ebustl-parser 0.7.0

//! Types and validation for GSI and TTI fields that are stored as they appear in the file.
//!
//! The stored values are kept so that files are written back unchanged, and so that a file with
//! an invalid value in one of these fields can still be read. The accessors of [`GsiBlock`] and
//! [`TtiBlock`] validate a single field at a time, using the functions here, and tell a blank
//! field apart from an invalid one.
//!
//! [`GsiBlock`]: crate::GsiBlock
//! [`TtiBlock`]: crate::TtiBlock

use thiserror::Error;

use crate::{DiskFormatCode, Time};

/// A field value that is invalid according to the spec, with the value as stored in the file.
///
/// `T` is the type of the stored value: `String` for fields in the GSI block, `u8` for single
/// byte fields and [`Time`] for time codes in the TTI blocks.
///
/// # Example
///
/// ```rust
/// let stl = ebustl_parser::parse_stl_from_file("stls/test.stl").unwrap();
/// let dfc = stl.gsi.disk_format_code();
/// for tti in &stl.ttis {
///     // Use the time code as stored, even if it is outside the ranges in the spec
///     let tci = tti.time_code_in(dfc).unwrap_or_else(|err| {
///         eprintln!("{err}");
///         err.into_value()
///     });
///     assert_eq!(10, tci.hours);
/// }
/// ```
#[derive(Debug, Clone, PartialEq, Eq, Error)]
#[error("Invalid {field}: {msg}")]
pub struct InvalidFieldValue<T> {
    field: &'static str,
    msg: String,
    value: T,
}

impl<T> InvalidFieldValue<T> {
    pub(crate) fn new(field: &'static str, msg: String, value: T) -> Self {
        Self { field, msg, value }
    }

    /// Mnemonic of the field in the spec, e.g. "CD".
    pub fn field(&self) -> &'static str {
        self.field
    }

    /// Why the value is invalid.
    pub fn msg(&self) -> &str {
        &self.msg
    }

    /// The value as stored in the file.
    pub fn value(&self) -> &T {
        &self.value
    }

    /// Takes the value as stored in the file.
    pub fn into_value(self) -> T {
        self.value
    }
}

/// Language code of the subtitle list (LC), Tech 3264 Appendix 3.
///
/// # Example
///
/// ```rust
/// use ebustl_parser::LanguageCode;
///
/// assert_eq!(Some("Swedish"), LanguageCode(0x28).name());
/// // Reserved for national assignment
/// assert_eq!(None, LanguageCode(0x30).name());
/// ```
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct LanguageCode(pub u8);

impl LanguageCode {
    /// The language as named in the spec, or `None` for a code that is unassigned or reserved
    /// for national assignment.
    pub fn name(self) -> Option<&'static str> {
        LATIN_LANGUAGES
            .get(usize::from(self.0))
            .copied()
            .or_else(|| {
                let index = usize::from(self.0).checked_sub(FIRST_OTHER_LANGUAGE)?;
                OTHER_LANGUAGES.get(index).copied()
            })
    }
}

/// European languages written in Latin-based alphabets, codes 00h-2Bh
const LATIN_LANGUAGES: [&str; 0x2C] = [
    "Unknown/not applicable",
    "Albanian",
    "Breton",
    "Catalan",
    "Croatian",
    "Welsh",
    "Czech",
    "Danish",
    "German",
    "English",
    "Spanish",
    "Esperanto",
    "Estonian",
    "Basque",
    "Faroese",
    "French",
    "Frisian",
    "Irish",
    "Gaelic",
    "Galician",
    "Icelandic",
    "Italian",
    "Lappish",
    "Latin",
    "Latvian",
    "Luxembourgian",
    "Lithuanian",
    "Hungarian",
    "Maltese",
    "Dutch",
    "Norwegian",
    "Occitan",
    "Polish",
    "Portugese",
    "Romanian",
    "Romansh",
    "Serbian",
    "Slovak",
    "Slovenian",
    "Finnish",
    "Swedish",
    "Turkish",
    "Flemish",
    "Wallon",
];

/// Code of the first entry in [`OTHER_LANGUAGES`]
const FIRST_OTHER_LANGUAGE: usize = 0x45;

