// Copyright 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 crate::data::*;
use crate::handles::*;
use crate::variant::*;
// Rust 1.14.0 requires the following despite the asterisk above.
use super::in_inclusive_range;
use super::in_inclusive_range16;

pub struct ShiftJisDecoder {
    lead: Option<u8>,
}

impl ShiftJisDecoder {
    pub fn new() -> VariantDecoder {
        VariantDecoder::ShiftJis(ShiftJisDecoder { lead: None })
    }

    pub fn in_neutral_state(&self) -> bool {
        self.lead.is_none()
    }

    fn plus_one_if_lead(&self, byte_length: usize) -> Option<usize> {
        byte_length.checked_add(match self.lead {
            None => 0,
            Some(_) => 1,
        })
    }

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

    pub fn max_utf8_buffer_length_without_replacement(&self, byte_length: usize) -> Option<usize> {
        // worst case: 1 to 3 (half-width katakana)
        self.max_utf8_buffer_length(byte_length)
    }

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

    ascii_compatible_two_byte_decoder_functions!(
        {
           // If lead is between 0x81 and 0x9F, inclusive,
           // subtract offset 0x81. Else if lead is
           // between 0xE0 and 0xFC, inclusive, subtract
           // offset 0xC1. Else if lead is between
           // 0xA1 and 0xDF, inclusive, map to half-width
           // Katakana. Else if lead is 0x80, pass through.
            let mut non_ascii_minus_offset =
                non_ascii.wrapping_sub(0x81);
            if non_ascii_minus_offset > (0x9F - 0x81) {
                let non_ascii_minus_range_start = non_ascii.wrapping_sub(0xE0);
                if non_ascii_minus_range_start > (0xFC - 0xE0) {
                    let non_ascii_minus_half_with_katakana_start = non_ascii.wrapping_sub(0xA1);
                    if non_ascii_minus_half_with_katakana_start > (0xDF - 0xA1) {
                        if non_ascii == 0x80 {
                            handle.write_mid_bmp(0x80);
                            // Not caring about optimizing subsequent non-ASCII
                            continue 'outermost;
                        }
                        return (DecoderResult::Malformed(1, 0),
                                source.consumed(),
                                handle.written());
                    }
                    handle.write_upper_bmp(0xFF61 + u16::from(non_ascii_minus_half_with_katakana_start));
                    // Not caring about optimizing subsequent non-ASCII
                    continue 'outermost;
                }
                non_ascii_minus_offset = non_ascii - 0xC1;
            }
            non_ascii_minus_offset
        },
        {
            // If trail is between 0x40 and 0x7E, inclusive,
            // subtract offset 0x40. Else if trail is
            // between 0x80 and 0xFC, inclusive, subtract
            // offset 0x41.
            // Fast-track Hiragana (60% according to Lunde)
            // and Katakana (10% acconding to Lunde).
            // Hiragana doesn't cross 0x7F, but Katakana does.
            // We can check for Hiragana before normalizing
            // trail.
            let trail_minus_hiragana = byte.wrapping_sub(0x9F);
            if lead_minus_offset == 0x01 && trail_minus_hiragana < 0x53 {
            // Hiragana
                handle.write_upper_bmp(0x3041 + u16::from(trail_minus_hiragana))
            } else {
                let mut trail_minus_offset =
                    byte.wrapping_sub(0x40);
                if trail_minus_offset > (0x7E - 0x40) {
                    let trail_minus_range_start =
                        byte.wrapping_sub(0x80);
                    if trail_minus_range_start > (0xFC - 0x80) {
                        if byte < 0x80 {
                            return (DecoderResult::Malformed(1, 0),
                                    unread_handle_trail.unread(),
                                    handle.written());
                        }
                        return (DecoderResult::Malformed(2, 0),
                                unread_handle_trail.consumed(),
                                handle.written());
                    }
                    trail_minus_offset = byte - 0x41;
                }
                if lead_minus_offset == 0x02 &&
                   trail_minus_offset < 0x56 {
                    // Katakana
                    handle.write_upper_bmp(0x30A1 + u16::from(trail_minus_offset))
                } else {
                    let pointer = lead_minus_offset as usize *
                                  188usize +
                                  trail_minus_offset as usize;
                    let level1_pointer = pointer.wrapping_sub(1410);
                    if level1_pointer < JIS0208_LEVEL1_KANJI.len() {
                        handle.write_upper_bmp(JIS0208_LEVEL1_KANJI[level1_pointer])
                    } else {
                        let level2_pointer = pointer.wrapping_sub(4418);
                        if level2_pointer <
                           JIS0208_LEVEL2_AND_ADDITIONAL_KANJI.len() {
                            handle.write_upper_bmp(JIS0208_LEVEL2_AND_ADDITIONAL_KANJI[level2_pointer])
                        } else {
                            let upper_ibm_pointer = pointer.wrapping_sub(10744);
                            if upper_ibm_pointer < IBM_KANJI.len() {
                                handle.write_upper_bmp(IBM_KANJI[upper_ibm_pointer])
                            } else {
                                let lower_ibm_pointer = pointer.wrapping_sub(8272);
                                if lower_ibm_pointer < IBM_KANJI.len() {
                                    handle.write_upper_bmp(IBM_KANJI[lower_ibm_pointer])
                                } else if in_inclusive_range(pointer, 8836, 10715) {
                                    handle.write_upper_bmp((0xE000 - 8836 + pointer) as u16)
                                } else if let Some(bmp) = jis0208_symbol_decode(pointer) {
                                    handle.write_bmp_excl_ascii(bmp)
                                } else if let Some(bmp) = jis0208_range_decode(pointer) {
                                    handle.write_bmp_excl_ascii(bmp)
                                } else {
                                    if byte < 0x80 {
                                        return (DecoderResult::Malformed(1, 0),
                                                unread_handle_trail.unread(),
                                                handle.written());
                                    }
                                    return (DecoderResult::Malformed(2, 0),
                                            unread_handle_trail.consumed(),
                                            handle.written());
                                }
                            }
                        }
                    }
                }
            }
        },
        self,
        non_ascii,
        byte,
        lead_minus_offset,
        unread_handle_trail,
        source,
        handle,
        'outermost,
        copy_ascii_from_check_space_bmp,
        check_space_bmp,
        false);
}

