10
|
1 /** Network Kanji Filter. (PDS Version)
|
|
2 ************************************************************************
|
|
3 ** Copyright (C) 1987, Fujitsu LTD. (Itaru ICHIKAWA)
|
|
4 ** $BO"Mm@h!'(B $B!J3t!KIY;NDL8&5f=j!!%=%U%H#38&!!;T@n!!;j(B
|
|
5 ** $B!J(BE-Mail Address: ichikawa@flab.fujitsu.co.jp$B!K(B
|
|
6 ** Copyright (C) 1996,1998
|
|
7 ** Copyright (C) 2002
|
|
8 ** $BO"Mm@h!'(B $BN05eBg3X>pJs9)3X2J(B $B2OLn(B $B??<#(B mime/X0208 support
|
|
9 ** $B!J(BE-Mail Address: kono@ie.u-ryukyu.ac.jp$B!K(B
|
|
10 ** $BO"Mm@h!'(B COW for DOS & Win16 & Win32 & OS/2
|
|
11 ** $B!J(BE-Mail Address: GHG00637@niftyserve.or.p$B!K(B
|
|
12 **
|
|
13 ** $B$3$N%=!<%9$N$$$+$J$kJ#<L!$2~JQ!$=$@5$b5vBz$7$^$9!#$?$@$7!"(B
|
|
14 ** $B$=$N:]$K$O!"C/$,9W8%$7$?$r<($9$3$NItJ,$r;D$9$3$H!#(B
|
|
15 ** $B:FG[I[$d;(;o$NIUO?$J$I$NLd$$9g$o$;$bI,MW$"$j$^$;$s!#(B
|
|
16 ** $B1DMxMxMQ$b>e5-$KH?$7$J$$HO0O$G5v2D$7$^$9!#(B
|
|
17 ** $B%P%$%J%j$NG[I[$N:]$K$O(Bversion message$B$rJ]B8$9$k$3$H$r>r7o$H$7$^$9!#(B
|
|
18 ** $B$3$N%W%m%0%i%`$K$D$$$F$OFC$K2?$NJ]>Z$b$7$J$$!"0-$7$+$i$:!#(B
|
|
19 **
|
|
20 ** Everyone is permitted to do anything on this program
|
|
21 ** including copying, modifying, improving,
|
|
22 ** as long as you don't try to pretend that you wrote it.
|
|
23 ** i.e., the above copyright notice has to appear in all copies.
|
|
24 ** Binary distribution requires original version messages.
|
|
25 ** You don't have to ask before copying, redistribution or publishing.
|
|
26 ** THE AUTHOR DISCLAIMS ALL WARRANTIES WITH REGARD TO THIS SOFTWARE.
|
|
27 ***********************************************************************/
|
|
28
|
|
29 /***********************************************************************
|
|
30 ** UTF-8 $B%5%]!<%H$K$D$$$F(B
|
|
31 ** $B=>Mh$N(B nkf $B$HF~$l$+$($F$=$N$^$^;H$($k$h$&$K$J$C$F$$$^$9(B
|
|
32 ** nkf -e $B$J$I$H$7$F5/F0$9$k$H!"<+F0H=JL$G(B UTF-8 $B$HH=Dj$5$l$l$P!"(B
|
|
33 ** $B$=$N$^$^(B euc-jp $B$KJQ49$5$l$^$9(B
|
|
34 **
|
|
35 ** $B$^$@%P%0$,$"$k2DG=@-$,9b$$$G$9!#(B
|
|
36 ** ($BFC$K<+F0H=JL!"%3!<%I:.:_!"%(%i!<=hM}7O(B)
|
|
37 **
|
|
38 ** $B2?$+LdBj$r8+$D$1$?$i!"(B
|
|
39 ** E-Mail: furukawa@tcp-ip.or.jp
|
|
40 ** $B$^$G8fO"Mm$r$*4j$$$7$^$9!#(B
|
|
41 ***********************************************************************/
|
|
42 /* $Id$ */
|
|
43 #define NKF_VERSION "2.0.5"
|
|
44 #define NKF_RELEASE_DATE "2005-04-10"
|
|
45 #include "config.h"
|
|
46
|
|
47 static char *CopyRight =
|
|
48 "Copyright (C) 1987, FUJITSU LTD. (I.Ichikawa),2000 S. Kono, COW, 2002-2005 Kono, Furukawa, Naruse";
|
|
49
|
|
50
|
|
51 /*
|
|
52 **
|
|
53 **
|
|
54 **
|
|
55 ** USAGE: nkf [flags] [file]
|
|
56 **
|
|
57 ** Flags:
|
|
58 ** b Output is buffered (DEFAULT)
|
|
59 ** u Output is unbuffered
|
|
60 **
|
|
61 ** t no operation
|
|
62 **
|
|
63 ** j Outout code is JIS 7 bit (DEFAULT SELECT)
|
|
64 ** s Output code is MS Kanji (DEFAULT SELECT)
|
|
65 ** e Output code is AT&T JIS (DEFAULT SELECT)
|
|
66 ** w Output code is AT&T JIS (DEFAULT SELECT)
|
|
67 ** l Output code is JIS 7bit and ISO8859-1 Latin-1
|
|
68 **
|
|
69 ** m MIME conversion for ISO-2022-JP
|
|
70 ** I Convert non ISO-2022-JP charactor to GETA by Pekoe <pekoe@lair.net>
|
|
71 ** i_ Output sequence to designate JIS-kanji (DEFAULT_J)
|
|
72 ** o_ Output sequence to designate single-byte roman characters (DEFAULT_R)
|
|
73 ** M MIME output conversion
|
|
74 **
|
|
75 ** r {de/en}crypt ROT13/47
|
|
76 **
|
|
77 ** v display Version
|
|
78 **
|
|
79 ** T Text mode output (for MS-DOS)
|
|
80 **
|
|
81 ** x Do not convert X0201 kana into X0208
|
|
82 ** Z Convert X0208 alphabet to ASCII
|
|
83 **
|
|
84 ** f60 fold option
|
|
85 **
|
|
86 ** m MIME decode
|
|
87 ** B try to fix broken JIS, missing Escape
|
|
88 ** B[1-9] broken level
|
|
89 **
|
|
90 ** O Output to 'nkf.out' file or last file name
|
|
91 ** d Delete \r in line feed
|
|
92 ** c Add \r in line feed
|
|
93 ** -- other long option
|
|
94 ** -- ignore following option (don't use with -O )
|
|
95 **
|
|
96 **/
|
|
97
|
|
98 #if (defined(__TURBOC__) || defined(_MSC_VER) || defined(LSI_C) || defined(__MINGW32__)) && !defined(MSDOS)
|
|
99 #define MSDOS
|
|
100 #if (defined(__Win32__) || defined(_WIN32)) && !defined(__WIN32__)
|
|
101 #define __WIN32__
|
|
102 #endif
|
|
103 #endif
|
|
104
|
|
105 #ifdef PERL_XS
|
|
106 #undef OVERWRITE
|
|
107 #endif
|
|
108
|
|
109 #ifndef PERL_XS
|
|
110 #include <stdio.h>
|
|
111 #endif
|
|
112
|
|
113 #include <stdlib.h>
|
|
114
|
|
115 #if defined(MSDOS) || defined(__OS2__)
|
|
116 #include <fcntl.h>
|
|
117 #include <io.h>
|
|
118 #endif
|
|
119
|
|
120 #ifdef MSDOS
|
|
121 #ifdef LSI_C
|
|
122 #define setbinmode(fp) fsetbin(fp)
|
|
123 #else /* Microsoft C, Turbo C */
|
|
124 #define setbinmode(fp) setmode(fileno(fp), O_BINARY)
|
|
125 #endif
|
|
126 #else /* UNIX,OS/2 */
|
|
127 #define setbinmode(fp)
|
|
128 #endif
|
|
129
|
|
130 #ifdef _IOFBF /* SysV and MSDOS, Windows */
|
|
131 #define setvbuffer(fp, buf, size) setvbuf(fp, buf, _IOFBF, size)
|
|
132 #else /* BSD */
|
|
133 #define setvbuffer(fp, buf, size) setbuffer(fp, buf, size)
|
|
134 #endif
|
|
135
|
|
136 /*Borland C++ 4.5 EasyWin*/
|
|
137 #if defined(__TURBOC__) && defined(_Windows) && !defined(__WIN32__) /*Easy Win */
|
|
138 #define EASYWIN
|
|
139 #ifndef __WIN16__
|
|
140 #define __WIN16__
|
|
141 #endif
|
|
142 #include <windows.h>
|
|
143 #endif
|
|
144
|
|
145 #ifdef OVERWRITE
|
|
146 /* added by satoru@isoternet.org */
|
|
147 #include <string.h>
|
|
148 #include <sys/stat.h>
|
|
149 #ifndef MSDOS /* UNIX, OS/2 */
|
|
150 #include <unistd.h>
|
|
151 #include <utime.h>
|
|
152 #else /* defined(MSDOS) */
|
|
153 #ifdef __WIN32__
|
|
154 #ifdef __BORLANDC__ /* BCC32 */
|
|
155 #include <utime.h>
|
|
156 #else /* !defined(__BORLANDC__) */
|
|
157 #include <sys/utime.h>
|
|
158 #endif /* (__BORLANDC__) */
|
|
159 #else /* !defined(__WIN32__) */
|
|
160 #if defined(_MSC_VER) || defined(__MINGW32__) /* VC++, MinGW */
|
|
161 #include <sys/utime.h>
|
|
162 #elif defined(__TURBOC__) /* BCC */
|
|
163 #include <utime.h>
|
|
164 #elif defined(LSI_C) /* LSI C */
|
|
165 #endif /* (__WIN32__) */
|
|
166 #endif
|
|
167 #endif
|
|
168 #endif
|
|
169
|
|
170 #ifdef INT_IS_SHORT
|
|
171 #define int long
|
|
172 #endif
|
|
173
|
|
174 #define FALSE 0
|
|
175 #define TRUE 1
|
|
176
|
|
177 /* state of output_mode and input_mode
|
|
178
|
|
179 c2 0 means ASCII
|
|
180 X0201
|
|
181 ISO8859_1
|
|
182 X0208
|
|
183 EOF all termination
|
|
184 c1 32bit data
|
|
185
|
|
186 */
|
|
187
|
|
188 #define ASCII 0
|
|
189 #define X0208 1
|
|
190 #define X0201 2
|
|
191 #define ISO8859_1 8
|
|
192 #define NO_X0201 3
|
|
193 #define X0212 16
|
|
194
|
|
195 /* Input Assumption */
|
|
196
|
|
197 #define JIS_INPUT 4
|
|
198 #define SJIS_INPUT 5
|
|
199 #define LATIN1_INPUT 6
|
|
200 #define FIXED_MIME 7
|
|
201 #define STRICT_MIME 8
|
|
202
|
|
203 /* MIME ENCODE */
|
|
204
|
|
205 #define ISO2022JP 9
|
|
206 #define JAPANESE_EUC 10
|
|
207 #define SHIFT_JIS 11
|
|
208
|
|
209 #define UTF8 12
|
|
210 #define UTF8_INPUT 13
|
|
211 #define UTF16BE_INPUT 14
|
|
212 #define UTF16LE_INPUT 15
|
|
213
|
|
214 #define WISH_TRUE 15
|
|
215
|
|
216 /* ASCII CODE */
|
|
217
|
|
218 #define BS 0x08
|
|
219 #define TAB 0x09
|
|
220 #define NL 0x0a
|
|
221 #define CR 0x0d
|
|
222 #define ESC 0x1b
|
|
223 #define SPACE 0x20
|
|
224 #define AT 0x40
|
|
225 #define SSP 0xa0
|
|
226 #define DEL 0x7f
|
|
227 #define SI 0x0f
|
|
228 #define SO 0x0e
|
|
229 #define SSO 0x8e
|
|
230 #define SS3 0x8f
|
|
231
|
|
232 #define is_alnum(c) \
|
|
233 (('a'<=c && c<='z')||('A'<= c && c<='Z')||('0'<=c && c<='9'))
|
|
234
|
|
235 #define HOLD_SIZE 1024
|
|
236 #define IOBUF_SIZE 16384
|
|
237
|
|
238 #define DEFAULT_J 'B'
|
|
239 #define DEFAULT_R 'B'
|
|
240
|
|
241 #define SJ0162 0x00e1 /* 01 - 62 ku offset */
|
|
242 #define SJ6394 0x0161 /* 63 - 94 ku offset */
|
|
243
|
|
244 #define RANGE_NUM_MAX 18
|
|
245 #define GETA1 0x22
|
|
246 #define GETA2 0x2e
|
|
247
|
|
248
|
|
249 #if defined( UTF8_OUTPUT_ENABLE ) || defined( UTF8_INPUT_ENABLE )
|
|
250 #define sizeof_euc_utf8 94
|
|
251 #define sizeof_euc_to_utf8_1byte 94
|
|
252 #define sizeof_euc_to_utf8_2bytes 94
|
|
253 #define sizeof_utf8_to_euc_C2 64
|
|
254 #define sizeof_utf8_to_euc_E5B8 64
|
|
255 #define sizeof_utf8_to_euc_2bytes 112
|
|
256 #define sizeof_utf8_to_euc_3bytes 112
|
|
257 #endif
|
|
258
|
|
259 /* MIME preprocessor */
|
|
260
|
|
261
|
|
262 #ifdef EASYWIN /*Easy Win */
|
|
263 extern POINT _BufferSize;
|
|
264 #endif
|
|
265
|
|
266 /* function prototype */
|
|
267
|
|
268 #ifdef ANSI_C_PROTOTYPE
|
|
269 #define PROTO(x) x
|
|
270 #define STATIC static
|
|
271 #else
|
|
272 #define PROTO(x) ()
|
|
273 #define STATIC
|
|
274 #endif
|
|
275
|
|
276 struct input_code{
|
|
277 char *name;
|
|
278 int stat;
|
|
279 int score;
|
|
280 int index;
|
|
281 int buf[3];
|
|
282 void (*status_func)PROTO((struct input_code *, int));
|
|
283 int (*iconv_func)PROTO((int c2, int c1, int c0));
|
|
284 int _file_stat;
|
|
285 };
|
|
286
|
|
287 STATIC char *input_codename = "";
|
|
288
|
|
289 STATIC int noconvert PROTO((FILE *f));
|
|
290 STATIC int kanji_convert PROTO((FILE *f));
|
|
291 STATIC int h_conv PROTO((FILE *f,int c2,int c1));
|
|
292 STATIC int push_hold_buf PROTO((int c2));
|
|
293 STATIC void set_iconv PROTO((int f, int (*iconv_func)(int c2,int c1,int c0)));
|
|
294 STATIC int s_iconv PROTO((int c2,int c1,int c0));
|
|
295 STATIC int s2e_conv PROTO((int c2, int c1, int *p2, int *p1));
|
|
296 STATIC int e_iconv PROTO((int c2,int c1,int c0));
|
|
297 #ifdef UTF8_INPUT_ENABLE
|
|
298 STATIC int w2e_conv PROTO((int c2,int c1,int c0,int *p2,int *p1));
|
|
299 STATIC int w_iconv PROTO((int c2,int c1,int c0));
|
|
300 STATIC int w_iconv16 PROTO((int c2,int c1,int c0));
|
|
301 STATIC int w_iconv_common PROTO((int c1,int c0,unsigned short **pp,int psize,int *p2,int *p1));
|
|
302 STATIC int ww16_conv PROTO((int c2, int c1, int c0));
|
|
303 #endif
|
|
304 #ifdef UTF8_OUTPUT_ENABLE
|
|
305 STATIC int e2w_conv PROTO((int c2,int c1));
|
|
306 STATIC void w_oconv PROTO((int c2,int c1));
|
|
307 STATIC void w_oconv16 PROTO((int c2,int c1));
|
|
308 #endif
|
|
309 STATIC void e_oconv PROTO((int c2,int c1));
|
|
310 STATIC int e2s_conv PROTO((int c2, int c1, int *p2, int *p1));
|
|
311 STATIC void s_oconv PROTO((int c2,int c1));
|
|
312 STATIC void j_oconv PROTO((int c2,int c1));
|
|
313 STATIC void fold_conv PROTO((int c2,int c1));
|
|
314 STATIC void cr_conv PROTO((int c2,int c1));
|
|
315 STATIC void z_conv PROTO((int c2,int c1));
|
|
316 STATIC void rot_conv PROTO((int c2,int c1));
|
|
317 STATIC void hira_conv PROTO((int c2,int c1));
|
|
318 STATIC void base64_conv PROTO((int c2,int c1));
|
|
319 STATIC void iso2022jp_check_conv PROTO((int c2,int c1));
|
|
320 STATIC void no_connection PROTO((int c2,int c1));
|
|
321 STATIC int no_connection2 PROTO((int c2,int c1,int c0));
|
|
322
|
|
323 STATIC void code_score PROTO((struct input_code *ptr));
|
|
324 STATIC void code_status PROTO((int c));
|
|
325
|
|
326 STATIC void std_putc PROTO((int c));
|
|
327 STATIC int std_getc PROTO((FILE *f));
|
|
328 STATIC int std_ungetc PROTO((int c,FILE *f));
|
|
329
|
|
330 STATIC int broken_getc PROTO((FILE *f));
|
|
331 STATIC int broken_ungetc PROTO((int c,FILE *f));
|
|
332
|
|
333 STATIC int mime_begin PROTO((FILE *f));
|
|
334 STATIC int mime_getc PROTO((FILE *f));
|
|
335 STATIC int mime_ungetc PROTO((int c,FILE *f));
|
|
336
|
|
337 STATIC int mime_begin_strict PROTO((FILE *f));
|
|
338 STATIC int mime_getc_buf PROTO((FILE *f));
|
|
339 STATIC int mime_ungetc_buf PROTO((int c,FILE *f));
|
|
340 STATIC int mime_integrity PROTO((FILE *f,unsigned char *p));
|
|
341
|
|
342 STATIC int base64decode PROTO((int c));
|
|
343 STATIC void mime_prechar PROTO((int c2, int c1));
|
|
344 STATIC void mime_putc PROTO((int c));
|
|
345 STATIC void open_mime PROTO((int c));
|
|
346 STATIC void close_mime PROTO(());
|
|
347 STATIC void usage PROTO(());
|
|
348 STATIC void version PROTO(());
|
|
349 STATIC void options PROTO((unsigned char *c));
|
|
350 #if defined(PERL_XS) || defined(WIN32DLL)
|
|
351 STATIC void reinit PROTO(());
|
|
352 #endif
|
|
353
|
|
354 /* buffers */
|
|
355
|
|
356 static unsigned char stdibuf[IOBUF_SIZE];
|
|
357 static unsigned char stdobuf[IOBUF_SIZE];
|
|
358 static unsigned char hold_buf[HOLD_SIZE*2];
|
|
359 static int hold_count;
|
|
360
|
|
361 /* MIME preprocessor fifo */
|
|
362
|
|
363 #define MIME_BUF_SIZE (1024) /* 2^n ring buffer */
|
|
364 #define MIME_BUF_MASK (MIME_BUF_SIZE-1)
|
|
365 #define Fifo(n) mime_buf[(n)&MIME_BUF_MASK]
|
|
366 static unsigned char mime_buf[MIME_BUF_SIZE];
|
|
367 static unsigned int mime_top = 0;
|
|
368 static unsigned int mime_last = 0; /* decoded */
|
|
369 static unsigned int mime_input = 0; /* undecoded */
|
|
370
|
|
371 /* flags */
|
|
372 static int unbuf_f = FALSE;
|
|
373 static int estab_f = FALSE;
|
|
374 static int nop_f = FALSE;
|
|
375 static int binmode_f = TRUE; /* binary mode */
|
|
376 static int rot_f = FALSE; /* rot14/43 mode */
|
|
377 static int hira_f = FALSE; /* hira/kata henkan */
|
|
378 static int input_f = FALSE; /* non fixed input code */
|
|
379 static int alpha_f = FALSE; /* convert JIx0208 alphbet to ASCII */
|
|
380 static int mime_f = STRICT_MIME; /* convert MIME B base64 or Q */
|
|
381 static int mime_decode_f = FALSE; /* mime decode is explicitly on */
|
|
382 static int mimebuf_f = FALSE; /* MIME buffered input */
|
|
383 static int broken_f = FALSE; /* convert ESC-less broken JIS */
|
|
384 static int iso8859_f = FALSE; /* ISO8859 through */
|
|
385 static int mimeout_f = FALSE; /* base64 mode */
|
|
386 #if defined(MSDOS) || defined(__OS2__)
|
|
387 static int x0201_f = TRUE; /* Assume JISX0201 kana */
|
|
388 #else
|
|
389 static int x0201_f = NO_X0201; /* Assume NO JISX0201 */
|
|
390 #endif
|
|
391 static int iso2022jp_f = FALSE; /* convert ISO-2022-JP */
|
|
392 #ifdef UTF8_OUTPUT_ENABLE
|
|
393 static int unicode_bom_f= 0; /* Output Unicode BOM */
|
|
394 static int w_oconv16_LE = 0; /* utf-16 little endian */
|
|
395 static int ms_ucs_map_f = FALSE; /* Microsoft UCS Mapping Compatible */
|
|
396 #endif
|
|
397
|
|
398
|
|
399 #ifdef NUMCHAR_OPTION
|
|
400
|
|
401 #define CLASS_MASK 0x0f000000
|
|
402 #define CLASS_UTF16 0x01000000
|
|
403 #endif
|
|
404
|
|
405 #ifdef INPUT_OPTION
|
|
406 static int cap_f = FALSE;
|
|
407 static int (*i_cgetc)PROTO((FILE *)) = std_getc; /* input of cgetc */
|
|
408 static int (*i_cungetc)PROTO((int c ,FILE *f)) = std_ungetc;
|
|
409 STATIC int cap_getc PROTO((FILE *f));
|
|
410 STATIC int cap_ungetc PROTO((int c,FILE *f));
|
|
411
|
|
412 static int url_f = FALSE;
|
|
413 static int (*i_ugetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
|
|
414 static int (*i_uungetc)PROTO((int c ,FILE *f)) = std_ungetc;
|
|
415 STATIC int url_getc PROTO((FILE *f));
|
|
416 STATIC int url_ungetc PROTO((int c,FILE *f));
|
|
417
|
|
418 static int numchar_f = FALSE;
|
|
419 static int (*i_ngetc)PROTO((FILE *)) = std_getc; /* input of ugetc */
|
|
420 static int (*i_nungetc)PROTO((int c ,FILE *f)) = std_ungetc;
|
|
421 STATIC int numchar_getc PROTO((FILE *f));
|
|
422 STATIC int numchar_ungetc PROTO((int c,FILE *f));
|
|
423 #endif
|
|
424
|
|
425 #ifdef CHECK_OPTION
|
|
426 static int noout_f = FALSE;
|
|
427 STATIC void no_putc PROTO((int c));
|
|
428 static int debug_f = FALSE;
|
|
429 STATIC void debug PROTO((char *str));
|
|
430 #endif
|
|
431
|
|
432 static int guess_f = FALSE;
|
|
433 STATIC void print_guessed_code PROTO((char *filename));
|
|
434 STATIC void set_input_codename PROTO((char *codename));
|
|
435 static int is_inputcode_mixed = FALSE;
|
|
436 static int is_inputcode_set = FALSE;
|
|
437
|
|
438 #ifdef EXEC_IO
|
|
439 static int exec_f = 0;
|
|
440 #endif
|
|
441
|
|
442 #ifdef SHIFTJIS_CP932
|
|
443 STATIC int cp932_f = TRUE;
|
|
444 #define CP932_TABLE_BEGIN (0xfa)
|
|
445 #define CP932_TABLE_END (0xfc)
|
|
446
|
|
447 STATIC int cp932inv_f = TRUE;
|
|
448 #define CP932INV_TABLE_BEGIN (0xed)
|
|
449 #define CP932INV_TABLE_END (0xee)
|
|
450
|
|
451 /* STATIC int cp932_conv PROTO((int c2, int c1)); */
|
|
452 #endif /* SHIFTJIS_CP932 */
|
|
453
|
|
454 #ifdef X0212_ENABLE
|
|
455 STATIC int x0212_f = FALSE;
|
|
456 static int x0212_shift PROTO((int c));
|
|
457 static int x0212_unshift PROTO((int c));
|
|
458 #endif
|
|
459
|
|
460 STATIC unsigned char prefix_table[256];
|
|
461
|
|
462 STATIC void e_status PROTO((struct input_code *, int));
|
|
463 STATIC void s_status PROTO((struct input_code *, int));
|
|
464
|
|
465 #ifdef UTF8_INPUT_ENABLE
|
|
466 STATIC void w_status PROTO((struct input_code *, int));
|
|
467 STATIC void w16_status PROTO((struct input_code *, int));
|
|
468 static int utf16_mode = UTF16BE_INPUT;
|
|
469 #endif
|
|
470
|
|
471 struct input_code input_code_list[] = {
|
|
472 {"EUC-JP", 0, 0, 0, {0, 0, 0}, e_status, e_iconv, 0},
|
|
473 {"Shift_JIS", 0, 0, 0, {0, 0, 0}, s_status, s_iconv, 0},
|
|
474 {"UTF-8", 0, 0, 0, {0, 0, 0}, w_status, w_iconv, 0},
|
|
475 {"UTF-16", 0, 0, 0, {0, 0, 0}, w16_status, w_iconv16, 0},
|
|
476 {0}
|
|
477 };
|
|
478
|
|
479 static int mimeout_mode = 0;
|
|
480 static int base64_count = 0;
|
|
481
|
|
482 /* X0208 -> ASCII converter */
|
|
483
|
|
484 /* fold parameter */
|
|
485 static int f_line = 0; /* chars in line */
|
|
486 static int f_prev = 0;
|
|
487 static int fold_preserve_f = FALSE; /* preserve new lines */
|
|
488 static int fold_f = FALSE;
|
|
489 static int fold_len = 0;
|
|
490
|
|
491 /* options */
|
|
492 static unsigned char kanji_intro = DEFAULT_J;
|
|
493 static unsigned char ascii_intro = DEFAULT_R;
|
|
494
|
|
495 /* Folding */
|
|
496
|
|
497 #define FOLD_MARGIN 10
|
|
498 #define DEFAULT_FOLD 60
|
|
499
|
|
500 static int fold_margin = FOLD_MARGIN;
|
|
501
|
|
502 /* converters */
|
|
503
|
|
504 #ifdef DEFAULT_CODE_JIS
|
|
505 # define DEFAULT_CONV j_oconv
|
|
506 #endif
|
|
507 #ifdef DEFAULT_CODE_SJIS
|
|
508 # define DEFAULT_CONV s_oconv
|
|
509 #endif
|
|
510 #ifdef DEFAULT_CODE_EUC
|
|
511 # define DEFAULT_CONV e_oconv
|
|
512 #endif
|
|
513 #ifdef DEFAULT_CODE_UTF8
|
|
514 # define DEFAULT_CONV w_oconv
|
|
515 #endif
|
|
516
|
|
517 /* process default */
|
|
518 static void (*output_conv)PROTO((int c2,int c1)) = DEFAULT_CONV;
|
|
519
|
|
520 static void (*oconv)PROTO((int c2,int c1)) = no_connection;
|
|
521 /* s_iconv or oconv */
|
|
522 static int (*iconv)PROTO((int c2,int c1,int c0)) = no_connection2;
|
|
523
|
|
524 static void (*o_zconv)PROTO((int c2,int c1)) = no_connection;
|
|
525 static void (*o_fconv)PROTO((int c2,int c1)) = no_connection;
|
|
526 static void (*o_crconv)PROTO((int c2,int c1)) = no_connection;
|
|
527 static void (*o_rot_conv)PROTO((int c2,int c1)) = no_connection;
|
|
528 static void (*o_hira_conv)PROTO((int c2,int c1)) = no_connection;
|
|
529 static void (*o_base64conv)PROTO((int c2,int c1)) = no_connection;
|
|
530 static void (*o_iso2022jp_check_conv)PROTO((int c2,int c1)) = no_connection;
