zhixi-document-runtime/crates/zx_document_core/src/document.rs

use serde::{Deserialize, Serialize};

use crate::material_type::{MaterialType, PreviewMode};

#[derive(Debug, Clone, Serialize, Deserialize, uniffi::Record)]
pub struct DocumentInfo {
    pub material_id: String,
    pub title: String,
    pub material_type: MaterialType,
    pub preview_mode: PreviewMode,
    pub file_size: u64,

    // Text stats
    pub page_count: Option<u32>,
    pub word_count: Option<u32>,
    pub char_count: Option<u32>,
    pub line_count: Option<u32>,

    // Image
    pub image_width: Option<u32>,
    pub image_height: Option<u32>,
    pub image_format: Option<String>,

    // EPUB
    pub epub_chapter_count: Option<u32>,

    // Capabilities
    pub is_searchable: bool,
    pub supports_position: bool,
    pub supports_anchor: bool,

    pub created_at: Option<String>,
}

impl DocumentInfo {
    pub fn new(material_id: String, title: String, material_type: MaterialType, file_size: u64) -> Self {
        let preview_mode = material_type.preview_mode();
        let (is_searchable, supports_position, supports_anchor) = match material_type {
            MaterialType::Markdown | MaterialType::Text | MaterialType::Pdf | MaterialType::Epub => (true, true, true),
            MaterialType::Image => (false, true, true),
            MaterialType::Word | MaterialType::Excel | MaterialType::PowerPoint => (false, false, true),
            MaterialType::Unknown => (false, false, false),
        };
        Self {
            material_id,
            title,
            material_type,
            preview_mode,
            file_size,
            page_count: None,
            word_count: None,
            char_count: None,
            line_count: None,
            image_width: None,
            image_height: None,
            image_format: None,
            epub_chapter_count: None,
            is_searchable,
            supports_position,
            supports_anchor,
            created_at: None,
        }
    }
}

/// Build a fully populated DocumentInfo from a file path.
pub fn build_document_info(
    file_path: &str,
    material_id: String,
    title: String,
) -> Result<DocumentInfo, crate::error::DocumentError> {
    use std::path::Path;
    use crate::material_type::detect_material_type;

    let path = Path::new(file_path);
    let mt = detect_material_type(file_path).unwrap_or(MaterialType::Unknown);
    let file_size = std::fs::metadata(path).map(|m| m.len()).unwrap_or(0);

    let mut info = DocumentInfo::new(material_id, title, mt, file_size);

    match &info.material_type {
        MaterialType::Markdown | MaterialType::Text => {
            if let Ok(content) = std::fs::read_to_string(path) {
                let stats = crate::text::text_stats(&content);
                info.word_count = Some(stats.word_count);
                info.line_count = Some(stats.line_count);
                info.char_count = Some(content.chars().count() as u32);
                info.page_count = Some((stats.line_count / 50).max(1));
            }
        }
        MaterialType::Image => {
            if let Ok(meta) = crate::image_meta::read_image_meta(file_path) {
                info.image_width = Some(meta.width);
                info.image_height = Some(meta.height);
                info.image_format = Some(meta.format);
            }
        }
        MaterialType::Pdf => {
            if let Ok(meta) = crate::pdf::read_pdf_metadata(path) {
                info.page_count = Some(meta.page_count);
            }
        }
        MaterialType::Epub => {
            if let Ok(meta) = crate::epub::read_epub_metadata(path) {
                info.epub_chapter_count = Some(meta.chapter_count);
            }
        }
        _ => {}
    }

    Ok(info)
}

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

    #[test]
    fn test_markdown_capabilities() {
        let info = DocumentInfo::new("m1".into(), "test.md".into(), MaterialType::Markdown, 1024);
        assert!(info.is_searchable);
        assert!(info.supports_position);
        assert_eq!(info.preview_mode, PreviewMode::NativeReader);
    }

    #[test]
    fn test_office_capabilities() {
        let info = DocumentInfo::new("m2".into(), "report.docx".into(), MaterialType::Word, 2048);
        assert!(!info.is_searchable);
        assert!(!info.supports_position);
        assert!(info.supports_anchor); // Material-level anchor only
        assert_eq!(info.preview_mode, PreviewMode::PlatformPreview);
    }

    #[test]
    fn test_unknown_capabilities() {
        let info = DocumentInfo::new("m3".into(), "file.bin".into(), MaterialType::Unknown, 0);
        assert!(!info.is_searchable);
        assert!(!info.supports_position);
        assert!(!info.supports_anchor);
    }

    #[test]
    fn test_text_capabilities() {
        let info = DocumentInfo::new("m4".into(), "notes.txt".into(), MaterialType::Text, 512);
        assert!(info.is_searchable);
        assert!(info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::NativeReader);
    }

    #[test]
    fn test_pdf_capabilities() {
        let info = DocumentInfo::new("m5".into(), "book.pdf".into(), MaterialType::Pdf, 2048);
        assert!(info.is_searchable);
        assert!(info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::PlatformPreview);
    }

    #[test]
    fn test_image_capabilities() {
        let info = DocumentInfo::new("m6".into(), "photo.jpg".into(), MaterialType::Image, 1024);
        assert!(!info.is_searchable);
        assert!(info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::NativeReader);
    }

    #[test]
    fn test_epub_capabilities() {
        let info = DocumentInfo::new("m7".into(), "book.epub".into(), MaterialType::Epub, 4096);
        assert!(info.is_searchable);
        assert!(info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::NativeReader);
    }

    #[test]
    fn test_excel_capabilities() {
        let info = DocumentInfo::new("m8".into(), "sheet.xlsx".into(), MaterialType::Excel, 2048);
        assert!(!info.is_searchable);
        assert!(!info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::PlatformPreview);
    }

    #[test]
    fn test_powerpoint_capabilities() {
        let info = DocumentInfo::new("m9".into(), "slides.pptx".into(), MaterialType::PowerPoint, 4096);
        assert!(!info.is_searchable);
        assert!(!info.supports_position);
        assert!(info.supports_anchor);
        assert_eq!(info.preview_mode, PreviewMode::ExternalOpen);
    }

    #[test]
    fn test_all_stats_fields_start_none() {
        let info = DocumentInfo::new("m10".into(), "any".into(), MaterialType::Markdown, 100);
        assert_eq!(info.page_count, None);
        assert_eq!(info.word_count, None);
        assert_eq!(info.char_count, None);
        assert_eq!(info.line_count, None);
        assert_eq!(info.image_width, None);
        assert_eq!(info.image_height, None);
        assert_eq!(info.image_format, None);
        assert_eq!(info.epub_chapter_count, None);
    }

    #[test]
    fn test_file_size_is_set() {
        let info = DocumentInfo::new("m11".into(), "f".into(), MaterialType::Markdown, 12345);
        assert_eq!(info.file_size, 12345);
    }

    #[test]
    fn test_material_type_matches_input() {
        let info = DocumentInfo::new("m12".into(), "t".into(), MaterialType::Pdf, 0);
        assert_eq!(info.material_type, MaterialType::Pdf);
    }
}