#[cfg(feature = "fast-kanji-encode")]
#[inline(always)]
fn encode_kanji(bmp: u16) -> Option<(u8, u8)> {
    jis0208_kanji_shift_jis_encode(bmp)
}

#[cfg(not(feature = "fast-kanji-encode"))]
#[inline(always)]
fn encode_kanji(bmp: u16) -> Option<(u8, u8)> {
    if let Some((lead, trail)) = jis0208_level1_kanji_shift_jis_encode(bmp) {
        return Some((lead, trail));
    }
    let pointer = if 0x4EDD == bmp {
        // Ideograph on the symbol row!
        23
    } else if let Some(pos) = jis0208_level2_and_additional_kanji_encode(bmp) {
        4418 + pos
    } else if let Some(pos) = position(&IBM_KANJI[..], bmp) {
        10744 + pos
    } else {
        return None;
    };
    let lead = pointer / 188;
    let lead_offset = if lead < 0x1F { 0x81usize } else { 0xC1usize };
    let trail = pointer % 188;
    let trail_offset = if trail < 0x3F { 0x40usize } else { 0x41usize };
    Some(((lead + lead_offset) as u8, (trail + trail_offset) as u8))
}

pub struct ShiftJisEncoder;

impl ShiftJisEncoder {
    pub fn new(encoding: &'static Encoding) -> Encoder {
        Encoder::new(encoding, VariantEncoder::ShiftJis(ShiftJisEncoder))
    }

    pub fn max_buffer_length_from_utf16_without_replacement(
        &self,
        u16_length: usize,
    ) -> Option<usize> {
        u16_length.checked_mul(2)
    }

    pub fn max_buffer_length_from_utf8_without_replacement(
        &self,
        byte_length: usize,
    ) -> Option<usize> {
        byte_length.checked_add(1)
    }

