docgen-core 0.8.0

use serde_yml::Value;

/// Result of splitting frontmatter from a markdown document.
#[derive(Debug, Clone, PartialEq)]
pub struct Parsed {
    pub frontmatter: Value,
    pub body: String,
}

/// Split an optional leading `---`-delimited YAML frontmatter block from the body.
/// On malformed YAML, frontmatter is `Value::Null` and the whole input is the body.
/// Handles both LF and CRLF line endings, empty frontmatter blocks, and requires the
/// closing fence to be a line containing exactly `---` (ignoring trailing whitespace).
pub fn parse_frontmatter(raw: &str) -> Parsed {
    parse_frontmatter_checked(raw).0
}

/// Like [`parse_frontmatter`], but also reports the YAML error message when a
/// frontmatter block *exists* but fails to parse (the `Parsed` still carries
/// `Value::Null`, so behavior is identical for callers that ignore the error).
/// An absent or empty frontmatter block is not an error.
pub fn parse_frontmatter_checked(raw: &str) -> (Parsed, Option<String>) {
    let input = raw.strip_prefix('\u{feff}').unwrap_or(raw);

    // Match an opening `---` fence followed by a line break (LF or CRLF).
    let after_open = input
        .strip_prefix("---\n")
        .or_else(|| input.strip_prefix("---\r\n"));

    if let Some(rest) = after_open {
        // Walk line by line looking for a closing fence that is exactly `---`.
        let mut offset = 0usize;
        for line in rest.split_inclusive('\n') {
            let trimmed = line.trim_end_matches('\n').trim_end_matches('\r');
            if trimmed.trim_end() == "---" {
                let yaml = &rest[..offset];
                let after = &rest[offset + line.len()..];
                let (frontmatter, err) = parse_yaml(yaml);
                return (
                    Parsed {
                        frontmatter,
                        body: after.to_string(),
                    },
                    err,
                );
            }
            offset += line.len();
        }
        // Also handle a closing fence with no trailing newline (EOF).
        let last = &rest[offset..];
        if last.trim_end_matches('\r').trim_end() == "---" {
            let yaml = &rest[..offset];
            let (frontmatter, err) = parse_yaml(yaml);
            return (
                Parsed {
                    frontmatter,
                    body: String::new(),
                },
                err,
            );
        }
    }

    (
        Parsed {
            frontmatter: Value::Null,
            body: input.to_string(),
        },
        None,
    )
}

/// Parse a frontmatter YAML block. On failure: `(Null, Some(error message))`.
/// An empty/whitespace-only block is `Null` with no error.
fn parse_yaml(yaml: &str) -> (Value, Option<String>) {
    if yaml.trim().is_empty() {
        return (Value::Null, None);
    }
    match serde_yml::from_str(yaml) {
        Ok(v) => (v, None),
        Err(e) => (Value::Null, Some(e.to_string())),
    }
}

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

    #[test]
    fn parses_yaml_frontmatter_and_body() {
        let raw = "---\ntitle: Hello\n---\n# Body\n";
        let parsed = parse_frontmatter(raw);
        assert_eq!(parsed.frontmatter["title"].as_str(), Some("Hello"));
        assert_eq!(parsed.body, "# Body\n");
    }

    #[test]
    fn no_frontmatter_returns_null_and_full_body() {
        let raw = "# Just body\n";
        let parsed = parse_frontmatter(raw);
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, "# Just body\n");
    }

    #[test]
    fn parses_crlf_frontmatter() {
        let raw = "---\r\ntitle: X\r\n---\r\nbody\r\n";
        let parsed = parse_frontmatter(raw);
        assert_eq!(parsed.frontmatter["title"].as_str(), Some("X"));
        assert_eq!(parsed.body, "body\r\n");
    }

    #[test]
    fn parses_empty_frontmatter_block() {
        let raw = "---\n---\nbody\n";
        let parsed = parse_frontmatter(raw);
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, "body\n");
    }

    #[test]
    fn longer_dash_run_is_not_a_closing_fence() {
        // A `----` line is not a bare `---` fence; it must not be treated as the close,
        // and no stray dash should leak into the body.
        let raw = "---\ntitle: X\n----\nbody\n";
        let parsed = parse_frontmatter(raw);
        // No valid closing fence -> whole input is body, frontmatter null.
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, raw);
    }

    #[test]
    fn malformed_yaml_falls_back_to_null_with_body() {
        let raw = "---\n: not: valid: yaml\n---\nbody\n";
        let parsed = parse_frontmatter(raw);
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, "body\n");
    }

    #[test]
    fn unterminated_block_returns_full_input_as_body() {
        let raw = "---\ntitle: X\n";
        let parsed = parse_frontmatter(raw);
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, raw);
    }

    #[test]
    fn checked_reports_no_error_for_valid_frontmatter() {
        let (parsed, err) = parse_frontmatter_checked("---\ntitle: Hello\n---\nbody\n");
        assert_eq!(parsed.frontmatter["title"].as_str(), Some("Hello"));
        assert!(err.is_none());
    }

    #[test]
    fn checked_reports_yaml_error_for_malformed_frontmatter() {
        let (parsed, err) = parse_frontmatter_checked("---\n: not: valid: yaml\n---\nbody\n");
        assert!(parsed.frontmatter.is_null());
        assert_eq!(parsed.body, "body\n");
        assert!(
            err.is_some(),
            "malformed YAML must surface an error message"
        );
        assert!(!err.unwrap().is_empty());
    }

    #[test]
    fn checked_reports_no_error_when_frontmatter_absent_or_empty() {
        let (parsed, err) = parse_frontmatter_checked("# Just body\n");
        assert!(parsed.frontmatter.is_null());
        assert!(err.is_none());

        let (parsed, err) = parse_frontmatter_checked("---\n---\nbody\n");
        assert!(parsed.frontmatter.is_null());
        assert!(err.is_none());
    }

    #[test]
    fn strips_leading_bom() {
        let raw = "\u{feff}---\ntitle: X\n---\nbody\n";
        let parsed = parse_frontmatter(raw);
        assert_eq!(parsed.frontmatter["title"].as_str(), Some("X"));
        assert_eq!(parsed.body, "body\n");
    }
}