|
|
531
|
|
532 /* static redirections */
|
|
533
|
|
534 static void (*o_putc)PROTO((int c)) = std_putc;
|
|
535
|
|
536 static int (*i_getc)PROTO((FILE *f)) = std_getc; /* general input */
|
|
537 static int (*i_ungetc)PROTO((int c,FILE *f)) =std_ungetc;
|
|
538
|
|
539 static int (*i_bgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
|
|
540 static int (*i_bungetc)PROTO((int c ,FILE *f)) = std_ungetc;
|
|
541
|
|
542 static void (*o_mputc)PROTO((int c)) = std_putc ; /* output of mputc */
|
|
543
|
|
544 static int (*i_mgetc)PROTO((FILE *)) = std_getc; /* input of mgetc */
|
|
545 static int (*i_mungetc)PROTO((int c ,FILE *f)) = std_ungetc;
|
|
546
|
|
547 /* for strict mime */
|
|
548 static int (*i_mgetc_buf)PROTO((FILE *)) = std_getc; /* input of mgetc_buf */
|
|
549 static int (*i_mungetc_buf)PROTO((int c,FILE *f)) = std_ungetc;
|
|
550
|
|
551 /* Global states */
|
|
552 static int output_mode = ASCII, /* output kanji mode */
|
|
553 input_mode = ASCII, /* input kanji mode */
|
|
554 shift_mode = FALSE; /* TRUE shift out, or X0201 */
|
|
555 static int mime_decode_mode = FALSE; /* MIME mode B base64, Q hex */
|
|
556
|
|
557 /* X0201 / X0208 conversion tables */
|
|
558
|
|
559 /* X0201 kana conversion table */
|
|
560 /* 90-9F A0-DF */
|
|
561 static
|
|
562 unsigned char cv[]= {
|
|
563 0x21,0x21,0x21,0x23,0x21,0x56,0x21,0x57,
|
|
564 0x21,0x22,0x21,0x26,0x25,0x72,0x25,0x21,
|
|
565 0x25,0x23,0x25,0x25,0x25,0x27,0x25,0x29,
|
|
566 0x25,0x63,0x25,0x65,0x25,0x67,0x25,0x43,
|
|
567 0x21,0x3c,0x25,0x22,0x25,0x24,0x25,0x26,
|
|
568 0x25,0x28,0x25,0x2a,0x25,0x2b,0x25,0x2d,
|
|
569 0x25,0x2f,0x25,0x31,0x25,0x33,0x25,0x35,
|
|
570 0x25,0x37,0x25,0x39,0x25,0x3b,0x25,0x3d,
|
|
571 0x25,0x3f,0x25,0x41,0x25,0x44,0x25,0x46,
|
|
572 0x25,0x48,0x25,0x4a,0x25,0x4b,0x25,0x4c,
|
|
573 0x25,0x4d,0x25,0x4e,0x25,0x4f,0x25,0x52,
|
|
574 0x25,0x55,0x25,0x58,0x25,0x5b,0x25,0x5e,
|
|
575 0x25,0x5f,0x25,0x60,0x25,0x61,0x25,0x62,
|
|
576 0x25,0x64,0x25,0x66,0x25,0x68,0x25,0x69,
|
|
577 0x25,0x6a,0x25,0x6b,0x25,0x6c,0x25,0x6d,
|
|
578 0x25,0x6f,0x25,0x73,0x21,0x2b,0x21,0x2c,
|
|
579 0x00,0x00};
|
|
580
|
|
581
|
|
582 /* X0201 kana conversion table for daguten */
|
|
583 /* 90-9F A0-DF */
|
|
584 static
|
|
585 unsigned char dv[]= {
|
|
586 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
587 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
588 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
589 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
590 0x00,0x00,0x00,0x00,0x00,0x00,0x25,0x74,
|
|
591 0x00,0x00,0x00,0x00,0x25,0x2c,0x25,0x2e,
|
|
592 0x25,0x30,0x25,0x32,0x25,0x34,0x25,0x36,
|
|
593 0x25,0x38,0x25,0x3a,0x25,0x3c,0x25,0x3e,
|
|
594 0x25,0x40,0x25,0x42,0x25,0x45,0x25,0x47,
|
|
595 0x25,0x49,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
596 0x00,0x00,0x00,0x00,0x25,0x50,0x25,0x53,
|
|
597 0x25,0x56,0x25,0x59,0x25,0x5c,0x00,0x00,
|
|
598 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
599 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
600 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
601 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
602 0x00,0x00};
|
|
603
|
|
604 /* X0201 kana conversion table for han-daguten */
|
|
605 /* 90-9F A0-DF */
|
|
606 static
|
|
607 unsigned char ev[]= {
|
|
608 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
609 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
610 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
611 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
612 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
613 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
614 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
615 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
616 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
617 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
618 0x00,0x00,0x00,0x00,0x25,0x51,0x25,0x54,
|
|
619 0x25,0x57,0x25,0x5a,0x25,0x5d,0x00,0x00,
|
|
620 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
621 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
622 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
623 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
624 0x00,0x00};
|
|
625
|
|
626
|
|
627 /* X0208 kigou conversion table */
|
|
628 /* 0x8140 - 0x819e */
|
|
629 static
|
|
630 unsigned char fv[] = {
|
|
631
|
|
632 0x00,0x00,0x00,0x00,0x2c,0x2e,0x00,0x3a,
|
|
633 0x3b,0x3f,0x21,0x00,0x00,0x27,0x60,0x00,
|
|
634 0x5e,0x00,0x5f,0x00,0x00,0x00,0x00,0x00,
|
|
635 0x00,0x00,0x00,0x00,0x00,0x2d,0x00,0x2f,
|
|
636 0x5c,0x00,0x00,0x7c,0x00,0x00,0x60,0x27,
|
|
637 0x22,0x22,0x28,0x29,0x00,0x00,0x5b,0x5d,
|
|
638 0x7b,0x7d,0x3c,0x3e,0x00,0x00,0x00,0x00,
|
|
639 0x00,0x00,0x00,0x00,0x2b,0x2d,0x00,0x00,
|
|
640 0x00,0x3d,0x00,0x3c,0x3e,0x00,0x00,0x00,
|
|
641 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00,
|
|
642 0x24,0x00,0x00,0x25,0x23,0x26,0x2a,0x40,
|
|
643 0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00
|
|
644 } ;
|
|
645
|
|
646
|
|
647 #define CRLF 1
|
|
648
|
|
649 static int file_out = FALSE;
|
|
650 #ifdef OVERWRITE
|
|
651 static int overwrite = FALSE;
|
|
652 #endif
|
|
653
|
|
654 static int crmode_f = 0; /* CR, NL, CRLF */
|
|
655 #ifdef EASYWIN /*Easy Win */
|
|
656 static int end_check;
|
|
657 #endif /*Easy Win */
|
|
658
|
|
659 #define STD_GC_BUFSIZE (256)
|
|
660 int std_gc_buf[STD_GC_BUFSIZE];
|
|
661 int std_gc_ndx;
|
|
662
|
|
663 #ifdef WIN32DLL
|
|
664 #include "nkf32dll.c"
|
|
665 #elif defined(PERL_XS)
|
|
666 #else /* WIN32DLL */
|
|
667 int
|
|
668 main(argc, argv)
|
|
669 int argc;
|
|
670 char **argv;
|
|
671 {
|
|
672 FILE *fin;
|
|
673 unsigned char *cp;
|
|
674
|
|
675 char *outfname = NULL;
|
|
676 char *origfname;
|
|
677
|
|
678 #ifdef EASYWIN /*Easy Win */
|
|
679 _BufferSize.y = 400;/*Set Scroll Buffer Size*/
|
|
680 #endif
|
|
681
|
|
682 for (argc--,argv++; (argc > 0) && **argv == '-'; argc--, argv++) {
|
|
683 cp = (unsigned char *)*argv;
|
|
684 options(cp);
|
|
685 #ifdef EXEC_IO
|
|
686 if (exec_f){
|
|
687 int fds[2], pid;
|
|
688 if (pipe(fds) < 0 || (pid = fork()) < 0){
|
|
689 abort();
|
|
690 }
|
|
691 if (pid == 0){
|
|
692 if (exec_f > 0){
|
|
693 close(fds[0]);
|
|
694 dup2(fds[1], 1);
|
|
695 }else{
|
|
696 close(fds[1]);
|
|
697 dup2(fds[0], 0);
|
|
698 }
|
|
699 execvp(argv[1], &argv[1]);
|
|
700 }
|
|
701 if (exec_f > 0){
|
|
702 close(fds[1]);
|
|
703 dup2(fds[0], 0);
|
|
704 }else{
|
|
705 close(fds[0]);
|
|
706 dup2(fds[1], 1);
|
|
707 }
|
|
708 argc = 0;
|
|
709 break;
|
|
710 }
|
|
711 #endif
|
|
712 }
|
|
713 if(x0201_f == WISH_TRUE)
|
|
714 x0201_f = ((!iso2022jp_f)? TRUE : NO_X0201);
|
|
715
|
|
716 if (binmode_f == TRUE)
|
|
717 #ifdef __OS2__
|
|
718 if (freopen("","wb",stdout) == NULL)
|
|
719 return (-1);
|
|
720 #else
|
|
721 setbinmode(stdout);
|
|
722 #endif
|
|
723
|
|
724 if (unbuf_f)
|
|
725 setbuf(stdout, (char *) NULL);
|
|
726 else
|
|
727 setvbuffer(stdout, stdobuf, IOBUF_SIZE);
|
|
728
|
|
729 if (argc == 0) {
|
|
730 if (binmode_f == TRUE)
|
|
731 #ifdef __OS2__
|
|
732 if (freopen("","rb",stdin) == NULL) return (-1);
|
|
733 #else
|
|
734 setbinmode(stdin);
|
|
735 #endif
|
|
736 setvbuffer(stdin, stdibuf, IOBUF_SIZE);
|
|
737 if (nop_f)
|
|
738 noconvert(stdin);
|
|
739 else {
|
|
740 kanji_convert(stdin);
|
|
741 if (guess_f) print_guessed_code(NULL);
|
|
742 }
|
|
743 } else {
|
|
744 int nfiles = argc;
|
|
745 while (argc--) {
|
|
746 is_inputcode_mixed = FALSE;
|
|
747 is_inputcode_set = FALSE;
|
|
748 input_codename = "";
|
|
749 if ((fin = fopen((origfname = *argv++), "r")) == NULL) {
|
|
750 perror(*--argv);
|
|
751 return(-1);
|
|
752 } else {
|
|
753 #ifdef OVERWRITE
|
|
754 int fd = 0;
|
|
755 int fd_backup = 0;
|
|
756 #endif
|
|
757
|
|
758 /* reopen file for stdout */
|
|
759 if (file_out == TRUE) {
|
|
760 #ifdef OVERWRITE
|
|
761 if (overwrite){
|
|
762 outfname = malloc(strlen(origfname)
|
|
763 + strlen(".nkftmpXXXXXX")
|
|
764 + 1);
|
|
765 if (!outfname){
|
|
766 perror(origfname);
|
|
767 return -1;
|
|
768 }
|
|
769 strcpy(outfname, origfname);
|
|
770 #ifdef MSDOS
|
|
771 {
|
|
772 int i;
|
|
773 for (i = strlen(outfname); i; --i){
|
|
774 if (outfname[i - 1] == '/'
|
|
775 || outfname[i - 1] == '\\'){
|
|
776 break;
|
|
777 }
|
|
778 }
|
|
779 outfname[i] = '\0';
|
|
780 }
|
|
781 strcat(outfname, "ntXXXXXX");
|
|
782 mktemp(outfname);
|
|
783 fd = open(outfname, O_WRONLY | O_CREAT | O_TRUNC,
|
|
784 S_IREAD | S_IWRITE);
|
|
785 #else
|
|
786 strcat(outfname, ".nkftmpXXXXXX");
|
|
787 fd = mkstemp(outfname);
|
|
788 #endif
|
|
789 if (fd < 0
|
|
790 || (fd_backup = dup(fileno(stdout))) < 0
|
|
791 || dup2(fd, fileno(stdout)) < 0
|
|
792 ){
|
|
793 perror(origfname);
|
|
794 return -1;
|
|
795 }
|
|
796 }else
|
|
797 #endif
|
|
798 if(argc == 1 ) {
|
|
799 outfname = *argv++;
|
|
800 argc--;
|
|
801 } else {
|
|
802 outfname = "nkf.out";
|
|
803 }
|
|
804
|
|
805 if(freopen(outfname, "w", stdout) == NULL) {
|
|
806 perror (outfname);
|
|
807 return (-1);
|
|
808 }
|
|
809 if (binmode_f == TRUE) {
|
|
810 #ifdef __OS2__
|
|
811 if (freopen("","wb",stdout) == NULL)
|
|
812 return (-1);
|
|
813 #else
|
|
814 setbinmode(stdout);
|
|
815 #endif
|
|
816 }
|
|
817 }
|
|
818 if (binmode_f == TRUE)
|
|
819 #ifdef __OS2__
|
|
820 if (freopen("","rb",fin) == NULL)
|
|
821 return (-1);
|
|
822 #else
|
|
823 setbinmode(fin);
|
|
824 #endif
|
|
825 setvbuffer(fin, stdibuf, IOBUF_SIZE);
|
|
826 if (nop_f)
|
|
827 noconvert(fin);
|
|
828 else {
|
|
829 char *filename = NULL;
|
|
830 kanji_convert(fin);
|
|
831 if (nfiles > 1) filename = origfname;
|
|
832 if (guess_f) print_guessed_code(filename);
|
|
833 }
|
|
834 fclose(fin);
|
|
835 #ifdef OVERWRITE
|
|
836 if (overwrite) {
|
|
837 struct stat sb;
|
|
838 #if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__)
|
|
839 time_t tb[2];
|
|
840 #else
|
|
841 struct utimbuf tb;
|
|
842 #endif
|
|
843
|
|
844 fflush(stdout);
|
|
845 close(fd);
|
|
846 if (dup2(fd_backup, fileno(stdout)) < 0){
|
|
847 perror("dup2");
|
|
848 }
|
|
849 if (stat(origfname, &sb)) {
|
|
850 fprintf(stderr, "Can't stat %s\n", origfname);
|
|
851 }
|
|
852 /* $B%Q!<%_%C%7%g%s$rI|85(B */
|
|
853 if (chmod(outfname, sb.st_mode)) {
|
|
854 fprintf(stderr, "Can't set permission %s\n", outfname);
|
|
855 }
|
|
856
|
|
857 /* $B%?%$%`%9%?%s%W$rI|85(B */
|
|
858 #if defined(MSDOS) && !defined(__MINGW32__) && !defined(__WIN32__)
|
|
859 tb[0] = tb[1] = sb.st_mtime;
|
|
860 if (utime(outfname, tb)) {
|
|
861 fprintf(stderr, "Can't set timestamp %s\n", outfname);
|
|
862 }
|
|
863 #else
|
|
864 tb.actime = sb.st_atime;
|
|
865 tb.modtime = sb.st_mtime;
|
|
866 if (utime(outfname, &tb)) {
|
|
867 fprintf(stderr, "Can't set timestamp %s\n", outfname);
|
|
868 }
|
|
869 #endif
|
|
870 #ifdef MSDOS
|
|
871 if (unlink(origfname)){
|
|
872 perror(origfname);
|
|
873 }
|
|
874 #endif
|
|
875 if (rename(outfname, origfname)) {
|
|
876 perror(origfname);
|
|
877 fprintf(stderr, "Can't rename %s to %s\n",
|
|
878 outfname, origfname);
|
|
879 }
|
|
880 free(outfname);
|
|
881 }
|
|
882 #endif
|
|
883 }
|
|
884 }
|
|
885 }
|
|
886 #ifdef EASYWIN /*Easy Win */
|
|
887 if (file_out == FALSE)
|
|
888 scanf("%d",&end_check);
|
|
889 else
|
|
890 fclose(stdout);
|
|
891 #else /* for Other OS */
|
|
892 if (file_out == TRUE)
|
|
893 fclose(stdout);
|
|
894 #endif /*Easy Win */
|
|
895 return (0);
|
|
896 }
|
|
897 #endif /* WIN32DLL */
|
|
898
|
|
899 static
|
|
900 struct {
|
|
901 char *name;
|
|
902 char *alias;
|
|
903 } long_option[] = {
|
|
904 {"base64","jMB"},
|
|
905 {"euc","e"},
|
|
906 {"euc-input","E"},
|
|
907 {"fj","jm"},
|
|
908 {"help","v"},
|
|
909 {"jis","j"},
|
|
910 {"jis-input","J"},
|
|
911 {"mac","sLm"},
|
|
912 {"mime","jM"},
|
|
913 {"mime-input","m"},
|
|
914 {"msdos","sLw"},
|
|
915 {"sjis","s"},
|
|
916 {"sjis-input","S"},
|
|
917 {"unix","eLu"},
|
|
918 {"version","V"},
|
|
919 {"windows","sLw"},
|
|
920 {"hiragana","h1"},
|
|
921 {"katakana","h2"},
|
|
922 {"katakana-hiragana","h3"},
|
|
923 {"guess", "g"},
|
|
924 {"cp932", ""},
|
|
925 {"no-cp932", ""},
|
|
926 #ifdef X0212_ENABLE
|
|
927 {"x0212", ""},
|
|
928 #endif
|
|
929 #ifdef UTF8_OUTPUT_ENABLE
|
|
930 {"utf8", "w"},
|
|
931 {"utf16", "w16"},
|
|
932 {"ms-ucs-map", ""},
|
|
933 #endif
|
|
934 #ifdef UTF8_INPUT_ENABLE
|
|
935 {"utf8-input", "W"},
|
|
936 {"utf16-input", "W16"},
|
|
937 #endif
|
|
938 #ifdef OVERWRITE
|
|
939 {"overwrite", ""},
|
|
940 #endif
|
|
941 #ifdef INPUT_OPTION
|
|
942 {"cap-input", ""},
|
|
943 {"url-input", ""},
|
|
944 #endif
|
|
945 #ifdef NUMCHAR_OPTION
|
|
946 {"numchar-input", ""},
|
|
947 #endif
|
|
948 #ifdef CHECK_OPTION
|
|
949 {"no-output", ""},
|
|
950 {"debug", ""},
|
|
951 #endif
|
|
952 #ifdef SHIFTJIS_CP932
|
|
953 {"cp932inv", ""},
|
|
954 #endif
|
|
955 #ifdef EXEC_IO
|
|
956 {"exec-in", ""},
|
|
957 {"exec-out", ""},
|
|
958 #endif
|
|
959 {"prefix=", ""},
|
|
960 };
|
|
961
|
|
962 static int option_mode = 0;
|
|
963
|
|
964 void
|
|
965 options(cp)
|
|
966 unsigned char *cp;
|
|
967 {
|
|
968 int i;
|
|
969 unsigned char *p = NULL;
|
|
970
|
|
971 if (option_mode==1)
|
|
972 return;
|
|
973 if (*cp++ != '-')
|
|
974 return;
|
|
975 while (*cp) {
|
|
976 if (p && !*cp) {
|
|
977 cp = p;
|
|
978 p = 0;
|
|
979 }
|
|
980 switch (*cp++) {
|
|
981 case '-': /* literal options */
|
|
982 if (!*cp) { /* ignore the rest of arguments */
|
|
983 option_mode = 1;
|
|
984 return;
|
|
985 }
|
|
986 for (i=0;i<sizeof(long_option)/sizeof(long_option[0]);i++) {
|
|
987 int j;
|
|
988 p = (unsigned char *)long_option[i].name;
|
|
989 for (j=0;*p && (*p != '=') && *p == cp[j];p++, j++);
|
|
990 if (*p == cp[j]){
|
|
991 p = &cp[j];
|
|
992 break;
|
|
993 }
|
|
994 p = 0;
|
|
995 }
|
|
996 if (p == 0) return;
|
|
997 cp = (unsigned char *)long_option[i].alias;
|
|
998 if (!*cp){
|
|
999 #ifdef OVERWRITE
|
|
1000 if (strcmp(long_option[i].name, "overwrite") == 0){
|
|
1001 file_out = TRUE;
|
|
1002 overwrite = TRUE;
|
|
1003 continue;
|
|
1004 }
|
|
1005 #endif
|
|
1006 #ifdef INPUT_OPTION
|
|
1007 if (strcmp(long_option[i].name, "cap-input") == 0){
|
|
1008 cap_f = TRUE;
|
|
1009 continue;
|
|
1010 }
|
|
1011 if (strcmp(long_option[i].name, "url-input") == 0){
|
|
1012 url_f = TRUE;
|
|
1013 continue;
|
|
1014 }
|
|
1015 #endif
|
|
1016 #ifdef NUMCHAR_OPTION
|
|
1017 if (strcmp(long_option[i].name, "numchar-input") == 0){
|
|
1018 numchar_f = TRUE;
|
|
1019 continue;
|
|
1020 }
|
|
1021 #endif
|
|
1022 #ifdef CHECK_OPTION
|
|
1023 if (strcmp(long_option[i].name, "no-output") == 0){
|
|
1024 noout_f = TRUE;
|
|
1025 continue;
|
|
1026 }
|
|
1027 if (strcmp(long_option[i].name, "debug") == 0){
|
|
1028 debug_f = TRUE;
|
|
1029 continue;
|
|
1030 }
|
|
1031 #endif
|
|
1032 if (strcmp(long_option[i].name, "cp932") == 0){
|
|
1033 #ifdef SHIFTJIS_CP932
|
|
1034 cp932_f = TRUE;
|
|
1035 cp932inv_f = TRUE;
|
|
1036 #endif
|
|
1037 #ifdef UTF8_OUTPUT_ENABLE
|
|
1038 ms_ucs_map_f = TRUE;
|
|
1039 #endif
|
|
1040 continue;
|
|
1041 }
|
|
1042 if (strcmp(long_option[i].name, "no-cp932") == 0){
|
|
1043 #ifdef SHIFTJIS_CP932
|
|
1044 cp932_f = FALSE;
|
|
1045 cp932inv_f = FALSE;
|
|
1046 #endif
|
|
1047 #ifdef UTF8_OUTPUT_ENABLE
|
|
1048 ms_ucs_map_f = FALSE;
|
|
1049 #endif
|
|
1050 continue;
|
|
1051 }
|
|
1052 #ifdef SHIFTJIS_CP932
|
|
1053 if (strcmp(long_option[i].name, "cp932inv") == 0){
|
|
1054 cp932inv_f = TRUE;
|
|
1055 continue;
|
|
1056 }
|
|
1057 #endif
|
|
1058
|
|
1059 #ifdef X0212_ENABLE
|
|
1060 if (strcmp(long_option[i].name, "x0212") == 0){
|
|
1061 x0212_f = TRUE;
|
|
1062 continue;
|
|
1063 }
|
|
1064 #endif
|
|
1065
|
|
1066 #ifdef EXEC_IO
|
|
1067 if (strcmp(long_option[i].name, "exec-in") == 0){
|
|
1068 exec_f = 1;
|
|
1069 return;
|
|
1070 }
|
|
1071 if (strcmp(long_option[i].name, "exec-out") == 0){
|
|
1072 exec_f = -1;
|
|
1073 return;
|
|
1074 }
|
|
1075 #endif
|
|
1076 #ifdef UTF8_OUTPUT_ENABLE
|
|
1077 if (strcmp(long_option[i].name, "ms-ucs-map") == 0){
|
|
1078 ms_ucs_map_f = TRUE;
|
|
1079 continue;
|
|
1080 }
|
|
1081 #endif
|
|
1082 if (strcmp(long_option[i].name, "prefix=") == 0){
|
|
1083 if (*p == '=' && ' ' < p[1] && p[1] < 128){
|
|
1084 for (i = 2; ' ' < p[i] && p[i] < 128; i++){
|
|
1085 prefix_table[p[i]] = p[1];
|
|
1086 }
|
|
1087 }
|
|
1088 continue;
|
|
1089 }
|
|
1090 }
|
|
1091 continue;
|
|
1092 case 'b': /* buffered mode */
|
|
1093 unbuf_f = FALSE;
|
|
1094 continue;
|
|
1095 case 'u': /* non bufferd mode */
|
|
1096 unbuf_f = TRUE;
|
|
1097 continue;
|
|
1098 case 't': /* transparent mode */
|
|
1099 nop_f = TRUE;
|
|
1100 continue;
|
|
1101 case 'j': /* JIS output */
|
|
1102 case 'n':
|
|
1103 output_conv = j_oconv;
|
|
1104 continue;
|
|
1105 case 'e': /* AT&T EUC output */
|
|
1106 output_conv = e_oconv;
|
|
1107 continue;
|
|
1108 case 's': /* SJIS output */
|
|
1109 output_conv = s_oconv;
|
|
1110 continue;
|
|
1111 case 'l': /* ISO8859 Latin-1 support, no conversion */
|
|
1112 iso8859_f = TRUE; /* Only compatible with ISO-2022-JP */
|
|
1113 input_f = LATIN1_INPUT;
|
|
1114 continue;
|
|
1115 case 'i': /* Kanji IN ESC-$-@/B */
|
|
1116 if (*cp=='@'||*cp=='B')
|
|
1117 kanji_intro = *cp++;
|
|
1118 continue;
|
|
1119 case 'o': /* ASCII IN ESC-(-J/B */
|
|
1120 if (*cp=='J'||*cp=='B'||*cp=='H')
|
|
1121 ascii_intro = *cp++;
|
|
1122 continue;
|
|
1123 case 'h':
|
|
1124 /*
|
|
1125 bit:1 hira -> kata
|
|
1126 bit:2 kata -> hira
|
|
1127 */
|
|
1128 if ('9'>= *cp && *cp>='0')
|
|
1129 hira_f |= (*cp++ -'0');
|
|
1130 else
|
|
1131 hira_f |= 1;
|
|
1132 continue;
|
|
1133 case 'r':
|
|
1134 rot_f = TRUE;
|
|
1135 continue;
|
|
1136 #if defined(MSDOS) || defined(__OS2__)
|
|
1137 case 'T':
|
|
1138 binmode_f = FALSE;
|
|
1139 continue;
|
|
1140 #endif
|
|
1141 #ifndef PERL_XS
|
|
1142 case 'V':
|
|
1143 version();
|
|
1144 exit(1);
|
|
1145 break;
|
|
1146 case 'v':
|
|
1147 usage();
|
|
1148 exit(1);
|
|
1149 break;
|
|
1150 #endif
|
|
1151 #ifdef UTF8_OUTPUT_ENABLE
|
|
1152 case 'w': /* UTF-8 output */
|
|
1153 if ('1'== cp[0] && '6'==cp[1]) {
|
|
1154 output_conv = w_oconv16; cp+=2;
|
|
1155 if (cp[0]=='L') {
|
|
1156 unicode_bom_f=2; cp++;
|
|
1157 w_oconv16_LE = 1;
|
|
1158 if (cp[0] == '0'){
|
|
1159 unicode_bom_f=1; cp++;
|
|
1160 }
|
|
1161 } else if (cp[0] == 'B') {
|
|
1162 unicode_bom_f=2; cp++;
|
|
1163 if (cp[0] == '0'){
|
|
1164 unicode_bom_f=1; cp++;
|
|
1165 }
|
|
1166 }
|
|
1167 } else if (cp[0] == '8') {
|
|
1168 output_conv = w_oconv; cp++;
|
|
1169 unicode_bom_f=2;
|
|
1170 if (cp[0] == '0'){
|
|
1171 unicode_bom_f=1; cp++;
|
|
1172 }
|
|
1173 } else
|
|
1174 output_conv = w_oconv;
|
|
1175 continue;
|
|
1176 #endif
|
|
1177 #ifdef UTF8_INPUT_ENABLE
|
|
1178 case 'W': /* UTF-8 input */
|
|
1179 if ('1'== cp[0] && '6'==cp[1]) {
|
|
1180 input_f = UTF16BE_INPUT;
|
|
1181 utf16_mode = UTF16BE_INPUT;
|
|
1182 cp += 2;
|
|
1183 if (cp[0]=='L') {
|
|
1184 cp++;
|
|
1185 input_f = UTF16LE_INPUT;
|
|
1186 utf16_mode = UTF16LE_INPUT;
|
|
1187 } else if (cp[0] == 'B') {
|
|
1188 cp++;
|
|
1189 input_f = UTF16BE_INPUT;
