zorto-core 0.20.2

//! Search index generation using SQLite.
//!
//! At build time, generates a `search.db` file containing a search index of all
//! site pages. Uses a regular table with LIKE queries for broad compatibility
//! (including sql.js WASM which lacks FTS5 support).

use std::path::Path;

use crate::content::{Page, Section};

/// Strip HTML tags from content, returning plain text.
fn strip_html(html: &str) -> String {
    let mut result = String::with_capacity(html.len());
    let mut in_tag = false;
    let mut in_entity = false;
    let mut entity = String::new();

    for ch in html.chars() {
        match ch {
            '<' => in_tag = true,
            '>' => in_tag = false,
            '&' if !in_tag => {
                in_entity = true;
                entity.clear();
                entity.push(ch);
            }
            ';' if in_entity => {
                in_entity = false;
                // Decode common entities
                match entity.as_str() {
                    "&amp" => result.push('&'),
                    "&lt" => result.push('<'),
                    "&gt" => result.push('>'),
                    "&quot" => result.push('"'),
                    "&#39" | "&apos" => result.push('\''),
                    "&nbsp" => result.push(' '),
                    _ => {} // skip unknown entities
                }
            }
            _ if in_entity => entity.push(ch),
            _ if !in_tag => result.push(ch),
            _ => {}
        }
    }

    // Collapse multiple whitespace into single spaces
    let mut collapsed = String::with_capacity(result.len());
    let mut prev_space = false;
    for ch in result.chars() {
        if ch.is_whitespace() {
            if !prev_space {
                collapsed.push(' ');
            }
            prev_space = true;
        } else {
            collapsed.push(ch);
            prev_space = false;
        }
    }

    collapsed.trim().to_string()
}

/// Generate a SQLite search index at `output_dir/search.db`.
///
/// The database contains a `pages` table with columns for display
/// (`title`, `url`, `description`, `content`) and pre-computed lowercase
/// columns (`title_lower`, `description_lower`, `content_lower`) for
/// efficient case-insensitive ranked search at query time.
/// Uses a regular table (not FTS5) for sql.js WASM compatibility.
/// Convert a permalink to a relative path by stripping the base URL.
fn to_relative_url(permalink: &str, base_url: &str) -> String {
    if let Some(path) = permalink.strip_prefix(base_url) {
        if path.is_empty() {
            "/".to_string()
        } else if path.starts_with('/') {
            path.to_string()
        } else {
            format!("/{path}")
        }
    } else {
        // Fallback: try to extract path from full URL
        if let Some(pos) = permalink.find("://") {
            if let Some(slash_pos) = permalink[pos + 3..].find('/') {
                return permalink[pos + 3 + slash_pos..].to_string();
            }
        }
        permalink.to_string()
    }
}

pub fn generate_search_index<'a>(
    pages: impl IntoIterator<Item = &'a Page>,
    sections: impl IntoIterator<Item = &'a Section>,
    base_url: &str,
    output_dir: &Path,
) -> anyhow::Result<()> {
    let db_path = output_dir.join("search.db");

    // Remove existing db if present
    if db_path.exists() {
        std::fs::remove_file(&db_path)?;
    }

    let conn = rusqlite::Connection::open(&db_path)?;

    // Create table with pre-computed lowercase columns for ranked search
    conn.execute_batch(
        "CREATE TABLE pages (
            title TEXT NOT NULL,
            url TEXT NOT NULL,
            description TEXT,
            content TEXT,
            title_lower TEXT,
            description_lower TEXT,
            content_lower TEXT
        );
        CREATE INDEX idx_pages_title_lower ON pages(title_lower);
        CREATE INDEX idx_pages_description_lower ON pages(description_lower);",
    )?;

    let mut stmt = conn.prepare(
        "INSERT INTO pages (title, url, description, content, title_lower, description_lower, content_lower)
         VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)",
    )?;

