third_party/rust/encoding_rs/src/single_byte.rs
author Chris AtLee <catlee@mozilla.com>
Mon, 30 Jul 2018 20:23:14 +0000
changeset 429247 76b7549059a4aaa3ce26587586be3e1ed3bbdbd1
parent 421942 466d43707902ec86b9729c86cd7e844b3af7f075
child 447691 3691033d6e78b3481438c6d75b6a1e4200591388
permissions -rw-r--r--
Bug 1466537: Disable full crashsymbols by branch r=ted Differential Revision: https://phabricator.services.mozilla.com/D1558

// Copyright 2015-2016 Mozilla Foundation. See the COPYRIGHT
// file at the top-level directory of this distribution.
//
// Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
// https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
// <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
// option. This file may not be copied, modified, or distributed
// except according to those terms.

use super::*;
use ascii::*;
use handles::*;
use variant::*;

pub struct SingleByteDecoder {
    table: &'static [u16; 128],
}

impl SingleByteDecoder {
    pub fn new(data: &'static [u16; 128]) -> VariantDecoder {
        VariantDecoder::SingleByte(SingleByteDecoder { table: data })
    }

    pub fn max_utf16_buffer_length(&self, byte_length: usize) -> Option<usize> {
        Some(byte_length)
    }

    pub fn max_utf8_buffer_length_without_replacement(&self, byte_length: usize) -> Option<usize> {
        byte_length.checked_mul(3)
    }

    pub fn max_utf8_buffer_length(&self, byte_length: usize) -> Option<usize> {
        byte_length.checked_mul(3)
    }

    pub fn decode_to_utf8_raw(
        &mut self,
        src: &[u8],
        dst: &mut [u8],
        _last: bool,
    ) -> (DecoderResult, usize, usize) {
        let mut source = ByteSource::new(src);
        let mut dest = Utf8Destination::new(dst);
        'outermost: loop {
            match dest.copy_ascii_from_check_space_bmp(&mut source) {
                CopyAsciiResult::Stop(ret) => return ret,
                CopyAsciiResult::GoOn((mut non_ascii, mut handle)) => 'middle: loop {
                    // Start non-boilerplate
                    //
                    // Since the non-ASCIIness of `non_ascii` is hidden from
                    // the optimizer, it can't figure out that it's OK to
                    // statically omit the bound check when accessing
                    // `[u16; 128]` with an index
                    // `non_ascii as usize - 0x80usize`.
                    let mapped =
                        unsafe { *(self.table.get_unchecked(non_ascii as usize - 0x80usize)) };
                    // let mapped = self.table[non_ascii as usize - 0x80usize];
                    if mapped == 0u16 {
                        return (
                            DecoderResult::Malformed(1, 0),
                            source.consumed(),
                            handle.written(),
                        );
                    }
                    let dest_again = handle.write_bmp_excl_ascii(mapped);
                    // End non-boilerplate
                    match source.check_available() {
                        Space::Full(src_consumed) => {
                            return (
                                DecoderResult::InputEmpty,
                                src_consumed,
                                dest_again.written(),
                            );
                        }
                        Space::Available(source_handle) => {
                            match dest_again.check_space_bmp() {
                                Space::Full(dst_written) => {
                                    return (
                                        DecoderResult::OutputFull,
                                        source_handle.consumed(),
                                        dst_written,
                                    );
                                }
                                Space::Available(mut destination_handle) => {
                                    let (mut b, unread_handle) = source_handle.read();
                                    let source_again = unread_handle.commit();
                                    'innermost: loop {
                                        if b > 127 {
                                            non_ascii = b;
                                            handle = destination_handle;
                                            continue 'middle;
                                        }
                                        // Testing on Haswell says that we should write the
                                        // byte unconditionally instead of trying to unread it
                                        // to make it part of the next SIMD stride.
                                        let dest_again_again = destination_handle.write_ascii(b);
                                        if b < 60 {
                                            // We've got punctuation
                                            match source_again.check_available() {
                                                Space::Full(src_consumed_again) => {
                                                    return (
                                                        DecoderResult::InputEmpty,
                                                        src_consumed_again,
                                                        dest_again_again.written(),
                                                    );
                                                }
                                                Space::Available(source_handle_again) => {
                                                    match dest_again_again.check_space_bmp() {
                                                        Space::Full(dst_written_again) => {
                                                            return (
                                                                DecoderResult::OutputFull,
                                                                source_handle_again.consumed(),
                                                                dst_written_again,
                                                            );
                                                        }
                                                        Space::Available(
                                                            destination_handle_again,
                                                        ) => {
                                                            let (b_again, _unread_handle_again) =
                                                                source_handle_again.read();
                                                            b = b_again;
                                                            destination_handle =
                                                                destination_handle_again;
                                                            continue 'innermost;
                                                        }
                                                    }
                                                }
                                            }
                                        }
                                        // We've got markup or ASCII text
                                        continue 'outermost;
                                    }
                                }
                            }
                        }
                    }
                },
            }
        }
    }