|
|
1190 utf16_mode = UTF16BE_INPUT;
|
|
1191 }
|
|
1192 } else if (cp[0] == '8') {
|
|
1193 cp++;
|
|
1194 input_f = UTF8_INPUT;
|
|
1195 } else
|
|
1196 input_f = UTF8_INPUT;
|
|
1197 continue;
|
|
1198 #endif
|
|
1199 /* Input code assumption */
|
|
1200 case 'J': /* JIS input */
|
|
1201 case 'E': /* AT&T EUC input */
|
|
1202 input_f = JIS_INPUT;
|
|
1203 continue;
|
|
1204 case 'S': /* MS Kanji input */
|
|
1205 input_f = SJIS_INPUT;
|
|
1206 if (x0201_f==NO_X0201) x0201_f=TRUE;
|
|
1207 continue;
|
|
1208 case 'Z': /* Convert X0208 alphabet to asii */
|
|
1209 /* bit:0 Convert X0208
|
|
1210 bit:1 Convert Kankaku to one space
|
|
1211 bit:2 Convert Kankaku to two spaces
|
|
1212 bit:3 Convert HTML Entity
|
|
1213 */
|
|
1214 if ('9'>= *cp && *cp>='0')
|
|
1215 alpha_f |= 1<<(*cp++ -'0');
|
|
1216 else
|
|
1217 alpha_f |= TRUE;
|
|
1218 continue;
|
|
1219 case 'x': /* Convert X0201 kana to X0208 or X0201 Conversion */
|
|
1220 x0201_f = FALSE; /* No X0201->X0208 conversion */
|
|
1221 /* accept X0201
|
|
1222 ESC-(-I in JIS, EUC, MS Kanji
|
|
1223 SI/SO in JIS, EUC, MS Kanji
|
|
1224 SSO in EUC, JIS, not in MS Kanji
|
|
1225 MS Kanji (0xa0-0xdf)
|
|
1226 output X0201
|
|
1227 ESC-(-I in JIS (0x20-0x5f)
|
|
1228 SSO in EUC (0xa0-0xdf)
|
|
1229 0xa0-0xd in MS Kanji (0xa0-0xdf)
|
|
1230 */
|
|
1231 continue;
|
|
1232 case 'X': /* Assume X0201 kana */
|
|
1233 /* Default value is NO_X0201 for EUC/MS-Kanji mix */
|
|
1234 x0201_f = TRUE;
|
|
1235 continue;
|
|
1236 case 'F': /* prserve new lines */
|
|
1237 fold_preserve_f = TRUE;
|
|
1238 case 'f': /* folding -f60 or -f */
|
|
1239 fold_f = TRUE;
|
|
1240 fold_len = 0;
|
|
1241 while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
|
|
1242 fold_len *= 10;
|
|
1243 fold_len += *cp++ - '0';
|
|
1244 }
|
|
1245 if (!(0<fold_len && fold_len<BUFSIZ))
|
|
1246 fold_len = DEFAULT_FOLD;
|
|
1247 if (*cp=='-') {
|
|
1248 fold_margin = 0;
|
|
1249 cp++;
|
|
1250 while('0'<= *cp && *cp <='9') { /* we don't use atoi here */
|
|
1251 fold_margin *= 10;
|
|
1252 fold_margin += *cp++ - '0';
|
|
1253 }
|
|
1254 }
|
|
1255 continue;
|
|
1256 case 'm': /* MIME support */
|
|
1257 /* mime_decode_f = TRUE; */ /* this has too large side effects... */
|
|
1258 if (*cp=='B'||*cp=='Q') {
|
|
1259 mime_decode_mode = *cp++;
|
|
1260 mimebuf_f = FIXED_MIME;
|
|
1261 } else if (*cp=='N') {
|
|
1262 mime_f = TRUE; cp++;
|
|
1263 } else if (*cp=='S') {
|
|
1264 mime_f = STRICT_MIME; cp++;
|
|
1265 } else if (*cp=='0') {
|
|
1266 mime_decode_f = FALSE;
|
|
1267 mime_f = FALSE; cp++;
|
|
1268 }
|
|
1269 continue;
|
|
1270 case 'M': /* MIME output */
|
|
1271 if (*cp=='B') {
|
|
1272 mimeout_mode = 'B';
|
|
1273 mimeout_f = FIXED_MIME; cp++;
|
|
1274 } else if (*cp=='Q') {
|
|
1275 mimeout_mode = 'Q';
|
|
1276 mimeout_f = FIXED_MIME; cp++;
|
|
1277 } else {
|
|
1278 mimeout_f = TRUE;
|
|
1279 }
|
|
1280 continue;
|
|
1281 case 'B': /* Broken JIS support */
|
|
1282 /* bit:0 no ESC JIS
|
|
1283 bit:1 allow any x on ESC-(-x or ESC-$-x
|
|
1284 bit:2 reset to ascii on NL
|
|
1285 */
|
|
1286 if ('9'>= *cp && *cp>='0')
|
|
1287 broken_f |= 1<<(*cp++ -'0');
|
|
1288 else
|
|
1289 broken_f |= TRUE;
|
|
1290 continue;
|
|
1291 #ifndef PERL_XS
|
|
1292 case 'O':/* for Output file */
|
|
1293 file_out = TRUE;
|
|
1294 continue;
|
|
1295 #endif
|
|
1296 case 'c':/* add cr code */
|
|
1297 crmode_f = CRLF;
|
|
1298 continue;
|
|
1299 case 'd':/* delete cr code */
|
|
1300 crmode_f = NL;
|
|
1301 continue;
|
|
1302 case 'I': /* ISO-2022-JP output */
|
|
1303 iso2022jp_f = TRUE;
|
|
1304 continue;
|
|
1305 case 'L': /* line mode */
|
|
1306 if (*cp=='u') { /* unix */
|
|
1307 crmode_f = NL; cp++;
|
|
1308 } else if (*cp=='m') { /* mac */
|
|
1309 crmode_f = CR; cp++;
|
|
1310 } else if (*cp=='w') { /* windows */
|
|
1311 crmode_f = CRLF; cp++;
|
|
1312 } else if (*cp=='0') { /* no conversion */
|
|
1313 crmode_f = 0; cp++;
|
|
1314 }
|
|
1315 continue;
|
|
1316 case 'g':
|
|
1317 #ifndef PERL_XS
|
|
1318 guess_f = TRUE;
|
|
1319 #endif
|
|
1320 continue;
|
|
1321 case ' ':
|
|
1322 /* module muliple options in a string are allowed for Perl moudle */
|
|
1323 while(*cp && *cp!='-') cp++;
|
|
1324 if(*cp=='-') cp++;
|
|
1325 continue;
|
|
1326 default:
|
|
1327 /* bogus option but ignored */
|
|
1328 continue;
|
|
1329 }
|
|
1330 }
|
|
1331 }
|
|
1332
|
|
1333 #ifdef ANSI_C_PROTOTYPE
|
|
1334 struct input_code * find_inputcode_byfunc(int (*iconv_func)(int c2,int c1,int c0))
|
|
1335 #else
|
|
1336 struct input_code * find_inputcode_byfunc(iconv_func)
|
|
1337 int (*iconv_func)();
|
|
1338 #endif
|
|
1339 {
|
|
1340 if (iconv_func){
|
|
1341 struct input_code *p = input_code_list;
|
|
1342 while (p->name){
|
|
1343 if (iconv_func == p->iconv_func){
|
|
1344 return p;
|
|
1345 }
|
|
1346 p++;
|
|
1347 }
|
|
1348 }
|
|
1349 return 0;
|
|
1350 }
|
|
1351
|
|
1352 #ifdef CHECK_OPTION
|
|
1353 static int (*iconv_for_check)() = 0;
|
|
1354 #endif
|
|
1355
|
|
1356 #ifdef ANSI_C_PROTOTYPE
|
|
1357 void set_iconv(int f, int (*iconv_func)(int c2,int c1,int c0))
|
|
1358 #else
|
|
1359 void set_iconv(f, iconv_func)
|
|
1360 int f;
|
|
1361 int (*iconv_func)();
|
|
1362 #endif
|
|
1363 {
|
|
1364 #ifdef INPUT_CODE_FIX
|
|
1365 if (f || !input_f)
|
|
1366 #endif
|
|
1367 if (estab_f != f){
|
|
1368 estab_f = f;
|
|
1369 }
|
|
1370
|
|
1371 if (iconv_func
|
|
1372 #ifdef INPUT_CODE_FIX
|
|
1373 && (f == -TRUE || !input_f) /* -TRUE means "FORCE" */
|
|
1374 #endif
|
|
1375 ){
|
|
1376 iconv = iconv_func;
|
|
1377 }
|
|
1378 #ifdef CHECK_OPTION
|
|
1379 if (estab_f && iconv_for_check != iconv){
|
|
1380 struct input_code *p = find_inputcode_byfunc(iconv);
|
|
1381 if (p){
|
|
1382 set_input_codename(p->name);
|
|
1383 debug(input_codename);
|
|
1384 }
|
|
1385 iconv_for_check = iconv;
|
|
1386 }
|
|
1387 #endif
|
|
1388 }
|
|
1389
|
|
1390 #define SCORE_L2 (1) /* $BBh(B2$B?e=`4A;z(B */
|
|
1391 #define SCORE_KANA (SCORE_L2 << 1) /* $B$$$o$f$kH>3Q%+%J(B */
|
|
1392 #define SCORE_DEPEND (SCORE_KANA << 1) /* $B5!<o0MB8J8;z(B */
|
|
1393 #ifdef SHIFTJIS_CP932
|
|
1394 #define SCORE_CP932 (SCORE_DEPEND << 1) /* CP932 $B$K$h$kFI$_49$((B */
|
|
1395 #define SCORE_NO_EXIST (SCORE_CP932 << 1) /* $BB8:_$7$J$$J8;z(B */
|
|
1396 #else
|
|
1397 #define SCORE_NO_EXIST (SCORE_DEPEND << 1) /* $BB8:_$7$J$$J8;z(B */
|
|
1398 #endif
|
|
1399 #define SCORE_iMIME (SCORE_NO_EXIST << 1) /* MIME $B$K$h$k;XDj(B */
|
|
1400 #define SCORE_ERROR (SCORE_iMIME << 1) /* $B%(%i!<(B */
|
|
1401
|
|
1402 #define SCORE_INIT (SCORE_iMIME)
|
|
1403
|
|
1404 int score_table_A0[] = {
|
|
1405 0, 0, 0, 0,
|
|
1406 0, 0, 0, 0,
|
|
1407 0, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
|
|
1408 SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_NO_EXIST,
|
|
1409 };
|
|
1410
|
|
1411 int score_table_F0[] = {
|
|
1412 SCORE_L2, SCORE_L2, SCORE_L2, SCORE_L2,
|
|
1413 SCORE_L2, SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST,
|
|
1414 SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND, SCORE_DEPEND,
|
|
1415 SCORE_DEPEND, SCORE_NO_EXIST, SCORE_NO_EXIST, SCORE_ERROR,
|
|
1416 };
|
|
1417
|
|
1418 void set_code_score(ptr, score)
|
|
1419 struct input_code *ptr;
|
|
1420 int score;
|
|
1421 {
|
|
1422 if (ptr){
|
|
1423 ptr->score |= score;
|
|
1424 }
|
|
1425 }
|
|
1426
|
|
1427 void clr_code_score(ptr, score)
|
|
1428 struct input_code *ptr;
|
|
1429 int score;
|
|
1430 {
|
|
1431 if (ptr){
|
|
1432 ptr->score &= ~score;
|
|
1433 }
|
|
1434 }
|
|
1435
|
|
1436 void code_score(ptr)
|
|
1437 struct input_code *ptr;
|
|
1438 {
|
|
1439 int c2 = ptr->buf[0];
|
|
1440 int c1 = ptr->buf[1];
|
|
1441 if (c2 < 0){
|
|
1442 set_code_score(ptr, SCORE_ERROR);
|
|
1443 }else if (c2 == SSO){
|
|
1444 set_code_score(ptr, SCORE_KANA);
|
|
1445 #ifdef UTF8_OUTPUT_ENABLE
|
|
1446 }else if (!e2w_conv(c2, c1)){
|
|
1447 set_code_score(ptr, SCORE_NO_EXIST);
|
|
1448 #endif
|
|
1449 }else if ((c2 & 0x70) == 0x20){
|
|
1450 set_code_score(ptr, score_table_A0[c2 & 0x0f]);
|
|
1451 }else if ((c2 & 0x70) == 0x70){
|
|
1452 set_code_score(ptr, score_table_F0[c2 & 0x0f]);
|
|
1453 }else if ((c2 & 0x70) >= 0x50){
|
|
1454 set_code_score(ptr, SCORE_L2);
|
|
1455 }
|
|
1456 }
|
|
1457
|
|
1458 void status_disable(ptr)
|
|
1459 struct input_code *ptr;
|
|
1460 {
|
|
1461 ptr->stat = -1;
|
|
1462 ptr->buf[0] = -1;
|
|
1463 code_score(ptr);
|
|
1464 if (iconv == ptr->iconv_func) set_iconv(FALSE, 0);
|
|
1465 }
|
|
1466
|
|
1467 void status_push_ch(ptr, c)
|
|
1468 struct input_code *ptr;
|
|
1469 int c;
|
|
1470 {
|
|
1471 ptr->buf[ptr->index++] = c;
|
|
1472 }
|
|
1473
|
|
1474 void status_clear(ptr)
|
|
1475 struct input_code *ptr;
|
|
1476 {
|
|
1477 ptr->stat = 0;
|
|
1478 ptr->index = 0;
|
|
1479 }
|
|
1480
|
|
1481 void status_reset(ptr)
|
|
1482 struct input_code *ptr;
|
|
1483 {
|
|
1484 status_clear(ptr);
|
|
1485 ptr->score = SCORE_INIT;
|
|
1486 }
|
|
1487
|
|
1488 void status_reinit(ptr)
|
|
1489 struct input_code *ptr;
|
|
1490 {
|
|
1491 status_reset(ptr);
|
|
1492 ptr->_file_stat = 0;
|
|
1493 }
|
|
1494
|
|
1495 void status_check(ptr, c)
|
|
1496 struct input_code *ptr;
|
|
1497 int c;
|
|
1498 {
|
|
1499 if (c <= DEL && estab_f){
|
|
1500 status_reset(ptr);
|
|
1501 }
|
|
1502 }
|
|
1503
|
|
1504 void s_status(ptr, c)
|
|
1505 struct input_code *ptr;
|
|
1506 int c;
|
|
1507 {
|
|
1508 switch(ptr->stat){
|
|
1509 case -1:
|
|
1510 status_check(ptr, c);
|
|
1511 break;
|
|
1512 case 0:
|
|
1513 if (c <= DEL){
|
|
1514 break;
|
|
1515 #ifdef NUMCHAR_OPTION
|
|
1516 }else if ((c & CLASS_MASK) == CLASS_UTF16){
|
|
1517 break;
|
|
1518 #endif
|
|
1519 }else if (0xa1 <= c && c <= 0xdf){
|
|
1520 status_push_ch(ptr, SSO);
|
|
1521 status_push_ch(ptr, c);
|
|
1522 code_score(ptr);
|
|
1523 status_clear(ptr);
|
|
1524 }else if ((0x81 <= c && c < 0xa0) || (0xe0 <= c && c <= 0xef)){
|
|
1525 ptr->stat = 1;
|
|
1526 status_push_ch(ptr, c);
|
|
1527 #ifdef SHIFTJIS_CP932
|
|
1528 }else if (cp932_f
|
|
1529 && CP932_TABLE_BEGIN <= c && c <= CP932_TABLE_END){
|
|
1530 ptr->stat = 2;
|
|
1531 status_push_ch(ptr, c);
|
|
1532 #endif /* SHIFTJIS_CP932 */
|
|
1533 #ifdef X0212_ENABLE
|
|
1534 }else if (x0212_f && 0xf0 <= c && c <= 0xfc){
|
|
1535 ptr->stat = 1;
|
|
1536 status_push_ch(ptr, c);
|
|
1537 #endif /* X0212_ENABLE */
|
|
1538 }else{
|
|
1539 status_disable(ptr);
|
|
1540 }
|
|
1541 break;
|
|
1542 case 1:
|
|
1543 if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
|
|
1544 status_push_ch(ptr, c);
|
|
1545 s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]);
|
|
1546 code_score(ptr);
|
|
1547 status_clear(ptr);
|
|
1548 }else{
|
|
1549 status_disable(ptr);
|
|
1550 }
|
|
1551 break;
|
|
1552 case 2:
|
|
1553 #ifdef SHIFTJIS_CP932
|
|
1554 if ((0x40 <= c && c <= 0x7e) || (0x80 <= c && c <= 0xfc)){
|
|
1555 status_push_ch(ptr, c);
|
|
1556 if (s2e_conv(ptr->buf[0], ptr->buf[1], &ptr->buf[0], &ptr->buf[1]) == 0){
|
|
1557 set_code_score(ptr, SCORE_CP932);
|
|
1558 status_clear(ptr);
|
|
1559 break;
|
|
1560 }
|
|
1561 }
|
|
1562 #endif /* SHIFTJIS_CP932 */
|
|
1563 #ifndef X0212_ENABLE
|
|
1564 status_disable(ptr);
|
|
1565 break;
|
|
1566 #endif
|
|
1567 }
|
|
1568 }
|
|
1569
|
|
1570 void e_status(ptr, c)
|
|
1571 struct input_code *ptr;
|
|
1572 int c;
|
|
1573 {
|
|
1574 switch (ptr->stat){
|
|
1575 case -1:
|
|
1576 status_check(ptr, c);
|
|
1577 break;
|
|
1578 case 0:
|
|
1579 if (c <= DEL){
|
|
1580 break;
|
|
1581 #ifdef NUMCHAR_OPTION
|
|
1582 }else if ((c & CLASS_MASK) == CLASS_UTF16){
|
|
1583 break;
|
|
1584 #endif
|
|
1585 }else if (SSO == c || (0xa1 <= c && c <= 0xfe)){
|
|
1586 ptr->stat = 1;
|
|
1587 status_push_ch(ptr, c);
|
|
1588 #ifdef X0212_ENABLE
|
|
1589 }else if (0x8f == c){
|
|
1590 ptr->stat = 2;
|
|
1591 status_push_ch(ptr, c);
|
|
1592 #endif /* X0212_ENABLE */
|
|
1593 }else{
|
|
1594 status_disable(ptr);
|
|
1595 }
|
|
1596 break;
|
|
1597 case 1:
|
|
1598 if (0xa1 <= c && c <= 0xfe){
|
|
1599 status_push_ch(ptr, c);
|
|
1600 code_score(ptr);
|
|
1601 status_clear(ptr);
|
|
1602 }else{
|
|
1603 status_disable(ptr);
|
|
1604 }
|
|
1605 break;
|
|
1606 #ifdef X0212_ENABLE
|
|
1607 case 2:
|
|
1608 if (0xa1 <= c && c <= 0xfe){
|
|
1609 ptr->stat = 1;
|
|
1610 status_push_ch(ptr, c);
|
|
1611 }else{
|
|
1612 status_disable(ptr);
|
|
1613 }
|
|
1614 #endif /* X0212_ENABLE */
|
|
1615 }
|
|
1616 }
|
|
1617
|
|
1618 #ifdef UTF8_INPUT_ENABLE
|
|
1619 void w16_status(ptr, c)
|
|
1620 struct input_code *ptr;
|
|
1621 int c;
|
|
1622 {
|
|
1623 switch (ptr->stat){
|
|
1624 case -1:
|
|
1625 break;
|
|
1626 case 0:
|
|
1627 if (ptr->_file_stat == 0){
|
|
1628 if (c == 0xfe || c == 0xff){
|
|
1629 ptr->stat = c;
|
|
1630 status_push_ch(ptr, c);
|
|
1631 ptr->_file_stat = 1;
|
|
1632 }else{
|
|
1633 status_disable(ptr);
|
|
1634 ptr->_file_stat = -1;
|
|
1635 }
|
|
1636 }else if (ptr->_file_stat > 0){
|
|
1637 ptr->stat = 1;
|
|
1638 status_push_ch(ptr, c);
|
|
1639 }else if (ptr->_file_stat < 0){
|
|
1640 status_disable(ptr);
|
|
1641 }
|
|
1642 break;
|
|
1643
|
|
1644 case 1:
|
|
1645 if (c == EOF){
|
|
1646 status_disable(ptr);
|
|
1647 ptr->_file_stat = -1;
|
|
1648 }else{
|
|
1649 status_push_ch(ptr, c);
|
|
1650 status_clear(ptr);
|
|
1651 }
|
|
1652 break;
|
|
1653
|
|
1654 case 0xfe:
|
|
1655 case 0xff:
|
|
1656 if (ptr->stat != c && (c == 0xfe || c == 0xff)){
|
|
1657 status_push_ch(ptr, c);
|
|
1658 status_clear(ptr);
|
|
1659 }else{
|
|
1660 status_disable(ptr);
|
|
1661 ptr->_file_stat = -1;
|
|
1662 }
|
|
1663 break;
|
|
1664 }
|
|
1665 }
|
|
1666
|
|
1667 void w_status(ptr, c)
|
|
1668 struct input_code *ptr;
|
|
1669 int c;
|
|
1670 {
|
|
1671 switch (ptr->stat){
|
|
1672 case -1:
|
|
1673 status_check(ptr, c);
|
|
1674 break;
|
|
1675 case 0:
|
|
1676 if (c <= DEL){
|
|
1677 break;
|
|
1678 #ifdef NUMCHAR_OPTION
|
|
1679 }else if ((c & CLASS_MASK) == CLASS_UTF16){
|
|
1680 break;
|
|
1681 #endif
|
|
1682 }else if (0xc0 <= c && c <= 0xdf){
|
|
1683 ptr->stat = 1;
|
|
1684 status_push_ch(ptr, c);
|
|
1685 }else if (0xe0 <= c && c <= 0xef){
|
|
1686 ptr->stat = 2;
|
|
1687 status_push_ch(ptr, c);
|
|
1688 }else{
|
|
1689 status_disable(ptr);
|
|
1690 }
|
|
1691 break;
|
|
1692 case 1:
|
|
1693 case 2:
|
|
1694 if (0x80 <= c && c <= 0xbf){
|
|
1695 status_push_ch(ptr, c);
|
|
1696 if (ptr->index > ptr->stat){
|
|
1697 int bom = (ptr->buf[0] == 0xef && ptr->buf[1] == 0xbb
|
|
1698 && ptr->buf[2] == 0xbf);
|
|
1699 w2e_conv(ptr->buf[0], ptr->buf[1], ptr->buf[2],
|
|
1700 &ptr->buf[0], &ptr->buf[1]);
|
|
1701 if (!bom){
|
|
1702 code_score(ptr);
|
|
1703 }
|
|
1704 status_clear(ptr);
|
|
1705 }
|
|
1706 }else{
|
|
1707 status_disable(ptr);
|
|
1708 }
|
|
1709 break;
|
|
1710 }
|
|
1711 }
|
|
1712 #endif
|
|
1713
|
|
1714 void
|
|
1715 code_status(c)
|
|
1716 int c;
|
|
1717 {
|
|
1718 int action_flag = 1;
|
|
1719 struct input_code *result = 0;
|
|
1720 struct input_code *p = input_code_list;
|
|
1721 while (p->name){
|
|
1722 (p->status_func)(p, c);
|
|
1723 if (p->stat > 0){
|
|
1724 action_flag = 0;
|
|
1725 }else if(p->stat == 0){
|
|
1726 if (result){
|
|
1727 action_flag = 0;
|
|
1728 }else{
|
|
1729 result = p;
|
|
1730 }
|
|
1731 }
|
|
1732 ++p;
|
|
1733 }
|
|
1734
|
|
1735 if (action_flag){
|
|
1736 if (result && !estab_f){
|
|
1737 set_iconv(TRUE, result->iconv_func);
|
|
1738 }else if (c <= DEL){
|
|
1739 struct input_code *ptr = input_code_list;
|
|
1740 while (ptr->name){
|
|
1741 status_reset(ptr);
|
|
1742 ++ptr;
|
|
1743 }
|
|
1744 }
|
|
1745 }
|
|
1746 }
|
|
1747
|
|
1748 #ifndef WIN32DLL
|
|
1749 int
|
|
1750 std_getc(f)
|
|
1751 FILE *f;
|
|
1752 {
|
|
1753 if (std_gc_ndx){
|
|
1754 return std_gc_buf[--std_gc_ndx];
|
|
1755 }
|
|
1756 return getc(f);
|
|
1757 }
|
|
1758 #endif /*WIN32DLL*/
|
|
1759
|
|
1760 int
|
|
1761 std_ungetc(c,f)
|
|
1762 int c;
|
|
1763 FILE *f;
|
|
1764 {
|
|
1765 if (std_gc_ndx == STD_GC_BUFSIZE){
|
|
1766 return EOF;
|
|
1767 }
|
|
1768 std_gc_buf[std_gc_ndx++] = c;
|
|
1769 return c;
|
|
1770 }
|
|
1771
|
|
1772 #ifndef WIN32DLL
|
|
1773 void
|
|
1774 std_putc(c)
|
|
1775 int c;
|
|
1776 {
|
|
1777 if(c!=EOF)
|
|
1778 putchar(c);
|
|
1779 }
|
|
1780 #endif /*WIN32DLL*/
|
|
1781
|
|
1782 int
|
|
1783 noconvert(f)
|
|
1784 FILE *f;
|
|
1785 {
|
|
1786 int c;
|
|
1787
|
|
1788 while ((c = (*i_getc)(f)) != EOF)
|
|
1789 (*o_putc)(c);
|
|
1790 return 1;
|
|
1791 }
|
|
1792
|
|
1793
|
|
1794 void
|
|
1795 module_connection()
|
|
1796 {
|
|
1797 oconv = output_conv;
|
|
1798 o_putc = std_putc;
|
|
1799
|
|
1800 /* replace continucation module, from output side */
|
|
1801
|
|
1802 /* output redicrection */
|
|
1803 #ifdef CHECK_OPTION
|
|
1804 if (noout_f || guess_f){
|
|
1805 o_putc = no_putc;
|
|
1806 }
|
|
1807 #endif
|
|
1808 if (mimeout_f) {
|
|
1809 o_mputc = o_putc;
|
|
1810 o_putc = mime_putc;
|
|
1811 if (mimeout_f == TRUE) {
|
|
1812 o_base64conv = oconv; oconv = base64_conv;
|
|
1813 }
|
|
1814 /* base64_count = 0; */
|
|
1815 }
|
|
1816
|
|
1817 if (crmode_f) {
|
|
1818 o_crconv = oconv; oconv = cr_conv;
|
|
1819 }
|
|
1820 if (rot_f) {
|
|
1821 o_rot_conv = oconv; oconv = rot_conv;
|
|
1822 }
|
|
1823 if (iso2022jp_f) {
|
|
1824 o_iso2022jp_check_conv = oconv; oconv = iso2022jp_check_conv;
|
|
1825 }
|
|
1826 if (hira_f) {
|
|
1827 o_hira_conv = oconv; oconv = hira_conv;
|
|
1828 }
|
|
1829 if (fold_f) {
|
|
1830 o_fconv = oconv; oconv = fold_conv;
|
|
1831 f_line = 0;
|
|
1832 }
|
|
1833 if (alpha_f || x0201_f) {
|
|
1834 o_zconv = oconv; oconv = z_conv;
|
|
1835 }
|
|
1836
|
|
1837 i_getc = std_getc;
|
|
1838 i_ungetc = std_ungetc;
|
|
1839 /* input redicrection */
|
|
1840 #ifdef INPUT_OPTION
|
|
1841 if (cap_f){
|
|
1842 i_cgetc = i_getc; i_getc = cap_getc;
|
|
1843 i_cungetc = i_ungetc; i_ungetc= cap_ungetc;
|
|
1844 }
|
|
1845 if (url_f){
|
|
1846 i_ugetc = i_getc; i_getc = url_getc;
|
|
1847 i_uungetc = i_ungetc; i_ungetc= url_ungetc;
|
|
1848 }
|
|
1849 #endif
|
|
1850 #ifdef NUMCHAR_OPTION
|
|
1851 if (numchar_f){
|
|
1852 i_ngetc = i_getc; i_getc = numchar_getc;
|
|
1853 i_nungetc = i_ungetc; i_ungetc= numchar_ungetc;
|
|
1854 }
|
|
1855 #endif
|
|
1856 if (mime_f && mimebuf_f==FIXED_MIME) {
|
|
1857 i_mgetc = i_getc; i_getc = mime_getc;
|
|
1858 i_mungetc = i_ungetc; i_ungetc = mime_ungetc;
|
|
1859 }
|
|
1860 if (broken_f & 1) {
|
|
1861 i_bgetc = i_getc; i_getc = broken_getc;
|
|
1862 i_bungetc = i_ungetc; i_ungetc = broken_ungetc;
|
|
1863 }
|
|
1864 if (input_f == JIS_INPUT || input_f == LATIN1_INPUT) {
|
|
1865 set_iconv(-TRUE, e_iconv);
|
|
1866 } else if (input_f == SJIS_INPUT) {
|
|
1867 set_iconv(-TRUE, s_iconv);
|
|
1868 #ifdef UTF8_INPUT_ENABLE
|
|
1869 } else if (input_f == UTF8_INPUT) {
|
|
1870 set_iconv(-TRUE, w_iconv);
|
|
1871 } else if (input_f == UTF16BE_INPUT) {
|
|
1872 set_iconv(-TRUE, w_iconv16);
|
|
1873 } else if (input_f == UTF16LE_INPUT) {
|
|
1874 set_iconv(-TRUE, w_iconv16);
|
|
1875 #endif
|
|
1876 } else {
|
|
1877 set_iconv(FALSE, e_iconv);
|
|
1878 }
|
|
1879
|
|
1880 {
|
|
1881 struct input_code *p = input_code_list;
|
|
1882 while (p->name){
|
|
1883 status_reinit(p++);