    // Index pages
    for page in pages {
        let plain_text = strip_html(&page.content);
        let description = page.description.as_deref().unwrap_or("");
        let url = to_relative_url(&page.permalink, base_url);
        stmt.execute(rusqlite::params![
            page.title,
            url,
            description,
            plain_text,
            page.title.to_lowercase(),
            description.to_lowercase(),
            plain_text.to_lowercase(),
        ])?;
    }

    // Index sections that have content
    for section in sections {
        if section.content.trim().is_empty() {
            continue;
        }
        let plain_text = strip_html(&section.content);
        let description = section.description.as_deref().unwrap_or("");
        let url = to_relative_url(&section.permalink, base_url);
        stmt.execute(rusqlite::params![
            section.title,
            url,
            description,
            plain_text,
            section.title.to_lowercase(),
            description.to_lowercase(),
            plain_text.to_lowercase(),
        ])?;
    }

    Ok(())
}

#[cfg(test)]
mod tests {
    use super::*;
    use std::collections::HashMap;

    fn make_page(title: &str, url: &str, description: Option<&str>, content: &str) -> Page {
        Page {
            title: title.to_string(),
            date: None,
            author: None,
            permalink: url.to_string(),
            description: description.map(|s| s.to_string()),
            content: content.to_string(),
            draft: false,
            slug: "test".to_string(),
            template: None,
            path: "/test/".to_string(),
            summary: None,
            raw_content: String::new(),
            taxonomies: HashMap::new(),
            extra: serde_json::Value::Null,
            aliases: vec![],
            word_count: 0,
            reading_time: 0,
            relative_path: "test.md".to_string(),
        }
    }

    /// Run the ranked search query (same logic as the JS) against a database.
    fn ranked_search(conn: &rusqlite::Connection, term: &str) -> Vec<(String, i64)> {
        let mut stmt = conn
            .prepare(
                "SELECT title,
                    CASE WHEN title_lower = ?1 THEN 100
                         WHEN title_lower LIKE ?1 || '%' THEN 80
                         WHEN title_lower LIKE '%' || ?1 || '%' THEN 60
                         ELSE 0 END +
                    CASE WHEN description_lower LIKE '%' || ?1 || '%' THEN 20
                         ELSE 0 END +
                    CASE WHEN content_lower LIKE '%' || ?1 || '%' THEN 10
                         ELSE 0 END as score
                 FROM pages
                 WHERE title_lower LIKE '%' || ?1 || '%'
                    OR description_lower LIKE '%' || ?1 || '%'
                    OR content_lower LIKE '%' || ?1 || '%'
                 ORDER BY score DESC
                 LIMIT 10",
            )
            .unwrap();
        let rows = stmt
            .query_map([term.to_lowercase()], |row| {
                Ok((
                    row.get::<_, String>(0).unwrap(),
                    row.get::<_, i64>(1).unwrap(),
                ))
            })
            .unwrap();
        rows.map(|r| r.unwrap()).collect()
    }

    #[test]
    fn test_strip_html_basic() {
        assert_eq!(strip_html("<p>Hello <b>world</b></p>"), "Hello world");
    }

    #[test]
    fn test_strip_html_entities() {
        assert_eq!(strip_html("&amp; &lt; &gt;"), "& < >");
    }

    #[test]
    fn test_strip_html_whitespace() {
        assert_eq!(strip_html("<p>  Hello   world  </p>"), "Hello world");
    }

    #[test]
    fn test_strip_html_empty() {
        assert_eq!(strip_html(""), "");
    }

    #[test]
    fn test_generate_search_index() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![make_page(
            "Test Page",
            "https://example.com/test/",
            Some("A test page"),
            "<p>Hello world</p>",
        )];
        let sections: Vec<Section> = vec![];

        generate_search_index(
            pages.iter(),
            sections.iter(),
            "https://example.com",
            tmp.path(),
        )
        .unwrap();

        let db_path = tmp.path().join("search.db");
        assert!(db_path.exists());

        let conn = rusqlite::Connection::open(&db_path).unwrap();
        let count: i64 = conn
            .query_row("SELECT count(*) FROM pages", [], |row| row.get(0))
            .unwrap();
        assert_eq!(count, 1);

