| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | /*
 | 
					
						
							|  |  |  |  * _codecs_jp.c: Codecs collection for Japanese encodings | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  * Written by Hye-Shik Chang <perky@FreeBSD.org> | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | #define USING_BINARY_PAIR_SEARCH
 | 
					
						
							|  |  |  | #define EMPBASE 0x20000
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | #include "cjkcodecs.h"
 | 
					
						
							|  |  |  | #include "mappings_jp.h"
 | 
					
						
							|  |  |  | #include "mappings_jisx0213_pair.h"
 | 
					
						
							|  |  |  | #include "alg_jisx0201.h"
 | 
					
						
							|  |  |  | #include "emu_jisx0213_2000.h"
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | /*
 | 
					
						
							|  |  |  |  * CP932 codec | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ENCODER(cp932) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |     while (*inpos < inlen) { | 
					
						
							|  |  |  |         Py_UCS4 c = INCHAR1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         DBCHAR code; | 
					
						
							|  |  |  |         unsigned char c1, c2; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (c <= 0x80) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE1((unsigned char)c); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xff61 && c <= 0xff9f) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE1(c - 0xfec0); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xf8f0 && c <= 0xf8f3) { | 
					
						
							|  |  |  |             /* Windows compatibility */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             REQUIRE_OUTBUF(1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c == 0xf8f0) | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |                 OUTBYTE1(0xa0); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else | 
					
						
							| 
									
										
										
										
											2018-11-09 10:12:06 +03:00
										 |  |  |                 OUTBYTE1(c - 0xf8f1 + 0xfd); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         if (c > 0xFFFF) | 
					
						
							|  |  |  |             return 1; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |         REQUIRE_OUTBUF(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-28 23:54:13 +01:00
										 |  |  |         if (TRYMAP_ENC(cp932ext, code, c)) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             OUTBYTE1(code >> 8); | 
					
						
							|  |  |  |             OUTBYTE2(code & 0xff); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							| 
									
										
										
										
											2013-10-28 23:54:13 +01:00
										 |  |  |         else if (TRYMAP_ENC(jisxcommon, code, c)) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (code & 0x8000) /* MSB set: JIS X 0212 */ | 
					
						
							|  |  |  |                 return 1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             /* JIS X 0208 */ | 
					
						
							|  |  |  |             c1 = code >> 8; | 
					
						
							|  |  |  |             c2 = code & 0xff; | 
					
						
							|  |  |  |             c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); | 
					
						
							|  |  |  |             c1 = (c1 - 0x21) >> 1; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1); | 
					
						
							|  |  |  |             OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xe000 && c < 0xe758) { | 
					
						
							|  |  |  |             /* User-defined area */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             c1 = (Py_UCS4)(c - 0xe000) / 188; | 
					
						
							|  |  |  |             c2 = (Py_UCS4)(c - 0xe000) % 188; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             OUTBYTE1(c1 + 0xf0); | 
					
						
							|  |  |  |             OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else | 
					
						
							|  |  |  |             return 1; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         NEXT(1, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | DECODER(cp932) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     while (inleft > 0) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         unsigned char c = INBYTE1, c2; | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |         Py_UCS4 decoded; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if (c <= 0x80) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             OUTCHAR(c); | 
					
						
							|  |  |  |             NEXT_IN(1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xa0 && c <= 0xdf) { | 
					
						
							|  |  |  |             if (c == 0xa0) | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xf8f0); /* half-width katakana */ | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xfec0 + c); | 
					
						
							|  |  |  |             NEXT_IN(1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xfd/* && c <= 0xff*/) { | 
					
						
							|  |  |  |             /* Windows compatibility */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             OUTCHAR(0xf8f1 - 0xfd + c); | 
					
						
							|  |  |  |             NEXT_IN(1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |         REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |         if (TRYMAP_DEC(cp932ext, decoded, c, c2)) | 
					
						
							|  |  |  |             OUTCHAR(decoded); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ | 
					