    pub fn decode_to_utf16_raw(
        &mut self,
        src: &[u8],
        dst: &mut [u16],
        _last: bool,
    ) -> (DecoderResult, usize, usize) {
        let (pending, length) = if dst.len() < src.len() {
            (DecoderResult::OutputFull, dst.len())
        } else {
            (DecoderResult::InputEmpty, src.len())
        };
        let mut converted = 0usize;
        'outermost: loop {
            match unsafe {
                ascii_to_basic_latin(
                    src.as_ptr().offset(converted as isize),
                    dst.as_mut_ptr().offset(converted as isize),
                    length - converted,
                )
            } {
                None => {
                    return (pending, length, length);
                }
                Some((mut non_ascii, consumed)) => {
                    converted += consumed;
                    'middle: loop {
                        // `converted` doesn't count the reading of `non_ascii` yet.
                        // Since the non-ASCIIness of `non_ascii` is hidden from
                        // the optimizer, it can't figure out that it's OK to
                        // statically omit the bound check when accessing
                        // `[u16; 128]` with an index
                        // `non_ascii as usize - 0x80usize`.
                        let mapped =
                            unsafe { *(self.table.get_unchecked(non_ascii as usize - 0x80usize)) };
                        // let mapped = self.table[non_ascii as usize - 0x80usize];
                        if mapped == 0u16 {
                            return (
                                DecoderResult::Malformed(1, 0),
                                converted + 1, // +1 `for non_ascii`
                                converted,
                            );
                        }
                        unsafe {
                            // The bound check has already been performed
                            *(dst.get_unchecked_mut(converted)) = mapped;
                        }
                        converted += 1;
                        // Next, handle ASCII punctuation and non-ASCII without
                        // going back to ASCII acceleration. Non-ASCII scripts
                        // use ASCII punctuation, so this avoid going to
                        // acceleration just for punctuation/space and then
                        // failing. This is a significant boost to non-ASCII
                        // scripts.
                        // TODO: Split out Latin converters without this part
                        // this stuff makes Latin script-conversion slower.
                        if converted == length {
                            return (pending, length, length);
                        }
                        let mut b = unsafe { *(src.get_unchecked(converted)) };
                        'innermost: loop {
                            if b > 127 {
                                non_ascii = b;
                                continue 'middle;
                            }
                            // Testing on Haswell says that we should write the
                            // byte unconditionally instead of trying to unread it
                            // to make it part of the next SIMD stride.
                            unsafe {
                                *(dst.get_unchecked_mut(converted)) = b as u16;
                            }
                            converted += 1;
                            if b < 60 {
                                // We've got punctuation
                                if converted == length {
                                    return (pending, length, length);
                                }
                                b = unsafe { *(src.get_unchecked(converted)) };
                                continue 'innermost;
                            }
                            // We've got markup or ASCII text
                            continue 'outermost;
                        }
                    }
                }
            }
        }
    }
}

pub struct SingleByteEncoder {
    table: &'static [u16; 128],
}

impl SingleByteEncoder {
    pub fn new(encoding: &'static Encoding, data: &'static [u16; 128]) -> Encoder {
        Encoder::new(
            encoding,
            VariantEncoder::SingleByte(SingleByteEncoder { table: data }),
        )
    }

    pub fn max_buffer_length_from_utf16_without_replacement(
        &self,
        u16_length: usize,
    ) -> Option<usize> {
        Some(u16_length)
    }

    pub fn max_buffer_length_from_utf8_without_replacement(
        &self,
        byte_length: usize,
    ) -> Option<usize> {
        Some(byte_length)
    }

    fn encode_u16(&self, code_unit: u16) -> Option<u8> {
        // We search the quadrants in reverse order, but we search forward
        // within each quadrant. For Windows and ISO encodings, this is
        // generally faster than just searching the whole table backwards.
        // (Exceptions: English, German, Czech.) This order is also OK for
        // KOI encodings. For IBM and Mac encodings, this order is bad,
        // but we don't really need to optimize for those encodings anyway.

