#include <ruby/ruby.h>
#include <ruby/oniguruma.h>
#include <ruby/encoding.h>

#include "transcode_data.h"

#if defined(__GNUC__)
#  define ARG_UNUSED  __attribute__ ((unused))
#else
#  define ARG_UNUSED
#endif

/*
 * Encode
 */

static int
stateless_TADTextBE_mbc_enc_len(const OnigUChar* p ARG_UNUSED, const OnigUChar* e ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    return 4;
}

static int
stateless_TADTextBE_is_mbc_newline(const OnigUChar* p, const OnigUChar* e, OnigEncoding enc ARG_UNUSED)
{
    if (p + 3 < e) {
        if (p[2] == 0 && (p[3] == 0x0a || p[3] == 0x0b || p[3] == 0x0c || p[3] == 0x0d))
            return 1;
    }
    return 0;
}

static OnigCodePoint
stateless_TADTextBE_mbc_to_code(const OnigUChar* p, const OnigUChar* e ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    return (OnigCodePoint)(((p[0] * 256 + p[1]) * 256 + p[2]) * 256 + p[3]);
}

static int
stateless_TADTextBE_code_to_mbclen(OnigCodePoint code ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    return 4;
}

static int
stateless_TADTextBE_code_to_mbc(OnigCodePoint code, OnigUChar *buf, OnigEncoding enc ARG_UNUSED)
{
    OnigUChar* p = buf;

    *p++ = (OnigUChar)((code & 0xff000000) >> 24);
    *p++ = (OnigUChar)((code & 0xff0000)   >> 16);
    *p++ = (OnigUChar)((code & 0xff00)     >>  8);
    *p++ = (OnigUChar) (code & 0xff);

    return 4;
}

static int
stateless_TADTextBE_mbc_case_fold(OnigCaseFoldType flag ARG_UNUSED, const OnigUChar** pp, const OnigUChar* e, OnigUChar* to, OnigEncoding enc)
{
    int len;
    const OnigUChar* p = *pp;
    int i;

    /* TODO fold case */
 
    len = stateless_TADTextBE_mbc_enc_len(p, e, enc);
    for (i = 0; i < len; ++i) {
        *to++ = *p++;
    }
    (*pp) += len;

    return len; /* return byte length of converted char to lower */
}

static int
stateless_TADTextBE_apply_all_case_fold(OnigCaseFoldType flag ARG_UNUSED, OnigApplyAllCaseFoldFunc f ARG_UNUSED, void* arg ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    /* TODO what to do here ? */
    return 0;
}

static int
stateless_TADTextBE_get_case_fold_codes_by_str(OnigCaseFoldType flag ARG_UNUSED, const OnigUChar* p ARG_UNUSED, const OnigUChar* e ARG_UNUSED, OnigCaseFoldCodeItem acs[] ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    /* TODO what to do here ? */
    return 0;
}

static int
stateless_TADTextBE_property_name_to_ctype(OnigEncoding enc ARG_UNUSED, OnigUChar* p ARG_UNUSED, OnigUChar* e ARG_UNUSED)
{
    /* don't support property name */
    return ONIGERR_INVALID_CHAR_PROPERTY_NAME;
}

static int
stateless_TADTextBE_is_code_ctype(OnigCodePoint code ARG_UNUSED, OnigCtype ctype ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    /* don't support ctype */
    return 0;
}

static int
stateless_TADTextBE_get_ctype_code_range(OnigCtype ctype ARG_UNUSED, OnigCodePoint* sb_out ARG_UNUSED, const OnigCodePoint* ranges[] ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    /* don't support ctype */
    return ONIGERR_TYPE_BUG;
}

static OnigUChar*
stateless_TADTextBE_left_adjust_char_head(const OnigUChar* start, const OnigUChar* s, const OnigUChar* e ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    int rem;

    if (s <= start)
        return (UChar*)s;

    rem = (s - start) % 4;

    return (OnigUChar*)(s - rem);
}

static int
stateless_TADTextBE_is_allowed_reverse_match(const OnigUChar* p ARG_UNUSED, const OnigUChar* e ARG_UNUSED, OnigEncoding enc ARG_UNUSED)
{
    /* always didn't allow */
    return 0;
}

