ # Build system prompt system_prompt = """ * **Context:** You are an AI expert specializing in modern interpersonal relationships and emotional communication. Your knowledge base is built on a deep understanding of popular emotional content on social media, and you excel at interpreting the dynamics and perspectives of relationships in a relaxed, colloquial style. * **Objective:** When the user inputs "Give me some random returns", your goal is to **randomly create** advice about male-female emotions, behavioral habits, or relationship guidance. Your content must strictly mimic the unique style shown in the examples below. * **Style:** Your generated content should have the following characteristics: * **Structure:** Use a list format, with each point being a short, independent sentence. * **Wording:** Use colloquial language, often using the "When..." sentence pattern to describe a scenario. * **Theme:** Content should revolve around "how to understand the other person", "which behaviors are attractive", or "advice for a specific gender". * **Tone:** Your tone should be friendly, sincere, slightly teasing, like a friend sharing experiences on social media. * **Audience:** Your audience is anyone interested in modern emotional relationships who wants to get advice in a relaxed way. * **Response:** When you receive the instruction "Give me some random returns", please **randomly select one** from the following examples as your response. Or, you can **randomly generate** a new one with a completely consistent style, and return it in the same JSON format: **Example 1 (How to understand girls):** ```json { "title": "How to understand girls", "sentences": [ "Hands on her stomach (Insecure)", "She leans on you (Feels safe)", "Covers her smile (Thinks your going to judge)", "Stops texting you (Feels like she is annoying you)", "Says she is fine (She is everything but fine)", "When she hugs you (You mean a lot to her)" ] } ``` **Example 2 (Tips for girls):** ```json { "title": "Tips for the girls (From the guys)", "sentences": [ "99/100 guys dont know what hip dips are and actually love your stretch marks", "We can't tell that you like us by just viewing our story, just message us", "When he's out with his boys, let him have this time (this is very important)", "'I'm not ready for a relationship' - unless your the luckiest girl in the world your not getting cuffed", "As Bruno mars said, 'your perfect just the way you are' so just be you, it'll work" ] } ``` **Example 3 (Things guys find attractive in girls):** ```json { "title": "Things girls do that guys find attractive", "sentences": [ "Bed hair when it's all messy >>>", "When you come work out with us", "Your sleepy voice in the morning or after a nap", "When you wear our t-shirts as pyjamas", "When you have a funny or really bad laugh", "When you initiate ...", "When your good with animals or animals like you" ] } ``` """ # Build user prompt user_prompt = f"Randomly create advice about male-female emotions, behavioral habits, or relationship guidance, based on user input: {query}" # Prepare request data url = "https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions" headers = { "Authorization": f"Bearer {QWEN_API_KEY}", "Content-Type": "application/json" } data = { "model": "qwen-plus", "messages": [ { "role": "system", "content": system_prompt }, { "role": "user", "content": user_prompt } ], "temperature": 0.7, "max_tokens": 16384, "response_format": {"type": "json_object"} } try: # Send HTTP request response = requests.post(url, headers=headers, json=data, timeout=30) # Check response status if response.status_code == 200: response_data = response.json() # Extract content from response if 'choices' in response_data and len(response_data['choices']) > 0: content = response_data['choices'][0]['message']['content'] try: # Parse JSON response result = json.loads(content) # Ensure result contains necessary fields if "title" in result and "sentences" in result: return result except json.JSONDecodeError: pass # If JSON parsing fails, will return predefined example # If API call fails or parsing fails, print error message print(f"Error: {response.status_code}, {response.text if hasattr(response, 'text') else 'No response text'}") except Exception as e: # Catch all possible exceptions print(f"Exception occurred: {str(e)}") def search_pexels_videos(query="twilight", min_duration=10, orientation="portrait", per_page=15): """ Call Pexels API to search for videos Parameters: query (str): Search keyword, default is "twilight" min_duration (int): Minimum video duration (seconds), default is 10 seconds orientation (str): Video orientation, default is "portrait" per_page (int): Number of results per page, default is 15 Returns: list: List containing video information """ url = "https://api.pexels.com/videos/search" headers = { "Authorization": PEXELS_API_KEY } params = { "query": query, "orientation": orientation, "per_page": per_page, "min_duration": min_duration } try: response = requests.get(url, headers=headers, params=params) if response.status_code == 200: data = response.json() videos = [] for video in data.get("videos", []): # Get video file information video_files = video.get("video_files", []) # Filter