insta-import-pixelfed/src/image_description.rs

use base64::{Engine as _, engine::general_purpose::STANDARD};


// module to hold all code for generating/fetching image descriptions
// Input is the image name
// Output is a String containing the image description
pub struct ChatGPTConfig {
    pub openai_api_key: String,
    pub openai_api_url: String,
    pub openai_model: String

}

pub struct OllamaConfig {
    pub ollama_api_key: String,
    pub ollama_api_url: String,
    pub ollama_model: String
}

pub struct FileConfig {
    pub caption_extension: String,
}

// fetch the imagedescription from a file named like the Image 
pub  fn get_description_from_file(image_name: String , file_config: FileConfig) -> Result<String, Box<dyn super::Error>>  {
    //read image caption from a local file that 
    //has the same name than the image with the extension ".caption.txt"
    let caption_extension = file_config.caption_extension;
    let captionname =   format!("{}{}", image_name, caption_extension);
    
    println!("Looking for {}",captionname);
    let caption_data = std::fs::read_to_string(captionname);
    
    
    Ok(caption_data.unwrap())
}

// fetch image description from ChatGPT
pub  fn get_description_from_chatgpt(image_name: String, chatgpt_config: self::ChatGPTConfig) -> Result<String, Box<dyn super::Error>> {
    // Read and encode image
    let image_data = std::fs::read(image_name)?;
        
    
    
    // Base64 encode the image for ChatGTP API
    let base64_image = STANDARD.encode(image_data);
    
    // Create ChatGPT API request
    let client = reqwest::blocking::Client::new();
    let response = client
        .post(chatgpt_config.openai_api_url)
        .header("Authorization", format!("Bearer {}", chatgpt_config.openai_api_key))
        .header("Content-Type", "application/json")
        .json(&super::json!({
            "model": chatgpt_config.openai_model,
            "max_tokens": 300,
            "messages": [
                {
                    "role": "user",
                    "content": [
                        {
                            "type": "text",
                            "text": "Please describe this image concisely for use as an alt text description. Focus on key visual elements and context."
                        },
                        {
                            "type": "image_url",
                            "image_url": {
                                "url": format!("data:image/jpeg;base64,{}", base64_image)
                            }
                            }
                            ]
                            }
                            ]
                            }))
                            .send();
                            
                            // Improved error handling for API response
                            //if !response.unwrap().status().is_success() {
                            //    let error_text = response.unwrap_err();
                            //    return Err(format!("OpenAI API error: ", std::error.box(error_text));
                            //}
                            
                            let result: super::Value = response.unwrap().json()?;
                            
                            // More detailed error handling for JSON parsing
                            let description = result["choices"]
                            .get(0)
                            .ok_or("No choices in response")?
                            ["message"]["content"]
                            .as_str()
                            .ok_or("Invalid content format in response")?
                            .to_string();
                            
                            Ok(description)
}

// fetch images description from own OLLAMA server
pub  fn get_description_from_ollama(image_name: String, ollama_config: OllamaConfig) -> Result<String, Box<dyn super::Error>> {
    Ok("Not implemented yet".to_string())
}