        // In Windows and ISO encodings, the fourth quadrant holds most of the
        // lower-case letters for bicameral scripts as well as the Hebrew
        // letters. There are some Thai letters and combining marks as well as
        // Thai numerals here. (In KOI8-R, the upper-case letters are here.)
        for i in 96..128 {
            if self.table[i] == code_unit {
                return Some((i + 128) as u8);
            }
        }

        // In Windows and ISO encodings, the third quadrant holds most of the
        // upper-case letters for bicameral scripts as well as most of the
        // Arabic letters. Searching this quadrant first would be better for
        // Arabic. There are a number of Thai letters and combining marks here.
        // (In KOI8-R, the lower-case letters are here.)
        for i in 64..96 {
            if self.table[i] == code_unit {
                return Some((i + 128) as u8);
            }
        }

        // In Windows and ISO encodings, the second quadrant hold most of the
        // Thai letters. In other scripts, there tends to be symbols here.
        // Even though the two quadrants above are relevant for Thai, for Thai
        // it would likely be optimal to search this quadrant first. :-(
        for i in 32..64 {
            if self.table[i] == code_unit {
                return Some((i + 128) as u8);
            }
        }

        // The first quadrant is useless in ISO encodings. In Windows encodings,
        // there is useful punctuation here that might warrant searching
        // before the symbols in the second quadrant, but the second quadrant
        // is searched before this one for the benefit of Thai.
        for i in 0..32 {
            if self.table[i] == code_unit {
                return Some((i + 128) as u8);
            }
        }
        None
    }

    ascii_compatible_bmp_encoder_function!(
        {
            match self.encode_u16(bmp) {
                Some(byte) => handle.write_one(byte),
                None => {
                    return (
                        EncoderResult::unmappable_from_bmp(bmp),
                        source.consumed(),
                        handle.written(),
                    );
                }
            }
        },
        bmp,
        self,
        source,
        handle,
        copy_ascii_to_check_space_one,
        check_space_one,
        encode_from_utf8_raw,
        str,
        Utf8Source,
        true
    );

