ai_synth/backend/src/services/prompts.rs

//! Prompt construction for the two-pass LLM generation pipeline.
//!
//! Builds system and user prompts for:
//! - **Search pass** (Pass 1): web search and initial article discovery
//! - **Rewrite pass** (Pass 2): rewrite summaries using scraped content
//!
//! Prompts are provider-agnostic and parameterized by user settings.

use crate::models::settings::UserSettings;
use crate::models::source::Source;
use crate::models::synthesis::ScrapedNewsItem;

/// Build the system prompt and user prompt for the search pass (Pass 1).
///
/// The search pass instructs the LLM to find recent news articles
/// matching the user's theme and categories, using web search grounding.
///
/// # Arguments
/// * `settings` — User's configured settings (theme, categories, etc.)
/// * `sources` — User's custom sources to prioritize
/// * `current_date` — Formatted date string for the prompt
pub fn build_search_prompt(
    settings: &UserSettings,
    sources: &[Source],
    current_date: &str,
) -> (String, String) {
    let sources_text = if sources.is_empty() {
        String::new()
    } else {
        let list = sources
            .iter()
            .map(|s| format!("- {} ({})", s.title, s.url))
            .collect::<Vec<_>>()
            .join("\n");
        format!(
            "\nEn plus des sources par defaut, tu DOIS imperativement consulter \
             et integrer les informations provenant de ces sources personnalisees :\n{}\n",
            list
        )
    };

    let categories_text = settings
        .categories
        .iter()
        .enumerate()
        .map(|(i, cat)| format!("{}. {}", i + 1, cat))
        .collect::<Vec<_>>()
        .join("\n");

    let behavior = if settings.search_agent_behavior.is_empty() {
        "Tu peux egalement utiliser d'autres sources pertinentes trouvees via la recherche Google."
            .to_string()
    } else {
        settings.search_agent_behavior.clone()
    };

    let system_prompt = format!(
        "Tu es un assistant IA precis. Tu dois TOUJOURS fournir des URLs completes et exactes. \
         Ne tronque jamais les URLs. Tu dois te concentrer UNIQUEMENT sur les actualites des {} \
         derniers jours.",
        settings.max_age_days
    );

    let user_prompt = format!(
        "Aujourd'hui, nous sommes le {date}.\n\
         Tu es un expert en analyse de l'actualite sur le theme : \"{theme}\".\n\
         Ta tache est de rechercher les actualites STRICTEMENT des {days} derniers jours.\n\
         Ne retourne AUCUNE actualite datant de plus de {days} jours.\n\n\
         Tu DOIS imperativement t'appuyer sur le contenu des sites web pertinents pour ce theme.\
         {sources}\
         {behavior}\n\n\
         La synthese doit etre divisee en {count} grandes sections :\n\
         {categories}\n\n\
         Pour chaque categorie, fournis au maximum {max_items} actualites.\n\
         Pour chaque actualite, fournis un titre provisoire, l'URL source exacte et complete, \
         et un resume provisoire.\n\
         Retourne le resultat au format JSON en utilisant les cles category_0, category_1, etc. \
         correspondant a l'ordre des sections ci-dessus.",
        date = current_date,
        theme = settings.theme,
        days = settings.max_age_days,
        sources = sources_text,
        behavior = behavior,
        count = settings.categories.len(),
        categories = categories_text,
        max_items = settings.max_items_per_category,
    );

    (system_prompt, user_prompt)
}

/// Build the system prompt and user prompt for the rewrite pass (Pass 2).
///
/// The rewrite pass takes scraped article content and asks the LLM to
/// rewrite titles and summaries to faithfully reflect the actual content.
///
/// # Arguments
/// * `scraped_data` — Map of category key to scraped news items with content
pub fn build_rewrite_prompt(
    scraped_data: &std::collections::HashMap<String, Vec<ScrapedNewsItem>>,
) -> (String, String) {
    let system_prompt =
        "Tu es un assistant IA precis. Tu dois generer des titres et resumes fideles \
         au contenu fourni."
            .to_string();

    let data_json = serde_json::to_string_pretty(scraped_data).unwrap_or_default();

    let user_prompt = format!(
        "Tu es un expert en analyse de l'actualite.\n\
         Voici une liste d'articles d'actualite classes par categorie, avec leur contenu textuel \
         brut extrait des sites web ('scrapedContent').\n\
         Ta tache est de reecrire le 'title' et le 'summary' (4 ou 5 lignes) pour chaque article \
         afin qu'ils refletent EXACTEMENT et FIDELEMENT le contenu textuel fourni.\n\
         Si le 'scrapedContent' est vide ou insuffisant, utilise le titre et le resume originaux \
         pour faire au mieux.\n\
         Conserve EXACTEMENT les memes URLs. Ne supprime aucun article de cette liste.\n\n\
         Donnees des articles :\n{data}",
        data = data_json,
    );

