ai_synth/backend/src/services/export.rs

//! Export service for generating Markdown and PDF from a synthesis.
//!
//! Provides functions to convert a synthesis (with its sections and items)
//! into downloadable Markdown or PDF format.

use printpdf::*;
use std::io::BufWriter;

use crate::errors::AppError;
use crate::models::synthesis::NewsSection;

/// Generate a Markdown representation of the synthesis.
///
/// Output format:
/// ```text
/// # Synthese de la Semaine 2026-W12
///
/// Generee le 2026-03-21
///
/// ## Section Title
///
/// ### Article Title
///
/// [Lien](https://example.com/article)
///
/// Summary text here.
/// ```
pub fn generate_markdown(week: &str, date: &str, sections: &[NewsSection]) -> String {
    let mut md = String::new();

    md.push_str(&format!("# Synthese de la Semaine {}\n\n", week));
    md.push_str(&format!("Generee le {}\n", date));

    for section in sections {
        md.push_str(&format!("\n## {}\n", section.title));

        for item in &section.items {
            md.push_str(&format!("\n### {}\n\n", item.title));
            md.push_str(&format!("[Lien]({})\n\n", item.url));
            md.push_str(&format!("{}\n", item.summary));
        }
    }

    md
}

/// Page dimensions (A4 in mm).
const PAGE_WIDTH_MM: f32 = 210.0;
const PAGE_HEIGHT_MM: f32 = 297.0;

/// Margins in mm.
const MARGIN_LEFT: f32 = 25.0;
const MARGIN_TOP: f32 = 25.0;
const MARGIN_RIGHT: f32 = 25.0;
const MARGIN_BOTTOM: f32 = 25.0;

/// Maximum text width in mm.
const TEXT_WIDTH: f32 = PAGE_WIDTH_MM - MARGIN_LEFT - MARGIN_RIGHT;

/// Font sizes in pt.
const TITLE_SIZE: f32 = 18.0;
const DATE_SIZE: f32 = 10.0;
const SECTION_SIZE: f32 = 14.0;
const ITEM_TITLE_SIZE: f32 = 11.0;
const URL_SIZE: f32 = 8.0;
const BODY_SIZE: f32 = 10.0;
const FOOTER_SIZE: f32 = 8.0;

/// Approximate width of a character in mm for a given font size.
/// This is a rough estimate for Helvetica (average char width ~0.5 * size in pt).
/// 1 pt = 0.3528 mm.
fn approx_char_width_mm(font_size: f32) -> f32 {
    font_size * 0.3528 * 0.5
}

/// Wrap text into lines that fit within the given width in mm.
fn wrap_text(text: &str, font_size: f32, max_width_mm: f32) -> Vec<String> {
    let char_width = approx_char_width_mm(font_size);
    let max_chars = (max_width_mm / char_width).floor() as usize;
    if max_chars == 0 {
        return vec![text.to_string()];
    }

    let mut lines = Vec::new();
    for paragraph in text.split('\n') {
        if paragraph.is_empty() {
            lines.push(String::new());
            continue;
        }

        let words: Vec<&str> = paragraph.split_whitespace().collect();
        if words.is_empty() {
            lines.push(String::new());
            continue;
        }

        let mut current_line = String::new();
        for word in words {
            if current_line.is_empty() {
                current_line = word.to_string();
            } else if current_line.len() + 1 + word.len() <= max_chars {
                current_line.push(' ');
                current_line.push_str(word);
            } else {
                lines.push(current_line);
                current_line = word.to_string();
            }
        }
        if !current_line.is_empty() {
            lines.push(current_line);
        }
    }

    lines
}

/// PDF writer state: tracks current Y position and manages page creation.
struct PdfWriter {
    doc: PdfDocumentReference,
    current_page: PdfPageIndex,
    current_layer: PdfLayerIndex,
    y_pos: f32,
    font_regular: IndirectFontRef,
    font_bold: IndirectFontRef,
    font_italic: IndirectFontRef,
}

impl PdfWriter {
    fn new(
        doc: PdfDocumentReference,
        page: PdfPageIndex,
        layer: PdfLayerIndex,
        font_regular: IndirectFontRef,
        font_bold: IndirectFontRef,
        font_italic: IndirectFontRef,
    ) -> Self {
        Self {
            doc,
            current_page: page,
            current_layer: layer,
            y_pos: PAGE_HEIGHT_MM - MARGIN_TOP,
            font_regular,
            font_bold,
            font_italic,
        }
    }

    /// Ensure there is enough vertical space; if not, add a new page.
    fn ensure_space(&mut self, needed_mm: f32) {
        if self.y_pos - needed_mm < MARGIN_BOTTOM {
            let (page, layer) = self.doc.add_page(
                Mm(PAGE_WIDTH_MM),
                Mm(PAGE_HEIGHT_MM),
                "Page",
            );
            self.current_page = page;
            self.current_layer = layer;
            self.y_pos = PAGE_HEIGHT_MM - MARGIN_TOP;
        }
    }

