rs-chunks 0.6.4

use serde_json::json;

use super::common::{
    current_section_heading, extract_heading_text, heading_level_txt, heading_path_strings,
    parse_txt_blocks, update_heading_stack, ChunkRecordInput, ContentType,
};

struct BlockUnit {
    content: String,
    section_heading: Option<String>,
    heading_path: Vec<String>,
}

pub fn build_sliding_window_chunks(
    bytes: &[u8],
    window_size: usize,
    overlap: usize,
) -> Result<Vec<ChunkRecordInput>, String> {
    if window_size == 0 {
        return Err("window_size must be greater than 0".to_string());
    }
    if overlap >= window_size {
        return Err("overlap must be less than window_size".to_string());
    }

    let text = crate::text_encoding::decode_text(bytes).0;
    // Empty input is not a failure. A blank or whitespace-only document parsed
    // perfectly well; it simply has nothing to chunk, so it returns `[]` like
    // docx/ppt/xlsx always have (TECH_DEBT T6). Reserving errors for genuine
    // parse failures is also what lets `epub::extract` stop swallowing them.
    if text.trim().is_empty() {
        return Ok(Vec::new());
    }

    let blocks = parse_txt_blocks(&text);
    let total = blocks.len();
    let mut heading_stack: Vec<(u8, String)> = Vec::new();
    let mut units: Vec<BlockUnit> = Vec::new();

    for block in &blocks {
        let content = if block.content_type == ContentType::HeadingSection {
            let level = heading_level_txt(&block.content);
            let text = extract_heading_text(&block.content);
            update_heading_stack(&mut heading_stack, level, text.clone());
            text
        } else {
            block.content.trim().to_string()
        };
        if content.is_empty() {
            continue;
        }
        units.push(BlockUnit {
            content,
            section_heading: current_section_heading(&heading_stack),
            heading_path: heading_path_strings(&heading_stack),
        });
    }

    // Same contract as the structural builders (TECH_DEBT T6): a document that
    // parsed cleanly but holds no content blocks is EMPTY, not broken. Returning
    // an error here is what made epub's per-chapter swallow necessary — an
    // image-only cover page hit this path (L14).
    if units.is_empty() {
        return Ok(Vec::new());
    }

    let step = window_size - overlap;
    let mut result: Vec<ChunkRecordInput> = Vec::new();
    let mut start = 0usize;
    let mut window_index = 0usize;

    loop {
        let end = (start + window_size).min(units.len());
        let window = &units[start..end];
        let content = window
            .iter()
            .map(|u| u.content.as_str())
            .collect::<Vec<_>>()
            .join("\n\n");
        if !content.is_empty() {
            result.push(ChunkRecordInput {
                content_type: ContentType::SlidingWindow,
                content,
                metadata: json!({
                    "window_size":      window_size,
                    "overlap":          overlap,
                    "window_index":     window_index,
                    "paragraph_range":  [start, end.saturating_sub(1)],
                    "paragraph_count":  window.len(),
                    "section_heading":  window[0].section_heading,
                    "heading_path":     window[0].heading_path,
                    "chunk_index":      window_index,
                    "document_metadata": { "source_type": "txt", "total_input_blocks": total }
                }),
            });
            window_index += 1;
        }
        if end >= units.len() {
            break;
        }
        start += step;
    }

    // Empty is not a failure (TECH_DEBT T6): the document parsed, this mode
    // simply produced nothing. Returning `[]` keeps every mode consistent with
    // docx/ppt/xlsx and lets epub distinguish an empty chapter from a broken
    // one without swallowing errors (L14).
    if result.is_empty() {
        return Ok(Vec::new());
    }
    Ok(result)
}