						
							|  |  |  |             if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |             c = (c < 0xe0 ? c - 0x81 : c - 0xc1); | 
					
						
							|  |  |  |             c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); | 
					
						
							|  |  |  |             c = (2 * c + (c2 < 0x5e ? 0 : 1) + 0x21); | 
					
						
							|  |  |  |             c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             if (TRYMAP_DEC(jisx0208, decoded, c, c2)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else if (c >= 0xf0 && c <= 0xf9) { | 
					
						
							|  |  |  |             if ((c2 >= 0x40 && c2 <= 0x7e) || | 
					
						
							|  |  |  |                 (c2 >= 0x80 && c2 <= 0xfc)) | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xe000 + 188 * (c - 0xf0) + | 
					
						
							|  |  |  |                     (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41)); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |             return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | /*
 | 
					
						
							|  |  |  |  * EUC-JIS-2004 codec | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ENCODER(euc_jis_2004) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |     while (*inpos < inlen) { | 
					
						
							|  |  |  |         Py_UCS4 c = INCHAR1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         DBCHAR code; | 
					
						
							|  |  |  |         Py_ssize_t insize; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (c < 0x80) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE1(c); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         insize = 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if (c <= 0xFFFF) { | 
					
						
							|  |  |  |             EMULATE_JISX0213_2000_ENCODE_BMP(code, c) | 
					
						
							| 
									
										
										
										
											2013-10-28 23:54:13 +01:00
										 |  |  |             else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 if (code == MULTIC) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                     if (inlen - *inpos < 2) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                         if (flags & MBENC_FLUSH) { | 
					
						
							|  |  |  |                             code = find_pairencmap( | 
					
						
							|  |  |  |                                 (ucs2_t)c, 0, | 
					
						
							|  |  |  |                               jisx0213_pair_encmap, | 
					
						
							|  |  |  |                                 JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                             if (code == DBCINV) | 
					
						
							|  |  |  |                                 return 1; | 
					
						
							|  |  |  |                         } | 
					
						
							|  |  |  |                         else | 
					
						
							|  |  |  |                             return MBERR_TOOFEW; | 
					
						
							|  |  |  |                     } | 
					
						
							|  |  |  |                     else { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                         Py_UCS4 c2 = INCHAR2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                         code = find_pairencmap( | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                             (ucs2_t)c, c2, | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                             jisx0213_pair_encmap, | 
					
						
							|  |  |  |                             JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                         if (code == DBCINV) { | 
					
						
							|  |  |  |                             code = find_pairencmap( | 
					
						
							|  |  |  |                                 (ucs2_t)c, 0, | 
					
						
							|  |  |  |                               jisx0213_pair_encmap, | 
					
						
							|  |  |  |                                 JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                             if (code == DBCINV) | 
					
						
							|  |  |  |                                 return 1; | 
					
						
							|  |  |  |                         } else | 
					
						
							|  |  |  |                             insize = 2; | 
					
						
							|  |  |  |                     } | 
					
						
							|  |  |  |                 } | 
					
						
							|  |  |  |             } | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             else if (TRYMAP_ENC(jisxcommon, code, c)) | 
					
						
							|  |  |  |                 ; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else if (c >= 0xff61 && c <= 0xff9f) { | 
					
						
							|  |  |  |                 /* JIS X 0201 half-width katakana */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |                 WRITEBYTE2(0x8e, c - 0xfec0); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                 NEXT(1, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							|  |  |  |             else if (c == 0xff3c) | 
					
						
							|  |  |  |                 /* F/W REVERSE SOLIDUS (see NOTES) */ | 
					
						
							|  |  |  |                 code = 0x2140; | 
					
						
							|  |  |  |             else if (c == 0xff5e) | 
					
						
							|  |  |  |                 /* F/W TILDE (see NOTES) */ | 
					
						
							|  |  |  |                 code = 0x2232; | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else if (c >> 16 == EMPBASE >> 16) { | 
					
						
							|  |  |  |             EMULATE_JISX0213_2000_ENCODE_EMP(code, c) | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             else if (TRYMAP_ENC(jisx0213_emp, code, c & 0xffff)) | 
					