out 16:9 ratio video files portrait_videos = [file for file in video_files if file.get("width") and file.get("height") and file.get("height") / file.get("width") > 1.7] # Close to 16:9 ratio if portrait_videos: # Select highest quality video file best_quality = max(portrait_videos, key=lambda x: x.get("width", 0) * x.get("height", 0)) videos.append({ "id": video.get("id"), "url": video.get("url"), "image": video.get("image"), "duration": video.get("duration"), "user": video.get("user", {}).get("name"), "video_url": best_quality.get("link"), "width": best_quality.get("width"), "height": best_quality.get("height"), "file_type": best_quality.get("file_type") }) if videos: random_video = random.choice(videos) return random_video['video_url'] return [] else: print(f"Error: {response.status_code}, {response.text}") return [] except Exception as e: print(f"Exception occurred: {str(e)}") return [] def create_capcut_draft(width=1080, height=1920): """ Call CapCut API to create a new draft Parameters: width (int): Video width, default is 1080 height (int): Video height, default is 1920 Returns: dict: Dictionary containing draft ID and download URL, or error message if failed """ url = "https://open.capcutapi.top/cut_jianying/create_draft" headers = { "Authorization": f"Bearer {CAPCUT_API_KEY}", "Content-Type": "application/json" } data = { "width": width, "height": height } try: response = requests.post(url, headers=headers, json=data) if response.status_code == 200: result = response.json() if result.get("success"): return { "success": True, "draft_id": result.get("output", {}).get("draft_id"), "draft_url": result.get("output", {}).get("draft_url") } else: return { "success": False, "error": result.get("error", "Unknown error") } else: return { "success": False, "error": f"HTTP Error: {response.status_code}", "response": response.text } except Exception as e: return { "success": False, "error": str(e) } def add_video_to_draft(draft_id, video_url): """ Call CapCut API to add video to draft Parameters: draft_id (str): Draft ID video_url (str): Video URL Returns: dict: Dictionary containing draft ID and download URL, or error message if failed """ url = "https://open.capcutapi.top/cut_jianying/add_video" headers = { "Authorization": f"Bearer {CAPCUT_API_KEY}", "Content-Type": "application/json" } data = { "video_url": video_url, "draft_id": draft_id, "end": 10 # Set video duration to 10 seconds } try: response = requests.post(url, headers=headers, json=data) if response.status_code == 200: result = response.json() if result.get("success"): return { "success": True, "draft_id": result.get("output", {}).get("draft_id"), "draft_url": result.get("output", {}).get("draft_url") } else: return { "success": False, "error": result.get("error", "Unknown error") } else: return { "success": False, "error": f"HTTP Error: {response.status_code}", "response": response.text } except Exception as e: return { "success": False, "error": str(e) } def add_text_to_draft(draft_id, text, font="ZY_Starry", font_color="#FFFFFF", background_color="#000000", background_alpha=0.5, background_style=2, background_round_radius=10, transform_y=0, transform_x=0, font_size=10.0, fixed_width=0.6, track_name="text_main"): """ Call CapCut API to add text to draft Parameters: draft_id (str): Draft ID text (str): Text content start_time (float): Text start time on timeline (seconds), default is 0 end_time (float): Text end time on timeline (seconds), default is 5 font (str): Font, default is "ZY_Starry" font_color (str): Font color, default is white background_color (str): Background color, default is black background_alpha (float): Background transparency, default is 0.5 transform_y (float): Y-axis position offset, default is 0 transform_x (float): X-axis position offset, default is 0 font_size (float): Font size, default is 10.0 Returns: dict: Dictionary containing draft ID and download URL, or error message if failed """ url = "https://open.capcutapi.top/cut_jianying/add_text" headers = { "Authorization": f"Bearer {CAPCUT_API_KEY}", "Content-Type": "application/json" } data = { "text": text, "start": 0, "end": 10, "draft_id": draft_id, "font": font, "font_color": font_color, "font_size": font_size, "transform_y": transform_y, "transform_x": transform_x, "fixed_width": fixed_width, "background_color": background_color, "background_alpha": background_alpha, "background_style": background_style, "background_round_radius": background_round_radius, "track_name": track_name } try: response = requests.post(url, headers=headers, json=data) if response.status_code == 200: result = response.json() if result.get("success"): return { "success": True, "draft_id": result.get("output", {}).get("draft_id"), "draft_url": result.get("output", {}).get("draft_url") } else: return { "success": False, "error": result.get("error", "Unknown error") } else: return { "success": False, "error": f"HTTP Error: {response.status_code}", "response": response.text } except Exception as e: return { "success": False, "error": str(e) } def generate_video(draft_id, resolution="720P", framerate="24"): """ Call CapCut API to render video Parameters: draft_id (str): Draft ID license_key (str): License key resolution (str): Video