|
|
1884 }
|
|
1885 }
|
|
1886 }
|
|
1887
|
|
1888 /*
|
|
1889 Conversion main loop. Code detection only.
|
|
1890 */
|
|
1891
|
|
1892 int
|
|
1893 kanji_convert(f)
|
|
1894 FILE *f;
|
|
1895 {
|
|
1896 int c1,
|
|
1897 c2, c3;
|
|
1898 int is_8bit = FALSE;
|
|
1899
|
|
1900 module_connection();
|
|
1901 c2 = 0;
|
|
1902
|
|
1903
|
|
1904 input_mode = ASCII;
|
|
1905 output_mode = ASCII;
|
|
1906 shift_mode = FALSE;
|
|
1907
|
|
1908 #define NEXT continue /* no output, get next */
|
|
1909 #define SEND ; /* output c1 and c2, get next */
|
|
1910 #define LAST break /* end of loop, go closing */
|
|
1911
|
|
1912 while ((c1 = (*i_getc)(f)) != EOF) {
|
|
1913 code_status(c1);
|
|
1914 if (c2) {
|
|
1915 /* second byte */
|
|
1916 if (c2 > DEL) {
|
|
1917 /* in case of 8th bit is on */
|
|
1918 if (!estab_f&&!mime_decode_mode) {
|
|
1919 /* in case of not established yet */
|
|
1920 /* It is still ambiguious */
|
|
1921 if (h_conv(f, c2, c1)==EOF)
|
|
1922 LAST;
|
|
1923 else
|
|
1924 c2 = 0;
|
|
1925 NEXT;
|
|
1926 } else
|
|
1927 /* in case of already established */
|
|
1928 if (c1 < AT) {
|
|
1929 /* ignore bogus code */
|
|
1930 c2 = 0;
|
|
1931 NEXT;
|
|
1932 } else
|
|
1933 SEND;
|
|
1934 } else
|
|
1935 /* second byte, 7 bit code */
|
|
1936 /* it might be kanji shitfted */
|
|
1937 if ((c1 == DEL) || (c1 <= SPACE)) {
|
|
1938 /* ignore bogus first code */
|
|
1939 c2 = 0;
|
|
1940 NEXT;
|
|
1941 } else
|
|
1942 SEND;
|
|
1943 } else {
|
|
1944 /* first byte */
|
|
1945 if (
|
|
1946 #ifdef UTF8_INPUT_ENABLE
|
|
1947 iconv == w_iconv16
|
|
1948 #else
|
|
1949 0
|
|
1950 #endif
|
|
1951 ) {
|
|
1952 c2 = c1;
|
|
1953 c1 = (*i_getc)(f);
|
|
1954 SEND;
|
|
1955 #ifdef NUMCHAR_OPTION
|
|
1956 } else if ((c1 & CLASS_MASK) == CLASS_UTF16){
|
|
1957 SEND;
|
|
1958 #endif
|
|
1959 } else if (c1 > DEL) {
|
|
1960 /* 8 bit code */
|
|
1961 if (!estab_f && !iso8859_f) {
|
|
1962 /* not established yet */
|
|
1963 if (!is_8bit) is_8bit = TRUE;
|
|
1964 c2 = c1;
|
|
1965 NEXT;
|
|
1966 } else { /* estab_f==TRUE */
|
|
1967 if (iso8859_f) {
|
|
1968 c2 = ISO8859_1;
|
|
1969 c1 &= 0x7f;
|
|
1970 SEND;
|
|
1971 } else if (SSP<=c1 && c1<0xe0 && iconv == s_iconv) {
|
|
1972 /* SJIS X0201 Case... */
|
|
1973 if(iso2022jp_f && x0201_f==NO_X0201) {
|
|
1974 (*oconv)(GETA1, GETA2);
|
|
1975 NEXT;
|
|
1976 } else {
|
|
1977 c2 = X0201;
|
|
1978 c1 &= 0x7f;
|
|
1979 SEND;
|
|
1980 }
|
|
1981 } else if (c1==SSO && iconv != s_iconv) {
|
|
1982 /* EUC X0201 Case */
|
|
1983 c1 = (*i_getc)(f); /* skip SSO */
|
|
1984 code_status(c1);
|
|
1985 if (SSP<=c1 && c1<0xe0) {
|
|
1986 if(iso2022jp_f && x0201_f==NO_X0201) {
|
|
1987 (*oconv)(GETA1, GETA2);
|
|
1988 NEXT;
|
|
1989 } else {
|
|
1990 c2 = X0201;
|
|
1991 c1 &= 0x7f;
|
|
1992 SEND;
|
|
1993 }
|
|
1994 } else { /* bogus code, skip SSO and one byte */
|
|
1995 NEXT;
|
|
1996 }
|
|
1997 } else {
|
|
1998 /* already established */
|
|
1999 c2 = c1;
|
|
2000 NEXT;
|
|
2001 }
|
|
2002 }
|
|
2003 } else if ((c1 > SPACE) && (c1 != DEL)) {
|
|
2004 /* in case of Roman characters */
|
|
2005 if (shift_mode) {
|
|
2006 /* output 1 shifted byte */
|
|
2007 if (iso8859_f) {
|
|
2008 c2 = ISO8859_1;
|
|
2009 SEND;
|
|
2010 } else if (SPACE<=c1 && c1<(0xe0&0x7f) ){
|
|
2011 /* output 1 shifted byte */
|
|
2012 if(iso2022jp_f && x0201_f==NO_X0201) {
|
|
2013 (*oconv)(GETA1, GETA2);
|
|
2014 NEXT;
|
|
2015 } else {
|
|
2016 c2 = X0201;
|
|
2017 SEND;
|
|
2018 }
|
|
2019 } else {
|
|
2020 /* look like bogus code */
|
|
2021 NEXT;
|
|
2022 }
|
|
2023 } else if (input_mode == X0208) {
|
|
2024 /* in case of Kanji shifted */
|
|
2025 c2 = c1;
|
|
2026 NEXT;
|
|
2027 } else if (c1 == '=' && mime_f && !mime_decode_mode ) {
|
|
2028 /* Check MIME code */
|
|
2029 if ((c1 = (*i_getc)(f)) == EOF) {
|
|
2030 (*oconv)(0, '=');
|
|
2031 LAST;
|
|
2032 } else if (c1 == '?') {
|
|
2033 /* =? is mime conversion start sequence */
|
|
2034 if(mime_f == STRICT_MIME) {
|
|
2035 /* check in real detail */
|
|
2036 if (mime_begin_strict(f) == EOF)
|
|
2037 LAST;
|
|
2038 else
|
|
2039 NEXT;
|
|
2040 } else if (mime_begin(f) == EOF)
|
|
2041 LAST;
|
|
2042 else
|
|
2043 NEXT;
|
|
2044 } else {
|
|
2045 (*oconv)(0, '=');
|
|
2046 (*i_ungetc)(c1,f);
|
|
2047 NEXT;
|
|
2048 }
|
|
2049 } else {
|
|
2050 /* normal ASCII code */
|
|
2051 SEND;
|
|
2052 }
|
|
2053 } else if (c1 == SI) {
|
|
2054 shift_mode = FALSE;
|
|
2055 NEXT;
|
|
2056 } else if (c1 == SO) {
|
|
2057 shift_mode = TRUE;
|
|
2058 NEXT;
|
|
2059 } else if (c1 == ESC ) {
|
|
2060 if ((c1 = (*i_getc)(f)) == EOF) {
|
|
2061 /* (*oconv)(0, ESC); don't send bogus code */
|
|
2062 LAST;
|
|
2063 } else if (c1 == '$') {
|
|
2064 if ((c1 = (*i_getc)(f)) == EOF) {
|
|
2065 /*
|
|
2066 (*oconv)(0, ESC); don't send bogus code
|
|
2067 (*oconv)(0, '$'); */
|
|
2068 LAST;
|
|
2069 } else if (c1 == '@'|| c1 == 'B') {
|
|
2070 /* This is kanji introduction */
|
|
2071 input_mode = X0208;
|
|
2072 shift_mode = FALSE;
|
|
2073 set_input_codename("ISO-2022-JP");
|
|
2074 debug(input_codename);
|
|
2075 NEXT;
|
|
2076 } else if (c1 == '(') {
|
|
2077 if ((c1 = (*i_getc)(f)) == EOF) {
|
|
2078 /* don't send bogus code
|
|
2079 (*oconv)(0, ESC);
|
|
2080 (*oconv)(0, '$');
|
|
2081 (*oconv)(0, '(');
|
|
2082 */
|
|
2083 LAST;
|
|
2084 } else if (c1 == '@'|| c1 == 'B') {
|
|
2085 /* This is kanji introduction */
|
|
2086 input_mode = X0208;
|
|
2087 shift_mode = FALSE;
|
|
2088 NEXT;
|
|
2089 #ifdef X0212_ENABLE
|
|
2090 } else if (c1 == 'D'){
|
|
2091 input_mode = X0212;
|
|
2092 shift_mode = FALSE;
|
|
2093 NEXT;
|
|
2094 #endif /* X0212_ENABLE */
|
|
2095 } else {
|
|
2096 /* could be some special code */
|
|
2097 (*oconv)(0, ESC);
|
|
2098 (*oconv)(0, '$');
|
|
2099 (*oconv)(0, '(');
|
|
2100 (*oconv)(0, c1);
|
|
2101 NEXT;
|
|
2102 }
|
|
2103 } else if (broken_f&0x2) {
|
|
2104 /* accept any ESC-(-x as broken code ... */
|
|
2105 input_mode = X0208;
|
|
2106 shift_mode = FALSE;
|
|
2107 NEXT;
|
|
2108 } else {
|
|
2109 (*oconv)(0, ESC);
|
|
2110 (*oconv)(0, '$');
|
|
2111 (*oconv)(0, c1);
|
|
2112 NEXT;
|
|
2113 }
|
|
2114 } else if (c1 == '(') {
|
|
2115 if ((c1 = (*i_getc)(f)) == EOF) {
|
|
2116 /* don't send bogus code
|
|
2117 (*oconv)(0, ESC);
|
|
2118 (*oconv)(0, '('); */
|
|
2119 LAST;
|
|
2120 } else {
|
|
2121 if (c1 == 'I') {
|
|
2122 /* This is X0201 kana introduction */
|
|
2123 input_mode = X0201; shift_mode = X0201;
|
|
2124 NEXT;
|
|
2125 } else if (c1 == 'B' || c1 == 'J' || c1 == 'H') {
|
|
2126 /* This is X0208 kanji introduction */
|
|
2127 input_mode = ASCII; shift_mode = FALSE;
|
|
2128 NEXT;
|
|
2129 } else if (broken_f&0x2) {
|
|
2130 input_mode = ASCII; shift_mode = FALSE;
|
|
2131 NEXT;
|
|
2132 } else {
|
|
2133 (*oconv)(0, ESC);
|
|
2134 (*oconv)(0, '(');
|
|
2135 /* maintain various input_mode here */
|
|
2136 SEND;
|
|
2137 }
|
|
2138 }
|
|
2139 } else if ( c1 == 'N' || c1 == 'n' ){
|
|
2140 /* SS2 */
|
|
2141 c3 = (*i_getc)(f); /* skip SS2 */
|
|
2142 if ( (SPACE<=c3 && c3 < 0x60) || (0xa0<=c3 && c3 < 0xe0)){
|
|
2143 c1 = c3;
|
|
2144 c2 = X0201;
|
|
2145 SEND;
|
|
2146 }else{
|
|
2147 (*i_ungetc)(c3, f);
|
|
2148 /* lonely ESC */
|
|
2149 (*oconv)(0, ESC);
|
|
2150 SEND;
|
|
2151 }
|
|
2152 } else {
|
|
2153 /* lonely ESC */
|
|
2154 (*oconv)(0, ESC);
|
|
2155 SEND;
|
|
2156 }
|
|
2157 } else if ((c1 == NL || c1 == CR) && broken_f&4) {
|
|
2158 input_mode = ASCII; set_iconv(FALSE, 0);
|
|
2159 SEND;
|
|
2160 } else if (c1 == NL && mime_decode_f && !mime_decode_mode ) {
|
|
2161 if ((c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
|
|
2162 i_ungetc(SPACE,f);
|
|
2163 continue;
|
|
2164 } else {
|
|
2165 i_ungetc(c1,f);
|
|
2166 }
|
|
2167 c1 = NL;
|
|
2168 SEND;
|
|
2169 } else if (c1 == CR && mime_decode_f && !mime_decode_mode ) {
|
|
2170 if ((c1=(*i_getc)(f))!=EOF) {
|
|
2171 if (c1==SPACE) {
|
|
2172 i_ungetc(SPACE,f);
|
|
2173 continue;
|
|
2174 } else if (c1 == NL && (c1=(*i_getc)(f))!=EOF && c1 == SPACE) {
|
|
2175 i_ungetc(SPACE,f);
|
|
2176 continue;
|
|
2177 } else {
|
|
2178 i_ungetc(c1,f);
|
|
2179 }
|
|
2180 i_ungetc(NL,f);
|
|
2181 } else {
|
|
2182 i_ungetc(c1,f);
|
|
2183 }
|
|
2184 c1 = CR;
|
|
2185 SEND;
|
|
2186 } else
|
|
2187 SEND;
|
|
2188 }
|
|
2189 /* send: */
|
|
2190 if (input_mode == X0208)
|
|
2191 (*oconv)(c2, c1); /* this is JIS, not SJIS/EUC case */
|
|
2192 #ifdef X0212_ENABLE
|
|
2193 else if (input_mode == X0212)
|
|
2194 (*oconv)((0x8f << 8) | c2, c1);
|
|
2195 #endif /* X0212_ENABLE */
|
|
2196 else if (input_mode)
|
|
2197 (*oconv)(input_mode, c1); /* other special case */
|
|
2198 else if ((*iconv)(c2, c1, 0) < 0){ /* can be EUC/SJIS */
|
|
2199 int c0 = (*i_getc)(f);
|
|
2200 if (c0 != EOF){
|
|
2201 code_status(c0);
|
|
2202 (*iconv)(c2, c1, c0);
|
|
2203 }
|
|
2204 }
|
|
2205
|
|
2206 c2 = 0;
|
|
2207 continue;
|
|
2208 /* goto next_word */
|
|
2209 }
|
|
2210
|
|
2211 /* epilogue */
|
|
2212 (*iconv)(EOF, 0, 0);
|
|
2213 if (!is_inputcode_set)
|
|
2214 {
|
|
2215 if (is_8bit) {
|
|
2216 struct input_code *p = input_code_list;
|
|
2217 struct input_code *result = p;
|
|
2218 while (p->name){
|
|
2219 if (p->score < result->score) result = p;
|
|
2220 ++p;
|
|
2221 }
|
|
2222 set_input_codename(result->name);
|
|
2223 }
|
|
2224 }
|
|
2225 return 1;
|
|
2226 }
|
|
2227
|
|
2228 int
|
|
2229 h_conv(f, c2, c1)
|
|
2230 FILE *f;
|
|
2231 int c1,
|
|
2232 c2;
|
|
2233 {
|
|
2234 int wc,c3;
|
|
2235
|
|
2236
|
|
2237 /** it must NOT be in the kanji shifte sequence */
|
|
2238 /** it must NOT be written in JIS7 */
|
|
2239 /** and it must be after 2 byte 8bit code */
|
|
2240
|
|
2241 hold_count = 0;
|
|
2242 push_hold_buf(c2);
|
|
2243 push_hold_buf(c1);
|
|
2244 c2 = 0;
|
|
2245
|
|
2246 while ((c1 = (*i_getc)(f)) != EOF) {
|
|
2247 if (c1 == ESC){
|
|
2248 (*i_ungetc)(c1,f);
|
|
2249 break;
|
|
2250 }
|
|
2251 code_status(c1);
|
|
2252 if (push_hold_buf(c1) == EOF || estab_f){
|
|
2253 break;
|
|
2254 }
|
|
2255 }
|
|
2256
|
|
2257 if (!estab_f){
|
|
2258 struct input_code *p = input_code_list;
|
|
2259 struct input_code *result = p;
|
|
2260 if (c1 == EOF){
|
|
2261 code_status(c1);
|
|
2262 }
|
|
2263 while (p->name){
|
|
2264 if (p->score < result->score){
|
|
2265 result = p;
|
|
2266 }
|
|
2267 ++p;
|
|
2268 }
|
|
2269 set_iconv(FALSE, result->iconv_func);
|
|
2270 }
|
|
2271
|
|
2272
|
|
2273 /** now,
|
|
2274 ** 1) EOF is detected, or
|
|
2275 ** 2) Code is established, or
|
|
2276 ** 3) Buffer is FULL (but last word is pushed)
|
|
2277 **
|
|
2278 ** in 1) and 3) cases, we continue to use
|
|
2279 ** Kanji codes by oconv and leave estab_f unchanged.
