snapper-fmt 0.9.1

use regex::Regex;
use std::sync::LazyLock;

use crate::parser::{FormatParser, Region, flush_prose};

/// Match `.. code-block:: LANG` or `.. sourcecode:: LANG` (or `.. code:: LANG`).
static CODE_DIRECTIVE_RE: LazyLock<Regex> = LazyLock::new(|| {
    Regex::new(r"^\s*\.\.\s+(?:code-block|sourcecode|code)::\s*([A-Za-z0-9_+.\-]+)?\s*$").unwrap()
});

pub struct RstParser;

impl FormatParser for RstParser {
    fn parse(&self, input: &str) -> Vec<Region> {
        parse_line_based(input)
    }
}

/// Line-based RST parser. Handles directives, literal blocks, sections,
/// field lists, comments, and tables as structure regions.
fn parse_line_based(input: &str) -> Vec<Region> {
    let mut regions = Vec::new();
    let mut current_prose = String::new();
    let mut in_literal_block = false;
    let mut literal_indent: usize = 0;
    let mut in_directive = false;
    let mut directive_indent: usize = 0;
    let mut pragma_off = false;

    // Code-block directive bookkeeping. Mutually exclusive with `in_directive`.
    let mut in_code_block = false;
    let mut code_indent: usize = 0;
    let mut code_lang: Option<String> = None;
    let mut code_header = String::new();
    let mut code_body = String::new();
    let mut code_footer_blanks = String::new();

    let lines: Vec<&str> = input.lines().collect();
    let total = lines.len();
    let mut i = 0;

    while i < total {
        let line = lines[i];

        // Pragma check; inside a code-block directive the per-language
        // reflow path handles pragmas instead.
        if !in_code_block {
            if let Some(on) = super::check_pragma(line) {
                flush_prose(&mut current_prose, &mut regions);
                pragma_off = !on;
                regions.push(Region::Structure(format!("{line}\n")));
                i += 1;
                continue;
            }

            if pragma_off {
                flush_prose(&mut current_prose, &mut regions);
                regions.push(Region::Structure(format!("{line}\n")));
                i += 1;
                continue;
            }
        }

        // Inside an rst code-block directive body.
        // The body consists of lines indented past `code_indent`, plus
        // interior blank lines. The block ends at a non-blank line whose
        // indent drops below `code_indent`.
        if in_code_block {
            let leading = line.len() - line.trim_start().len();
            if line.trim().is_empty() {
                // Could be interior blank or end-of-block; buffer and look ahead.
                code_footer_blanks.push_str(line);
                code_footer_blanks.push('\n');
                i += 1;
                continue;
            }
            if leading >= code_indent {
                // Promote any buffered interior blanks into the body.
                if !code_footer_blanks.is_empty() {
                    code_body.push_str(&code_footer_blanks);
                    code_footer_blanks.clear();
                }
                // Strip the directive's option indent if present? RST options
                // are keyed `:option: value` at code_indent before the blank
                // line. We've already passed those into the body verbatim
                // since they look like normal indented lines; harmless.
                code_body.push_str(line);
                code_body.push('\n');
                i += 1;
                continue;
            }
            // Less-indented non-blank line: close the code block.
            in_code_block = false;
            regions.push(Region::Code {
                lang: code_lang.take(),
                header: std::mem::take(&mut code_header),
                body: std::mem::take(&mut code_body),
                footer: std::mem::take(&mut code_footer_blanks),
            });
            // Fall through to reprocess this line as normal.
        }

        // Inside literal block
        if in_literal_block {
            let leading = line.len() - line.trim_start().len();
            if line.trim().is_empty() || leading >= literal_indent {
                regions.push(Region::Structure(format!("{line}\n")));
                i += 1;
                continue;
            }
            in_literal_block = false;
        }

        // Inside directive body
        if in_directive {
            let leading = line.len() - line.trim_start().len();
            if line.trim().is_empty() || leading >= directive_indent {
                regions.push(Region::Structure(format!("{line}\n")));
                i += 1;
                continue;
            }
            in_directive = false;
        }

        // Blank line
        if line.trim().is_empty() {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::BlankLines(format!("{line}\n")));
            i += 1;
            continue;
        }

        // RST code-block directive (.. code-block:: LANG)
        if let Some(caps) = CODE_DIRECTIVE_RE.captures(line) {
            flush_prose(&mut current_prose, &mut regions);
            code_lang = caps.get(1).map(|m| m.as_str().to_string());
            code_header = format!("{line}\n");
            // Body indent: directive_indent + 3 spaces is the rst convention;
            // be liberal and accept any deeper indent of the first body line.
            let leading = line.len() - line.trim_start().len();
            code_indent = leading + 3;
            code_body.clear();
            code_footer_blanks.clear();
            in_code_block = true;
            i += 1;
            continue;
        }

