pipedash-plugin-gitlab 0.8.0

use chrono::{
    DateTime,
    Utc,
};
use pipedash_plugin_api::{
    LineMarker,
    LogDialect,
    ParsedLine,
};

const HEADER_LEN: usize = 32;
const DATETIME_LEN: usize = 27;
const SECTION_START: &[u8] = b"section_start:";
const SECTION_END: &[u8] = b"section_end:";
const CLEAR_LINE: &[u8] = b"\r\x1b[0K";

pub struct GitlabLogDialect;

struct Marker<'a> {
    start: bool,
    unix: i64,
    name: &'a [u8],
    collapsed: bool,
    rest: &'a [u8],
}

impl LogDialect for GitlabLogDialect {
    fn parse<'a>(&self, line: &'a [u8]) -> ParsedLine<'a> {
        let (header_time, rest) = split_header(line);
        let plain = ParsedLine {
            timestamp: header_time,
            marker: LineMarker::Plain,
            body: rest,
        };

        let mut text = skip_csi(rest);
        let mut started = None;
        let mut ended = None;
        let mut body = rest;
        while let Some(marker) = parse_marker(text) {
            if marker.start {
                started = Some((marker.name, marker.unix, marker.collapsed));
            } else {
                ended = Some(marker.unix);
            }
            body = trim_carriage_returns(marker.rest);
            text = skip_csi(marker.rest);
        }

        if let Some((name, unix, collapsed)) = started {
            return ParsedLine {
                timestamp: header_time.or_else(|| from_unix(unix)),
                marker: LineMarker::SectionStart { open: !collapsed },
                body: if is_blank(body) { name } else { body },
            };
        }
        let Some(unix) = ended else {
            return plain;
        };
        ParsedLine {
            timestamp: header_time.or_else(|| from_unix(unix)),
            marker: LineMarker::GroupEnd,
            body: if is_blank(body) { &[] } else { body },
        }
    }
}

fn split_header(line: &[u8]) -> (Option<DateTime<Utc>>, &[u8]) {
    let Some(header) = line.get(..HEADER_LEN) else {
        return (None, line);
    };
    let well_formed = header[DATETIME_LEN - 1] == b'Z'
        && header[DATETIME_LEN] == b' '
        && header[DATETIME_LEN + 1..DATETIME_LEN + 3]
            .iter()
            .all(|b| b.is_ascii_digit() || (b'a'..=b'f').contains(b))
        && matches!(header[DATETIME_LEN + 3], b'E' | b'O')
        && matches!(header[DATETIME_LEN + 4], b'+' | b' ');
    let time = well_formed
        .then(|| std::str::from_utf8(&header[..DATETIME_LEN]).ok())
        .flatten()
        .and_then(|text| DateTime::parse_from_rfc3339(text).ok());
    match time {
        Some(time) => (Some(time.with_timezone(&Utc)), &line[HEADER_LEN..]),
        None => (None, line),
    }
}

fn parse_marker(text: &[u8]) -> Option<Marker<'_>> {
    let (start, rest) = match text.strip_prefix(SECTION_START) {
        Some(rest) => (true, rest),
        None => (false, text.strip_prefix(SECTION_END)?),
    };
    let digits = rest.iter().take_while(|b| b.is_ascii_digit()).count();
    if digits == 0 || rest.get(digits) != Some(&b':') {
        return None;
    }
    let unix = std::str::from_utf8(&rest[..digits]).ok()?.parse().ok()?;
    let rest = &rest[digits + 1..];
    let name_len = rest
        .iter()
        .take_while(|b| b.is_ascii_alphanumeric() || matches!(**b, b'_' | b'.' | b'-'))
        .count();
    if name_len == 0 {
        return None;
    }
    let (name, mut rest) = rest.split_at(name_len);
    let mut collapsed = false;
    if rest.first() == Some(&b'[') {
        let close = rest.iter().position(|b| *b == b']')?;
        collapsed = rest[1..close]
            .split(|b| *b == b',')
            .any(|option| option.trim_ascii() == b"collapsed=true");
        rest = &rest[close + 1..];
    }
    Some(Marker {
        start,
        unix,
        name,
        collapsed,
        rest: rest.strip_prefix(CLEAR_LINE)?,
    })
}