|
|
2280 **/
|
|
2281
|
|
2282 c3=c1;
|
|
2283 wc = 0;
|
|
2284 while (wc < hold_count){
|
|
2285 c2 = hold_buf[wc++];
|
|
2286 if (c2 <= DEL
|
|
2287 #ifdef NUMCHAR_OPTION
|
|
2288 || (c2 & CLASS_MASK) == CLASS_UTF16
|
|
2289 #endif
|
|
2290 ){
|
|
2291 (*iconv)(0, c2, 0);
|
|
2292 continue;
|
|
2293 }else if (iconv == s_iconv && 0xa1 <= c2 && c2 <= 0xdf){
|
|
2294 (*iconv)(X0201, c2, 0);
|
|
2295 continue;
|
|
2296 }
|
|
2297 if (wc < hold_count){
|
|
2298 c1 = hold_buf[wc++];
|
|
2299 }else{
|
|
2300 c1 = (*i_getc)(f);
|
|
2301 if (c1 == EOF){
|
|
2302 c3 = EOF;
|
|
2303 break;
|
|
2304 }
|
|
2305 code_status(c1);
|
|
2306 }
|
|
2307 if ((*iconv)(c2, c1, 0) < 0){
|
|
2308 int c0;
|
|
2309 if (wc < hold_count){
|
|
2310 c0 = hold_buf[wc++];
|
|
2311 }else{
|
|
2312 c0 = (*i_getc)(f);
|
|
2313 if (c0 == EOF){
|
|
2314 c3 = EOF;
|
|
2315 break;
|
|
2316 }
|
|
2317 code_status(c0);
|
|
2318 }
|
|
2319 (*iconv)(c2, c1, c0);
|
|
2320 c1 = c0;
|
|
2321 }
|
|
2322 }
|
|
2323 return c3;
|
|
2324 }
|
|
2325
|
|
2326
|
|
2327
|
|
2328 int
|
|
2329 push_hold_buf(c2)
|
|
2330 int c2;
|
|
2331 {
|
|
2332 if (hold_count >= HOLD_SIZE*2)
|
|
2333 return (EOF);
|
|
2334 hold_buf[hold_count++] = c2;
|
|
2335 return ((hold_count >= HOLD_SIZE*2) ? EOF : hold_count);
|
|
2336 }
|
|
2337
|
|
2338 int s2e_conv(c2, c1, p2, p1)
|
|
2339 int c2, c1;
|
|
2340 int *p2, *p1;
|
|
2341 {
|
|
2342 int val;
|
|
2343 #ifdef SHIFTJIS_CP932
|
|
2344 if (cp932_f && CP932_TABLE_BEGIN <= c2 && c2 <= CP932_TABLE_END){
|
|
2345 extern unsigned short shiftjis_cp932[3][189];
|
|
2346 val = shiftjis_cp932[c2 - CP932_TABLE_BEGIN][c1 - 0x40];
|
|
2347 if (val){
|
|
2348 c2 = val >> 8;
|
|
2349 c1 = val & 0xff;
|
|
2350 }
|
|
2351 }
|
|
2352 #endif /* SHIFTJIS_CP932 */
|
|
2353 #ifdef X0212_ENABLE
|
|
2354 if (x0212_f && 0xfa <= c2 && c2 <= 0xfc){
|
|
2355 extern unsigned short shiftjis_x0212[3][189];
|
|
2356 val = shiftjis_x0212[c2 - 0xfa][c1 - 0x40];
|
|
2357 if (val){
|
|
2358 if (val & 0x8000){
|
|
2359 c2 = (0x8f << 8) | (val >> 8);
|
|
2360 c1 = val & 0xff;
|
|
2361 }else{
|
|
2362 c2 = val >> 8;
|
|
2363 c1 = val & 0xff;
|
|
2364 }
|
|
2365 if (p2) *p2 = c2;
|
|
2366 if (p1) *p1 = c1;
|
|
2367 return 0;
|
|
2368 }
|
|
2369 }
|
|
2370 #endif
|
|
2371 c2 = c2 + c2 - ((c2 <= 0x9f) ? SJ0162 : SJ6394);
|
|
2372 if (c1 < 0x9f)
|
|
2373 c1 = c1 - ((c1 > DEL) ? SPACE : 0x1f);
|
|
2374 else {
|
|
2375 c1 = c1 - 0x7e;
|
|
2376 c2++;
|
|
2377 }
|
|
2378
|
|
2379 #ifdef X0212_ENABLE
|
|
2380 if (x0212_f){
|
|
2381 c2 = x0212_unshift(c2);
|
|
2382 }
|
|
2383 #endif
|
|
2384 if (p2) *p2 = c2;
|
|
2385 if (p1) *p1 = c1;
|
|
2386 return 0;
|
|
2387 }
|
|
2388
|
|
2389 int
|
|
2390 s_iconv(c2, c1, c0)
|
|
2391 int c2,
|
|
2392 c1, c0;
|
|
2393 {
|
|
2394 if (c2 == X0201) {
|
|
2395 c1 &= 0x7f;
|
|
2396 } else if ((c2 == EOF) || (c2 == 0) || c2 < SPACE) {
|
|
2397 /* NOP */
|
|
2398 } else {
|
|
2399 int ret = s2e_conv(c2, c1, &c2, &c1);
|
|
2400 if (ret) return ret;
|
|
2401 }
|
|
2402 (*oconv)(c2, c1);
|
|
2403 return 0;
|
|
2404 }
|
|
2405
|
|
2406 int
|
|
2407 e_iconv(c2, c1, c0)
|
|
2408 int c2,
|
|
2409 c1, c0;
|
|
2410 {
|
|
2411 if (c2 == X0201) {
|
|
2412 c1 &= 0x7f;
|
|
2413 #ifdef X0212_ENABLE
|
|
2414 }else if (c2 == 0x8f){
|
|
2415 if (c0 == 0){
|
|
2416 return -1;
|
|
2417 }
|
|
2418 c2 = (c2 << 8) | (c1 & 0x7f);
|
|
2419 c1 = c0 & 0x7f;
|
|
2420 #ifdef SHIFTJIS_CP932
|
|
2421 if (cp932_f){
|
|
2422 int s2, s1;
|
|
2423 if (e2s_conv(c2, c1, &s2, &s1) == 0){
|
|
2424 s2e_conv(s2, s1, &c2, &c1);
|
|
2425 if ((c2 & 0xff00) == 0){
|
|
2426 c1 &= 0x7f;
|
|
2427 c2 &= 0x7f;
|
|
2428 }
|
|
2429 }
|
|
2430 }
|
|
2431 #endif /* SHIFTJIS_CP932 */
|
|
2432 #endif /* X0212_ENABLE */
|
|
2433 } else if (c2 == SSO){
|
|
2434 c2 = X0201;
|
|
2435 c1 &= 0x7f;
|
|
2436 } else if ((c2 == EOF) || (c2 == 0) || c2 < SPACE) {
|
|
2437 /* NOP */
|
|
2438 } else {
|
|
2439 c1 &= 0x7f;
|
|
2440 c2 &= 0x7f;
|
|
2441 }
|
|
2442 (*oconv)(c2, c1);
|
|
2443 return 0;
|
|
2444 }
|
|
2445
|
|
2446 #ifdef UTF8_INPUT_ENABLE
|
|
2447 int
|
|
2448 w2e_conv(c2, c1, c0, p2, p1)
|
|
2449 int c2, c1, c0;
|
|
2450 int *p2, *p1;
|
|
2451 {
|
|
2452 extern unsigned short * utf8_to_euc_2bytes[];
|
|
2453 extern unsigned short ** utf8_to_euc_3bytes[];
|
|
2454 int ret = 0;
|
|
2455
|
|
2456 if (0xc0 <= c2 && c2 <= 0xef) {
|
|
2457 unsigned short **pp;
|
|
2458
|
|
2459 if (0xe0 <= c2) {
|
|
2460 if (c0 == 0) return -1;
|
|
2461 pp = utf8_to_euc_3bytes[c2 - 0x80];
|
|
2462 ret = w_iconv_common(c1, c0, pp, sizeof_utf8_to_euc_C2, p2, p1);
|
|
2463 } else {
|
|
2464 ret = w_iconv_common(c2, c1, utf8_to_euc_2bytes, sizeof_utf8_to_euc_2bytes, p2, p1);
|
|
2465 }
|
|
2466 #ifdef NUMCHAR_OPTION
|
|
2467 if (ret){
|
|
2468 if (p2) *p2 = 0;
|
|
2469 if (p1) *p1 = CLASS_UTF16 | ww16_conv(c2, c1, c0);
|
|
2470 ret = 0;
|
|
2471 }
|
|
2472 #endif
|
|
2473 return ret;
|
|
2474 } else if (c2 == X0201) {
|
|
2475 c1 &= 0x7f;
|
|
2476 }
|
|
2477 if (p2) *p2 = c2;
|
|
2478 if (p1) *p1 = c1;
|
|
2479 return ret;
|
|
2480 }
|
|
2481
|
|
2482 int
|
|
2483 w_iconv(c2, c1, c0)
|
|
2484 int c2,
|
|
2485 c1, c0;
|
|
2486 {
|
|
2487 int ret = w2e_conv(c2, c1, c0, &c2, &c1);
|
|
2488 if (ret == 0){
|
|
2489 (*oconv)(c2, c1);
|
|
2490 }
|
|
2491 return ret;
|
|
2492 }
|
|
2493
|
|
2494 void
|
|
2495 w16w_conv(val, p2, p1, p0)
|
|
2496 unsigned short val;
|
|
2497 int *p2, *p1, *p0;
|
|
2498 {
|
|
2499 if (val < 0x80){
|
|
2500 *p2 = val;
|
|
2501 *p1 = 0;
|
|
2502 *p0 = 0;
|
|
2503 }else if (val < 0x800){
|
|
2504 *p2 = 0xc0 | (val >> 6);
|
|
2505 *p1 = 0x80 | (val & 0x3f);
|
|
2506 *p0 = 0;
|
|
2507 }else{
|
|
2508 *p2 = 0xe0 | (val >> 12);
|
|
2509 *p1 = 0x80 | ((val >> 6) & 0x3f);
|
|
2510 *p0 = 0x80 | (val & 0x3f);
|
|
2511 }
|
|
2512 }
|
|
2513
|
|
2514 int
|
|
2515 ww16_conv(c2, c1, c0)
|
|
2516 int c2, c1, c0;
|
|
2517 {
|
|
2518 unsigned short val;
|
|
2519 if (c2 >= 0xe0){
|
|
2520 val = (c2 & 0x0f) << 12;
|
|
2521 val |= (c1 & 0x3f) << 6;
|
|
2522 val |= (c0 & 0x3f);
|
|
2523 }else if (c2 >= 0xc0){
|
|
2524 val = (c2 & 0x1f) << 6;
|
|
2525 val |= (c1 & 0x3f);
|
|
2526 }else{
|
|
2527 val = c2;
|
|
2528 }
|
|
2529 return val;
|
|
2530 }
|
|
2531
|
|
2532 int
|
|
2533 w16e_conv(val, p2, p1)
|
|
2534 unsigned short val;
|
|
2535 int *p2, *p1;
|
|
2536 {
|
|
2537 extern unsigned short * utf8_to_euc_2bytes[];
|
|
2538 extern unsigned short ** utf8_to_euc_3bytes[];
|
|
2539 int c2, c1, c0;
|
|
2540 unsigned short **pp;
|
|
2541 int psize;
|
|
2542 int ret = 0;
|
|
2543
|
|
2544 w16w_conv(val, &c2, &c1, &c0);
|
|
2545 if (c1){
|
|
2546 if (c0){
|
|
2547 pp = utf8_to_euc_3bytes[c2 - 0x80];
|
|
2548 psize = sizeof_utf8_to_euc_C2;
|
|
2549 ret = w_iconv_common(c1, c0, pp, psize, p2, p1);
|
|
2550 }else{
|
|
2551 pp = utf8_to_euc_2bytes;
|
|
2552 psize = sizeof_utf8_to_euc_2bytes;
|
|
2553 ret = w_iconv_common(c2, c1, pp, psize, p2, p1);
|
|
2554 }
|
|
2555 #ifdef NUMCHAR_OPTION
|
|
2556 if (ret){
|
|
2557 *p2 = 0;
|
|
2558 *p1 = CLASS_UTF16 | val;
|
|
2559 ret = 0;
|
|
2560 }
|
|
2561 #endif
|
|
2562 }else{
|
|
2563 *p2 = 0;
|
|
2564 *p1 = c2;
|
|
2565 }
|
|
2566 return ret;
|
|
2567 }
|
|
2568
|
|
2569 int
|
|
2570 w_iconv16(c2, c1, c0)
|
|
2571 int c2, c1,c0;
|
|
2572 {
|
|
2573 int ret;
|
|
2574
|
|
2575 if (c2==0376 && c1==0377){
|
|
2576 utf16_mode = UTF16BE_INPUT;
|
|
2577 return 0;
|
|
2578 } else if (c2==0377 && c1==0376){
|
|
2579 utf16_mode = UTF16LE_INPUT;
|
|
2580 return 0;
|
|
2581 }
|
|
2582 if (c2 != EOF && utf16_mode == UTF16LE_INPUT) {
|
|
2583 int tmp;
|
|
2584 tmp=c1; c1=c2; c2=tmp;
|
|
2585 }
|
|
2586 if ((c2==0 && c1 < 0x80) || c2==EOF) {
|
|
2587 (*oconv)(c2, c1);
|
|
2588 return 0;
|
|
2589 }
|
|
2590 ret = w16e_conv(((c2<<8)&0xff00) + c1, &c2, &c1);
|
|
2591 if (ret) return ret;
|
|
2592 (*oconv)(c2, c1);
|
|
2593 return 0;
|
|
2594 }
|
|
2595
|
|
2596 int
|
|
2597 w_iconv_common(c1, c0, pp, psize, p2, p1)
|
|
2598 int c1,c0;
|
|
2599 unsigned short **pp;
|
|
2600 int psize;
|
|
2601 int *p2, *p1;
|
|
2602 {
|
|
2603 int c2;
|
|
2604 unsigned short *p ;
|
|
2605 unsigned short val;
|
|
2606
|
|
2607 if (pp == 0) return 1;
|
|
2608
|
|
2609 c1 -= 0x80;
|
|
2610 if (c1 < 0 || psize <= c1) return 1;
|
|
2611 p = pp[c1];
|
|
2612 if (p == 0) return 1;
|
|
2613
|
|
2614 c0 -= 0x80;
|
|
2615 if (c0 < 0 || sizeof_utf8_to_euc_E5B8 <= c0) return 1;
|
|
2616 val = p[c0];
|
|
2617 if (val == 0) return 1;
|
|
2618
|
|
2619 c2 = val >> 8;
|
|
2620 if (val & 0x8000){
|
|
2621 c2 &= 0x7f;
|
|
2622 c2 |= 0x8f00;
|
|
2623 }
|
|
2624 if (c2 == SO) c2 = X0201;
|
|
2625 c1 = val & 0x7f;
|
|
2626 if (p2) *p2 = c2;
|
|
2627 if (p1) *p1 = c1;
|
|
2628 return 0;
|
|
2629 }
|
|
2630
|
|
2631 #endif
|
|
2632
|
|
2633 #ifdef UTF8_OUTPUT_ENABLE
|
|
2634 int
|
|
2635 e2w_conv(c2, c1)
|
|
2636 int c2, c1;
|
|
2637 {
|
|
2638 extern unsigned short euc_to_utf8_1byte[];
|
|
2639 extern unsigned short * euc_to_utf8_2bytes[];
|
|
2640 extern unsigned short * euc_to_utf8_2bytes_ms[];
|
|
2641 unsigned short *p;
|
|
2642
|
|
2643 if (c2 == X0201) {
|
|
2644 p = euc_to_utf8_1byte;
|
|
2645 #ifdef X0212_ENABLE
|
|
2646 } else if (c2 >> 8 == 0x8f){
|
|
2647 extern unsigned short * x0212_to_utf8_2bytes[];
|
|
2648 c2 = (c2&0x7f) - 0x21;
|
|
2649 if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
|
|
2650 p = x0212_to_utf8_2bytes[c2];
|
|
2651 else
|
|
2652 return 0;
|
|
2653 #endif
|
|
2654 } else {
|
|
2655 c2 &= 0x7f;
|
|
2656 c2 = (c2&0x7f) - 0x21;
|
|
2657 if (0<=c2 && c2<sizeof_euc_to_utf8_2bytes)
|
|
2658 p = ms_ucs_map_f ? euc_to_utf8_2bytes_ms[c2] : euc_to_utf8_2bytes[c2];
|
|
2659 else
|
|
2660 return 0;
|
|
2661 }
|
|
2662 if (!p) return 0;
|
|
2663 c1 = (c1 & 0x7f) - 0x21;
|
|
2664 if (0<=c1 && c1<sizeof_euc_to_utf8_1byte)
|
|
2665 return p[c1];
|
|
2666 return 0;
|
|
2667 }
|
|
2668
|
|
2669 void
|
|
2670 w_oconv(c2, c1)
|
|
2671 int c2,
|
|
2672 c1;
|
|
2673 {
|
|
2674 int c0;
|
|
2675 if (c2 == EOF) {
|
|
2676 (*o_putc)(EOF);
|
|
2677 return;
|
|
2678 }
|
|
2679
|
|
2680 if (unicode_bom_f==2) {
|
|
2681 (*o_putc)('\357');
|
|
2682 (*o_putc)('\273');
|
|
2683 (*o_putc)('\277');
|
|
2684 unicode_bom_f=1;
|
|
2685 }
|
|
2686
|
|
2687 #ifdef NUMCHAR_OPTION
|
|
2688 if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
|
|
2689 w16w_conv(c1, &c2, &c1, &c0);
|
|
2690 (*o_putc)(c2);
|
|
2691 if (c1){
|
|
2692 (*o_putc)(c1);
|
|
2693 if (c0) (*o_putc)(c0);
|
|
2694 }
|
|
2695 return;
|
|
2696 }
|
|
2697 #endif
|
|
2698
|
|
2699 if (c2 == 0) {
|
|
2700 output_mode = ASCII;
|
|
2701 (*o_putc)(c1);
|
|
2702 } else if (c2 == ISO8859_1) {
|
|
2703 output_mode = ISO8859_1;
|
|
2704 (*o_putc)(c1 | 0x080);
|
|
2705 } else {
|
|
2706 unsigned short val;
|
|
2707 output_mode = UTF8;
|
|
2708 val = e2w_conv(c2, c1);
|
|
2709 if (val){
|
|
2710 w16w_conv(val, &c2, &c1, &c0);
|
|
2711 (*o_putc)(c2);
|
|
2712 if (c1){
|
|
2713 (*o_putc)(c1);
|
|
2714 if (c0) (*o_putc)(c0);
|
|
2715 }
|
|
2716 }
|
|
2717 }
|
|
2718 }
|
|
2719
|
|
2720 void
|
|
2721 w_oconv16(c2, c1)
|
|
2722 int c2,
|
|
2723 c1;
|
|
2724 {
|
|
2725 if (c2 == EOF) {
|
|
2726 (*o_putc)(EOF);
|
|
2727 return;
|
|
2728 }
|
|
2729
|
|
2730 if (unicode_bom_f==2) {
|
|
2731 if (w_oconv16_LE){
|
|
2732 (*o_putc)((unsigned char)'\377');
|
|
2733 (*o_putc)('\376');
|
|
2734 }else{
|
|
2735 (*o_putc)('\376');
|
|
2736 (*o_putc)((unsigned char)'\377');
|
|
2737 }
|
|
2738 unicode_bom_f=1;
|
|
2739 }
|
|
2740
|
|
2741 if (c2 == ISO8859_1) {
|
|
2742 c2 = 0;
|
|
2743 c1 |= 0x80;
|
|
2744 #ifdef NUMCHAR_OPTION
|
|
2745 } else if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16) {
|
|
2746 c2 = (c1 >> 8) & 0xff;
|
|
2747 c1 &= 0xff;
|
|
2748 #endif
|
|
2749 } else if (c2) {
|
|
2750 unsigned short val = e2w_conv(c2, c1);
|
|
2751 c2 = (val >> 8) & 0xff;
|
|
2752 c1 = val & 0xff;
|
|
2753 }
|
|
2754 if (w_oconv16_LE){
|
|
2755 (*o_putc)(c1);
|
|
2756 (*o_putc)(c2);
|
|
2757 }else{
|
|
2758 (*o_putc)(c2);
|
|
2759 (*o_putc)(c1);
|
|
2760 }
|
|
2761 }
|
|
2762
|
|
2763 #endif
|
|
2764
|
|
2765 void
|
|
2766 e_oconv(c2, c1)
|
|
2767 int c2,
|
|
2768 c1;
|
|
2769 {
|
|
2770 #ifdef NUMCHAR_OPTION
|
|
2771 if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
|
|
2772 w16e_conv(c1, &c2, &c1);
|
|
2773 if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
|
|
2774 return;
|
|
2775 }
|
|
2776 }
|
|
2777 #endif
|
|
2778 if (c2 == EOF) {
|
|
2779 (*o_putc)(EOF);
|
|
2780 return;
|
|
2781 } else if (c2 == 0) {
|
|
2782 output_mode = ASCII;
|
|
2783 (*o_putc)(c1);
|
|
2784 } else if (c2 == X0201) {
|
|
2785 output_mode = JAPANESE_EUC;
|
|
2786 (*o_putc)(SSO); (*o_putc)(c1|0x80);
|
|
2787 } else if (c2 == ISO8859_1) {
|
|
2788 output_mode = ISO8859_1;
|
|
2789 (*o_putc)(c1 | 0x080);
|
|
2790 #ifdef X0212_ENABLE
|
|
2791 } else if ((c2 & 0xff00) >> 8 == 0x8f){
|
|
2792 output_mode = JAPANESE_EUC;
|
|
2793 #ifdef SHIFTJIS_CP932
|
|
2794 if (cp932_f){
|
|
2795 int s2, s1;
|
|
2796 if (e2s_conv(c2, c1, &s2, &s1) == 0){
|
|
2797 s2e_conv(s2, s1, &c2, &c1);
|
|
2798 }
|
|
2799 }
|
|
2800 #endif
|
|
2801 if ((c2 & 0xff00) >> 8 == 0x8f){
|
|
2802 if (x0212_f){
|
|
2803 (*o_putc)(0x8f);
|
|
2804 (*o_putc)((c2 & 0x7f) | 0x080);
|
|
2805 (*o_putc)(c1 | 0x080);
|
|
2806 }
|
|
2807 }else{
|
|
2808 (*o_putc)((c2 & 0x7f) | 0x080);
|
|
2809 (*o_putc)(c1 | 0x080);
|
|
2810 }
|
|
2811 #endif
|
|
2812 } else {
|
|
2813 if ((c1<0x21 || 0x7e<c1) ||
|
|
2814 (c2<0x21 || 0x7e<c2)) {
|
|
2815 set_iconv(FALSE, 0);
|
|
2816 return; /* too late to rescue this char */
|
|
2817 }
|
|
2818 output_mode = JAPANESE_EUC;
|
|
2819 (*o_putc)(c2 | 0x080);
|
|
2820 (*o_putc)(c1 | 0x080);
|
|
2821 }
|
|
2822 }
|
|
2823
|
|
2824 #ifdef X0212_ENABLE
|
|
2825 int x0212_shift(c)
|
|
2826 int c;
|
|
2827 {
|
|
2828 int ret = c;
|
|
2829 c &= 0x7f;
|
|
2830 if ((ret & 0xff00) == 0x8f00){
|
|
2831 if (0x75 <= c && c <= 0x7f){
|
|
2832 ret = c + (0x109 - 0x75);
|
|
2833 }
|
|
2834 }else{
|
|
2835 if (0x75 <= c && c <= 0x7f){
|
|
2836 ret = c + (0x113 - 0x75);
|
|
2837 }
|
|
2838 }
|
|
2839 return ret;
|
|
2840 }
|
|
2841
|
|
2842
|
|
2843 int x0212_unshift(c)
|
|
2844 int c;
|
|
2845 {
|
|
2846 int ret = c;
|
|
2847 if (0x7f <= c && c <= 0x88){
|
|
2848 ret = c + (0x75 - 0x7f);
|
|
2849 }else if (0x89 <= c && c <= 0x92){
|
|
2850 ret = (0x8f << 8) | 0x80 | (c + (0x75 - 0x89));
|
|
2851 }
|
|
2852 return ret;
|
|
2853 }
|
|
2854 #endif /* X0212_ENABLE */
|
|
2855
|
|
2856 int
|
|
2857 e2s_conv(c2, c1, p2, p1)
|
|
2858 int c2, c1, *p2, *p1;
|
|
2859 {
|
|
2860 int val = 0;
|
|
2861 unsigned short *ptr;
|
|
2862 int ndx;
|
|
2863 extern unsigned short *x0212_shiftjis[];
|
|
2864 #ifdef X0212_ENABLE
|
|
2865 if ((c2 & 0xff00) == 0x8f00){
|
|
2866 ndx = c2 & 0x7f;
|
|
2867 if (0x21 <= ndx && ndx <= 0x7e){
|
|
2868 ptr = x0212_shiftjis[ndx - 0x21];
|
|
2869 if (ptr){
|
|
2870 val = ptr[(c1 & 0x7f) - 0x21];
|
|
2871 }
|
|
2872 if (val){
|
|
2873 c2 = val >> 8;
|
|
2874 c1 = val & 0xff;
|
|
2875 if (p2) *p2 = c2;
|
|
2876 if (p1) *p1 = c1;
|
|
2877 return 0;
|
|
2878 }
|
|
2879 }
|
|
2880 c2 = x0212_shift(c2);
|
|
2881 }
|
|
2882 #endif /* X0212_ENABLE */
|
|
2883 if ((c2 & 0xff00) == 0x8f00){
|
|
2884 return 1;
|
|
2885 }
|
|
2886 if (p2) *p2 = ((c2 - 1) >> 1) + ((c2 <= 0x5e) ? 0x71 : 0xb1);
|
|
2887 if (p1) *p1 = c1 + ((c2 & 1) ? ((c1 < 0x60) ? 0x1f : 0x20) : 0x7e);
|
|
2888 return 0;
|
|
2889 }
|
|
2890
|
|
2891 void
|
|
2892 s_oconv(c2, c1)
|
|
2893 int c2,
|
|
2894 c1;
|
|
2895 {
|
|
2896 #ifdef NUMCHAR_OPTION
|
|
2897 if (c2 == 0 && (c1 & CLASS_MASK) == CLASS_UTF16){
|
|
2898 w16e_conv(c1, &c2, &c1);
|
|
2899 }
|
|
2900 #endif
|
|
2901 if (c2 == EOF) {
|
|
2902 (*o_putc)(EOF);
|
|
2903 return;
|
|
2904 } else if (c2 == 0) {
|
|
2905 output_mode = ASCII;
|
|
2906 (*o_putc)(c1);
|
|
2907 } else if (c2 == X0201) {
|
|
2908 output_mode = SHIFT_JIS;
|
|
2909 (*o_putc)(c1|0x80);
|
|
2910 } else if (c2 == ISO8859_1) {
|
|
2911 output_mode = ISO8859_1;
|
|
2912 (*o_putc)(c1 | 0x080);
|
|
2913 #ifdef X0212_ENABLE
|
|
2914 } else if ((c2 & 0xff00) >> 8 == 0x8f){
|
|
2915 output_mode = SHIFT_JIS;
|
|
2916 if (e2s_conv(c2, c1, &c2, &c1) == 0){
|
|
2917 (*o_putc)(c2);
|
|
2918 (*o_putc)(c1);
|
|
2919 }
|
|
2920 #endif
|
|
2921 } else {
|
|
2922 if ((c1<0x20 || 0x7e<c1) ||
|
|
2923 (c2<0x20 || 0x7e<c2)) {
|
|
2924 set_iconv(FALSE, 0);
|
|
2925 return; /* too late to rescue this char */
|
|
2926 }
|
|
2927 output_mode = SHIFT_JIS;
|
|
2928 e2s_conv(c2, c1, &c2, &c1);
|
|
2929
|
|
2930 #ifdef SHIFTJIS_CP932
|
|
2931 if (cp932inv_f
|
|
2932 && CP932INV_TABLE_BEGIN <= c2 && c2 <= CP932INV_TABLE_END){
|
|
2933 extern unsigned short cp932inv[2][189];
|
|
2934 int c = cp932inv[c2 - CP932INV_TABLE_BEGIN][c1 - 0x40];
|
|
2935 if (c){
|
|
2936 c2 = c >> 8;
|
|
2937 c1 = c & 0xff;
|
|
2938 }
|
|
2939 }
|
|
2940 #endif /* SHIFTJIS_CP932 */
|
|
2941
|
|
2942 (*o_putc)(c2);
|
|
2943 if (prefix_table[(unsigned char)c1]){
|
|
2944 (*o_putc)(prefix_table[(unsigned char)c1]);
|
|
2945 }
|
|
2946 (*o_putc)(c1);
|
|
2947 }
|
|
2948 }
|
|
2949
|
|
2950 void
|
|
2951 j_oconv(c2, c1)
|
|
2952 int c2,
|
|
2953 c1;
|
|
2954 {
|
|
2955 #ifdef NUMCHAR_OPTION
|
|
2956 if ((c1 & CLASS_MASK) == CLASS_UTF16){
|
|
2957 w16e_conv(c1, &c2, &c1);
|
|
2958 }
|
|
2959 #endif
|
|
2960 if (c2 == EOF) {
|
|
2961 if (output_mode !=ASCII && output_mode!=ISO8859_1) {
|
|
2962 (*o_putc)(ESC);
|
|
2963 (*o_putc)('(');
|
|
2964 (*o_putc)(ascii_intro);
|
|
2965 output_mode = ASCII;
|
|
2966 }
|
|
2967 (*o_putc)(EOF);
|
|
2968 #ifdef X0212_ENABLE
|
|
2969 } else if ((c2 & 0xff00) >> 8 == 0x8f){
|
|
2970 if (output_mode!=X0212) {
|
|
2971 output_mode = X0212;
|
|
2972 (*o_putc)(ESC);
|
|
2973 (*o_putc)('$');
|
|
2974 (*o_putc)('(');
|
|
2975 (*o_putc)('D');
|
|
2976 }
|
|
2977 (*o_putc)(c2 & 0x7f);
|
|
2978 (*o_putc)(c1);
|
|
2979 #endif
|
|
2980 } else if (c2==X0201) {
|
|
2981 if (output_mode!=X0201) {
|
|
2982 output_mode = X0201;
|
|
2983 (*o_putc)(ESC);
|
|
2984 (*o_putc)('(');
|
|
2985 (*o_putc)('I');
|
|
2986 }
|
|
2987 (*o_putc)(c1);
|
|
2988 } else if (c2==ISO8859_1) {
|
|
2989 /* iso8859 introduction, or 8th bit on */
|
|
2990 /* Can we convert in 7bit form using ESC-'-'-A ?