    ascii_compatible_bmp_encoder_functions!(
        {
            // Lunde says 60% Hiragana, 30% Kanji, 10% Katakana
            let bmp_minus_hiragana = bmp.wrapping_sub(0x3041);
            if bmp_minus_hiragana < 0x53 {
                handle.write_two(0x82, 0x9F + bmp_minus_hiragana as u8)
            } else if in_inclusive_range16(bmp, 0x4E00, 0x9FA0) {
                if let Some((lead, trail)) = encode_kanji(bmp) {
                    handle.write_two(lead, trail)
                } else {
                    return (
                        EncoderResult::unmappable_from_bmp(bmp),
                        source.consumed(),
                        handle.written(),
                    );
                }
            } else {
                let bmp_minus_katakana = bmp.wrapping_sub(0x30A1);
                if bmp_minus_katakana < 0x56 {
                    let trail_offset = if bmp_minus_katakana < 0x3F {
                        0x40
                    } else {
                        0x41
                    };
                    handle.write_two(0x83, (trail_offset + bmp_minus_katakana) as u8)
                } else {
                    let bmp_minus_space = bmp.wrapping_sub(0x3000);
                    if bmp_minus_space < 3 {
                        // fast-track common punctuation
                        handle.write_two(0x81, 0x40 + bmp_minus_space as u8)
                    } else if bmp == 0xA5 {
                        handle.write_one(0x5Cu8)
                    } else if bmp == 0x80 {
                        handle.write_one(0x80u8)
                    } else if bmp == 0x203E {
                        handle.write_one(0x7Eu8)
                    } else if in_inclusive_range16(bmp, 0xFF61, 0xFF9F) {
                        handle.write_one((bmp - (0xFF61 - 0xA1)) as u8)
                    } else if bmp == 0x2212 {
                        handle.write_two(0x81u8, 0x7Cu8)
                    } else {
                        let bmp_minus_roman = bmp.wrapping_sub(0x2170);
                        let pointer = if bmp_minus_roman <= (0x2179 - 0x2170) {
                            10716 + bmp_minus_roman as usize
                        } else if let Some(pointer) = jis0208_range_encode(bmp) {
                            pointer
                        } else if in_inclusive_range16(bmp, 0xFA0E, 0xFA2D)
                            || bmp == 0xF929
                            || bmp == 0xF9DC
                        {
                            // Guaranteed to be found in IBM_KANJI
                            let pos = position(&IBM_KANJI[..], bmp).unwrap();
                            10744 + pos
                        } else if let Some(pointer) = jis0208_symbol_encode(bmp) {
                            pointer
                        } else {
                            return (
                                EncoderResult::unmappable_from_bmp(bmp),
                                source.consumed(),
                                handle.written(),
                            );
                        };
                        let lead = pointer / 188;
                        let lead_offset = if lead < 0x1F { 0x81usize } else { 0xC1usize };
                        let trail = pointer % 188;
                        let trail_offset = if trail < 0x3F { 0x40usize } else { 0x41usize };
                        handle.write_two((lead + lead_offset) as u8, (trail + trail_offset) as u8)
                    }
                }
            }
        },
        bmp,
        self,
        source,
        handle,
        copy_ascii_to_check_space_two,
        check_space_two,
        false
    );
}

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

#[cfg(all(test, feature = "alloc"))]
mod tests {
    use super::super::testing::*;
    use super::super::*;

    fn decode_shift_jis(bytes: &[u8], expect: &str) {
        decode(SHIFT_JIS, bytes, expect);
    }

    fn encode_shift_jis(string: &str, expect: &[u8]) {
        encode(SHIFT_JIS, string, expect);
    }

    #[test]
    fn test_shift_jis_decode() {
        // Empty
        decode_shift_jis(b"", &"");

        // ASCII
        decode_shift_jis(b"\x61\x62", "\u{0061}\u{0062}");

        // Half-width
        decode_shift_jis(b"\xA1", "\u{FF61}");
        decode_shift_jis(b"\xDF", "\u{FF9F}");
        decode_shift_jis(b"\xA0", "\u{FFFD}");
        decode_shift_jis(b"\xE0", "\u{FFFD}");
        decode_shift_jis(b"\xA0+", "\u{FFFD}+");
        decode_shift_jis(b"\xE0+", "\u{FFFD}+");