						
							|  |  |  |                 ; | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 return insize; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else | 
					
						
							|  |  |  |             return insize; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (code & 0x8000) { | 
					
						
							|  |  |  |             /* Codeset 2 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(insize, 3); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } else { | 
					
						
							|  |  |  |             /* Codeset 1 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(insize, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | DECODER(euc_jis_2004) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     while (inleft > 0) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         unsigned char c = INBYTE1; | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |         Py_UCS4 code, decoded; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if (c < 0x80) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             OUTCHAR(c); | 
					
						
							|  |  |  |             NEXT_IN(1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (c == 0x8e) { | 
					
						
							|  |  |  |             /* JIS X 0201 half-width katakana */ | 
					
						
							|  |  |  |             unsigned char c2; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c2 >= 0xa1 && c2 <= 0xdf) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xfec0 + c2); | 
					
						
							|  |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else if (c == 0x8f) { | 
					
						
							|  |  |  |             unsigned char c2, c3; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(3); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2 ^ 0x80; | 
					
						
							|  |  |  |             c3 = INBYTE3 ^ 0x80; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |             /* JIS X 0213 Plane 2 or JIS X 0212 (see NOTES) */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             EMULATE_JISX0213_2000_DECODE_PLANE2(writer, c2, c3) | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c2, c3)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							|  |  |  |             else if (TRYMAP_DEC(jisx0213_2_emp, code, c2, c3)) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(EMPBASE | code); | 
					
						
							|  |  |  |                 NEXT_IN(3); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             else if (TRYMAP_DEC(jisx0212, decoded, c2, c3)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             NEXT_IN(3); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else { | 
					
						
							|  |  |  |             unsigned char c2; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             c ^= 0x80; | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2 ^ 0x80; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |             /* JIS X 0213 Plane 1 */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             EMULATE_JISX0213_2000_DECODE_PLANE1(writer, c, c2) | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             else if (c == 0x21 && c2 == 0x40) | 
					
						
							|  |  |  |                 OUTCHAR(0xff3c); | 
					
						
							|  |  |  |             else if (c == 0x22 && c2 == 0x32) | 
					
						
							|  |  |  |                 OUTCHAR(0xff5e); | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             else if (TRYMAP_DEC(jisx0208, decoded, c, c2)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							|  |  |  |             else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c, c2)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							|  |  |  |             else if (TRYMAP_DEC(jisx0213_1_emp, code, c, c2)) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(EMPBASE | code); | 
					
						
							|  |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             else if (TRYMAP_DEC(jisx0213_pair, code, c, c2)) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR2(code >> 16, code & 0xffff); | 
					
						
							|  |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | /*
 | 
					
						
							|  |  |  |  * EUC-JP codec | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ENCODER(euc_jp) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |     while (*inpos < inlen) { | 
					
						
							|  |  |  |         Py_UCS4 c = INCHAR1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         DBCHAR code; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (c < 0x80) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE1((unsigned char)c); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         if (c > 0xFFFF) | 
					
						
							|  |  |  |             return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         if (TRYMAP_ENC(jisxcommon, code, c)) | 
					
						
							|  |  |  |             ; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else if (c >= 0xff61 && c <= 0xff9f) { | 
					
						
							|  |  |  |             /* JIS X 0201 half-width katakana */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE2(0x8e, c - 0xfec0); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #ifndef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else if (c == 0xff3c) /* FULL-WIDTH REVERSE SOLIDUS */ | 
					
						
							|  |  |  |             code = 0x2140; | 
					
						
							|  |  |  |         else if (c == 0xa5) { /* YEN SIGN */ | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             WRITEBYTE1(0x5c); | 
					
						
							|  |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } else if (c == 0x203e) { /* OVERLINE */ | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             WRITEBYTE1(0x7e); | 
					
						
							|  |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else | 
					
						
							|  |  |  |             return 1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (code & 0x8000) { | 
					
