Hye-Shik Chang | 3e2a306 | 2004-01-17 14:29:29 +0000 | [diff] [blame^] | 1 | /* |
| 2 | * codeccommon.h: Common Codec Routines |
| 3 | * |
| 4 | * Written by Hye-Shik Chang <perky@FreeBSD.org> |
| 5 | * $CJKCodecs: codeccommon.h,v 1.4 2003/12/31 05:46:55 perky Exp $ |
| 6 | */ |
| 7 | |
| 8 | #include "Python.h" |
| 9 | #include "multibytecodec.h" |
| 10 | |
| 11 | |
| 12 | #ifdef STRICT_BUILD |
| 13 | #define STRICT_SUFX "_strict" |
| 14 | #else |
| 15 | #define STRICT_SUFX |
| 16 | #endif |
| 17 | |
| 18 | #define ENCMAP(encoding) \ |
| 19 | const static encode_map *encoding##encmap; |
| 20 | #define DECMAP(encoding) \ |
| 21 | const static decode_map *encoding##decmap; |
| 22 | |
| 23 | #define ENCODER_INIT(encoding) \ |
| 24 | static int encoding##_encode_init( \ |
| 25 | MultibyteCodec_State *state) |
| 26 | #define ENCODER(encoding) \ |
| 27 | static int encoding##_encode( \ |
| 28 | MultibyteCodec_State *state, \ |
| 29 | const Py_UNICODE **inbuf, size_t inleft, \ |
| 30 | unsigned char **outbuf, size_t outleft, int flags) |
| 31 | #define ENCODER_RESET(encoding) \ |
| 32 | static int encoding##_encode_reset( \ |
| 33 | MultibyteCodec_State *state, \ |
| 34 | unsigned char **outbuf, size_t outleft) |
| 35 | |
| 36 | #define DECODER_INIT(encoding) \ |
| 37 | static int encoding##_decode_init( \ |
| 38 | MultibyteCodec_State *state) |
| 39 | #define DECODER(encoding) \ |
| 40 | static int encoding##_decode( \ |
| 41 | MultibyteCodec_State *state, \ |
| 42 | const unsigned char **inbuf, size_t inleft, \ |
| 43 | Py_UNICODE **outbuf, size_t outleft) |
| 44 | #define DECODER_RESET(encoding) \ |
| 45 | static int encoding##_decode_reset( \ |
| 46 | MultibyteCodec_State *state) |
| 47 | |
| 48 | #if Py_UNICODE_SIZE == 4 |
| 49 | #define UCS4INVALID(code) \ |
| 50 | if ((code) > 0xFFFF) \ |
| 51 | return 1; |
| 52 | #else |
| 53 | #define UCS4INVALID(code) \ |
| 54 | if (0) ; |
| 55 | #endif |
| 56 | |
| 57 | #define NEXT_IN(i) \ |
| 58 | (*inbuf) += (i); \ |
| 59 | (inleft) -= (i); |
| 60 | #define NEXT_OUT(o) \ |
| 61 | (*outbuf) += (o); \ |
| 62 | (outleft) -= (o); |
| 63 | #define NEXT(i, o) NEXT_IN(i) NEXT_OUT(o) |
| 64 | |
| 65 | #define RESERVE_INBUF(n) \ |
| 66 | if (inleft < (n)) \ |
| 67 | return MBERR_TOOFEW; |
| 68 | #define RESERVE_OUTBUF(n) \ |
| 69 | if (outleft < (n)) \ |
| 70 | return MBERR_TOOSMALL; |
| 71 | |
| 72 | #define IN1 ((*inbuf)[0]) |
| 73 | #define IN2 ((*inbuf)[1]) |
| 74 | #define IN3 ((*inbuf)[2]) |
| 75 | #define IN4 ((*inbuf)[3]) |