    /// Write a single line of text at the current position.
    fn write_line(&mut self, text: &str, font: &IndirectFontRef, size: f32, color: (f32, f32, f32)) {
        let line_height_mm = size * 0.3528 * 1.4;
        self.ensure_space(line_height_mm);

        let layer = self.doc.get_page(self.current_page).get_layer(self.current_layer);
        layer.set_fill_color(Color::Rgb(Rgb::new(color.0, color.1, color.2, None)));
        layer.use_text(text, size, Mm(MARGIN_LEFT), Mm(self.y_pos), font);

        self.y_pos -= line_height_mm;
    }

    /// Write wrapped text (may span multiple lines).
    fn write_wrapped(
        &mut self,
        text: &str,
        font: &IndirectFontRef,
        size: f32,
        color: (f32, f32, f32),
    ) {
        let lines = wrap_text(text, size, TEXT_WIDTH);
        for line in &lines {
            self.write_line(line, font, size, color);
        }
    }

    /// Add vertical spacing in mm.
    fn space(&mut self, mm: f32) {
        self.y_pos -= mm;
    }
}

/// Indigo color (R, G, B) normalized to 0.0-1.0 for PDF.
const COLOR_INDIGO: (f32, f32, f32) = (0.263, 0.275, 0.898); // #4338CA
const COLOR_DARK: (f32, f32, f32) = (0.133, 0.133, 0.133); // #222222
const COLOR_GRAY: (f32, f32, f32) = (0.4, 0.4, 0.4); // #666666
const COLOR_LIGHT_GRAY: (f32, f32, f32) = (0.6, 0.6, 0.6); // #999999

/// Generate a PDF document from the synthesis, returned as raw bytes.
///
/// Uses the `printpdf` crate with built-in PDF fonts (Helvetica family).
/// No external font files are required.
///
/// The PDF contains:
/// - A title with "AI Weekly Synth - Semaine {week}"
/// - The generation date
/// - Section headers and items with titles, URLs, and summaries
/// - Automatic page breaks when content exceeds page height
pub fn generate_pdf(
    week: &str,
    date: &str,
    sections: &[NewsSection],
) -> Result<Vec<u8>, AppError> {
    let (doc, page, layer) = PdfDocument::new(
        format!("AI Weekly Synth - Semaine {}", week),
        Mm(PAGE_WIDTH_MM),
        Mm(PAGE_HEIGHT_MM),
        "Page 1",
    );

    let font_regular = doc.add_builtin_font(BuiltinFont::Helvetica).map_err(|e| {
        AppError::Internal(anyhow::anyhow!("Failed to load Helvetica font: {}", e))
    })?;
    let font_bold = doc
        .add_builtin_font(BuiltinFont::HelveticaBold)
        .map_err(|e| {
            AppError::Internal(anyhow::anyhow!("Failed to load Helvetica-Bold font: {}", e))
        })?;
    let font_italic = doc
        .add_builtin_font(BuiltinFont::HelveticaOblique)
        .map_err(|e| {
            AppError::Internal(anyhow::anyhow!(
                "Failed to load Helvetica-Oblique font: {}",
                e
            ))
        })?;

    let mut writer = PdfWriter::new(doc, page, layer, font_regular, font_bold, font_italic);

    // Title
    let title = format!("AI Weekly Synth - Semaine {}", week);
    writer.write_line(&title, &writer.font_bold.clone(), TITLE_SIZE, COLOR_INDIGO);
    writer.space(3.0);

    // Date
    let date_line = format!("Generee le {}", date);
    writer.write_line(
        &date_line,
        &writer.font_italic.clone(),
        DATE_SIZE,
        COLOR_GRAY,
    );
    writer.space(8.0);

    // Sections
    for section in sections {
        // Section title
        writer.write_line(
            &section.title,
            &writer.font_bold.clone(),
            SECTION_SIZE,
            COLOR_INDIGO,
        );
        writer.space(3.0);

        for item in &section.items {
            // Item title (bold)
            writer.write_wrapped(
                &item.title,
                &writer.font_bold.clone(),
                ITEM_TITLE_SIZE,
                COLOR_DARK,
            );

            // URL (italic, gray)
            writer.write_wrapped(
                &item.url,
                &writer.font_italic.clone(),
                URL_SIZE,
                COLOR_LIGHT_GRAY,
            );
            writer.space(1.0);

            // Summary
            writer.write_wrapped(
                &item.summary,
                &writer.font_regular.clone(),
                BODY_SIZE,
                COLOR_DARK,
            );
            writer.space(4.0);
        }

        writer.space(4.0);
    }

    // Footer
    writer.space(8.0);
    writer.write_line(
        "Genere par AI Weekly Synth",
        &writer.font_italic.clone(),
        FOOTER_SIZE,
        COLOR_LIGHT_GRAY,
    );