|
|
2991 Is this popular? */
|
|
2992 output_mode = ISO8859_1;
|
|
2993 (*o_putc)(c1|0x80);
|
|
2994 } else if (c2 == 0) {
|
|
2995 if (output_mode !=ASCII && output_mode!=ISO8859_1) {
|
|
2996 (*o_putc)(ESC);
|
|
2997 (*o_putc)('(');
|
|
2998 (*o_putc)(ascii_intro);
|
|
2999 output_mode = ASCII;
|
|
3000 }
|
|
3001 (*o_putc)(c1);
|
|
3002 } else {
|
|
3003 if (output_mode != X0208) {
|
|
3004 output_mode = X0208;
|
|
3005 (*o_putc)(ESC);
|
|
3006 (*o_putc)('$');
|
|
3007 (*o_putc)(kanji_intro);
|
|
3008 }
|
|
3009 if (c1<0x20 || 0x7e<c1)
|
|
3010 return;
|
|
3011 if (c2<0x20 || 0x7e<c2)
|
|
3012 return;
|
|
3013 (*o_putc)(c2);
|
|
3014 (*o_putc)(c1);
|
|
3015 }
|
|
3016 }
|
|
3017
|
|
3018 void
|
|
3019 base64_conv(c2, c1)
|
|
3020 int c2,
|
|
3021 c1;
|
|
3022 {
|
|
3023 mime_prechar(c2, c1);
|
|
3024 (*o_base64conv)(c2,c1);
|
|
3025 }
|
|
3026
|
|
3027
|
|
3028 static int broken_buf[3];
|
|
3029 static int broken_counter = 0;
|
|
3030 static int broken_last = 0;
|
|
3031 int
|
|
3032 broken_getc(f)
|
|
3033 FILE *f;
|
|
3034 {
|
|
3035 int c,c1;
|
|
3036
|
|
3037 if (broken_counter>0) {
|
|
3038 return broken_buf[--broken_counter];
|
|
3039 }
|
|
3040 c= (*i_bgetc)(f);
|
|
3041 if (c=='$' && broken_last != ESC
|
|
3042 && (input_mode==ASCII || input_mode==X0201)) {
|
|
3043 c1= (*i_bgetc)(f);
|
|
3044 broken_last = 0;
|
|
3045 if (c1=='@'|| c1=='B') {
|
|
3046 broken_buf[0]=c1; broken_buf[1]=c;
|
|
3047 broken_counter=2;
|
|
3048 return ESC;
|
|
3049 } else {
|
|
3050 (*i_bungetc)(c1,f);
|
|
3051 return c;
|
|
3052 }
|
|
3053 } else if (c=='(' && broken_last != ESC
|
|
3054 && (input_mode==X0208 || input_mode==X0201)) { /* ) */
|
|
3055 c1= (*i_bgetc)(f);
|
|
3056 broken_last = 0;
|
|
3057 if (c1=='J'|| c1=='B') {
|
|
3058 broken_buf[0]=c1; broken_buf[1]=c;
|
|
3059 broken_counter=2;
|
|
3060 return ESC;
|
|
3061 } else {
|
|
3062 (*i_bungetc)(c1,f);
|
|
3063 return c;
|
|
3064 }
|
|
3065 } else {
|
|
3066 broken_last = c;
|
|
3067 return c;
|
|
3068 }
|
|
3069 }
|
|
3070
|
|
3071 int
|
|
3072 broken_ungetc(c,f)
|
|
3073 int c;
|
|
3074 FILE *f;
|
|
3075 {
|
|
3076 if (broken_counter<2)
|
|
3077 broken_buf[broken_counter++]=c;
|
|
3078 return c;
|
|
3079 }
|
|
3080
|
|
3081 static int prev_cr = 0;
|
|
3082
|
|
3083 void
|
|
3084 cr_conv(c2,c1)
|
|
3085 int c2,c1;
|
|
3086 {
|
|
3087 if (prev_cr) {
|
|
3088 prev_cr = 0;
|
|
3089 if (! (c2==0&&c1==NL) ) {
|
|
3090 cr_conv(0,'\n');
|
|
3091 }
|
|
3092 }
|
|
3093 if (c2) {
|
|
3094 (*o_crconv)(c2,c1);
|
|
3095 } else if (c1=='\r') {
|
|
3096 prev_cr = c1;
|
|
3097 } else if (c1=='\n') {
|
|
3098 if (crmode_f==CRLF) {
|
|
3099 (*o_crconv)(0,'\r');
|
|
3100 } else if (crmode_f==CR) {
|
|
3101 (*o_crconv)(0,'\r');
|
|
3102 return;
|
|
3103 }
|
|
3104 (*o_crconv)(0,NL);
|
|
3105 } else if (c1!='\032' || crmode_f!=NL){
|
|
3106 (*o_crconv)(c2,c1);
|
|
3107 }
|
|
3108 }
|
|
3109
|
|
3110 /*
|
|
3111 Return value of fold_conv()
|
|
3112
|
|
3113 \n add newline and output char
|
|
3114 \r add newline and output nothing
|
|
3115 ' ' space
|
|
3116 0 skip
|
|
3117 1 (or else) normal output
|
|
3118
|
|
3119 fold state in prev (previous character)
|
|
3120
|
|
3121 >0x80 Japanese (X0208/X0201)
|
|
3122 <0x80 ASCII
|
|
3123 \n new line
|
|
3124 ' ' space
|
|
3125
|
|
3126 This fold algorthm does not preserve heading space in a line.
|
|
3127 This is the main difference from fmt.
|
|
3128 */
|
|
3129
|
|
3130 #define char_size(c2,c1) (c2?2:1)
|
|
3131
|
|
3132 void
|
|
3133 fold_conv(c2,c1)
|
|
3134 int c2,c1;
|
|
3135 {
|
|
3136 int prev0;
|
|
3137 int fold_state=0;
|
|
3138
|
|
3139 if (c1== '\r' && !fold_preserve_f) {
|
|
3140 fold_state=0; /* ignore cr */
|
|
3141 }else if (c1== '\n'&&f_prev=='\r' && fold_preserve_f) {
|
|
3142 f_prev = '\n';
|
|
3143 fold_state=0; /* ignore cr */
|
|
3144 } else if (c1== BS) {
|
|
3145 if (f_line>0) f_line--;
|
|
3146 fold_state = 1;
|
|
3147 } else if (c2==EOF && f_line != 0) { /* close open last line */
|
|
3148 fold_state = '\n';
|
|
3149 } else if ((c1=='\n' && !fold_preserve_f)
|
|
3150 || ((c1=='\r'||(c1=='\n'&&f_prev!='\r'))
|
|
3151 && fold_preserve_f)) {
|
|
3152 /* new line */
|
|
3153 if (fold_preserve_f) {
|
|
3154 f_prev = c1;
|
|
3155 f_line = 0;
|
|
3156 fold_state = '\r';
|
|
3157 } else if ((f_prev == c1 && !fold_preserve_f)
|
|
3158 || (f_prev == '\n' && fold_preserve_f)
|
|
3159 ) { /* duplicate newline */
|
|
3160 if (f_line) {
|
|
3161 f_line = 0;
|
|
3162 fold_state = '\n'; /* output two newline */
|
|
3163 } else {
|
|
3164 f_line = 0;
|
|
3165 fold_state = 1;
|
|
3166 }
|
|
3167 } else {
|
|
3168 if (f_prev&0x80) { /* Japanese? */
|
|
3169 f_prev = c1;
|
|
3170 fold_state = 0; /* ignore given single newline */
|
|
3171 } else if (f_prev==' ') {
|
|
3172 fold_state = 0;
|
|
3173 } else {
|
|
3174 f_prev = c1;
|
|
3175 if (++f_line<=fold_len)
|
|
3176 fold_state = ' ';
|
|
3177 else {
|
|
3178 f_line = 0;
|
|
3179 fold_state = '\r'; /* fold and output nothing */
|
|
3180 }
|
|
3181 }
|
|
3182 }
|
|
3183 } else if (c1=='\f') {
|
|
3184 f_prev = '\n';
|
|
3185 if (f_line==0)
|
|
3186 fold_state = 1;
|
|
3187 f_line = 0;
|
|
3188 fold_state = '\n'; /* output newline and clear */
|
|
3189 } else if ( (c2==0 && c1==' ')||
|
|
3190 (c2==0 && c1=='\t')||
|
|
3191 (c2=='!'&& c1=='!')) {
|
|
3192 /* X0208 kankaku or ascii space */
|
|
3193 if (f_prev == ' ') {
|
|
3194 fold_state = 0; /* remove duplicate spaces */
|
|
3195 } else {
|
|
3196 f_prev = ' ';
|
|
3197 if (++f_line<=fold_len)
|
|
3198 fold_state = ' '; /* output ASCII space only */
|
|
3199 else {
|
|
3200 f_prev = ' '; f_line = 0;
|
|
3201 fold_state = '\r'; /* fold and output nothing */
|
|
3202 }
|
|
3203 }
|
|
3204 } else {
|
|
3205 prev0 = f_prev; /* we still need this one... , but almost done */
|
|
3206 f_prev = c1;
|
|
3207 if (c2 || c2==X0201)
|
|
3208 f_prev |= 0x80; /* this is Japanese */
|
|
3209 f_line += char_size(c2,c1);
|
|
3210 if (f_line<=fold_len) { /* normal case */
|
|
3211 fold_state = 1;
|
|
3212 } else {
|
|
3213 if (f_line>=fold_len+fold_margin) { /* too many kinsou suspension */
|
|
3214 f_line = char_size(c2,c1);
|
|
3215 fold_state = '\n'; /* We can't wait, do fold now */
|
|
3216 } else if (c2==X0201) {
|
|
3217 /* simple kinsoku rules return 1 means no folding */
|
|
3218 if (c1==(0xde&0x7f)) fold_state = 1; /* $B!+(B*/
|
|
3219 else if (c1==(0xdf&0x7f)) fold_state = 1; /* $B!,(B*/
|
|
3220 else if (c1==(0xa4&0x7f)) fold_state = 1; /* $B!#(B*/
|
|
3221 else if (c1==(0xa3&0x7f)) fold_state = 1; /* $B!$(B*/
|
|
3222 else if (c1==(0xa1&0x7f)) fold_state = 1; /* $B!W(B*/
|
|
3223 else if (c1==(0xb0&0x7f)) fold_state = 1; /* - */
|
|
3224 else if (SPACE<=c1 && c1<=(0xdf&0x7f)) { /* X0201 */
|
|
3225 f_line = 1;
|
|
3226 fold_state = '\n';/* add one new f_line before this character */
|
|
3227 } else {
|
|
3228 f_line = 1;
|
|
3229 fold_state = '\n';/* add one new f_line before this character */
|
|
3230 }
|
|
3231 } else if (c2==0) {
|
|
3232 /* kinsoku point in ASCII */
|
|
3233 if ( c1==')'|| /* { [ ( */
|
|
3234 c1==']'||
|
|
3235 c1=='}'||
|
|
3236 c1=='.'||
|
|
3237 c1==','||
|
|
3238 c1=='!'||
|
|
3239 c1=='?'||
|
|
3240 c1=='/'||
|
|
3241 c1==':'||
|
|
3242 c1==';' ) {
|
|
3243 fold_state = 1;
|
|
3244 /* just after special */
|
|
3245 } else if (!is_alnum(prev0)) {
|
|
3246 f_line = char_size(c2,c1);
|
|
3247 fold_state = '\n';
|
|
3248 } else if ((prev0==' ') || /* ignored new f_line */
|
|
3249 (prev0=='\n')|| /* ignored new f_line */
|
|
3250 (prev0&0x80)) { /* X0208 - ASCII */
|
|
3251 f_line = char_size(c2,c1);
|
|
3252 fold_state = '\n';/* add one new f_line before this character */
|
|
3253 } else {
|
|
3254 fold_state = 1; /* default no fold in ASCII */
|
|
3255 }
|
|
3256 } else {
|
|
3257 if (c2=='!') {
|
|
3258 if (c1=='"') fold_state = 1; /* $B!"(B */
|
|
3259 else if (c1=='#') fold_state = 1; /* $B!#(B */
|
|
3260 else if (c1=='W') fold_state = 1; /* $B!W(B */
|
|
3261 else if (c1=='K') fold_state = 1; /* $B!K(B */
|
|
3262 else if (c1=='$') fold_state = 1; /* $B!$(B */
|
|
3263 else if (c1=='%') fold_state = 1; /* $B!%(B */
|
|
3264 else if (c1=='\'') fold_state = 1; /* $B!\(B */
|
|
3265 else if (c1=='(') fold_state = 1; /* $B!((B */
|
|
3266 else if (c1==')') fold_state = 1; /* $B!)(B */
|
|
3267 else if (c1=='*') fold_state = 1; /* $B!*(B */
|
|
3268 else if (c1=='+') fold_state = 1; /* $B!+(B */
|
|
3269 else if (c1==',') fold_state = 1; /* $B!,(B */
|
|
3270 /* default no fold in kinsoku */
|
|
3271 else {
|
|
3272 fold_state = '\n';
|
|
3273 f_line = char_size(c2,c1);
|
|
3274 /* add one new f_line before this character */
|
|
3275 }
|
|
3276 } else {
|
|
3277 f_line = char_size(c2,c1);
|
|
3278 fold_state = '\n';
|
|
3279 /* add one new f_line before this character */
|
|
3280 }
|
|
3281 }
|
|
3282 }
|
|
3283 }
|
|
3284 /* terminator process */
|
|
3285 switch(fold_state) {
|
|
3286 case '\n':
|
|
3287 (*o_fconv)(0,'\n');
|
|
3288 (*o_fconv)(c2,c1);
|
|
3289 break;
|
|
3290 case 0:
|
|
3291 return;
|
|
3292 case '\r':
|
|
3293 (*o_fconv)(0,'\n');
|
|
3294 break;
|
|
3295 case '\t':
|
|
3296 case ' ':
|
|
3297 (*o_fconv)(0,' ');
|
|
3298 break;
|
|
3299 default:
|
|
3300 (*o_fconv)(c2,c1);
|
|
3301 }
|
|
3302 }
|
|
3303
|
|
3304 int z_prev2=0,z_prev1=0;
|
|
3305
|
|
3306 void
|
|
3307 z_conv(c2,c1)
|
|
3308 int c2,c1;
|
|
3309 {
|
|
3310
|
|
3311 /* if (c2) c1 &= 0x7f; assertion */
|
|
3312
|
|
3313 if (x0201_f && z_prev2==X0201) { /* X0201 */
|
|
3314 if (c1==(0xde&0x7f)) { /* $BByE@(B */
|
|
3315 z_prev2=0;
|
|
3316 (*o_zconv)(dv[(z_prev1-SPACE)*2],dv[(z_prev1-SPACE)*2+1]);
|
|
3317 return;
|
|
3318 } else if (c1==(0xdf&0x7f)&&ev[(z_prev1-SPACE)*2]) { /* $BH>ByE@(B */
|
|
3319 z_prev2=0;
|
|
3320 (*o_zconv)(ev[(z_prev1-SPACE)*2],ev[(z_prev1-SPACE)*2+1]);
|
|
3321 return;
|
|
3322 } else {
|
|
3323 z_prev2=0;
|
|
3324 (*o_zconv)(cv[(z_prev1-SPACE)*2],cv[(z_prev1-SPACE)*2+1]);
|
|
3325 }
|
|
3326 }
|
|
3327
|
|
3328 if (c2==EOF) {
|
|
3329 (*o_zconv)(c2,c1);
|
|
3330 return;
|
|
3331 }
|
|
3332
|
|
3333 if (x0201_f && c2==X0201) {
|
|
3334 if (dv[(c1-SPACE)*2]||ev[(c1-SPACE)*2]) {
|
|
3335 /* wait for $BByE@(B or $BH>ByE@(B */
|
|
3336 z_prev1 = c1; z_prev2 = c2;
|
|
3337 return;
|
|
3338 } else {
|
|
3339 (*o_zconv)(cv[(c1-SPACE)*2],cv[(c1-SPACE)*2+1]);
|
|
3340 return;
|
|
3341 }
|
|
3342 }
|
|
3343
|
|
3344 /* JISX0208 Alphabet */
|
|
3345 if (alpha_f && c2 == 0x23 ) {
|
|
3346 c2 = 0;
|
|
3347 } else if (alpha_f && c2 == 0x21 ) {
|
|
3348 /* JISX0208 Kigou */
|
|
3349 if (0x21==c1) {
|
|
3350 if (alpha_f&0x2) {
|
|
3351 c1 = ' ';
|
|
3352 c2 = 0;
|
|
3353 } else if (alpha_f&0x4) {
|
|
3354 (*o_zconv)(0,' ');
|
|
3355 (*o_zconv)(0,' ');
|
|
3356 return;
|
|
3357 }
|
|
3358 } else if (0x20<c1 && c1<0x7f && fv[c1-0x20]) {
|
|
3359 c1 = fv[c1-0x20];
|
|
3360 c2 = 0;
|
|
3361 if (alpha_f&0x8) {
|
|
3362 char *entity = 0;
|
|
3363 switch (c1){
|
|
3364 case '>': entity = ">"; break;
|
|
3365 case '<': entity = "<"; break;
|
|
3366 case '\"': entity = """; break;
|
|
3367 case '&': entity = "&"; break;
|
|
3368 }
|
|
3369 if (entity){
|
|
3370 while (*entity) (*o_zconv)(0, *entity++);
|
|
3371 return;
|
|
3372 }
|
|
3373 }
|
|
3374 }
|
|
3375 }
|
|
3376 (*o_zconv)(c2,c1);
|
|
3377 }
|
|
3378
|
|
3379
|
|
3380 #define rot13(c) ( \
|
|
3381 ( c < 'A' ) ? c: \
|
|
3382 (c <= 'M') ? (c + 13): \
|
|
3383 (c <= 'Z') ? (c - 13): \
|
|
3384 (c < 'a') ? (c): \
|
|
3385 (c <= 'm') ? (c + 13): \
|
|
3386 (c <= 'z') ? (c - 13): \
|
|
3387 (c) \
|
|
3388 )
|
|
3389
|
|
3390 #define rot47(c) ( \
|
|
3391 ( c < '!' ) ? c: \
|
|
3392 ( c <= 'O' ) ? (c + 47) : \
|
|
3393 ( c <= '~' ) ? (c - 47) : \
|
|
3394 c \
|
|
3395 )
|
|
3396
|
|
3397 void
|
|
3398 rot_conv(c2,c1)
|
|
3399 int c2,c1;
|
|
3400 {
|
|
3401 if (c2==0 || c2==X0201 || c2==ISO8859_1) {
|
|
3402 c1 = rot13(c1);
|
|
3403 } else if (c2) {
|
|
3404 c1 = rot47(c1);
|
|
3405 c2 = rot47(c2);
|
|
3406 }
|
|
3407 (*o_rot_conv)(c2,c1);
|
|
3408 }
|
|
3409
|
|
3410 void
|
|
3411 hira_conv(c2,c1)
|
|
3412 int c2,c1;
|
|
3413 {
|
|
3414 if ((hira_f & 1) && c2==0x25 && 0x20<c1 && c1<0x74) {
|
|
3415 c2 = 0x24;
|
|
3416 } else if ((hira_f & 2) && c2==0x24 && 0x20<c1 && c1<0x74) {
|
|
3417 c2 = 0x25;
|
|
3418 }
|
|
3419 (*o_hira_conv)(c2,c1);
|
|
3420 }
|
|
3421
|
|
3422
|
|
3423 void
|
|
3424 iso2022jp_check_conv(c2,c1)
|
|
3425 int c2, c1;
|
|
3426 {
|
|
3427 static int range[RANGE_NUM_MAX][2] = {
|
|
3428 {0x222f, 0x2239,},
|
|
3429 {0x2242, 0x2249,},
|
|
3430 {0x2251, 0x225b,},
|
|
3431 {0x226b, 0x2271,},
|
|
3432 {0x227a, 0x227d,},
|
|
3433 {0x2321, 0x232f,},
|
|
3434 {0x233a, 0x2340,},
|
|
3435 {0x235b, 0x2360,},
|
|
3436 {0x237b, 0x237e,},
|
|
3437 {0x2474, 0x247e,},
|
|
3438 {0x2577, 0x257e,},
|
|
3439 {0x2639, 0x2640,},
|
|
3440 {0x2659, 0x267e,},
|
|
3441 {0x2742, 0x2750,},
|
|
3442 {0x2772, 0x277e,},
|
|
3443 {0x2841, 0x287e,},
|
|
3444 {0x4f54, 0x4f7e,},
|
|
3445 {0x7425, 0x747e},
|
|
3446 };
|
|
3447 int i;
|
|
3448 int start, end, c;
|
|
3449
|
|
3450 if(c2 >= 0x00 && c2 <= 0x20 && c1 >= 0x7f && c1 <= 0xff) {
|
|
3451 c2 = GETA1;
|
|
3452 c1 = GETA2;
|
|
3453 }
|
|
3454 if((c2 >= 0x29 && c2 <= 0x2f) || (c2 >= 0x75 && c2 <= 0x7e)) {
|
|
3455 c2 = GETA1;
|
|
3456 c1 = GETA2;
|
|
3457 }
|
|
3458
|
|
3459 for (i = 0; i < RANGE_NUM_MAX; i++) {
|
|
3460 start = range[i][0];
|
|
3461 end = range[i][1];
|
|
3462 c = (c2 << 8) + c1;
|
|
3463 if (c >= start && c <= end) {
|
|
3464 c2 = GETA1;
|
|
3465 c1 = GETA2;
|
|
3466 }
|
|
3467 }
|
|
3468 (*o_iso2022jp_check_conv)(c2,c1);
|
|
3469 }
|
|
3470
|
|
3471
|
|
3472 /* This converts =?ISO-2022-JP?B?HOGE HOGE?= */
|
|
3473
|
|
3474 unsigned char *mime_pattern[] = {
|
|
3475 (unsigned char *)"\075?EUC-JP?B?",
|
|
3476 (unsigned char *)"\075?SHIFT_JIS?B?",
|
|
3477 (unsigned char *)"\075?ISO-8859-1?Q?",
|
|
3478 (unsigned char *)"\075?ISO-8859-1?B?",
|
|
3479 (unsigned char *)"\075?ISO-2022-JP?B?",
|
|
3480 (unsigned char *)"\075?ISO-2022-JP?Q?",
|
|
3481 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
|
|
3482 (unsigned char *)"\075?UTF-8?B?",
|
|
3483 (unsigned char *)"\075?UTF-8?Q?",
|
|
3484 #endif
|
|
3485 (unsigned char *)"\075?US-ASCII?Q?",
|
|
3486 NULL
|
|
3487 };
|
|
3488
|
|
3489
|
|
3490 /* $B3:Ev$9$k%3!<%I$NM%@hEY$r>e$2$k$?$a$NL\0u(B */
|
|
3491 int (*mime_priority_func[])PROTO((int c2, int c1, int c0)) = {
|
|
3492 e_iconv, s_iconv, 0, 0, 0, 0,
|
|
3493 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
|
|
3494 w_iconv, w_iconv,
|
|
3495 #endif
|
|
3496 0,
|
|
3497 };
|
|
3498
|
|
3499 int mime_encode[] = {
|
|
3500 JAPANESE_EUC, SHIFT_JIS,ISO8859_1, ISO8859_1, X0208, X0201,
|
|
3501 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
|
|
3502 UTF8, UTF8,
|
|
3503 #endif
|
|
3504 ASCII,
|
|
3505 0
|
|
3506 };
|
|
3507
|
|
3508 int mime_encode_method[] = {
|
|
3509 'B', 'B','Q', 'B', 'B', 'Q',
|
|
3510 #if defined(UTF8_INPUT_ENABLE) || defined(UTF8_OUTPUT_ENABLE)
|
|
3511 'B', 'Q',
|
|
3512 #endif
|
|
3513 'Q',
|
|
3514 0
|
|
3515 };
|
|
3516
|
|
3517
|
|
3518 #define MAXRECOVER 20
|
|
3519
|
|
3520 /* I don't trust portablity of toupper */
|
|
3521 #define nkf_toupper(c) (('a'<=c && c<='z')?(c-('a'-'A')):c)
|
|
3522 #define nkf_isdigit(c) ('0'<=c && c<='9')
|
|
3523 #define nkf_isxdigit(c) (nkf_isdigit(c) || ('a'<=c && c<='f') || ('A'<=c && c <= 'F'))
|
|
3524 #define nkf_isblank(c) (c == SPACE || c == TAB)
|
|
3525 #define nkf_isspace(c) (nkf_isblank(c) || c == CR || c == NL)
|
|
3526 #define nkf_isalpha(c) (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z'))
|
|
3527 #define nkf_isalnum(c) (nkf_isdigit(c) || nkf_isalpha(c))
|
|
3528
|
|
3529 void
|
|
3530 switch_mime_getc()
|
|
3531 {
|
|
3532 if (i_getc!=mime_getc) {
|
|
3533 i_mgetc = i_getc; i_getc = mime_getc;
|
|
3534 i_mungetc = i_ungetc; i_ungetc = mime_ungetc;
|
|
3535 if(mime_f==STRICT_MIME) {
|
|
3536 i_mgetc_buf = i_mgetc; i_mgetc = mime_getc_buf;
|
|
3537 i_mungetc_buf = i_mungetc; i_mungetc = mime_ungetc_buf;
|
|
3538 }
|
|
3539 }
|
|
3540 }
|
|
3541
|
|
3542 void
|
|
3543 unswitch_mime_getc()
|
|
3544 {
|
|
3545 if(mime_f==STRICT_MIME) {
|
|
3546 i_mgetc = i_mgetc_buf;
|
|
3547 i_mungetc = i_mungetc_buf;
|
|
3548 }
|
|
3549 i_getc = i_mgetc;
|
|
3550 i_ungetc = i_mungetc;
|
|
3551 }
|
|
3552
|
|
3553 int
|
|
3554 mime_begin_strict(f)
|
|
3555 FILE *f;
|
|
3556 {
|
|
3557 int c1 = 0;
|
|
3558 int i,j,k;
|
|
3559 unsigned char *p,*q;
|
|
3560 int r[MAXRECOVER]; /* recovery buffer, max mime pattern lenght */
|
|
3561
|
|
3562 mime_decode_mode = FALSE;
|
|
3563 /* =? has been checked */
|
|
3564 j = 0;
|
|
3565 p = mime_pattern[j];
|
|
3566 r[0]='='; r[1]='?';
|
|
3567
|
|
3568 for(i=2;p[i]>' ';i++) { /* start at =? */
|
|
3569 if ( ((r[i] = c1 = (*i_getc)(f))==EOF) || nkf_toupper(c1) != p[i] ) {
|
|
3570 /* pattern fails, try next one */
|
|
3571 q = p;
|
|
3572 while ((p = mime_pattern[++j])) {
|
|
3573 for(k=2;k<i;k++) /* assume length(p) > i */
|
|
3574 if (p[k]!=q[k]) break;
|
|
3575 if (k==i && nkf_toupper(c1)==p[k]) break;
|
|
3576 }
|
|
3577 if (p) continue; /* found next one, continue */
|
|
3578 /* all fails, output from recovery buffer */
|
|
3579 (*i_ungetc)(c1,f);
|
|
3580 for(j=0;j<i;j++) {
|
|
3581 (*oconv)(0,r[j]);
|
|
3582 }
|
|
3583 return c1;
|
|
3584 }
|
|
3585 }
|
|
3586 mime_decode_mode = p[i-2];
|
|
3587
|
|
3588 set_iconv(FALSE, mime_priority_func[j]);
|
|
3589 clr_code_score(find_inputcode_byfunc(mime_priority_func[j]), SCORE_iMIME);
|
|
3590
|
|
3591 if (mime_decode_mode=='B') {
|
|
3592 mimebuf_f = unbuf_f;
|
|
3593 if (!unbuf_f) {
|
|
3594 /* do MIME integrity check */
|
|
3595 return mime_integrity(f,mime_pattern[j]);
|
|
3596 }
|
|
3597 }
|
|
3598 switch_mime_getc();
|
|
3599 mimebuf_f = TRUE;
|
|
3600 return c1;
|
|
3601 }
|
|
3602
|
|
3603 int
|
|
3604 mime_getc_buf(f)
|
|
3605 FILE *f;
|
|
3606 {
|
|
3607 /* we don't keep eof of Fifo, becase it contains ?= as
|
|
3608 a terminator. It was checked in mime_integrity. */
|
|
3609 return ((mimebuf_f)?