| 76 | |
| 77 | #define OUT1(c) ((*outbuf)[0]) = (c); |
| 78 | #define OUT2(c) ((*outbuf)[1]) = (c); |
| 79 | #define OUT3(c) ((*outbuf)[2]) = (c); |
| 80 | #define OUT4(c) ((*outbuf)[3]) = (c); |
| 81 | |
| 82 | #define WRITE1(c1) \ |
| 83 | RESERVE_OUTBUF(1) \ |
| 84 | (*outbuf)[0] = (c1); |
| 85 | #define WRITE2(c1, c2) \ |
| 86 | RESERVE_OUTBUF(2) \ |
| 87 | (*outbuf)[0] = (c1); \ |
| 88 | (*outbuf)[1] = (c2); |
| 89 | #define WRITE3(c1, c2, c3) \ |
| 90 | RESERVE_OUTBUF(3) \ |
| 91 | (*outbuf)[0] = (c1); \ |
| 92 | (*outbuf)[1] = (c2); \ |
| 93 | (*outbuf)[2] = (c3); |
| 94 | #define WRITE4(c1, c2, c3, c4) \ |
| 95 | RESERVE_OUTBUF(4) \ |
| 96 | (*outbuf)[0] = (c1); \ |
| 97 | (*outbuf)[1] = (c2); \ |
| 98 | (*outbuf)[2] = (c3); \ |
| 99 | (*outbuf)[3] = (c4); |
| 100 | |
| 101 | #if Py_UNICODE_SIZE == 2 |
| 102 | # define PUTUCS4(c) \ |
| 103 | RESERVE_OUTBUF(2) \ |
| 104 | (*outbuf)[0] = 0xd800 + (((c) - 0x10000) >> 10); \ |
| 105 | (*outbuf)[1] = 0xdc00 + (((c) - 0x10000) & 0x3ff); \ |
| 106 | NEXT_OUT(2) |
| 107 | #else |
| 108 | # define PUTUCS4(c) \ |
| 109 | RESERVE_OUTBUF(1) \ |
| 110 | **outbuf = (Py_UNICODE)(c); \ |
| 111 | NEXT_OUT(1) |
| 112 | #endif |
| 113 | |
| 114 | #define _TRYMAP_ENC(m, assi, val) \ |
| 115 | if ((m)->map != NULL && (val) >= (m)->bottom && \ |
| 116 | (val)<= (m)->top && ((assi) = (m)->map[(val) - \ |
| 117 | (m)->bottom]) != NOCHAR) |
| 118 | #define TRYMAP_ENC(charset, assi, uni) \ |
| 119 | _TRYMAP_ENC(&charset##encmap[(uni) >> 8], assi, (uni) & 0xff) |
| 120 | #define _TRYMAP_DEC(m, assi, val) \ |
| 121 | if ((m)->map != NULL && (val) >= (m)->bottom && \ |
| 122 | (val)<= (m)->top && ((assi) = (m)->map[(val) - \ |
| 123 | (m)->bottom]) != UNIINV) |
| 124 | #define TRYMAP_DEC(charset, assi, c1, c2) \ |
| 125 | _TRYMAP_DEC(&charset##decmap[c1], assi, c2) |
| 126 | |
| 127 | #if Py_UNICODE_SIZE == 2 |
| 128 | #define DECODE_SURROGATE(c) \ |
| 129 | if (c >> 10 == 0xd800 >> 10) { /* high surrogate */ \ |
| 130 | RESERVE_INBUF(2) \ |
| 131 | if (IN2 >> 10 == 0xdc00 >> 10) { /* low surrogate */ \ |
| 132 | c = 0x10000 + ((ucs4_t)(c - 0xd800) << 10) + \ |
| 133 | ((ucs4_t)(IN2) - 0xdc00); \ |
| 134 | } \ |
| 135 | } |
| 136 | #define GET_INSIZE(c) ((c) > 0xffff ? 2 : 1) |
| 137 | #else |
| 138 | #define DECODE_SURROGATE(c) {;} |
| 139 | #define GET_INSIZE(c) 1 |
| 140 | #endif |
| 141 | |
| 142 | #ifdef USING_BINARY_PAIR_SEARCH |
| 143 | static DBCHAR find_pairencmap(ucs2_t , ucs2_t, struct pair_encodemap *, int); |
| 144 | #endif |