						
							|  |  |  |             /* JIS X 0212 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE3(0x8f, code >> 8, (code & 0xFF) | 0x80); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 3); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } else { | 
					
						
							|  |  |  |             /* JIS X 0208 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE2((code >> 8) | 0x80, (code & 0xFF) | 0x80); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | DECODER(euc_jp) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     while (inleft > 0) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         unsigned char c = INBYTE1; | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |         Py_UCS4 decoded; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-06-03 23:34:09 +02:00
										 |  |  |         if (c < 0x80) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             OUTCHAR(c); | 
					
						
							|  |  |  |             NEXT_IN(1); | 
					
						
							| 
									
										
										
										
											2011-06-03 23:34:09 +02:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if (c == 0x8e) { | 
					
						
							|  |  |  |             /* JIS X 0201 half-width katakana */ | 
					
						
							|  |  |  |             unsigned char c2; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c2 >= 0xa1 && c2 <= 0xdf) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xfec0 + c2); | 
					
						
							|  |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else if (c == 0x8f) { | 
					
						
							|  |  |  |             unsigned char c2, c3; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(3); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							|  |  |  |             c3 = INBYTE3; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             /* JIS X 0212 */ | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             if (TRYMAP_DEC(jisx0212, decoded, c2 ^ 0x80, c3 ^ 0x80)) { | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 NEXT_IN(3); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else { | 
					
						
							|  |  |  |             unsigned char c2; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             /* JIS X 0208 */ | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #ifndef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c == 0xa1 && c2 == 0xc0) | 
					
						
							|  |  |  |                 /* FULL-WIDTH REVERSE SOLIDUS */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xff3c); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             if (TRYMAP_DEC(jisx0208, decoded, c ^ 0x80, c2 ^ 0x80)) | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |             NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | /*
 | 
					
						
							|  |  |  |  * SHIFT_JIS codec | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ENCODER(shift_jis) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |     while (*inpos < inlen) { | 
					
						
							|  |  |  |         Py_UCS4 c = INCHAR1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         DBCHAR code; | 
					
						
							|  |  |  |         unsigned char c1, c2; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  | #ifdef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         JISX0201_R_ENCODE(c, code) | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #else
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         if (c < 0x80) | 
					
						
							|  |  |  |             code = c; | 
					
						
							|  |  |  |         else if (c == 0x00a5) | 
					
						
							|  |  |  |             code = 0x5c; /* YEN SIGN */ | 
					
						
							|  |  |  |         else if (c == 0x203e) | 
					
						
							|  |  |  |             code = 0x7e; /* OVERLINE */ | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else JISX0201_K_ENCODE(c, code) | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         else if (c > 0xFFFF) | 
					
						
							|  |  |  |             return 1; | 
					
						
							|  |  |  |         else | 
					
						
							|  |  |  |             code = NOCHAR; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             REQUIRE_OUTBUF(1); | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             OUTBYTE1((unsigned char)code); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |         REQUIRE_OUTBUF(2); | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         if (code == NOCHAR) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             if (TRYMAP_ENC(jisxcommon, code, c)) | 
					
						
							|  |  |  |                 ; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #ifndef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else if (c == 0xff3c) | 
					
						
							|  |  |  |                 code = 0x2140; /* FULL-WIDTH REVERSE SOLIDUS */ | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             else | 
					
						
							|  |  |  |                 return 1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             if (code & 0x8000) /* MSB set: JIS X 0212 */ | 
					
						
							|  |  |  |                 return 1; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         c1 = code >> 8; | 
					
						
							|  |  |  |         c2 = code & 0xff; | 
					
						
							|  |  |  |         c2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21); | 
					
						
							|  |  |  |         c1 = (c1 - 0x21) >> 1; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |         OUTBYTE1(c1 < 0x1f ? c1 + 0x81 : c1 + 0xc1); | 
					