    pub fn encode_from_utf16_raw(
        &mut self,
        src: &[u16],
        dst: &mut [u8],
        _last: bool,
    ) -> (EncoderResult, usize, usize) {
        let (pending, length) = if dst.len() < src.len() {
            (EncoderResult::OutputFull, dst.len())
        } else {
            (EncoderResult::InputEmpty, src.len())
        };
        let mut converted = 0usize;
        'outermost: loop {
            match unsafe {
                basic_latin_to_ascii(
                    src.as_ptr().offset(converted as isize),
                    dst.as_mut_ptr().offset(converted as isize),
                    length - converted,
                )
            } {
                None => {
                    return (pending, length, length);
                }
                Some((mut non_ascii, consumed)) => {
                    converted += consumed;
                    'middle: loop {
                        // `converted` doesn't count the reading of `non_ascii` yet.
                        match self.encode_u16(non_ascii) {
                            Some(byte) => {
                                unsafe {
                                    *(dst.get_unchecked_mut(converted)) = byte;
                                }
                                converted += 1;
                            }
                            None => {
                                // At this point, we need to know if we
                                // have a surrogate.
                                let high_bits = non_ascii & 0xFC00u16;
                                if high_bits == 0xD800u16 {
                                    // high surrogate
                                    if converted + 1 == length {
                                        // End of buffer. This surrogate is unpaired.
                                        return (
                                            EncoderResult::Unmappable('\u{FFFD}'),
                                            converted + 1, // +1 `for non_ascii`
                                            converted,
                                        );
                                    }
                                    let second =
                                        unsafe { *src.get_unchecked(converted + 1) } as u32;
                                    if second & 0xFC00u32 != 0xDC00u32 {
                                        return (
                                            EncoderResult::Unmappable('\u{FFFD}'),
                                            converted + 1, // +1 `for non_ascii`
                                            converted,
                                        );
                                    }
                                    // The next code unit is a low surrogate.
                                    let astral: char = unsafe {
                                        ::std::mem::transmute(
                                            ((non_ascii as u32) << 10) + second
                                                - (((0xD800u32 << 10) - 0x10000u32) + 0xDC00u32),
                                        )
                                    };
                                    return (
                                        EncoderResult::Unmappable(astral),
                                        converted + 2, // +2 `for non_ascii` and `second`
                                        converted,
                                    );
                                }
                                if high_bits == 0xDC00u16 {
                                    // Unpaired low surrogate
                                    return (
                                        EncoderResult::Unmappable('\u{FFFD}'),
                                        converted + 1, // +1 `for non_ascii`
                                        converted,
                                    );
                                }
                                let thirty_two = non_ascii as u32;
                                let bmp: char = unsafe { ::std::mem::transmute(thirty_two) };
                                return (
                                    EncoderResult::Unmappable(bmp),
                                    converted + 1, // +1 `for non_ascii`
                                    converted,
                                );
                            }
                        }
                        // Next, handle ASCII punctuation and non-ASCII without
                        // going back to ASCII acceleration. Non-ASCII scripts
                        // use ASCII punctuation, so this avoid going to
                        // acceleration just for punctuation/space and then
                        // failing. This is a significant boost to non-ASCII
                        // scripts.
                        // TODO: Split out Latin converters without this part
                        // this stuff makes Latin script-conversion slower.
                        if converted == length {
                            return (pending, length, length);
                        }
                        let mut unit = unsafe { *(src.get_unchecked(converted)) };
                        'innermost: loop {
                            if unit > 127 {
                                non_ascii = unit;
                                continue 'middle;
                            }
                            // Testing on Haswell says that we should write the
                            // byte unconditionally instead of trying to unread it
                            // to make it part of the next SIMD stride.
                            unsafe {
                                *(dst.get_unchecked_mut(converted)) = unit as u8;
                            }
                            converted += 1;
                            if unit < 60 {
                                // We've got punctuation
                                if converted == length {
                                    return (pending, length, length);
                                }
                                unit = unsafe { *(src.get_unchecked(converted)) };
                                continue 'innermost;
                            }
                            // We've got markup or ASCII text
                            continue 'outermost;
                        }
                    }
                }
            }
        }
    }
}

// Any copyright to the test code below this comment is dedicated to the
// Public Domain. http://creativecommons.org/publicdomain/zero/1.0/

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

    #[test]
    fn test_windows_1255_ca() {
        decode(WINDOWS_1255, b"\xCA", "\u{05BA}");
        encode(WINDOWS_1255, "\u{05BA}", b"\xCA");
    }

    #[test]
    fn test_ascii_punctuation() {
        let bytes = b"\xC1\xF5\xF4\xFC \xE5\xDF\xED\xE1\xE9 \xDD\xED\xE1 \xF4\xE5\xF3\xF4. \xC1\xF5\xF4\xFC \xE5\xDF\xED\xE1\xE9 \xDD\xED\xE1 \xF4\xE5\xF3\xF4.";
        let characters = "\u{0391}\u{03C5}\u{03C4}\u{03CC} \
                          \u{03B5}\u{03AF}\u{03BD}\u{03B1}\u{03B9} \u{03AD}\u{03BD}\u{03B1} \
                          \u{03C4}\u{03B5}\u{03C3}\u{03C4}. \u{0391}\u{03C5}\u{03C4}\u{03CC} \
                          \u{03B5}\u{03AF}\u{03BD}\u{03B1}\u{03B9} \u{03AD}\u{03BD}\u{03B1} \
                          \u{03C4}\u{03B5}\u{03C3}\u{03C4}.";
        decode(WINDOWS_1253, bytes, characters);
        encode(WINDOWS_1253, characters, bytes);
    }

    #[test]
    fn test_decode_malformed() {
        decode(
            WINDOWS_1253,
            b"\xC1\xF5\xD2\xF4\xFC",
            "\u{0391}\u{03C5}\u{FFFD}\u{03C4}\u{03CC}",
        );
    }