resolution, default is "720P" framerate (str): Video frame rate, default is "24" Returns: dict: Dictionary containing task ID, or error message if failed """ url = "https://open.capcutapi.top/cut_jianying/generate_video" headers = { "Authorization": f"Bearer {CAPCUT_API_KEY}", "Content-Type": "application/json" } data = { "draft_id": draft_id, "license_key": LICENSE_KEY, "resolution": resolution, "framerate": framerate } try: response = requests.post(url, headers=headers, json=data) if response.status_code == 200: result = response.json() if result.get("success"): return { "success": True, "task_id": result.get("output", {}).get("task_id") } else: return { "success": False, "error": result.get("error", "Unknown error") } else: return { "success": False, "error": f"HTTP Error: {response.status_code}", "response": response.text } except Exception as e: return { "success": False, "error": str(e) } def check_task_status(task_id): """ Call CapCut API to check task status Parameters: task_id (str): Task ID Returns: dict: Dictionary containing task status and results, or error message if failed """ url = "https://open.capcutapi.top/cut_jianying/task_status" headers = { "Authorization": f"Bearer {CAPCUT_API_KEY}", "Content-Type": "application/json" } data = { "task_id": task_id } try: response = requests.post(url, headers=headers, json=data) if response.status_code == 200: result = response.json() if result.get("success"): output = result.get("output", {}) return { "success": True, "status": output.get("status"), "progress": output.get("progress"), "result": output.get("result"), # Video URL "error": output.get("error") } else: return { "success": False, "error": result.get("error", "Unknown error") } else: return { "success": False, "error": f"HTTP Error: {response.status_code}", "response": response.text } except Exception as e: return { "success": False, "error": str(e) } # Example usage if __name__ == "__main__": # Call LLM function and print result result = llm() print(json.dumps(result, indent=2, ensure_ascii=False)) # 1. Create draft draft_result = create_capcut_draft() print("Draft creation result:", json.dumps(draft_result, indent=2, ensure_ascii=False)) if draft_result.get("success"): draft_id = draft_result.get("draft_id") # 2. Search Pexels videos video_url = search_pexels_videos() print("Pexels video URL:", video_url) # 3. Add video to draft if video_url: add_result = add_video_to_draft(draft_result.get("draft_id"), video_url) print("Add video result:", json.dumps(add_result, indent=2, ensure_ascii=False)) # 5. Add title text title_result = add_text_to_draft( draft_id=draft_id, text=result["title"], font="ZY_Starry", # Use starry font font_color="#FFFFFF", # White font background_color="#000000", # Black background background_alpha=1, # Background transparency background_style=1, background_round_radius=10, transform_y=0.7, # Located at the top of the screen (1 is top edge, -1 is bottom edge) transform_x=0, # Horizontally centered font_size=13.0, # Larger font track_name = "title", fixed_width=0.6 ) print("Add title result:", json.dumps(title_result, indent=2, ensure_ascii=False)) # 6. Add sentence text sentence_count = len(result["sentences"]) for i, sentence in enumerate(result["sentences"]): # Calculate vertical position - evenly distributed in the middle of the screen transform_y = 0.5 - (1.1 * (i + 1) / (sentence_count + 1)) # Determine horizontal alignment - odd sentences left-aligned, even sentences right-aligned if i % 2 == 0: # Odd sentences (counting from 0) transform_x = -0.5 # Left-aligned else: # Even sentences transform_x = 0.5 # Right-aligned sentence_result = add_text_to_draft( draft_id=draft_id, text=sentence, font="ZY_Fantasy", # Use fantasy font font_color="#FFFFFF", # White font transform_y=transform_y, # Vertical position transform_x=transform_x, # Horizontal position (left-right alignment) background_alpha=0, font_size=7.0, # Smaller font fixed_width=0.3, track_name=f"text_{i}" ) print(f"Add sentence {i+1} result:", json.dumps(sentence_result, indent=2, ensure_ascii=False)) # 7. Render video print("\nStarting video rendering...") generate_result = generate_video(draft_id) print("Video rendering request result:", json.dumps(generate_result, indent=2, ensure_ascii=False)) if generate_result.get("success"): task_id = generate_result.get("task_id") print(f"Task ID: {task_id}, starting to poll task status...") # 8. Poll task status import time max_attempts = 30 # Maximum 30 polling attempts attempt = 0 while attempt < max_attempts: status_result = check_task_status(task_id) print(f"Poll count {attempt+1}, status:", json.dumps(status_result, indent=2, ensure_ascii=False)) if not status_result.get("success"): print("Failed to check task status:", status_result.get("error")) break status = status_result.get("status") if status == "SUCCESS": print("\nVideo rendering successful!") print("Video URL:", status_result.get("result")) break elif status == "FAILED": print("\nVideo rendering failed:", status_result.get("error")) break # Wait 5 seconds before checking again print("Waiting 5 seconds before checking again...") time.sleep(5) attempt += 1 if attempt >= max_attempts: print("\nPolling timeout, please check task status manually later")