/// Other languages, codes 45h-7Fh. The spec lists them from 7Fh downwards.
const OTHER_LANGUAGES: [&str; 0x80 - FIRST_OTHER_LANGUAGE] = [
    "Zulu",
    "Vietnamese",
    "Uzbek",
    "Urdu",
    "Ukrainian",
    "Thai",
    "Telugu",
    "Tatar",
    "Tamil",
    "Tadzhik",
    "Swahili",
    "Sranan Tongo",
    "Somali",
    "Sinhalese",
    "Shona",
    "Serbo-croat",
    "Ruthenian",
    "Russian",
    "Quechua",
    "Pushtu",
    "Punjabi",
    "Persian",
    "Papamiento",
    "Oriya",
    "Nepali",
    "Ndebele",
    "Marathi",
    "Moldavian",
    "Malaysian",
    "Malagasay",
    "Macedonian",
    "Laotian",
    "Korean",
    "Khmer",
    "Kazakh",
    "Kannada",
    "Japanese",
    "Indonesian",
    "Hindi",
    "Hebrew",
    "Hausa",
    "Gurani",
    "Gujurati",
    "Greek",
    "Georgian",
    "Fulani",
    "Dari",
    "Churash",
    "Chinese",
    "Burmese",
    "Bulgarian",
    "Bengali",
    "Bielorussian",
    "Bambora",
    "Azerbaijani",
    "Assamese",
    "Armenian",
    "Arabic",
    "Amharic",
];

/// A date in the GSI block (CD, RD), stored as YYMMDD.
///
/// Two-digit years 80-99 are 1980-1999 and 00-79 are 2000-2079.
///
/// # Example
///
/// ```rust
/// use ebustl_parser::Date;
///
/// let date = Date { year: 2013, month: 10, day: 10 };
/// assert_eq!(10, date.month);
/// ```
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub struct Date {
    /// The year with four digits, e.g. 1991
    pub year: u16,
    /// The month, 1-12
    pub month: u8,
    /// The day of the month, 1-31
    pub day: u8,
}

/// Two-digit years from this one are in the 20th century
const FIRST_YEAR_IN_1900S: u16 = 80;

/// Parses a field of decimal digits only, e.g. no sign.
pub(crate) fn parse_digits(value: &str) -> Option<u32> {
    if value.is_empty() || !value.bytes().all(|b| b.is_ascii_digit()) {
        return None;
    }
    value.parse().ok()
}

/// Returns `None` for a blank field, i.e. only spaces, which is how unused GSI bytes are set.
fn non_blank(value: &str) -> Option<&str> {
    if value.chars().all(|c| c == ' ') {
        None
    } else {
        Some(value)
    }
}

/// Parses a blank field as `Ok(None)`, and other values with `parse`, which returns why the
/// value is invalid.
pub(crate) fn parse_field<'a, T>(
    field: &'static str,
    value: &'a str,
    parse: impl FnOnce(&'a str) -> Result<T, String>,
) -> Result<Option<T>, InvalidFieldValue<String>> {
    match non_blank(value) {
        None => Ok(None),
        Some(value) => parse(value)
            .map(Some)
            .map_err(|msg| InvalidFieldValue::new(field, msg, value.to_string())),
    }
}

pub(crate) fn parse_language_code(value: &str) -> Result<LanguageCode, String> {
    if value.len() != 2 || !value.bytes().all(|b| b.is_ascii_hexdigit()) {
        return Err(format!("{value:?} is not two hexadecimal digits"));
    }
    let code = u8::from_str_radix(value, 16).map_err(|err| err.to_string())?;
    // Appendix 3 defines codes 00h-7Fh
    if code >= 0x80 {
        return Err(format!("{code:02X}h is outside 00h-7Fh"));
    }
    Ok(LanguageCode(code))
}

pub(crate) fn parse_date(value: &str) -> Result<Date, String> {
    let date = two_digit_date(value)
        .ok_or_else(|| format!("{value:?} is not a date in the format YYMMDD"))?;
    if !date_exists(date) {
        return Err(format!("{value:?} is not an existing date"));
    }
    Ok(date)
}

/// Splits a YYMMDD value into a date, without checking that it exists.
fn two_digit_date(value: &str) -> Option<Date> {
    // ASCII only, so that slicing by byte index cannot split a character
    if value.len() != 6 || !value.is_ascii() {
        return None;
    }
    let two_digit_year = u16::try_from(parse_digits(&value[0..2])?).ok()?;
    let month = u8::try_from(parse_digits(&value[2..4])?).ok()?;
    let day = u8::try_from(parse_digits(&value[4..6])?).ok()?;
    let year = if two_digit_year >= FIRST_YEAR_IN_1900S {
        1900 + two_digit_year
    } else {
        2000 + two_digit_year
    };
    Some(Date { year, month, day })
}

fn date_exists(date: Date) -> bool {
    let (Ok(year), Ok(month), Ok(day)) = (
        i16::try_from(date.year),
        i8::try_from(date.month),
        i8::try_from(date.day),
    ) else {
        return false;
    };
    jiff::civil::Date::new(year, month, day).is_ok()
}

pub(crate) fn parse_revision_number(value: &str) -> Result<u8, String> {
    parse_digits(value.trim_matches(' '))
        .and_then(|number| u8::try_from(number).ok())
        .ok_or_else(|| format!("{value:?} is not a number 0-99"))
}