    #[test]
    fn test_encode_unmappables() {
        encode(
            WINDOWS_1253,
            "\u{0391}\u{03C5}\u{2603}\u{03C4}\u{03CC}",
            b"\xC1\xF5&#9731;\xF4\xFC",
        );
        encode(
            WINDOWS_1253,
            "\u{0391}\u{03C5}\u{1F4A9}\u{03C4}\u{03CC}",
            b"\xC1\xF5&#128169;\xF4\xFC",
        );
    }

    #[test]
    fn test_encode_unpaired_surrogates() {
        encode_from_utf16(
            WINDOWS_1253,
            &[0x0391u16, 0x03C5u16, 0xDCA9u16, 0x03C4u16, 0x03CCu16],
            b"\xC1\xF5&#65533;\xF4\xFC",
        );
        encode_from_utf16(
            WINDOWS_1253,
            &[0x0391u16, 0x03C5u16, 0xD83Du16, 0x03C4u16, 0x03CCu16],
            b"\xC1\xF5&#65533;\xF4\xFC",
        );
        encode_from_utf16(
            WINDOWS_1253,
            &[0x0391u16, 0x03C5u16, 0x03C4u16, 0x03CCu16, 0xD83Du16],
            b"\xC1\xF5\xF4\xFC&#65533;",
        );
    }

    pub const HIGH_BYTES: &'static [u8; 128] = &[
        0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E,
        0x8F, 0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97, 0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D,
        0x9E, 0x9F, 0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7, 0xA8, 0xA9, 0xAA, 0xAB, 0xAC,
        0xAD, 0xAE, 0xAF, 0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7, 0xB8, 0xB9, 0xBA, 0xBB,
        0xBC, 0xBD, 0xBE, 0xBF, 0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7, 0xC8, 0xC9, 0xCA,
        0xCB, 0xCC, 0xCD, 0xCE, 0xCF, 0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7, 0xD8, 0xD9,
        0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF, 0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7, 0xE8,
        0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF, 0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7,
        0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF,
    ];

    fn decode_single_byte(encoding: &'static Encoding, data: &'static [u16; 128]) {
        let mut with_replacement = [0u16; 128];
        let mut it = data.iter().enumerate();
        loop {
            match it.next() {
                Some((i, code_point)) => {
                    if *code_point == 0 {
                        with_replacement[i] = 0xFFFD;
                    } else {
                        with_replacement[i] = *code_point;
                    }
                }
                None => {
                    break;
                }
            }
        }

        decode_to_utf16(encoding, HIGH_BYTES, &with_replacement[..]);
    }

    fn encode_single_byte(encoding: &'static Encoding, data: &'static [u16; 128]) {
        let mut with_zeros = [0u8; 128];
        let mut it = data.iter().enumerate();
        loop {
            match it.next() {
                Some((i, code_point)) => {
                    if *code_point == 0 {
                        with_zeros[i] = 0;
                    } else {
                        with_zeros[i] = HIGH_BYTES[i];
                    }
                }
                None => {
                    break;
                }
            }
        }

        encode_from_utf16(encoding, data, &with_zeros[..]);
    }

    #[test]
    fn test_single_byte_from_two_low_surrogates() {
        let expectation = b"&#65533;&#65533;";
        let mut output = [0u8; 40];
        let mut encoder = WINDOWS_1253.new_encoder();
        let (result, read, written, had_errors) =
            encoder.encode_from_utf16(&[0xDC00u16, 0xDEDEu16], &mut output[..], true);
        assert_eq!(result, CoderResult::InputEmpty);
        assert_eq!(read, 2);
        assert_eq!(written, expectation.len());
        assert!(had_errors);
        assert_eq!(&output[..written], expectation);
    }

    // These tests are so self-referential that they are pretty useless.