|
|
3610 (*i_mgetc_buf)(f):Fifo(mime_input++));
|
|
3611 }
|
|
3612
|
|
3613 int
|
|
3614 mime_ungetc_buf(c,f)
|
|
3615 FILE *f;
|
|
3616 int c;
|
|
3617 {
|
|
3618 if (mimebuf_f)
|
|
3619 (*i_mungetc_buf)(c,f);
|
|
3620 else
|
|
3621 Fifo(--mime_input)=c;
|
|
3622 return c;
|
|
3623 }
|
|
3624
|
|
3625 int
|
|
3626 mime_begin(f)
|
|
3627 FILE *f;
|
|
3628 {
|
|
3629 int c1;
|
|
3630 int i,k;
|
|
3631
|
|
3632 /* In NONSTRICT mode, only =? is checked. In case of failure, we */
|
|
3633 /* re-read and convert again from mime_buffer. */
|
|
3634
|
|
3635 /* =? has been checked */
|
|
3636 k = mime_last;
|
|
3637 Fifo(mime_last++)='='; Fifo(mime_last++)='?';
|
|
3638 for(i=2;i<MAXRECOVER;i++) { /* start at =? */
|
|
3639 /* We accept any character type even if it is breaked by new lines */
|
|
3640 c1 = (*i_getc)(f); Fifo(mime_last++)= c1 ;
|
|
3641 if (c1=='\n'||c1==' '||c1=='\r'||
|
|
3642 c1=='-'||c1=='_'||is_alnum(c1) ) continue;
|
|
3643 if (c1=='=') {
|
|
3644 /* Failed. But this could be another MIME preemble */
|
|
3645 (*i_ungetc)(c1,f);
|
|
3646 mime_last--;
|
|
3647 break;
|
|
3648 }
|
|
3649 if (c1!='?') break;
|
|
3650 else {
|
|
3651 /* c1=='?' */
|
|
3652 c1 = (*i_getc)(f); Fifo(mime_last++) = c1;
|
|
3653 if (!(++i<MAXRECOVER) || c1==EOF) break;
|
|
3654 if (c1=='b'||c1=='B') {
|
|
3655 mime_decode_mode = 'B';
|
|
3656 } else if (c1=='q'||c1=='Q') {
|
|
3657 mime_decode_mode = 'Q';
|
|
3658 } else {
|
|
3659 break;
|
|
3660 }
|
|
3661 c1 = (*i_getc)(f); Fifo(mime_last++) = c1;
|
|
3662 if (!(++i<MAXRECOVER) || c1==EOF) break;
|
|
3663 if (c1!='?') {
|
|
3664 mime_decode_mode = FALSE;
|
|
3665 }
|
|
3666 break;
|
|
3667 }
|
|
3668 }
|
|
3669 switch_mime_getc();
|
|
3670 if (!mime_decode_mode) {
|
|
3671 /* false MIME premble, restart from mime_buffer */
|
|
3672 mime_decode_mode = 1; /* no decode, but read from the mime_buffer */
|
|
3673 /* Since we are in MIME mode until buffer becomes empty, */
|
|
3674 /* we never go into mime_begin again for a while. */
|
|
3675 return c1;
|
|
3676 }
|
|
3677 /* discard mime preemble, and goto MIME mode */
|
|
3678 mime_last = k;
|
|
3679 /* do no MIME integrity check */
|
|
3680 return c1; /* used only for checking EOF */
|
|
3681 }
|
|
3682
|
|
3683 #ifdef CHECK_OPTION
|
|
3684 void
|
|
3685 no_putc(c)
|
|
3686 int c;
|
|
3687 {
|
|
3688 ;
|
|
3689 }
|
|
3690
|
|
3691 void debug(str)
|
|
3692 char *str;
|
|
3693 {
|
|
3694 if (debug_f){
|
|
3695 fprintf(stderr, "%s\n", str);
|
|
3696 }
|
|
3697 }
|
|
3698 #endif
|
|
3699
|
|
3700 void
|
|
3701 set_input_codename (codename)
|
|
3702 char *codename;
|
|
3703 {
|
|
3704 if (guess_f &&
|
|
3705 is_inputcode_set &&
|
|
3706 strcmp(codename, "") != 0 &&
|
|
3707 strcmp(codename, input_codename) != 0)
|
|
3708 {
|
|
3709 is_inputcode_mixed = TRUE;
|
|
3710 }
|
|
3711 input_codename = codename;
|
|
3712 is_inputcode_set = TRUE;
|
|
3713 }
|
|
3714
|
|
3715 #ifndef WIN32DLL
|
|
3716 void
|
|
3717 print_guessed_code (filename)
|
|
3718 char *filename;
|
|
3719 {
|
|
3720 char *codename = "BINARY";
|
|
3721 if (!is_inputcode_mixed) {
|
|
3722 if (strcmp(input_codename, "") == 0) {
|
|
3723 codename = "ASCII";
|
|
3724 } else {
|
|
3725 codename = input_codename;
|
|
3726 }
|
|
3727 }
|
|
3728 if (filename != NULL) printf("%s:", filename);
|
|
3729 printf("%s\n", codename);
|
|
3730 }
|
|
3731 #endif /*WIN32DLL*/
|
|
3732
|
|
3733 int
|
|
3734 hex2bin(x)
|
|
3735 int x;
|
|
3736 {
|
|
3737 if (nkf_isdigit(x)) return x - '0';
|
|
3738 return nkf_toupper(x) - 'A' + 10;
|
|
3739 }
|
|
3740
|
|
3741 #ifdef INPUT_OPTION
|
|
3742
|
|
3743 #ifdef ANSI_C_PROTOTYPE
|
|
3744 int hex_getc(int ch, FILE *f, int (*g)(FILE *f), int (*u)(int c, FILE *f))
|
|
3745 #else
|
|
3746 int
|
|
3747 hex_getc(ch, f, g, u)
|
|
3748 int ch;
|
|
3749 FILE *f;
|
|
3750 int (*g)();
|
|
3751 int (*u)();
|
|
3752 #endif
|
|
3753 {
|
|
3754 int c1, c2, c3;
|
|
3755 c1 = (*g)(f);
|
|
3756 if (c1 != ch){
|
|
3757 return c1;
|
|
3758 }
|
|
3759 c2 = (*g)(f);
|
|
3760 if (!nkf_isxdigit(c2)){
|
|
3761 (*u)(c2, f);
|
|
3762 return c1;
|
|
3763 }
|
|
3764 c3 = (*g)(f);
|
|
3765 if (!nkf_isxdigit(c3)){
|
|
3766 (*u)(c2, f);
|
|
3767 (*u)(c3, f);
|
|
3768 return c1;
|
|
3769 }
|
|
3770 return (hex2bin(c2) << 4) | hex2bin(c3);
|
|
3771 }
|
|
3772
|
|
3773 int
|
|
3774 cap_getc(f)
|
|
3775 FILE *f;
|
|
3776 {
|
|
3777 return hex_getc(':', f, i_cgetc, i_cungetc);
|
|
3778 }
|
|
3779
|
|
3780 int
|
|
3781 cap_ungetc(c, f)
|
|
3782 int c;
|
|
3783 FILE *f;
|
|
3784 {
|
|
3785 return (*i_cungetc)(c, f);
|
|
3786 }
|
|
3787
|
|
3788 int
|
|
3789 url_getc(f)
|
|
3790 FILE *f;
|
|
3791 {
|
|
3792 return hex_getc('%', f, i_ugetc, i_uungetc);
|
|
3793 }
|
|
3794
|
|
3795 int
|
|
3796 url_ungetc(c, f)
|
|
3797 int c;
|
|
3798 FILE *f;
|
|
3799 {
|
|
3800 return (*i_uungetc)(c, f);
|
|
3801 }
|
|
3802 #endif
|
|
3803
|
|
3804 #ifdef NUMCHAR_OPTION
|
|
3805 int
|
|
3806 numchar_getc(f)
|
|
3807 FILE *f;
|
|
3808 {
|
|
3809 int (*g)() = i_ngetc;
|
|
3810 int (*u)() = i_nungetc;
|
|
3811 int i = 0, j;
|
|
3812 int buf[8];
|
|
3813 long c = -1;
|
|
3814
|
|
3815 buf[i] = (*g)(f);
|
|
3816 if (buf[i] == '&'){
|
|
3817 buf[++i] = (*g)(f);
|
|
3818 if (buf[i] == '#'){
|
|
3819 c = 0;
|
|
3820 buf[++i] = (*g)(f);
|
|
3821 if (buf[i] == 'x' || buf[i] == 'X'){
|
|
3822 for (j = 0; j < 5; j++){
|
|
3823 buf[++i] = (*g)(f);
|
|
3824 if (!nkf_isxdigit(buf[i])){
|
|
3825 if (buf[i] != ';'){
|
|
3826 c = -1;
|
|
3827 }
|
|
3828 break;
|
|
3829 }
|
|
3830 c <<= 4;
|
|
3831 c |= hex2bin(buf[i]);
|
|
3832 }
|
|
3833 }else{
|
|
3834 for (j = 0; j < 6; j++){
|
|
3835 if (j){
|
|
3836 buf[++i] = (*g)(f);
|
|
3837 }
|
|
3838 if (!nkf_isdigit(buf[i])){
|
|
3839 if (buf[i] != ';'){
|
|
3840 c = -1;
|
|
3841 }
|
|
3842 break;
|
|
3843 }
|
|
3844 c *= 10;
|
|
3845 c += hex2bin(buf[i]);
|
|
3846 }
|
|
3847 }
|
|
3848 }
|
|
3849 }
|
|
3850 if (c != -1){
|
|
3851 return CLASS_UTF16 | c;
|
|
3852 }
|
|
3853 while (i > 0){
|
|
3854 (*u)(buf[i], f);
|
|
3855 --i;
|
|
3856 }
|
|
3857 return buf[0];
|
|
3858 }
|
|
3859
|
|
3860 int
|
|
3861 numchar_ungetc(c, f)
|
|
3862 int c;
|
|
3863 FILE *f;
|
|
3864 {
|
|
3865 return (*i_nungetc)(c, f);
|
|
3866 }
|
|
3867 #endif
|
|
3868
|
|
3869
|
|
3870 int
|
|
3871 mime_getc(f)
|
|
3872 FILE *f;
|
|
3873 {
|
|
3874 int c1, c2, c3, c4, cc;
|
|
3875 int t1, t2, t3, t4, mode, exit_mode;
|
|
3876 int lwsp_count;
|
|
3877 char *lwsp_buf;
|
|
3878 char *lwsp_buf_new;
|
|
3879 int lwsp_size = 128;
|
|
3880
|
|
3881 if (mime_top != mime_last) { /* Something is in FIFO */
|
|
3882 return Fifo(mime_top++);
|
|
3883 }
|
|
3884 if (mime_decode_mode==1 ||mime_decode_mode==FALSE) {
|
|
3885 mime_decode_mode=FALSE;
|
|
3886 unswitch_mime_getc();
|
|
3887 return (*i_getc)(f);
|
|
3888 }
|
|
3889
|
|
3890 if (mimebuf_f == FIXED_MIME)
|
|
3891 exit_mode = mime_decode_mode;
|
|
3892 else
|
|
3893 exit_mode = FALSE;
|
|
3894 if (mime_decode_mode == 'Q') {
|
|
3895 if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF);
|
|
3896 restart_mime_q:
|
|
3897 if (c1=='_') return ' ';
|
|
3898 if (c1!='=' && c1!='?') {
|
|
3899 return c1;
|
|
3900 }
|
|
3901
|
|
3902 mime_decode_mode = exit_mode; /* prepare for quit */
|
|
3903 if (c1<=' ') return c1;
|
|
3904 if ((c2 = (*i_mgetc)(f)) == EOF) return (EOF);
|
|
3905 if (c1=='?'&&c2=='=' && mimebuf_f != FIXED_MIME) {
|
|
3906 /* end Q encoding */
|
|
3907 input_mode = exit_mode;
|
|
3908 lwsp_count = 0;
|
|
3909 lwsp_buf = malloc((lwsp_size+5)*sizeof(char));
|
|
3910 if (lwsp_buf==NULL) {
|
|
3911 perror("can't malloc");
|
|
3912 return -1;
|
|
3913 }
|
|
3914 while ((c1=(*i_getc)(f))!=EOF) {
|
|
3915 switch (c1) {
|
|
3916 case NL:
|
|
3917 case CR:
|
|
3918 if (c1==NL) {
|
|
3919 if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) {
|
|
3920 i_ungetc(SPACE,f);
|
|
3921 continue;
|
|
3922 } else {
|
|
3923 i_ungetc(c1,f);
|
|
3924 }
|
|
3925 c1 = NL;
|
|
3926 } else {
|
|
3927 if ((c1=(*i_getc)(f))!=EOF && c1 == NL) {
|
|
3928 if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) {
|
|
3929 i_ungetc(SPACE,f);
|
|
3930 continue;
|
|
3931 } else {
|
|
3932 i_ungetc(c1,f);
|
|
3933 }
|
|
3934 i_ungetc(NL,f);
|
|
3935 } else {
|
|
3936 i_ungetc(c1,f);
|
|
3937 }
|
|
3938 c1 = CR;
|
|
3939 }
|
|
3940 break;
|
|
3941 case SPACE:
|
|
3942 case TAB:
|
|
3943 lwsp_buf[lwsp_count] = c1;
|
|
3944 if (lwsp_count++>lwsp_size){
|
|
3945 lwsp_size *= 2;
|
|
3946 lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
|
|
3947 if (lwsp_buf_new==NULL) {
|
|
3948 free(lwsp_buf);
|
|
3949 lwsp_buf = NULL;
|
|
3950 perror("can't realloc");
|
|
3951 return -1;
|
|
3952 }
|
|
3953 lwsp_buf = lwsp_buf_new;
|
|
3954 }
|
|
3955 continue;
|
|
3956 }
|
|
3957 break;
|
|
3958 }
|
|
3959 if (lwsp_count > 0) {
|
|
3960 if (c1=='=' && (lwsp_buf[lwsp_count-1]==SPACE||lwsp_buf[lwsp_count-1]==TAB)) {
|
|
3961 lwsp_count = 0;
|
|
3962 } else {
|
|
3963 i_ungetc(c1,f);
|
|
3964 for(lwsp_count--;lwsp_count>0;lwsp_count--)
|
|
3965 i_ungetc(lwsp_buf[lwsp_count],f);
|
|
3966 c1 = lwsp_buf[0];
|
|
3967 }
|
|
3968 }
|
|
3969 free(lwsp_buf);
|
|
3970 lwsp_buf = NULL;
|
|
3971 return c1;
|
|
3972 }
|
|
3973 if (c1=='='&&c2<' ') { /* this is soft wrap */
|
|
3974 while((c1 = (*i_mgetc)(f)) <=' ') {
|
|
3975 if ((c1 = (*i_mgetc)(f)) == EOF) return (EOF);
|
|
3976 }
|
|
3977 mime_decode_mode = 'Q'; /* still in MIME */
|
|
3978 goto restart_mime_q;
|
|
3979 }
|
|
3980 if (c1=='?') {
|
|
3981 mime_decode_mode = 'Q'; /* still in MIME */
|
|
3982 (*i_mungetc)(c2,f);
|
|
3983 return c1;
|
|
3984 }
|
|
3985 if ((c3 = (*i_mgetc)(f)) == EOF) return (EOF);
|
|
3986 if (c2<=' ') return c2;
|
|
3987 mime_decode_mode = 'Q'; /* still in MIME */
|
|
3988 #define hex(c) (('0'<=c&&c<='9')?(c-'0'):\
|
|
3989 ('A'<=c&&c<='F')?(c-'A'+10):('a'<=c&&c<='f')?(c-'a'+10):0)
|
|
3990 return ((hex(c2)<<4) + hex(c3));
|
|
3991 }
|
|
3992
|
|
3993 if (mime_decode_mode != 'B') {
|
|
3994 mime_decode_mode = FALSE;
|
|
3995 return (*i_mgetc)(f);
|
|
3996 }
|
|
3997
|
|
3998
|
|
3999 /* Base64 encoding */
|
|
4000 /*
|
|
4001 MIME allows line break in the middle of
|
|
4002 Base64, but we are very pessimistic in decoding
|
|
4003 in unbuf mode because MIME encoded code may broken by
|
|
4004 less or editor's control sequence (such as ESC-[-K in unbuffered
|
|
4005 mode. ignore incomplete MIME.
|
|
4006 */
|
|
4007 mode = mime_decode_mode;
|
|
4008 mime_decode_mode = exit_mode; /* prepare for quit */
|
|
4009
|
|
4010 while ((c1 = (*i_mgetc)(f))<=' ') {
|
|
4011 if (c1==EOF)
|
|
4012 return (EOF);
|
|
4013 }
|
|
4014 mime_c2_retry:
|
|
4015 if ((c2 = (*i_mgetc)(f))<=' ') {
|
|
4016 if (c2==EOF)
|
|
4017 return (EOF);
|
|
4018 if (mime_f != STRICT_MIME) goto mime_c2_retry;
|
|
4019 if (mimebuf_f!=FIXED_MIME) input_mode = ASCII;
|
|
4020 return c2;
|
|
4021 }
|
|
4022 if ((c1 == '?') && (c2 == '=')) {
|
|
4023 input_mode = ASCII;
|
|
4024 lwsp_count = 0;
|
|
4025 lwsp_buf = malloc((lwsp_size+5)*sizeof(char));
|
|
4026 if (lwsp_buf==NULL) {
|
|
4027 perror("can't malloc");
|
|
4028 return -1;
|
|
4029 }
|
|
4030 while ((c1=(*i_getc)(f))!=EOF) {
|
|
4031 switch (c1) {
|
|
4032 case NL:
|
|
4033 case CR:
|
|
4034 if (c1==NL) {
|
|
4035 if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) {
|
|
4036 i_ungetc(SPACE,f);
|
|
4037 continue;
|
|
4038 } else {
|
|
4039 i_ungetc(c1,f);
|
|
4040 }
|
|
4041 c1 = NL;
|
|
4042 } else {
|
|
4043 if ((c1=(*i_getc)(f))!=EOF) {
|
|
4044 if (c1==SPACE) {
|
|
4045 i_ungetc(SPACE,f);
|
|
4046 continue;
|
|
4047 } else if ((c1=(*i_getc)(f))!=EOF && (c1==SPACE||c1==TAB)) {
|
|
4048 i_ungetc(SPACE,f);
|
|
4049 continue;
|
|
4050 } else {
|
|
4051 i_ungetc(c1,f);
|
|
4052 }
|
|
4053 i_ungetc(NL,f);
|
|
4054 } else {
|
|
4055 i_ungetc(c1,f);
|
|
4056 }
|
|
4057 c1 = CR;
|
|
4058 }
|
|
4059 break;
|
|
4060 case SPACE:
|
|
4061 case TAB:
|
|
4062 lwsp_buf[lwsp_count] = c1;
|
|
4063 if (lwsp_count++>lwsp_size){
|
|
4064 lwsp_size *= 2;
|
|
4065 lwsp_buf_new = realloc(lwsp_buf, (lwsp_size+5)*sizeof(char));
|
|
4066 if (lwsp_buf_new==NULL) {
|
|
4067 free(lwsp_buf);
|
|
4068 lwsp_buf = NULL;
|
|
4069 perror("can't realloc");
|
|
4070 return -1;
|
|
4071 }
|
|
4072 lwsp_buf = lwsp_buf_new;
|
|
4073 }
|
|
4074 continue;
|
|
4075 }
|
|
4076 break;
|
|
4077 }
|
|
4078 if (lwsp_count > 0) {
|
|
4079 if (c1=='=' && (lwsp_buf[lwsp_count-1]==SPACE||lwsp_buf[lwsp_count-1]==TAB)) {
|
|
4080 lwsp_count = 0;
|
|
4081 } else {
|
|
4082 i_ungetc(c1,f);
|
|
4083 for(lwsp_count--;lwsp_count>0;lwsp_count--)
|
|
4084 i_ungetc(lwsp_buf[lwsp_count],f);
|
|
4085 c1 = lwsp_buf[0];
|
|
4086 }
|
|
4087 }
|
|
4088 free(lwsp_buf);
|
|
4089 lwsp_buf = NULL;
|
|
4090 return c1;
|
|
4091 }
|
|
4092 mime_c3_retry:
|
|
4093 if ((c3 = (*i_mgetc)(f))<=' ') {
|
|
4094 if (c3==EOF)
|
|
4095 return (EOF);
|
|
4096 if (mime_f != STRICT_MIME) goto mime_c3_retry;
|
|
4097 if (mimebuf_f!=FIXED_MIME) input_mode = ASCII;
|
|
4098 return c3;
|
|
4099 }
|
|
4100 mime_c4_retry:
|
|
4101 if ((c4 = (*i_mgetc)(f))<=' ') {
|
|
4102 if (c4==EOF)
|
|
4103 return (EOF);
|
|
4104 if (mime_f != STRICT_MIME) goto mime_c4_retry;
|
|
4105 if (mimebuf_f!=FIXED_MIME) input_mode = ASCII;
|
|
4106 return c4;
|
|
4107 }
|
|
4108
|
|
4109 mime_decode_mode = mode; /* still in MIME sigh... */
|
|
4110
|
|
4111 /* BASE 64 decoding */
|
|
4112
|
|
4113 t1 = 0x3f & base64decode(c1);
|
|
4114 t2 = 0x3f & base64decode(c2);
|
|
4115 t3 = 0x3f & base64decode(c3);
|
|
4116 t4 = 0x3f & base64decode(c4);
|
|
4117 cc = ((t1 << 2) & 0x0fc) | ((t2 >> 4) & 0x03);
|
|
4118 if (c2 != '=') {
|
|
4119 Fifo(mime_last++) = cc;
|
|
4120 cc = ((t2 << 4) & 0x0f0) | ((t3 >> 2) & 0x0f);
|
|
4121 if (c3 != '=') {
|
|
4122 Fifo(mime_last++) = cc;
|
|
4123 cc = ((t3 << 6) & 0x0c0) | (t4 & 0x3f);
|
|
4124 if (c4 != '=')
|
|
4125 Fifo(mime_last++) = cc;
|
|
4126 }
|
|
4127 } else {
|
|
4128 return c1;
|
|
4129 }
|
|
4130 return Fifo(mime_top++);
|
|
4131 }
|
|
4132
|
|
4133 int
|
|
4134 mime_ungetc(c,f)
|
|
4135 int c;
|
|
4136 FILE *f;
|
|
4137 {
|
|
4138 Fifo(--mime_top) = c;
|
|
4139 return c;
|
|
4140 }
|
|
4141
|
|
4142 int
|
|
4143 mime_integrity(f,p)
|
|
4144 FILE *f;
|
|
4145 unsigned char *p;
|
|
4146 {
|
|
4147 int c,d;
|
|
4148 unsigned int q;
|
|
4149 /* In buffered mode, read until =? or NL or buffer full
|
|
4150 */
|
|
4151 mime_input = mime_top;
|
|
4152 mime_last = mime_top;
|
|
4153 while(*p) Fifo(mime_input++) = *p++;
|
|
4154 d = 0;
|
|
4155 q = mime_input;
|
|
4156 while((c=(*i_getc)(f))!=EOF) {
|
|
4157 if (((mime_input-mime_top)&MIME_BUF_MASK)==0) {
|
|
4158 break; /* buffer full */
|
|
4159 }
|
|
4160 if (c=='=' && d=='?') {
|
|
4161 /* checked. skip header, start decode */
|
|
4162 Fifo(mime_input++) = c;
|
|
4163 /* mime_last_input = mime_input; */
|
|
4164 mime_input = q;
|
|
4165 switch_mime_getc();
|
|
4166 return 1;
|
|
4167 }
|
|
4168 if (!( (c=='+'||c=='/'|| c=='=' || c=='?' || is_alnum(c))))
|
|
4169 break;
|
|
4170 /* Should we check length mod 4? */
|
|
4171 Fifo(mime_input++) = c;
|
|
4172 d=c;
|
|
4173 }
|
|
4174 /* In case of Incomplete MIME, no MIME decode */
|
|
4175 Fifo(mime_input++) = c;
|
|
4176 mime_last = mime_input; /* point undecoded buffer */
|
|
4177 mime_decode_mode = 1; /* no decode on Fifo last in mime_getc */
|
|
4178 switch_mime_getc(); /* anyway we need buffered getc */
|
|
4179 return 1;
|
|
4180 }
|
|
4181
|
|
4182 int
|
|
4183 base64decode(c)
|
|
4184 int c;
|
|
4185 {
|
|
4186 int i;
|
|
4187 if (c > '@') {
|
|
4188 if (c < '[') {
|
|
4189 i = c - 'A'; /* A..Z 0-25 */
|
|
4190 } else {
|
|
4191 i = c - 'G' /* - 'a' + 26 */ ; /* a..z 26-51 */
|
|
4192 }
|
|
4193 } else if (c > '/') {
|
|
4194 i = c - '0' + '4' /* - '0' + 52 */ ; /* 0..9 52-61 */
|
|
4195 } else if (c == '+') {
|
|
4196 i = '>' /* 62 */ ; /* + 62 */
|
|
4197 } else {
|
|
4198 i = '?' /* 63 */ ; /* / 63 */
|
|
4199 }
|
|
4200 return (i);
|
|
4201 }
|
|
4202
|
|
4203 static char basis_64[] =
|
|
4204 "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
|
|
4205
|
|
4206 static int b64c;
|
|
4207 #define MIMEOUT_BUF_LENGTH (60)
|
|
4208 char mimeout_buf[MIMEOUT_BUF_LENGTH+1];
|
|
4209 int mimeout_buf_count = 0;
|
|
4210 int mimeout_preserve_space = 0;
|
|
4211 #define itoh4(c) (c>=10?c+'A'-10:c+'0')
|
|
4212
|
|
4213 void
|
|
4214 open_mime(mode)
|
|
4215 int mode;
|
|
4216 {
|
|
4217 unsigned char *p;
|
|
4218 int i;
|
|
4219 int j;
|
|
4220 p = mime_pattern[0];
|
|
4221 for(i=0;mime_encode[i];i++) {
|
|
4222 if (mode == mime_encode[i]) {
|
|
4223 p = mime_pattern[i];
|
|
4224 break;
|
|
4225 }
|
|
4226 }
|
|
4227 mimeout_mode = mime_encode_method[i];
|
|
4228
|
|
4229 i = 0;
|
|
4230 if (base64_count>45) {
|
|
4231 if (mimeout_buf_count>0 && nkf_isblank(mimeout_buf[i])){
|
|
4232 (*o_mputc)(mimeout_buf[i]);
|
|
4233 i++;
|
|
4234 }
|
|
4235 (*o_mputc)(NL);
|
|
4236 (*o_mputc)(SPACE);
|
|
4237 base64_count = 1;
|
|
4238 if (!mimeout_preserve_space && mimeout_buf_count>0
|
|
4239 && (mimeout_buf[i]==SPACE || mimeout_buf[i]==TAB
|
|
4240 || mimeout_buf[i]==CR || mimeout_buf[i]==NL )) {
|
|
4241 i++;
|
|
4242 }
|
|