        // EUDC
        decode_shift_jis(b"\xF0\x40", "\u{E000}");
        decode_shift_jis(b"\xF9\xFC", "\u{E757}");
        decode_shift_jis(b"\xEF\xFC", "\u{FFFD}");
        decode_shift_jis(b"\xFA\x40", "\u{2170}");

        // JIS 0208
        decode_shift_jis(b"\x81\x40", "\u{3000}");
        decode_shift_jis(b"\x81\x3F", "\u{FFFD}?");
        decode_shift_jis(b"\xEE\xFC", "\u{FF02}");
        decode_shift_jis(b"\xEE\xFD", "\u{FFFD}");
        decode_shift_jis(b"\xFA\x40", "\u{2170}");
        decode_shift_jis(b"\xFA\x3F", "\u{FFFD}?");
        decode_shift_jis(b"\xFC\x4B", "\u{9ED1}");
        decode_shift_jis(b"\xFC\x4C", "\u{FFFD}L");
        //
    }

    #[test]
    fn test_shift_jis_encode() {
        // Empty
        encode_shift_jis("", b"");

        // ASCII
        encode_shift_jis("\u{0061}\u{0062}", b"\x61\x62");

        // Exceptional code points
        encode_shift_jis("\u{0080}", b"\x80");
        encode_shift_jis("\u{00A5}", b"\x5C");
        encode_shift_jis("\u{203E}", b"\x7E");
        encode_shift_jis("\u{2212}", b"\x81\x7C");

        // Half-width
        encode_shift_jis("\u{FF61}", b"\xA1");
        encode_shift_jis("\u{FF9F}", b"\xDF");

        // EUDC
        encode_shift_jis("\u{E000}", b"");
        encode_shift_jis("\u{E757}", b"");

        // JIS 0212
        encode_shift_jis("\u{02D8}", b"˘");

        // JIS 0208
        encode_shift_jis("\u{3000}", b"\x81\x40");
        encode_shift_jis("\u{FF02}", b"\xFA\x57");
        encode_shift_jis("\u{2170}", b"\xFA\x40");
        encode_shift_jis("\u{9ED1}", b"\xFC\x4B");
    }

    #[test]
    #[cfg_attr(miri, ignore)] // Miri is too slow
    fn test_shift_jis_decode_all() {
        let input = include_bytes!("test_data/shift_jis_in.txt");
        let expectation = include_str!("test_data/shift_jis_in_ref.txt");
        let (cow, had_errors) = SHIFT_JIS.decode_without_bom_handling(input);
        assert!(had_errors, "Should have had errors.");
        assert_eq!(&cow[..], expectation);
    }

    #[test]
    #[cfg_attr(miri, ignore)] // Miri is too slow
    fn test_shift_jis_encode_all() {
        let input = include_str!("test_data/shift_jis_out.txt");
        let expectation = include_bytes!("test_data/shift_jis_out_ref.txt");
        let (cow, encoding, had_errors) = SHIFT_JIS.encode(input);
        assert!(!had_errors, "Should not have had errors.");
        assert_eq!(encoding, SHIFT_JIS);
        assert_eq!(&cow[..], &expectation[..]);
    }

    #[test]
    fn test_shift_jis_half_width_katakana_length() {
        let mut output = [0u8; 20];
        let mut decoder = SHIFT_JIS.new_decoder();
        {
            let needed = decoder
                .max_utf8_buffer_length_without_replacement(1)
                .unwrap();
            let (result, read, written) =
                decoder.decode_to_utf8_without_replacement(b"\xA1", &mut output[..needed], true);
            assert_eq!(result, DecoderResult::InputEmpty);
            assert_eq!(read, 1);
            assert_eq!(written, 3);
            assert_eq!(output[0], 0xEF);
            assert_eq!(output[1], 0xBD);
            assert_eq!(output[2], 0xA1);
        }
    }
}