						
							|  |  |  |         OUTBYTE2(c2 < 0x3f ? c2 + 0x40 : c2 + 0x41); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         NEXT(1, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | DECODER(shift_jis) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     while (inleft > 0) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         unsigned char c = INBYTE1; | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |         Py_UCS4 decoded; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  | #ifdef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         JISX0201_R_DECODE(c, writer) | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #else
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         if (c < 0x80) | 
					
						
							|  |  |  |             OUTCHAR(c); | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         else JISX0201_K_DECODE(c, writer) | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xea)){ | 
					
						
							|  |  |  |             unsigned char c1, c2; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); | 
					
						
							|  |  |  |             c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); | 
					
						
							|  |  |  |             c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1) + 0x21); | 
					
						
							|  |  |  |             c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  | #ifndef STRICT_BUILD
 | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c1 == 0x21 && c2 == 0x40) { | 
					
						
							|  |  |  |                 /* FULL-WIDTH REVERSE SOLIDUS */ | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 OUTCHAR(0xff3c); | 
					
						
							|  |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | #endif
 | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) { | 
					
						
							|  |  |  |                 OUTCHAR(decoded); | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 continue; | 
					
						
							|  |  |  |             } | 
					
						
							|  |  |  |             else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							|  |  |  |         else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |             return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         NEXT_IN(1); /* JIS X 0201 */ | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | /*
 | 
					
						
							|  |  |  |  * SHIFT_JIS-2004 codec | 
					
						
							|  |  |  |  */ | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | ENCODER(shift_jis_2004) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |     while (*inpos < inlen) { | 
					
						
							|  |  |  |         Py_UCS4 c = INCHAR1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         DBCHAR code = NOCHAR; | 
					
						
							|  |  |  |         int c1, c2; | 
					
						
							|  |  |  |         Py_ssize_t insize; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         JISX0201_ENCODE(c, code) | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         if (code < 0x80 || (code >= 0xa1 && code <= 0xdf)) { | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |             WRITEBYTE1((unsigned char)code); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             NEXT(1, 1); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |         REQUIRE_OUTBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         insize = 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |         if (code == NOCHAR) { | 
					
						
							|  |  |  |             if (c <= 0xffff) { | 
					
						
							|  |  |  |                 EMULATE_JISX0213_2000_ENCODE_BMP(code, c) | 
					
						
							| 
									
										
										
										
											2013-10-28 23:54:13 +01:00
										 |  |  |                 else if (TRYMAP_ENC(jisx0213_bmp, code, c)) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                     if (code == MULTIC) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                         if (inlen - *inpos < 2) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                             if (flags & MBENC_FLUSH) { | 
					
						
							|  |  |  |                             code = find_pairencmap | 
					
						
							|  |  |  |                                 ((ucs2_t)c, 0, | 
					
						
							|  |  |  |                               jisx0213_pair_encmap, | 
					
						
							|  |  |  |                                 JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                             if (code == DBCINV) | 
					
						
							|  |  |  |                                 return 1; | 
					
						
							|  |  |  |                             } | 
					
						
							|  |  |  |                             else | 
					
						
							|  |  |  |                                 return MBERR_TOOFEW; | 
					
						
							|  |  |  |                         } | 
					
						
							|  |  |  |                         else { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                             Py_UCS4 ch2 = INCHAR2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                             code = find_pairencmap( | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |                                 (ucs2_t)c, ch2, | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                               jisx0213_pair_encmap, | 
					
						
							|  |  |  |                                 JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                             if (code == DBCINV) { | 
					
						
							|  |  |  |                             code = find_pairencmap( | 
					
						
							|  |  |  |                                 (ucs2_t)c, 0, | 
					
						
							|  |  |  |                               jisx0213_pair_encmap, | 
					
						
							|  |  |  |                                 JISX0213_ENCPAIRS); | 
					
						
							|  |  |  |                             if (code == DBCINV) | 
					
						
							|  |  |  |                                 return 1; | 
					
						
							|  |  |  |                             } | 
					
						
							|  |  |  |                             else | 
					
						
							|  |  |  |                                 insize = 2; | 
					
						
							|  |  |  |                         } | 
					
						
							|  |  |  |                     } | 
					
						
							|  |  |  |                 } | 
					
						
							| 
									
										
										