/// Parses a time code stored as HHMMSSFF. Hours may exceed 23, e.g. for a programme that
/// continues past midnight.
pub(crate) fn parse_time_code(value: &str, fps: usize) -> Result<Time, String> {
    let time = split_time_code(value)
        .ok_or_else(|| format!("{value:?} is not a time code in the format HHMMSSFF"))?;
    match time.range_error(fps, false) {
        Some(msg) => Err(msg),
        None => Ok(time),
    }
}

/// Splits a HHMMSSFF value into a time, without checking the ranges.
fn split_time_code(value: &str) -> Option<Time> {
    // ASCII only, so that slicing by byte index cannot split a character
    if value.len() != 8 || !value.is_ascii() {
        return None;
    }
    let element = |range| u8::try_from(parse_digits(&value[range])?).ok();
    Some(Time {
        hours: element(0..2)?,
        minutes: element(2..4)?,
        seconds: element(4..6)?,
        frames: element(6..8)?,
    })
}

/// Checks a Time Code In or Out against the ranges in the spec, see [`Time::is_valid`].
pub(crate) fn checked_time_code(
    field: &'static str,
    time: &Time,
    dfc: &DiskFormatCode,
) -> Result<Time, InvalidFieldValue<Time>> {
    match time.range_error(dfc.fps(), true) {
        Some(msg) => Err(InvalidFieldValue::new(field, msg, *time)),
        None => Ok(*time),
    }
}

pub(crate) fn parse_country_code(value: &str) -> Result<&str, String> {
    if value.len() == 3 && value.bytes().all(|b| b.is_ascii_uppercase()) {
        Ok(value)
    } else {
        Err(format!("{value:?} is not three upper case letters"))
    }
}

/// Free text without the trailing space padding, `None` if the field is blank.
pub(crate) fn text(value: &str) -> Option<&str> {
    non_blank(value).map(|value| value.trim_end_matches(' '))
}

/// Justification Code (JC), the horizontal alignment of a subtitle.
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum Justification {
    /// Positioned by the spaces and control codes in the Text Field
    Unchanged,
    /// Left-justified text
    Left,
    /// Centred text
    Centered,
    /// Right-justified text
    Right,
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::{GsiBlock, parse_stl_from_file};

    fn gsi() -> GsiBlock {
        parse_stl_from_file("stls/test.stl").unwrap().gsi
    }

    /// Asserts that `result` is an error for `field` that carries `value`
    fn assert_invalid<T: std::fmt::Debug, V: std::fmt::Debug + PartialEq>(
        result: Result<T, InvalidFieldValue<V>>,
        field: &str,
        value: V,
    ) {
        let err = result.expect_err("invalid value");
        assert_eq!(field, err.field());
        assert_eq!(&value, err.value());
        assert!(!err.msg().is_empty());
    }

    fn time(hours: u8, minutes: u8, seconds: u8, frames: u8) -> Time {
        Time {
            hours,
            minutes,
            seconds,
            frames,
        }
    }

    #[test]
    fn language_names() {
        assert_eq!(Some("Unknown/not applicable"), LanguageCode(0x00).name());
        assert_eq!(Some("French"), LanguageCode(0x0F).name());
        assert_eq!(Some("Wallon"), LanguageCode(0x2B).name());
        assert_eq!(Some("Zulu"), LanguageCode(0x45).name());
        assert_eq!(Some("Russian"), LanguageCode(0x56).name());
        assert_eq!(Some("Amharic"), LanguageCode(0x7F).name());
    }

    #[test]
    fn unassigned_language_codes_have_no_name() {
        for code in [0x2C, 0x2F, 0x3F, 0x40, 0x44, 0x80, 0xFF] {
            assert_eq!(None, LanguageCode(code).name(), "code {code:02X}");
        }
    }