4243 }
|
|
4244 if (!mimeout_preserve_space) {
|
|
4245 for (;i<mimeout_buf_count;i++) {
|
|
4246 if (mimeout_buf[i]==SPACE || mimeout_buf[i]==TAB
|
|
4247 || mimeout_buf[i]==CR || mimeout_buf[i]==NL ) {
|
|
4248 (*o_mputc)(mimeout_buf[i]);
|
|
4249 base64_count ++;
|
|
4250 } else {
|
|
4251 break;
|
|
4252 }
|
|
4253 }
|
|
4254 }
|
|
4255 mimeout_preserve_space = FALSE;
|
|
4256
|
|
4257 while(*p) {
|
|
4258 (*o_mputc)(*p++);
|
|
4259 base64_count ++;
|
|
4260 }
|
|
4261 j = mimeout_buf_count;
|
|
4262 mimeout_buf_count = 0;
|
|
4263 for (;i<j;i++) {
|
|
4264 mime_putc(mimeout_buf[i]);
|
|
4265 }
|
|
4266 }
|
|
4267
|
|
4268 void
|
|
4269 close_mime()
|
|
4270 {
|
|
4271 (*o_mputc)('?');
|
|
4272 (*o_mputc)('=');
|
|
4273 base64_count += 2;
|
|
4274 mimeout_mode = 0;
|
|
4275 }
|
|
4276
|
|
4277 void
|
|
4278 eof_mime()
|
|
4279 {
|
|
4280 switch(mimeout_mode) {
|
|
4281 case 'Q':
|
|
4282 case 'B':
|
|
4283 break;
|
|
4284 case 2:
|
|
4285 (*o_mputc)(basis_64[((b64c & 0x3)<< 4)]);
|
|
4286 (*o_mputc)('=');
|
|
4287 (*o_mputc)('=');
|
|
4288 base64_count += 3;
|
|
4289 break;
|
|
4290 case 1:
|
|
4291 (*o_mputc)(basis_64[((b64c & 0xF) << 2)]);
|
|
4292 (*o_mputc)('=');
|
|
4293 base64_count += 2;
|
|
4294 break;
|
|
4295 }
|
|
4296 if (mimeout_mode) {
|
|
4297 if (mimeout_f!=FIXED_MIME) {
|
|
4298 close_mime();
|
|
4299 } else if (mimeout_mode != 'Q')
|
|
4300 mimeout_mode = 'B';
|
|
4301 }
|
|
4302 }
|
|
4303
|
|
4304 void
|
|
4305 mimeout_addchar(c)
|
|
4306 int c;
|
|
4307 {
|
|
4308 switch(mimeout_mode) {
|
|
4309 case 'Q':
|
|
4310 if(c==SPACE){
|
|
4311 (*o_mputc)('_');
|
|
4312 base64_count++;
|
|
4313 } else if (c==CR||c==NL) {
|
|
4314 (*o_mputc)(c);
|
|
4315 base64_count = 0;
|
|
4316 } else if(c<SPACE||c=='='||c=='?'||c=='_'||DEL<=c) {
|
|
4317 (*o_mputc)('=');
|
|
4318 (*o_mputc)(itoh4(((c>>4)&0xf)));
|
|
4319 (*o_mputc)(itoh4((c&0xf)));
|
|
4320 base64_count += 3;
|
|
4321 } else {
|
|
4322 (*o_mputc)(c);
|
|
4323 base64_count++;
|
|
4324 }
|
|
4325 break;
|
|
4326 case 'B':
|
|
4327 b64c=c;
|
|
4328 (*o_mputc)(basis_64[c>>2]);
|
|
4329 mimeout_mode=2;
|
|
4330 base64_count ++;
|
|
4331 break;
|
|
4332 case 2:
|
|
4333 (*o_mputc)(basis_64[((b64c & 0x3)<< 4) | ((c & 0xF0) >> 4)]);
|
|
4334 b64c=c;
|
|
4335 mimeout_mode=1;
|
|
4336 base64_count ++;
|
|
4337 break;
|
|
4338 case 1:
|
|
4339 (*o_mputc)(basis_64[((b64c & 0xF) << 2) | ((c & 0xC0) >>6)]);
|
|
4340 (*o_mputc)(basis_64[c & 0x3F]);
|
|
4341 mimeout_mode='B';
|
|
4342 base64_count += 2;
|
|
4343 break;
|
|
4344 default:
|
|
4345 (*o_mputc)(c);
|
|
4346 base64_count++;
|
|
4347 break;
|
|
4348 }
|
|
4349 }
|
|
4350
|
|
4351 int mime_lastchar2, mime_lastchar1;
|
|
4352
|
|
4353 void mime_prechar(c2, c1)
|
|
4354 int c2, c1;
|
|
4355 {
|
|
4356 if (mimeout_mode){
|
|
4357 if (c2){
|
|
4358 if (base64_count + mimeout_buf_count/3*4> 66){
|
|
4359 (*o_base64conv)(EOF,0);
|
|
4360 (*o_base64conv)(0,NL);
|
|
4361 (*o_base64conv)(0,SPACE);
|
|
4362 }
|
|
4363 }/*else if (mime_lastchar2){
|
|
4364 if (c1 <=DEL && !nkf_isspace(c1)){
|
|
4365 (*o_base64conv)(0,SPACE);
|
|
4366 }
|
|
4367 }*/
|
|
4368 }/*else{
|
|
4369 if (c2 && mime_lastchar2 == 0
|
|
4370 && mime_lastchar1 && !nkf_isspace(mime_lastchar1)){
|
|
4371 (*o_base64conv)(0,SPACE);
|
|
4372 }
|
|
4373 }*/
|
|
4374 mime_lastchar2 = c2;
|
|
4375 mime_lastchar1 = c1;
|
|
4376 }
|
|
4377
|
|
4378 void
|
|
4379 mime_putc(c)
|
|
4380 int c;
|
|
4381 {
|
|
4382 int i = 0;
|
|
4383 int j = 0;
|
|
4384 int lastchar;
|
|
4385
|
|
4386 if (mimeout_f == FIXED_MIME){
|
|
4387 if (mimeout_mode == 'Q'){
|
|
4388 if (base64_count > 71){
|
|
4389 if (c!=CR && c!=NL) {
|
|
4390 (*o_mputc)('=');
|
|
4391 (*o_mputc)(NL);
|
|
4392 }
|
|
4393 base64_count = 0;
|
|
4394 }
|
|
4395 }else{
|
|
4396 if (base64_count > 71){
|
|
4397 eof_mime();
|
|
4398 (*o_mputc)(NL);
|
|
4399 base64_count = 0;
|
|
4400 }
|
|
4401 if (c == EOF) { /* c==EOF */
|
|
4402 eof_mime();
|
|
4403 }
|
|
4404 }
|
|
4405 if (c != EOF) { /* c==EOF */
|
|
4406 mimeout_addchar(c);
|
|
4407 }
|
|
4408 return;
|
|
4409 }
|
|
4410
|
|
4411 /* mimeout_f != FIXED_MIME */
|
|
4412
|
|
4413 if (c == EOF) { /* c==EOF */
|
|
4414 j = mimeout_buf_count;
|
|
4415 mimeout_buf_count = 0;
|
|
4416 i = 0;
|
|
4417 for (;i<j;i++) {
|
|
4418 /*if (nkf_isspace(mimeout_buf[i])){
|
|
4419 break;
|
|
4420 }*/
|
|
4421 mimeout_addchar(mimeout_buf[i]);
|
|
4422 }
|
|
4423 eof_mime();
|
|
4424 for (;i<j;i++) {
|
|
4425 (*o_mputc)(mimeout_buf[i]);
|
|
4426 base64_count++;
|
|
4427 }
|
|
4428 return;
|
|
4429 }
|
|
4430
|
|
4431 if (mimeout_mode=='Q') {
|
|
4432 if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
|
|
4433 if (c <= SPACE) {
|
|
4434 close_mime();
|
|
4435 (*o_mputc)(SPACE);
|
|
4436 base64_count++;
|
|
4437 }
|
|
4438 (*o_mputc)(c);
|
|
4439 base64_count++;
|
|
4440 }
|
|
4441 return;
|
|
4442 }
|
|
4443
|
|
4444 if (mimeout_buf_count > 0){
|
|
4445 lastchar = mimeout_buf[mimeout_buf_count - 1];
|
|
4446 }else{
|
|
4447 lastchar = -1;
|
|
4448 }
|
|
4449
|
|
4450 if (!mimeout_mode) {
|
|
4451 if (c <= DEL && (output_mode==ASCII ||output_mode == ISO8859_1)) {
|
|
4452 if (nkf_isspace(c)) {
|
|
4453 if (c==CR || c==NL) {
|
|
4454 base64_count=0;
|
|
4455 }
|
|
4456 for (i=0;i<mimeout_buf_count;i++) {
|
|
4457 (*o_mputc)(mimeout_buf[i]);
|
|
4458 if (mimeout_buf[i] == CR || mimeout_buf[i] == NL){
|
|
4459 base64_count = 0;
|
|
4460 }else{
|
|
4461 base64_count++;
|
|
4462 }
|
|
4463 }
|
|
4464 mimeout_buf[0] = c;
|
|
4465 mimeout_buf_count = 1;
|
|
4466 }else{
|
|
4467 if (base64_count > 1
|
|
4468 && base64_count + mimeout_buf_count > 76){
|
|
4469 (*o_mputc)(NL);
|
|
4470 base64_count = 0;
|
|
4471 if (!nkf_isspace(mimeout_buf[0])){
|
|
4472 (*o_mputc)(SPACE);
|
|
4473 base64_count++;
|
|
4474 }
|
|
4475 }
|
|
4476 mimeout_buf[mimeout_buf_count++] = c;
|
|
4477 if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
|
|
4478 open_mime(output_mode);
|
|
4479 }
|
|
4480 }
|
|
4481 return;
|
|
4482 }else{
|
|
4483 if (lastchar==CR || lastchar == NL){
|
|
4484 for (i=0;i<mimeout_buf_count;i++) {
|
|
4485 (*o_mputc)(mimeout_buf[i]);
|
|
4486 }
|
|
4487 base64_count = 0;
|
|
4488 mimeout_buf_count = 0;
|
|
4489 }
|
|
4490 if (lastchar==SPACE) {
|
|
4491 for (i=0;i<mimeout_buf_count-1;i++) {
|
|
4492 (*o_mputc)(mimeout_buf[i]);
|
|
4493 base64_count++;
|
|
4494 }
|
|
4495 mimeout_buf[0] = SPACE;
|
|
4496 mimeout_buf_count = 1;
|
|
4497 }
|
|
4498 open_mime(output_mode);
|
|
4499 }
|
|
4500 }else{
|
|
4501 /* mimeout_mode == 'B', 1, 2 */
|
|
4502 if ( c<=DEL && (output_mode==ASCII ||output_mode == ISO8859_1 ) ) {
|
|
4503 if (lastchar == CR || lastchar == NL){
|
|
4504 if (nkf_isblank(c)) {
|
|
4505 for (i=0;i<mimeout_buf_count;i++) {
|
|
4506 mimeout_addchar(mimeout_buf[i]);
|
|
4507 }
|
|
4508 mimeout_buf_count = 0;
|
|
4509 } else if (SPACE<c && c<DEL) {
|
|
4510 eof_mime();
|
|
4511 for (i=0;i<mimeout_buf_count;i++) {
|
|
4512 (*o_mputc)(mimeout_buf[i]);
|
|
4513 }
|
|
4514 base64_count = 0;
|
|
4515 mimeout_buf_count = 0;
|
|
4516 }
|
|
4517 }
|
|
4518 if (c==SPACE || c==TAB || c==CR || c==NL) {
|
|
4519 for (i=0;i<mimeout_buf_count;i++) {
|
|
4520 if (SPACE<mimeout_buf[i] && mimeout_buf[i]<DEL) {
|
|
4521 eof_mime();
|
|
4522 for (i=0;i<mimeout_buf_count;i++) {
|
|
4523 (*o_mputc)(mimeout_buf[i]);
|
|
4524 base64_count++;
|
|
4525 }
|
|
4526 mimeout_buf_count = 0;
|
|
4527 }
|
|
4528 }
|
|
4529 mimeout_buf[mimeout_buf_count++] = c;
|
|
4530 if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
|
|
4531 eof_mime();
|
|
4532 for (i=0;i<mimeout_buf_count;i++) {
|
|
4533 (*o_mputc)(mimeout_buf[i]);
|
|
4534 base64_count++;
|
|
4535 }
|
|
4536 mimeout_buf_count = 0;
|
|
4537 }
|
|
4538 return;
|
|
4539 }
|
|
4540 if (mimeout_buf_count>0 && SPACE<c && c!='=') {
|
|
4541 mimeout_buf[mimeout_buf_count++] = c;
|
|
4542 if (mimeout_buf_count>MIMEOUT_BUF_LENGTH) {
|
|
4543 j = mimeout_buf_count;
|
|
4544 mimeout_buf_count = 0;
|
|
4545 for (i=0;i<j;i++) {
|
|
4546 mimeout_addchar(mimeout_buf[i]);
|
|
4547 }
|
|
4548 }
|
|
4549 return;
|
|
4550 }
|
|
4551 }
|
|
4552 }
|
|
4553 if (mimeout_buf_count>0) {
|
|
4554 j = mimeout_buf_count;
|
|
4555 mimeout_buf_count = 0;
|
|
4556 for (i=0;i<j;i++) {
|
|
4557 if (mimeout_buf[i]==CR || mimeout_buf[i]==NL)
|
|
4558 break;
|
|
4559 mimeout_addchar(mimeout_buf[i]);
|
|
4560 }
|
|
4561 if (i<j) {
|
|
4562 eof_mime();
|
|
4563 base64_count=0;
|
|
4564 for (;i<j;i++) {
|
|
4565 (*o_mputc)(mimeout_buf[i]);
|
|
4566 }
|
|
4567 open_mime(output_mode);
|
|
4568 }
|
|
4569 }
|
|
4570 mimeout_addchar(c);
|
|
4571 }
|
|
4572
|
|
4573
|
|
4574 #if defined(PERL_XS) || defined(WIN32DLL)
|
|
4575 void
|
|
4576 reinit()
|
|
4577 {
|
|
4578 {
|
|
4579 struct input_code *p = input_code_list;
|
|
4580 while (p->name){
|
|
4581 status_reinit(p++);
|
|
4582 }
|
|
4583 }
|
|
4584 unbuf_f = FALSE;
|
|
4585 estab_f = FALSE;
|
|
4586 nop_f = FALSE;
|
|
4587 binmode_f = TRUE;
|
|
4588 rot_f = FALSE;
|
|
4589 hira_f = FALSE;
|
|
4590 input_f = FALSE;
|
|
4591 alpha_f = FALSE;
|
|
4592 mime_f = STRICT_MIME;
|
|
4593 mime_decode_f = FALSE;
|
|
4594 mimebuf_f = FALSE;
|
|
4595 broken_f = FALSE;
|
|
4596 iso8859_f = FALSE;
|
|
4597 mimeout_f = FALSE;
|
|
4598 #if defined(MSDOS) || defined(__OS2__)
|
|
4599 x0201_f = TRUE;
|
|
4600 #else
|
|
4601 x0201_f = NO_X0201;
|
|
4602 #endif
|
|
4603 iso2022jp_f = FALSE;
|
|
4604 #ifdef UTF8_OUTPUT_ENABLE
|
|
4605 unicode_bom_f = 0;
|
|
4606 w_oconv16_LE = 0;
|
|
4607 ms_ucs_map_f = FALSE;
|
|
4608 #endif
|
|
4609 #ifdef INPUT_OPTION
|
|
4610 cap_f = FALSE;
|
|
4611 url_f = FALSE;
|
|
4612 numchar_f = FALSE;
|
|
4613 #endif
|
|
4614 #ifdef CHECK_OPTION
|
|
4615 noout_f = FALSE;
|
|
4616 debug_f = FALSE;
|
|
4617 #endif
|
|
4618 guess_f = FALSE;
|
|
4619 is_inputcode_mixed = FALSE;
|
|
4620 is_inputcode_set = FALSE;
|
|
4621 #ifdef EXEC_IO
|
|
4622 exec_f = 0;
|
|
4623 #endif
|
|
4624 #ifdef SHIFTJIS_CP932
|
|
4625 cp932_f = TRUE;
|
|
4626 cp932inv_f = TRUE;
|
|
4627 #endif
|
|
4628 {
|
|
4629 int i;
|
|
4630 for (i = 0; i < 256; i++){
|
|
4631 prefix_table[i] = 0;
|
|
4632 }
|
|
4633 }
|
|
4634 #ifdef UTF8_INPUT_ENABLE
|
|
4635 utf16_mode = UTF16BE_INPUT;
|
|
4636 #endif
|
|
4637 mimeout_buf_count = 0;
|
|
4638 mimeout_mode = 0;
|
|
4639 base64_count = 0;
|
|
4640 f_line = 0;
|
|
4641 f_prev = 0;
|
|
4642 fold_preserve_f = FALSE;
|
|
4643 fold_f = FALSE;
|
|
4644 fold_len = 0;
|
|
4645 kanji_intro = DEFAULT_J;
|
|
4646 ascii_intro = DEFAULT_R;
|
|
4647 fold_margin = FOLD_MARGIN;
|
|
4648 output_conv = DEFAULT_CONV;
|
|
4649 oconv = DEFAULT_CONV;
|
|
4650 o_zconv = no_connection;
|
|
4651 o_fconv = no_connection;
|
|
4652 o_crconv = no_connection;
|
|
4653 o_rot_conv = no_connection;
|
|
4654 o_hira_conv = no_connection;
|
|
4655 o_base64conv = no_connection;
|
|
4656 o_iso2022jp_check_conv = no_connection;
|
|
4657 o_putc = std_putc;
|
|
4658 i_getc = std_getc;
|
|
4659 i_ungetc = std_ungetc;
|
|
4660 i_bgetc = std_getc;
|
|
4661 i_bungetc = std_ungetc;
|
|
4662 o_mputc = std_putc;
|
|
4663 i_mgetc = std_getc;
|
|
4664 i_mungetc = std_ungetc;
|
|
4665 i_mgetc_buf = std_getc;
|
|
4666 i_mungetc_buf = std_ungetc;
|
|
4667 output_mode = ASCII;
|
|
4668 input_mode = ASCII;
|
|
4669 shift_mode = FALSE;
|
|
4670 mime_decode_mode = FALSE;
|
|
4671 file_out = FALSE;
|
|
4672 crmode_f = 0;
|
|
4673 option_mode = 0;
|
|
4674 broken_counter = 0;
|
|
4675 broken_last = 0;
|
|
4676 z_prev2=0,z_prev1=0;
|
|
4677 #ifdef CHECK_OPTION
|
|
4678 iconv_for_check = 0;
|
|
4679 #endif
|
|
4680 input_codename = "";
|
|
4681 #ifdef WIN32DLL
|
|
4682 reinitdll();
|
|
4683 #endif /*WIN32DLL*/
|
|
4684 }
|
|
4685 #endif
|
|
4686
|
|
4687 void
|
|
4688 no_connection(c2,c1)
|
|
4689 int c2,c1;
|
|
4690 {
|
|
4691 no_connection2(c2,c1,0);
|
|
4692 }
|
|
4693
|
|
4694 int
|
|
4695 no_connection2(c2,c1,c0)
|
|
4696 int c2,c1,c0;
|
|
4697 {
|
|
4698 fprintf(stderr,"nkf internal module connection failure.\n");
|
|
4699 exit(1);
|
|
4700 return 0; /* LINT */
|
|
4701 }
|
|
4702
|
|
4703 #ifndef PERL_XS
|
|
4704 #ifdef WIN32DLL
|
|
4705 #define fprintf dllprintf
|
|
4706 #endif
|
|
4707 void
|
|
4708 usage()
|
|
4709 {
|
|
4710 fprintf(stderr,"USAGE: nkf(nkf32,wnkf,nkf2) -[flags] [in file] .. [out file for -O flag]\n");
|
|
4711 fprintf(stderr,"Flags:\n");
|
|
4712 fprintf(stderr,"b,u Output is buffered (DEFAULT),Output is unbuffered\n");
|
|
4713 #ifdef DEFAULT_CODE_SJIS
|
|
4714 fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS (DEFAULT), AT&T JIS (EUC), UTF-8\n");
|
|
4715 #endif
|
|
4716 #ifdef DEFAULT_CODE_JIS
|
|
4717 fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit (DEFAULT), Shift JIS, AT&T JIS (EUC), UTF-8\n");
|
|
4718 #endif
|
|
4719 #ifdef DEFAULT_CODE_EUC
|
|
4720 fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC) (DEFAULT), UTF-8\n");
|
|
4721 #endif
|
|
4722 #ifdef DEFAULT_CODE_UTF8
|
|
4723 fprintf(stderr,"j,s,e,w Outout code is JIS 7 bit, Shift JIS, AT&T JIS (EUC), UTF-8 (DEFAULT)\n");
|
|
4724 #endif
|
|
4725 #ifdef UTF8_OUTPUT_ENABLE
|
|
4726 fprintf(stderr," After 'w' you can add more options. (80?|16((B|L)0?)?) \n");
|
|
4727 #endif
|
|
4728 fprintf(stderr,"J,S,E,W Input assumption is JIS 7 bit , Shift JIS, AT&T JIS (EUC), UTF-8\n");
|
|
4729 #ifdef UTF8_INPUT_ENABLE
|
|
4730 fprintf(stderr," After 'W' you can add more options. (8|16(B|L)?) \n");
|
|
4731 #endif
|
|
4732 fprintf(stderr,"t no conversion\n");
|
|
4733 fprintf(stderr,"i_/o_ Output sequence to designate JIS-kanji/ASCII (DEFAULT B)\n");
|
|
4734 fprintf(stderr,"r {de/en}crypt ROT13/47\n");
|
|
4735 fprintf(stderr,"h 1 hirakana->katakana, 2 katakana->hirakana,3 both\n");
|
|
4736 fprintf(stderr,"v Show this usage. V: show version\n");
|
|
4737 fprintf(stderr,"m[BQN0] MIME decode [B:base64,Q:quoted,N:non-strict,0:no decode]\n");
|
|
4738 fprintf(stderr,"M[BQ] MIME encode [B:base64 Q:quoted]\n");
|
|
4739 fprintf(stderr,"l ISO8859-1 (Latin-1) support\n");
|
|
4740 fprintf(stderr,"f/F Folding: -f60 or -f or -f60-10 (fold margin 10) F preserve nl\n");
|
|
4741 fprintf(stderr,"Z[0-3] Convert X0208 alphabet to ASCII 1: Kankaku to space,2: 2 spaces,\n");
|
|
4742 fprintf(stderr," 3: Convert HTML Entity\n");
|
|
4743 fprintf(stderr,"X,x Assume X0201 kana in MS-Kanji, -x preserves X0201\n");
|
|
4744 fprintf(stderr,"B[0-2] Broken input 0: missing ESC,1: any X on ESC-[($]-X,2: ASCII on NL\n");
|
|
4745 #ifdef MSDOS
|
|
4746 fprintf(stderr,"T Text mode output\n");
|
|
4747 #endif
|
|
4748 fprintf(stderr,"O Output to File (DEFAULT 'nkf.out')\n");
|
|
4749 fprintf(stderr,"d,c Delete \\r in line feed and \\032, Add \\r in line feed\n");
|
|
4750 fprintf(stderr,"I Convert non ISO-2022-JP charactor to GETA\n");
|
|
4751 fprintf(stderr,"-L[uwm] line mode u:LF w:CRLF m:CR (DEFAULT noconversion)\n");
|
|
4752 fprintf(stderr,"long name options\n");
|
|
4753 fprintf(stderr," --fj,--unix,--mac,--windows convert for the system\n");
|
|
4754 fprintf(stderr," --jis,--euc,--sjis,--utf8,--utf16,--mime,--base64 convert for the code\n");
|
|
4755 fprintf(stderr," --hiragana, --katakana Hiragana/Katakana Conversion\n");
|
|
4756 fprintf(stderr," --x0212 Convert JISX0212\n");
|
|
4757 fprintf(stderr," --cp932, --no-cp932 CP932 compatibility\n");
|
|
4758 #ifdef INPUT_OPTION
|
|
4759 fprintf(stderr," --cap-input, --url-input Convert hex after ':' or '%%'\n");
|
|
4760 #endif
|
|
4761 #ifdef NUMCHAR_OPTION
|
|
4762 fprintf(stderr," --numchar-input Convert Unicode Character Reference\n");
|
|
4763 #endif
|
|
4764 #ifdef UTF8_OUTPUT_ENABLE
|
|
4765 fprintf(stderr," --ms-ucs-map Microsoft UCS Mapping Compatible\n");
|
|
4766 #endif
|
|
4767 #ifdef OVERWRITE
|
|
4768 fprintf(stderr," --overwrite Overwrite original listed files by filtered result\n");
|
|
4769 #endif
|
|
4770 fprintf(stderr," -g, --guess Guess the input code\n");
|
|
4771 fprintf(stderr," --help,--version\n");
|
|
4772 version();
|
|
4773 }
|
|
4774
|
|
4775 void
|
|
4776 version()
|
|
4777 {
|
|
4778 fprintf(stderr,"Network Kanji Filter Version %s (%s) "
|
|
4779 #if defined(MSDOS) && !defined(__WIN32__) && !defined(__WIN16__)
|
|
4780 "for DOS"
|
|
4781 #endif
|
|
4782 #if defined(MSDOS) && defined(__WIN16__)
|
|
4783 "for Win16"
|
|
4784 #endif
|
|
4785 #if defined(MSDOS) && defined(__WIN32__)
|
|
4786 "for Win32"
|
|
4787 #endif
|
|
4788 #ifdef __OS2__
|
|
4789 "for OS/2"
|
|
4790 #endif
|
|
4791 ,NKF_VERSION,NKF_RELEASE_DATE);
|
|
4792 fprintf(stderr,"\n%s\n",CopyRight);
|
|
4793 }
|
|
4794 #endif /*PERL_XS*/
|
|
4795
|
|
4796 /**
|
|
4797 ** $B%Q%C%A@):n<T(B
|
|
4798 ** void@merope.pleiades.or.jp (Kusakabe Youichi)
|
|
4799 ** NIDE Naoyuki <nide@ics.nara-wu.ac.jp>
|
|
4800 ** ohta@src.ricoh.co.jp (Junn Ohta)
|
|
4801 ** inouet@strl.nhk.or.jp (Tomoyuki Inoue)
|
|
4802 ** kiri@pulser.win.or.jp (Tetsuaki Kiriyama)
|
|
4803 ** Kimihiko Sato <sato@sail.t.u-tokyo.ac.jp>
|
|
4804 ** a_kuroe@kuroe.aoba.yokohama.jp (Akihiko Kuroe)
|
|
4805 ** kono@ie.u-ryukyu.ac.jp (Shinji Kono)
|
|
4806 ** GHG00637@nifty-serve.or.jp (COW)
|
|
4807 **
|
|
4808 **/
|
|
4809
|
|
4810 /* end */
|