										
											2013-10-28 23:54:13 +01:00
										 |  |  |                 else if (TRYMAP_ENC(jisxcommon, code, c)) { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                     /* abandon JIS X 0212 codes */ | 
					
						
							|  |  |  |                     if (code & 0x8000) | 
					
						
							|  |  |  |                         return 1; | 
					
						
							|  |  |  |                 } | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |                 else | 
					
						
							|  |  |  |                     return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else if (c >> 16 == EMPBASE >> 16) { | 
					
						
							|  |  |  |                 EMULATE_JISX0213_2000_ENCODE_EMP(code, c) | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |                 else if (TRYMAP_ENC(jisx0213_emp, code, c&0xffff)) | 
					
						
							|  |  |  |                     ; | 
					
						
							|  |  |  |                 else | 
					
						
							|  |  |  |                     return insize; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 return insize; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         c1 = code >> 8; | 
					
						
							|  |  |  |         c2 = (code & 0xff) - 0x21; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |         if (c1 & 0x80) { | 
					
						
							|  |  |  |             /* Plane 2 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |             if (c1 >= 0xee) | 
					
						
							|  |  |  |                 c1 -= 0x87; | 
					
						
							|  |  |  |             else if (c1 >= 0xac || c1 == 0xa8) | 
					
						
							|  |  |  |                 c1 -= 0x49; | 
					
						
							|  |  |  |             else | 
					
						
							|  |  |  |                 c1 -= 0x43; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         } | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         else { | 
					
						
							|  |  |  |             /* Plane 1 */ | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             c1 -= 0x21; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         } | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |         if (c1 & 1) | 
					
						
							|  |  |  |             c2 += 0x5e; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         c1 >>= 1; | 
					
						
							| 
									
										
										
										
											2013-10-29 00:19:27 +01:00
										 |  |  |         OUTBYTE1(c1 + (c1 < 0x1f ? 0x81 : 0xc1)); | 
					
						
							|  |  |  |         OUTBYTE2(c2 + (c2 < 0x3f ? 0x40 : 0x41)); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         NEXT(insize, 2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | DECODER(shift_jis_2004) | 
					
						
							|  |  |  | { | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     while (inleft > 0) { | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |         unsigned char c = INBYTE1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         JISX0201_DECODE(c, writer) | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |         else if ((c >= 0x81 && c <= 0x9f) || (c >= 0xe0 && c <= 0xfc)){ | 
					
						
							|  |  |  |             unsigned char c1, c2; | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |             Py_UCS4 code, decoded; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-10-29 00:59:44 +01:00
										 |  |  |             REQUIRE_INBUF(2); | 
					
						
							| 
									
										
										
										
											2013-04-14 02:06:32 +02:00
										 |  |  |             c2 = INBYTE2; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             if (c2 < 0x40 || (c2 > 0x7e && c2 < 0x80) || c2 > 0xfc) | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                 return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							|  |  |  |             c1 = (c < 0xe0 ? c - 0x81 : c - 0xc1); | 
					
						
							|  |  |  |             c2 = (c2 < 0x80 ? c2 - 0x40 : c2 - 0x41); | 
					
						
							|  |  |  |             c1 = (2 * c1 + (c2 < 0x5e ? 0 : 1)); | 
					
						
							|  |  |  |             c2 = (c2 < 0x5e ? c2 : c2 - 0x5e) + 0x21; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |             if (c1 < 0x5e) { /* Plane 1 */ | 
					
						
							|  |  |  |                 c1 += 0x21; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 EMULATE_JISX0213_2000_DECODE_PLANE1(writer, | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                                 c1, c2) | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |                 else if (TRYMAP_DEC(jisx0208, decoded, c1, c2)) | 
					
						
							|  |  |  |                     OUTCHAR(decoded); | 
					
						
							|  |  |  |                 else if (TRYMAP_DEC(jisx0213_1_bmp, decoded, c1, c2)) | 
					
						
							|  |  |  |                     OUTCHAR(decoded); | 
					
						
							|  |  |  |                 else if (TRYMAP_DEC(jisx0213_1_emp, code, c1, c2)) | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                     OUTCHAR(EMPBASE | code); | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |                 else if (TRYMAP_DEC(jisx0213_pair, code, c1, c2)) | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                     OUTCHAR2(code >> 16, code & 0xffff); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                 else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                     return 1; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             else { /* Plane 2 */ | 
					
						
							| 
									
										
										