fn from_unix(seconds: i64) -> Option<DateTime<Utc>> {
    DateTime::from_timestamp(seconds, 0)
}

fn skip_csi(mut text: &[u8]) -> &[u8] {
    while let Some(rest) = text.strip_prefix(b"\x1b[") {
        let Some(end) = rest.iter().position(|b| (0x40..=0x7e).contains(b)) else {
            break;
        };
        text = &rest[end + 1..];
    }
    text
}

fn is_blank(mut text: &[u8]) -> bool {
    loop {
        text = skip_csi(text);
        match text.split_first() {
            None => return true,
            Some((byte, rest)) if byte.is_ascii_whitespace() => text = rest,
            Some(_) => return false,
        }
    }
}

fn trim_carriage_returns(mut text: &[u8]) -> &[u8] {
    while let Some(rest) = text.strip_suffix(b"\r") {
        text = rest;
    }
    text
}

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

    fn parse(line: &[u8]) -> ParsedLine<'_> {
        GitlabLogDialect.parse(line)
    }

    fn millis(parsed: &ParsedLine<'_>) -> Option<i64> {
        parsed.timestamp.map(|time| time.timestamp_millis())
    }

    #[test]
    fn runner_timestamp_header_is_parsed_and_stripped() {
        for (line, body) in [
            (
                b"2024-05-22T08:46:32.123456Z 00O Running with gitlab-runner 17.0.0".as_slice(),
                b"Running with gitlab-runner 17.0.0".as_slice(),
            ),
            (b"2024-05-22T08:46:32.123456Z 01E boom", b"boom"),
            (
                b"2024-05-22T08:46:32.123456Z 00O+\x1b[0K\x1b[36;1mPreparing\x1b[0;m",
                b"\x1b[0K\x1b[36;1mPreparing\x1b[0;m",
            ),
            (b"2024-05-22T08:46:32.123456Z 00O ", b""),
        ] {
            let parsed = parse(line);
            assert_eq!(parsed.marker, LineMarker::Plain, "{line:?}");
            assert_eq!(parsed.body, body, "{line:?}");
            assert_eq!(millis(&parsed), Some(1_716_367_592_123), "{line:?}");
        }
    }

    #[test]
    fn lines_without_a_valid_header_keep_their_bytes() {
        for line in [
            b"plain output".as_slice(),
            b"",
            b"2024-05-22T08:46:32.123456Z 00O",
            b"2024-05-22T08:46:32.123456Z 0zO text",
            b"2024-05-22T08:46:32.123456Z 00X text",
            b"2024-05-22T08:46:32.123456Z 00O?text",
            b"2024-05-22T08:46:32.123Z 00O text 12345",
            b"2024-13-22T08:46:32.123456Z 00O text",
        ] {
            let parsed = parse(line);
            assert_eq!(parsed.timestamp, None, "{line:?}");
            assert_eq!(parsed.marker, LineMarker::Plain, "{line:?}");
            assert_eq!(parsed.body, line);
        }
    }

    #[test]
    fn section_start_opens_a_section_with_its_header() {
        let parsed = parse(
            b"section_start:1700000000:prepare_executor\r\x1b[0K\x1b[0K\x1b[36;1mPreparing the \"docker\" executor\x1b[0;m",
        );
        assert_eq!(parsed.marker, LineMarker::SectionStart { open: true });
        assert_eq!(
            parsed.body,
            b"\x1b[0K\x1b[36;1mPreparing the \"docker\" executor\x1b[0;m"
        );
        assert_eq!(millis(&parsed), Some(1_700_000_000_000));
    }