    (system_prompt, user_prompt)
}

#[cfg(test)]
mod tests {
    use super::*;
    use chrono::Utc;
    use uuid::Uuid;

    fn test_settings() -> UserSettings {
        UserSettings {
            user_id: Uuid::nil(),
            theme: "Intelligence Artificielle".to_string(),
            max_age_days: 7,
            categories: vec![
                "Annonces majeures".to_string(),
                "Recherche et innovation".to_string(),
            ],
            max_items_per_category: 4,
            search_agent_behavior: String::new(),
            ai_provider: String::new(),
            ai_model: String::new(),
            ai_model_writing: String::new(),
            rate_limit_max_requests: None,
            rate_limit_time_window_seconds: None,
            updated_at: Utc::now(),
        }
    }

    #[test]
    fn search_prompt_includes_theme() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("Intelligence Artificielle"));
    }

    #[test]
    fn search_prompt_includes_date() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("lundi 21 mars 2026"));
    }

    #[test]
    fn search_prompt_includes_max_age() {
        let settings = test_settings();
        let (system, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("7 derniers jours"));
        assert!(system.contains("7"));
    }

    #[test]
    fn search_prompt_includes_categories() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("1. Annonces majeures"));
        assert!(user_prompt.contains("2. Recherche et innovation"));
        assert!(user_prompt.contains("2 grandes sections"));
    }

    #[test]
    fn search_prompt_includes_max_items() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("4 actualites"));
    }

    #[test]
    fn search_prompt_includes_custom_sources() {
        let settings = test_settings();
        let sources = vec![
            Source {
                id: Uuid::nil(),
                user_id: Uuid::nil(),
                title: "TechCrunch".into(),
                url: "https://techcrunch.com".into(),
                created_at: Utc::now(),
            },
            Source {
                id: Uuid::nil(),
                user_id: Uuid::nil(),
                title: "The Verge".into(),
                url: "https://theverge.com".into(),
                created_at: Utc::now(),
            },
        ];

        let (_, user_prompt) = build_search_prompt(&settings, &sources, "lundi 21 mars 2026");
        assert!(user_prompt.contains("TechCrunch (https://techcrunch.com)"));
        assert!(user_prompt.contains("The Verge (https://theverge.com)"));
        assert!(user_prompt.contains("sources personnalisees"));
    }

    #[test]
    fn search_prompt_no_sources_no_section() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(!user_prompt.contains("sources personnalisees"));
    }

    #[test]
    fn search_prompt_custom_behavior() {
        let mut settings = test_settings();
        settings.search_agent_behavior =
            "Concentre-toi sur les sources europeennes.".to_string();

        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("Concentre-toi sur les sources europeennes."));
        assert!(!user_prompt.contains("recherche Google"));
    }

    #[test]
    fn search_prompt_default_behavior_when_empty() {
        let settings = test_settings();
        let (_, user_prompt) = build_search_prompt(&settings, &[], "lundi 21 mars 2026");
        assert!(user_prompt.contains("recherche Google"));
    }

    #[test]
    fn rewrite_prompt_includes_instructions() {
        let mut data = std::collections::HashMap::new();
        data.insert(
            "category_0".to_string(),
            vec![ScrapedNewsItem {
                title: "Test Article".into(),
                url: "https://example.com".into(),
                summary: "A summary".into(),
                scraped_content: "Full article text here...".into(),
            }],
        );

        let (system, user_prompt) = build_rewrite_prompt(&data);
        assert!(system.contains("fideles"));
        assert!(user_prompt.contains("scrapedContent"));
        assert!(user_prompt.contains("Test Article"));
        assert!(user_prompt.contains("https://example.com"));
        assert!(user_prompt.contains("Ne supprime aucun article"));
    }

    #[test]
    fn rewrite_prompt_with_empty_data() {
        let data = std::collections::HashMap::new();
        let (_, user_prompt) = build_rewrite_prompt(&data);
        // Should still produce a valid prompt with empty data
        assert!(user_prompt.contains("Donnees des articles"));
    }
}