										
											2013-10-29 00:09:41 +01:00
										 |  |  |                 if (c1 >= 0x67) | 
					
						
							|  |  |  |                     c1 += 0x07; | 
					
						
							|  |  |  |                 else if (c1 >= 0x63 || c1 == 0x5f) | 
					
						
							|  |  |  |                     c1 -= 0x37; | 
					
						
							|  |  |  |                 else | 
					
						
							|  |  |  |                     c1 -= 0x3d; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 EMULATE_JISX0213_2000_DECODE_PLANE2(writer, | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                                 c1, c2) | 
					
						
							| 
									
										
										
										
											2013-10-28 23:18:39 +01:00
										 |  |  |                 else if (TRYMAP_DEC(jisx0213_2_bmp, decoded, c1, c2)) | 
					
						
							|  |  |  |                     OUTCHAR(decoded); | 
					
						
							|  |  |  |                 else if (TRYMAP_DEC(jisx0213_2_emp, code, c1, c2)) { | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                     OUTCHAR(EMPBASE | code); | 
					
						
							|  |  |  |                     NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |                     continue; | 
					
						
							|  |  |  |                 } | 
					
						
							|  |  |  |                 else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |                     return 1; | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |                 NEXT_IN(2); | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |             } | 
					
						
							|  |  |  |             continue; | 
					
						
							|  |  |  |         } | 
					
						
							|  |  |  |         else | 
					
						
							| 
									
										
										
										
											2011-07-08 01:45:13 +02:00
										 |  |  |             return 1; | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-04-11 22:09:04 +02:00
										 |  |  |         NEXT_IN(1); /* JIS X 0201 */ | 
					
						
							| 
									
										
										
										
											2010-05-09 15:52:27 +00:00
										 |  |  |     } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |     return 0; | 
					
						
							| 
									
										
										
										
											2004-07-18 03:06:29 +00:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | BEGIN_MAPPINGS_LIST | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0208) | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0212) | 
					
						
							|  |  |  |   MAPPING_ENCONLY(jisxcommon) | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0213_1_bmp) | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0213_2_bmp) | 
					
						
							|  |  |  |   MAPPING_ENCONLY(jisx0213_bmp) | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0213_1_emp) | 
					
						
							|  |  |  |   MAPPING_DECONLY(jisx0213_2_emp) | 
					
						
							|  |  |  |   MAPPING_ENCONLY(jisx0213_emp) | 
					
						
							|  |  |  |   MAPPING_ENCDEC(jisx0213_pair) | 
					
						
							|  |  |  |   MAPPING_ENCDEC(cp932ext) | 
					
						
							|  |  |  | END_MAPPINGS_LIST | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | BEGIN_CODECS_LIST | 
					
						
							|  |  |  |   CODEC_STATELESS(shift_jis) | 
					
						
							|  |  |  |   CODEC_STATELESS(cp932) | 
					
						
							|  |  |  |   CODEC_STATELESS(euc_jp) | 
					
						
							|  |  |  |   CODEC_STATELESS(shift_jis_2004) | 
					
						
							|  |  |  |   CODEC_STATELESS(euc_jis_2004) | 
					
						
							|  |  |  |   { "euc_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(euc_jis_2004) }, | 
					
						
							|  |  |  |   { "shift_jisx0213", (void *)2000, NULL, _STATELESS_METHODS(shift_jis_2004) }, | 
					
						
							|  |  |  | END_CODECS_LIST | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | I_AM_A_MODULE_FOR(jp) |