        // RST directive (.. something::)
        let trimmed = line.trim_start();
        if trimmed.starts_with(".. ") && trimmed.contains("::") {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            let leading = line.len() - trimmed.len();
            directive_indent = leading + 3;
            in_directive = true;
            i += 1;
            continue;
        }

        // RST comment (.. without directive)
        if trimmed.starts_with(".. ") && !trimmed.contains("::") {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            i += 1;
            continue;
        }

        // Section underline
        if is_underline(line) {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            i += 1;
            continue;
        }

        // Section title (next line is underline)
        if i + 1 < total && is_underline(lines[i + 1]) {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            i += 1;
            continue;
        }

        // Field list (:field: value)
        if trimmed.starts_with(':') && trimmed.len() > 2 {
            if let Some(colon_pos) = trimmed[1..].find(':') {
                if colon_pos > 0 && colon_pos < trimmed.len() - 2 {
                    flush_prose(&mut current_prose, &mut regions);
                    regions.push(Region::Structure(format!("{line}\n")));
                    i += 1;
                    continue;
                }
            }
        }

        // Literal block intro (line ending with ::)
        if trimmed.ends_with("::") {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            // Find indent of next non-blank line
            let mut j = i + 1;
            while j < total && lines[j].trim().is_empty() {
                j += 1;
            }
            if j < total {
                let next_indent = lines[j].len() - lines[j].trim_start().len();
                if next_indent > 0 {
                    literal_indent = next_indent;
                    in_literal_block = true;
                }
            }
            i += 1;
            continue;
        }

        // Grid/simple table rows
        if trimmed.starts_with('|') || trimmed.starts_with('+') {
            flush_prose(&mut current_prose, &mut regions);
            regions.push(Region::Structure(format!("{line}\n")));
            i += 1;
            continue;
        }

        // Regular prose
        if !current_prose.is_empty() {
            current_prose.push(' ');
        }
        current_prose.push_str(trimmed);
        i += 1;
    }

    flush_prose(&mut current_prose, &mut regions);
    if in_code_block {
        regions.push(Region::Code {
            lang: code_lang.take(),
            header: std::mem::take(&mut code_header),
            body: std::mem::take(&mut code_body),
            footer: std::mem::take(&mut code_footer_blanks),
        });
    }
    regions
}

/// Check if a line is a section underline (2+ repeated punctuation chars).
fn is_underline(line: &str) -> bool {
    let trimmed = line.trim();
    if trimmed.len() < 2 {
        return false;
    }
    let first = trimmed.as_bytes()[0];
    matches!(first, b'=' | b'-' | b'~' | b'^' | b'"' | b'#' | b'*' | b'+')
        && trimmed.bytes().all(|b| b == first)
}

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

    #[test]
    fn simple_prose() {
        let input = "Hello world. This is a test.\nAnother line here.";
        let regions = RstParser.parse(input);
        assert!(
            regions
                .iter()
                .any(|r| matches!(r, Region::Prose(s) if s.contains("Hello world.")))
        );
    }

    #[test]
    fn directive_preserved() {
        let input = "Some prose.\n\n.. code-block:: python\n\n   print('hello')\n\nMore prose.";
        let regions = RstParser.parse(input);
        let prose_count = regions
            .iter()
            .filter(|r| matches!(r, Region::Prose(_)))
            .count();
        assert_eq!(prose_count, 2);
        // The code block surfaces as Region::Code with lang=python.
        let code = regions.iter().find_map(|r| match r {
            Region::Code { lang, body, .. } => Some((lang.clone(), body.clone())),
            _ => None,
        });
        let (lang, body) = code.expect("expected one Region::Code");
        assert_eq!(lang.as_deref(), Some("python"));
        assert!(body.contains("print('hello')"));
    }

    #[test]
    fn section_title_preserved() {
        let input = "My Title\n========\n\nSome text here.";
        let regions = RstParser.parse(input);
        assert!(
            regions
                .iter()
                .any(|r| matches!(r, Region::Structure(s) if s.contains("My Title")))
        );
        assert!(
            regions
                .iter()
                .any(|r| matches!(r, Region::Structure(s) if s.contains("====")))
        );
    }

    #[test]
    fn literal_block_preserved() {
        let input = "Example::\n\n   some code\n   more code\n\nBack to prose.";
        let regions = RstParser.parse(input);
        let structure_count = regions
            .iter()
            .filter(|r| matches!(r, Region::Structure(_)))
            .count();
        assert!(structure_count >= 3);
    }

    #[test]
    fn field_list_preserved() {
        let input = ":Author: Someone\n:Date: 2026\n\nParagraph text.";
        let regions = RstParser.parse(input);
        assert!(
            regions
                .iter()
                .any(|r| matches!(r, Region::Structure(s) if s.contains("Author")))
        );
    }
}