    #[test]
    fn language_code() {
        let mut gsi = gsi();
        gsi.lc = "0F".to_string();
        assert_eq!(Ok(Some(LanguageCode(0x0F))), gsi.language());
        gsi.lc = "7f".to_string();
        assert_eq!(Ok(Some(LanguageCode(0x7F))), gsi.language());
        gsi.lc = "  ".to_string();
        assert_eq!(Ok(None), gsi.language());
        for lc in ["80", "+F", "F ", "XY", "\0\0"] {
            gsi.lc = lc.to_string();
            assert_invalid(gsi.language(), "LC", lc.to_string());
        }
    }

    #[test]
    fn dates() {
        let mut gsi = gsi();
        gsi.cd = "910215".to_string();
        let date = Date {
            year: 1991,
            month: 2,
            day: 15,
        };
        assert_eq!(Ok(Some(date)), gsi.creation_date());
        gsi.rd = "240229".to_string();
        let date = Date {
            year: 2024,
            month: 2,
            day: 29,
        };
        assert_eq!(Ok(Some(date)), gsi.revision_date());
        gsi.rd = "      ".to_string();
        assert_eq!(Ok(None), gsi.revision_date());
    }

    #[test]
    fn century_of_two_digit_year() {
        let mut gsi = gsi();
        gsi.cd = "790101".to_string();
        assert_eq!(2079, gsi.creation_date().unwrap().unwrap().year);
        gsi.cd = "800101".to_string();
        assert_eq!(1980, gsi.creation_date().unwrap().unwrap().year);
    }

    #[test]
    fn invalid_dates() {
        let mut gsi = gsi();
        for cd in [
            "911341",
            "910230",
            "230229",
            "9102  ",
            "91-2-1",
            "\0\0\0\0\0\0",
            // Six bytes, where slicing by byte index would split 'é'
            "1é234",
        ] {
            gsi.cd = cd.to_string();
            assert_invalid(gsi.creation_date(), "CD", cd.to_string());
        }
    }

    #[test]
    fn revision_number() {
        let mut gsi = gsi();
        gsi.rn = "07".to_string();
        assert_eq!(Ok(Some(7)), gsi.revision_number());
        gsi.rn = " 7".to_string();
        assert_eq!(Ok(Some(7)), gsi.revision_number());
        for rn in ["-1", "+1", "x1"] {
            gsi.rn = rn.to_string();
            assert_invalid(gsi.revision_number(), "RN", rn.to_string());
        }
    }

    #[test]
    fn time_codes() {
        let mut gsi = gsi();
        gsi.tcp = "23595924".to_string();
        assert_eq!(
            Ok(Some(time(23, 59, 59, 24))),
            gsi.time_code_start_of_programme()
        );
        gsi.tcf = "        ".to_string();
        assert_eq!(Ok(None), gsi.time_code_first_in_cue());
        for tcp in [
            "00600000", "00006000", "00000025", "1000000", "10:00:00", "1é00000",
        ] {
            gsi.tcp = tcp.to_string();
            assert_invalid(gsi.time_code_start_of_programme(), "TCP", tcp.to_string());
        }
    }

    #[test]
    fn start_of_programme_and_first_in_cue_allow_hours_past_midnight() {
        let mut gsi = gsi();
        gsi.tcp = "24000000".to_string();
        assert_eq!(
            Ok(Some(time(24, 0, 0, 0))),
            gsi.time_code_start_of_programme()
        );
        gsi.tcf = "99595924".to_string();
        assert_eq!(Ok(Some(time(99, 59, 59, 24))), gsi.time_code_first_in_cue());
    }

    #[test]
    fn time_code_in_and_out() {
        let mut stl = parse_stl_from_file("stls/test.stl").unwrap();
        let tti = &mut stl.ttis[0];
        tti.tci = time(23, 59, 59, 24);
        tti.tco = time(0, 0, 0, 0);
        assert_eq!(Ok(time(23, 59, 59, 24)), tti.time_code_in(&stl.gsi.dfc));
        assert_eq!(Ok(time(0, 0, 0, 0)), tti.time_code_out(&stl.gsi.dfc));
    }