        // Verify lowercase columns are populated
        let title_lower: String = conn
            .query_row("SELECT title_lower FROM pages", [], |row| row.get(0))
            .unwrap();
        assert_eq!(title_lower, "test page");

        // Verify URL is stored as relative path
        let url: String = conn
            .query_row("SELECT url FROM pages", [], |row| row.get(0))
            .unwrap();
        assert_eq!(url, "/test/");
    }

    #[test]
    fn test_title_matches_rank_higher_than_content_only() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![
            make_page(
                "Themes",
                "/themes/",
                Some("Browse all themes"),
                "<p>Pick a theme for your site.</p>",
            ),
            make_page(
                "Getting Started",
                "/getting-started/",
                None,
                "<p>You can customize themes in config.</p>",
            ),
        ];

        generate_search_index(pages.iter(), std::iter::empty(), "", tmp.path()).unwrap();
        let conn = rusqlite::Connection::open(tmp.path().join("search.db")).unwrap();

        let results = ranked_search(&conn, "theme");
        assert_eq!(results.len(), 2);
        // "Themes" page should rank first (title contains "theme")
        assert_eq!(results[0].0, "Themes");
        assert!(results[0].1 > results[1].1);
    }

    #[test]
    fn test_exact_title_match_ranks_highest() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![
            make_page("Themes", "/themes/", None, "<p>All themes.</p>"),
            make_page(
                "Custom Themes Guide",
                "/custom-themes/",
                None,
                "<p>How to build themes.</p>",
            ),
            make_page(
                "Getting Started",
                "/start/",
                Some("themes overview"),
                "<p>Intro.</p>",
            ),
        ];

        generate_search_index(pages.iter(), std::iter::empty(), "", tmp.path()).unwrap();
        let conn = rusqlite::Connection::open(tmp.path().join("search.db")).unwrap();

        let results = ranked_search(&conn, "themes");
        // Exact title match "Themes" = "themes" should be first (score 100+10)
        assert_eq!(results[0].0, "Themes");
        // "Custom Themes Guide" title contains "themes" (score 60+10)
        assert_eq!(results[1].0, "Custom Themes Guide");
    }

    #[test]
    fn test_case_insensitive_matching() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![make_page(
            "Hello World",
            "/hello/",
            None,
            "<p>HELLO WORLD content</p>",
        )];

        generate_search_index(pages.iter(), std::iter::empty(), "", tmp.path()).unwrap();
        let conn = rusqlite::Connection::open(tmp.path().join("search.db")).unwrap();

        // Search with different cases
        for term in &["hello", "HELLO", "Hello", "hElLo"] {
            let results = ranked_search(&conn, term);
            assert_eq!(results.len(), 1, "Failed for term: {term}");
        }
    }

    #[test]
    fn test_empty_query_returns_nothing() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![make_page("Test", "/test/", None, "<p>Content</p>")];

        generate_search_index(pages.iter(), std::iter::empty(), "", tmp.path()).unwrap();
        let conn = rusqlite::Connection::open(tmp.path().join("search.db")).unwrap();

        let results = ranked_search(&conn, "");
        // Empty string matches everything via LIKE '%%', but the JS guards
        // against empty queries before executing. Verify the query at least
        // doesn't crash.
        assert!(!results.is_empty() || results.is_empty());
    }

    #[test]
    fn test_special_characters_dont_break_queries() {
        let tmp = tempfile::TempDir::new().unwrap();
        let pages = vec![make_page(
            "C++ Guide",
            "/cpp/",
            Some("Learn C++ basics"),
            "<p>Using % and ' and \" in code</p>",
        )];

        generate_search_index(pages.iter(), std::iter::empty(), "", tmp.path()).unwrap();
        let conn = rusqlite::Connection::open(tmp.path().join("search.db")).unwrap();

        // These should not crash
        for term in &["c++", "%", "'", "\"", "'; DROP TABLE", "test%test"] {
            let _results = ranked_search(&conn, term);
        }

        // Verify actual match works with special chars
        let results = ranked_search(&conn, "c++");
        assert!(!results.is_empty());
    }
}