    #[test]
    fn collapsed_option_closes_the_section() {
        let cases: [(&[u8], bool); 5] = [
            (b"section_start:1:a[collapsed=true]\r\x1b[0KA", false),
            (b"section_start:1:a[collapsed=false]\r\x1b[0KA", true),
            (
                b"section_start:1:a[hide_duration=true, collapsed=true]\r\x1b[0KA",
                false,
            ),
            (
                b"section_start:1:a[collapsed=true,hide_duration=true]\r\x1b[0KA",
                false,
            ),
            (b"section_start:1:a[hide_duration=true]\r\x1b[0KA", true),
        ];
        for (line, open) in cases {
            let parsed = parse(line);
            assert_eq!(parsed.marker, LineMarker::SectionStart { open }, "{line:?}");
            assert_eq!(parsed.body, b"A", "{line:?}");
        }
    }

    #[test]
    fn section_start_without_a_header_is_titled_by_its_name() {
        let parsed =
            parse(b"2024-05-22T08:46:32.123456Z 00O section_start:1:prepare_script\r\x1b[0K");
        assert_eq!(parsed.marker, LineMarker::SectionStart { open: true });
        assert_eq!(parsed.body, b"prepare_script");
        assert_eq!(millis(&parsed), Some(1_716_367_592_123));
    }

    #[test]
    fn section_end_closes_the_section() {
        let parsed = parse(b"section_end:1700000007:step_script\r\x1b[0K");
        assert_eq!(parsed.marker, LineMarker::GroupEnd);
        assert_eq!(parsed.body, b"");
        assert_eq!(millis(&parsed), Some(1_700_000_007_000));

        let stamped = parse(b"2024-05-22T08:46:32.123456Z 00O \x1b[0Ksection_end:1:a\r\x1b[0K\r");
        assert_eq!(stamped.marker, LineMarker::GroupEnd);
        assert_eq!(millis(&stamped), Some(1_716_367_592_123));
    }

    #[test]
    fn end_and_start_on_one_line_start_the_next_section() {
        let parsed = parse(
            b"section_end:1700000005:prepare_executor\r\x1b[0Ksection_start:1700000006:prepare_script\r\x1b[0KPreparing environment\r",
        );
        assert_eq!(parsed.marker, LineMarker::SectionStart { open: true });
        assert_eq!(parsed.body, b"Preparing environment");
        assert_eq!(millis(&parsed), Some(1_700_000_006_000));
    }

    #[test]
    fn text_after_a_section_end_closes_the_section_and_stays_a_line() {
        let parsed = parse(
            b"section_end:1700000010:cleanup_file_variables\r\x1b[0K\x1b[32;1mJob succeeded\x1b[0;m",
        );
        assert_eq!(parsed.marker, LineMarker::GroupEnd);
        assert_eq!(parsed.body, b"\x1b[32;1mJob succeeded\x1b[0;m");
    }

    #[test]
    fn markers_follow_color_resets_but_not_output() {
        let parsed = parse(b"\x1b[0;msection_start:1:build\r\x1b[0KBuild");
        assert_eq!(parsed.marker, LineMarker::SectionStart { open: true });
        assert_eq!(parsed.body, b"Build");

        let output = b"echo section_start:1:build\r\x1b[0KBuild";
        assert_eq!(parse(output).marker, LineMarker::Plain);
        assert_eq!(parse(output).body, output);
    }

    #[test]
    fn malformed_markers_are_output() {
        for line in [
            b"section_start:1:my_$ection\r\x1b[0Khello".as_slice(),
            b"section_start:1:<a_tag>\r\x1b[0Khello",
            b"section_start:1:name hello",
            b"section_start:1:name\rhello",
            b"section_start::name\r\x1b[0Khello",
            b"section_start:1:\r\x1b[0Khello",
            b"section_start:1:name[collapsed=true\r\x1b[0Khello",
            b"section_end:x:name\r\x1b[0K",
        ] {
            let parsed = parse(line);
            assert_eq!(parsed.marker, LineMarker::Plain, "{line:?}");
            assert_eq!(parsed.body, line);
        }
    }
}