    #[test]
    fn time_code_in_and_out_outside_spec_are_errors() {
        let mut stl = parse_stl_from_file("stls/test.stl").unwrap();
        let dfc = DiskFormatCode::STL25_01;
        let tti = &mut stl.ttis[0];
        for (tci, msg) in [
            (time(24, 0, 0, 0), "24:00:00:00 has hours 24 above 23"),
            (time(0, 60, 0, 0), "00:60:00:00 has minutes 60 above 59"),
            (time(0, 0, 60, 0), "00:00:60:00 has seconds 60 above 59"),
            (
                time(0, 0, 0, 25),
                "00:00:00:25 has frame 25 above 24 at 25 frames per second",
            ),
        ] {
            tti.tci = tci;
            let err = tti.time_code_in(&dfc).unwrap_err();
            assert_eq!(("TCI", msg, &tci), (err.field(), err.msg(), err.value()));
        }
        tti.tco = time(0, 0, 0, 25);
        assert_invalid(tti.time_code_out(&dfc), "TCO", time(0, 0, 0, 25));
        // Frames up to 29 at 30 frames per second
        assert!(tti.time_code_out(&DiskFormatCode::STL30_01).is_ok());
    }

    #[test]
    fn time_code_frames_depend_on_disk_format_code() {
        let mut gsi = gsi();
        gsi.tcf = "00000029".to_string();
        assert!(gsi.time_code_first_in_cue().is_err());
        gsi.dfc = DiskFormatCode::STL30_01;
        assert_eq!(Ok(Some(time(0, 0, 0, 29))), gsi.time_code_first_in_cue());
        gsi.tcf = "00000030".to_string();
        assert!(gsi.time_code_first_in_cue().is_err());
    }

    #[test]
    fn country_of_origin() {
        let mut gsi = gsi();
        gsi.co = "   ".to_string();
        assert_eq!(Ok(None), gsi.country_of_origin());
        for co in ["fra", "FR ", "F1A"] {
            gsi.co = co.to_string();
            assert_invalid(gsi.country_of_origin(), "CO", co.to_string());
        }
    }

    #[test]
    fn free_text_keeps_leading_spaces() {
        let mut gsi = gsi();
        gsi.opt = "  Title with leading spaces     ".to_string();
        assert_eq!(
            Some("  Title with leading spaces"),
            gsi.original_programme_title()
        );
    }

    #[test]
    fn justification_code() {
        let mut stl = parse_stl_from_file("stls/test.stl").unwrap();
        let tti = &mut stl.ttis[0];
        for (jc, expected) in [
            (0, Justification::Unchanged),
            (1, Justification::Left),
            (2, Justification::Centered),
            (3, Justification::Right),
        ] {
            tti.jc = jc;
            assert_eq!(Ok(expected), tti.justification());
        }
        tti.jc = 0x20;
        assert_invalid(tti.justification(), "JC", 0x20);
    }

    #[test]
    fn comment_flag() {
        let mut stl = parse_stl_from_file("stls/test.stl").unwrap();
        let tti = &mut stl.ttis[0];
        tti.cf = 1;
        assert_eq!(Ok(true), tti.comment_flag());
        tti.cf = 2;
        assert_invalid(tti.comment_flag(), "CF", 2);
    }

    #[test]
    fn error_messages() {
        let mut gsi = gsi();
        gsi.cd = "911341".to_string();
        assert_eq!(
            "Invalid CD: \"911341\" is not an existing date",
            gsi.creation_date().unwrap_err().to_string()
        );
        gsi.cd = "91-2-1".to_string();
        assert_eq!(
            "Invalid CD: \"91-2-1\" is not a date in the format YYMMDD",
            gsi.creation_date().unwrap_err().to_string()
        );
        gsi.lc = "80".to_string();
        assert_eq!(
            "Invalid LC: 80h is outside 00h-7Fh",
            gsi.language().unwrap_err().to_string()
        );
        gsi.tcp = "00000025".to_string();
        assert_eq!(
            "Invalid TCP: 00:00:00:25 has frame 25 above 24 at 25 frames per second",
            gsi.time_code_start_of_programme().unwrap_err().to_string()
        );
    }

    #[test]
    fn invalid_value_can_be_taken() {
        let mut stl = parse_stl_from_file("stls/test.stl").unwrap();
        let tti = &mut stl.ttis[0];
        tti.tci = time(10, 2, 31, 25);

        let tci = tti
            .time_code_in(&stl.gsi.dfc)
            .unwrap_or_else(|err| err.into_value());
        assert_eq!(time(10, 2, 31, 25), tci);
    }

    #[test]
    fn invalid_value_converts_into_parse_error() {
        let mut gsi = gsi();
        gsi.co = "fra".to_string();

        let err = gsi.country_of_origin().unwrap_err();
        let msg = err.to_string();
        assert!(matches!(
            crate::ParseError::from(err),
            crate::ParseError::InvalidFieldValue(converted) if converted == msg
        ));
    }
}