    // BEGIN GENERATED CODE. PLEASE DO NOT EDIT.
    // Instead, please regenerate using generate-encoding-data.py

    #[test]
    fn test_single_byte_decode() {
        decode_single_byte(IBM866, IBM866_DATA);
        decode_single_byte(ISO_8859_10, ISO_8859_10_DATA);
        decode_single_byte(ISO_8859_13, ISO_8859_13_DATA);
        decode_single_byte(ISO_8859_14, ISO_8859_14_DATA);
        decode_single_byte(ISO_8859_15, ISO_8859_15_DATA);
        decode_single_byte(ISO_8859_16, ISO_8859_16_DATA);
        decode_single_byte(ISO_8859_2, ISO_8859_2_DATA);
        decode_single_byte(ISO_8859_3, ISO_8859_3_DATA);
        decode_single_byte(ISO_8859_4, ISO_8859_4_DATA);
        decode_single_byte(ISO_8859_5, ISO_8859_5_DATA);
        decode_single_byte(ISO_8859_6, ISO_8859_6_DATA);
        decode_single_byte(ISO_8859_7, ISO_8859_7_DATA);
        decode_single_byte(ISO_8859_8, ISO_8859_8_DATA);
        decode_single_byte(KOI8_R, KOI8_R_DATA);
        decode_single_byte(KOI8_U, KOI8_U_DATA);
        decode_single_byte(MACINTOSH, MACINTOSH_DATA);
        decode_single_byte(WINDOWS_1250, WINDOWS_1250_DATA);
        decode_single_byte(WINDOWS_1251, WINDOWS_1251_DATA);
        decode_single_byte(WINDOWS_1252, WINDOWS_1252_DATA);
        decode_single_byte(WINDOWS_1253, WINDOWS_1253_DATA);
        decode_single_byte(WINDOWS_1254, WINDOWS_1254_DATA);
        decode_single_byte(WINDOWS_1255, WINDOWS_1255_DATA);
        decode_single_byte(WINDOWS_1256, WINDOWS_1256_DATA);
        decode_single_byte(WINDOWS_1257, WINDOWS_1257_DATA);
        decode_single_byte(WINDOWS_1258, WINDOWS_1258_DATA);
        decode_single_byte(WINDOWS_874, WINDOWS_874_DATA);
        decode_single_byte(X_MAC_CYRILLIC, X_MAC_CYRILLIC_DATA);
    }

    #[test]
    fn test_single_byte_encode() {
        encode_single_byte(IBM866, IBM866_DATA);
        encode_single_byte(ISO_8859_10, ISO_8859_10_DATA);
        encode_single_byte(ISO_8859_13, ISO_8859_13_DATA);
        encode_single_byte(ISO_8859_14, ISO_8859_14_DATA);
        encode_single_byte(ISO_8859_15, ISO_8859_15_DATA);
        encode_single_byte(ISO_8859_16, ISO_8859_16_DATA);
        encode_single_byte(ISO_8859_2, ISO_8859_2_DATA);
        encode_single_byte(ISO_8859_3, ISO_8859_3_DATA);
        encode_single_byte(ISO_8859_4, ISO_8859_4_DATA);
        encode_single_byte(ISO_8859_5, ISO_8859_5_DATA);
        encode_single_byte(ISO_8859_6, ISO_8859_6_DATA);
        encode_single_byte(ISO_8859_7, ISO_8859_7_DATA);
        encode_single_byte(ISO_8859_8, ISO_8859_8_DATA);
        encode_single_byte(KOI8_R, KOI8_R_DATA);
        encode_single_byte(KOI8_U, KOI8_U_DATA);
        encode_single_byte(MACINTOSH, MACINTOSH_DATA);
        encode_single_byte(WINDOWS_1250, WINDOWS_1250_DATA);
        encode_single_byte(WINDOWS_1251, WINDOWS_1251_DATA);
        encode_single_byte(WINDOWS_1252, WINDOWS_1252_DATA);
        encode_single_byte(WINDOWS_1253, WINDOWS_1253_DATA);
        encode_single_byte(WINDOWS_1254, WINDOWS_1254_DATA);
        encode_single_byte(WINDOWS_1255, WINDOWS_1255_DATA);
        encode_single_byte(WINDOWS_1256, WINDOWS_1256_DATA);
        encode_single_byte(WINDOWS_1257, WINDOWS_1257_DATA);
        encode_single_byte(WINDOWS_1258, WINDOWS_1258_DATA);
        encode_single_byte(WINDOWS_874, WINDOWS_874_DATA);
        encode_single_byte(X_MAC_CYRILLIC, X_MAC_CYRILLIC_DATA);
    }
    // END GENERATED CODE

}