    // Render to bytes
    let buf = Vec::new();
    let mut buf_writer = BufWriter::new(buf);
    writer
        .doc
        .save(&mut buf_writer)
        .map_err(|e| AppError::Internal(anyhow::anyhow!("Failed to render PDF: {}", e)))?;

    let bytes = buf_writer
        .into_inner()
        .map_err(|e| AppError::Internal(anyhow::anyhow!("Failed to flush PDF buffer: {}", e)))?;

    Ok(bytes)
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::models::synthesis::{NewsItem, NewsSection};

    fn sample_sections() -> Vec<NewsSection> {
        vec![
            NewsSection {
                title: "Annonces Majeures".into(),
                items: vec![
                    NewsItem {
                        title: "OpenAI lance GPT-5".into(),
                        url: "https://openai.com/gpt5".into(),
                        summary:
                            "OpenAI a annonce la sortie de GPT-5 avec des capacites ameliorees."
                                .into(),
                        date: None,
                    },
                    NewsItem {
                        title: "Google DeepMind publie Gemini 3".into(),
                        url: "https://deepmind.google/gemini3".into(),
                        summary:
                            "DeepMind presente Gemini 3, son nouveau modele multimodal.".into(),
                        date: None,
                    },
                ],
            },
            NewsSection {
                title: "Recherche".into(),
                items: vec![NewsItem {
                    title: "Nouveau papier sur le RLHF".into(),
                    url: "https://arxiv.org/abs/2026.12345".into(),
                    summary: "Une nouvelle approche du RLHF prometteuse.".into(),
                    date: None,
                }],
            },
        ]
    }

    // --- Markdown tests ---

    #[test]
    fn markdown_contains_title() {
        let md = generate_markdown("2026-W12", "2026-03-21", &sample_sections());
        assert!(md.starts_with("# Synthese de la Semaine 2026-W12\n"));
    }

    #[test]
    fn markdown_contains_date() {
        let md = generate_markdown("2026-W12", "2026-03-21", &sample_sections());
        assert!(md.contains("Generee le 2026-03-21"));
    }

    #[test]
    fn markdown_contains_section_titles() {
        let md = generate_markdown("2026-W12", "2026-03-21", &sample_sections());
        assert!(md.contains("## Annonces Majeures"));
        assert!(md.contains("## Recherche"));
    }

    #[test]
    fn markdown_contains_item_titles_and_links() {
        let md = generate_markdown("2026-W12", "2026-03-21", &sample_sections());
        assert!(md.contains("### OpenAI lance GPT-5"));
        assert!(md.contains("[Lien](https://openai.com/gpt5)"));
        assert!(md.contains("### Nouveau papier sur le RLHF"));
        assert!(md.contains("[Lien](https://arxiv.org/abs/2026.12345)"));
    }

    #[test]
    fn markdown_contains_summaries() {
        let md = generate_markdown("2026-W12", "2026-03-21", &sample_sections());
        assert!(md.contains("OpenAI a annonce la sortie de GPT-5"));
        assert!(md.contains("Une nouvelle approche du RLHF prometteuse."));
    }

    #[test]
    fn markdown_empty_sections_produces_header_only() {
        let md = generate_markdown("2026-W12", "2026-03-21", &[]);
        assert!(md.contains("# Synthese de la Semaine 2026-W12"));
        assert!(md.contains("Generee le 2026-03-21"));
        // No section headers
        assert!(!md.contains("##"));
    }

    // --- PDF tests ---

    #[test]
    fn pdf_output_is_nonempty_and_starts_with_pdf_header() {
        let bytes = generate_pdf("2026-W12", "2026-03-21", &sample_sections()).unwrap();
        assert!(!bytes.is_empty(), "PDF output should not be empty");
        assert!(
            bytes.starts_with(b"%PDF"),
            "PDF output should start with %PDF magic bytes"
        );
    }

    #[test]
    fn pdf_empty_sections_still_produces_valid_pdf() {
        let bytes = generate_pdf("2026-W12", "2026-03-21", &[]).unwrap();
        assert!(!bytes.is_empty());
        assert!(bytes.starts_with(b"%PDF"));
    }

    #[test]
    fn pdf_contains_title_text() {
        let bytes = generate_pdf("2026-W12", "2026-03-21", &sample_sections()).unwrap();
        // The PDF binary content should contain the title string somewhere
        let content = String::from_utf8_lossy(&bytes);
        assert!(
            content.contains("AI Weekly Synth"),
            "PDF should contain the application name"
        );
    }

    // --- Text wrapping tests ---

    #[test]
    fn wrap_text_short_line_fits() {
        let lines = wrap_text("Hello world", 10.0, 160.0);
        assert_eq!(lines.len(), 1);
        assert_eq!(lines[0], "Hello world");
    }

    #[test]
    fn wrap_text_long_line_wraps() {
        // Very narrow width to force wrapping
        let lines = wrap_text("Hello world from the test", 10.0, 20.0);
        assert!(lines.len() > 1, "Long text should wrap into multiple lines");
    }

    #[test]
    fn wrap_text_empty_input() {
        let lines = wrap_text("", 10.0, 160.0);
        assert_eq!(lines.len(), 1);
        assert!(lines[0].is_empty());
    }
}