static OnigEncodingType stateless_TADTextBE = {
    &stateless_TADTextBE_mbc_enc_len,
    "stateless-TADTextBE",  /* name */
    4,  /* max byte length */
    4,  /* min byte length */
    &stateless_TADTextBE_is_mbc_newline,
    &stateless_TADTextBE_mbc_to_code,
    &stateless_TADTextBE_code_to_mbclen,
    &stateless_TADTextBE_code_to_mbc,
    &stateless_TADTextBE_mbc_case_fold,
    &stateless_TADTextBE_apply_all_case_fold,
    &stateless_TADTextBE_get_case_fold_codes_by_str,
    &stateless_TADTextBE_property_name_to_ctype,
    &stateless_TADTextBE_is_code_ctype,
    &stateless_TADTextBE_get_ctype_code_range,
    &stateless_TADTextBE_left_adjust_char_head,
    &stateless_TADTextBE_is_allowed_reverse_match
};

/*
 * Transcode
 */

<%
  map = {
    "00{00-20}" => :func_so,  # Control
    "{21-7e}{21-7e}" => :func_so,  # A Zone
    "{80-fd}{21-7e}" => :func_so,  # B Zone
    "{21-7e}{80-fd}" => :func_so,  # C Zone
    "{80-fd}{80-fd}" => :func_so,  # D Zone
    "fe{21-7e,80-fd}" => :func_so,  # Script/Language Change
    "fefe" => :func_si,  # Script/Language Escape
  }
  transcode_generate_node(ActionMap.parse(map), "TADTextBE_decoder")

  map = {
    "{00,fe}{21-7e,80-fd}" => :func_so,  # Script/Language Change
  }
  transcode_generate_node(ActionMap.parse(map), "TADTextBE_decoder_lang_escape")
%>

<%= transcode_generated_code %>

static int
tadtext_init(void *statep)
{
    uint16_t *sp = statep;

    *sp = 0x21;  /* System Script */

    return 0;
}

/*
 * fe21      - 0021
 * fe7e      - 007e
 * fefe 0021 - 0121
 * fefe 007e - 017e
 * fefe fe21 - 0221
 * fefe fe7e - 027e
 * fefe fefe 0021 - 0321
 * fefe fefe 007e - 037e
 */

static VALUE
fun_si_TADTextBE_decoder(void *statep, const unsigned char *s ARG_UNUSED, size_t l ARG_UNUSED)
{
    uint16_t *sp = statep;

    if (*sp & 0x8000) {
        *sp += 0x200;
        *sp |= 0x8000;  /* prevent overflow */
    }
    else {
        *sp = 0x8100;
    }

    return (VALUE)TADTextBE_decoder_lang_escape;
}

static ssize_t
fun_so_TADTextBE_decoder(void *statep, const unsigned char *s, size_t l ARG_UNUSED, unsigned char* o, size_t osize ARG_UNUSED)
{
    uint16_t *sp = statep;

    if (*sp & 0x8000) {
        *sp &= 0x7f00;
        if (s[2] == 0xfe) {
            *sp += 0x100;
        }
        *sp |= s[3];

        return 0;
    }

    if (s[0] == 0xfe) {
        *sp = s[1];

        return 0;
    }
    else {
        o[0] = *sp >> 8;
        o[1] = *sp & 0xff;
        o[2] = s[0];
        o[3] = s[1];

        return 4;
    }
}

static const rb_transcoder
rb_TADTextBE_decoder = {
    "TADTextBE", "stateless-TADTextBE", TADTextBE_decoder,
    TRANSCODE_TABLE_INFO,
    2, /* input_unit_length */
    4, /* max_input */
    4, /* max_output */
    incompat_converter, /* asciicompat_type */
    2, tadtext_init, tadtext_init, /* state_size, state_init, state_fini */
    NULL, fun_si_TADTextBE_decoder, NULL, fun_so_TADTextBE_decoder
};

/*
 * Initialize
 */

void
Init_troncode(void)
{
    rb_define_dummy_encoding("TADTextBE");
    rb_define_dummy_encoding("TADTextLE");
#if 0
    /*
     * ロードしなくてもEncoding.listやEncoding.name_listを出せるようにするために
     * rb_encdb_declare はあるので、後付けの場合は必要ない
     */
    rb_encdb_declare("stateless-TADTextBE");  /* いらない */
#endif
    rb_enc_register("stateless-TADTextBE", &stateless_TADTextBE);

    rb_register_transcoder(&rb_TADTextBE_decoder);
}