1 941e1713 2006-02-15 devnull /* Quick and dirty RFC 2047 */
  2 941e1713 2006-02-15 devnull 
  3 941e1713 2006-02-15 devnull #include "a.h"
  4 941e1713 2006-02-15 devnull 
  5 941e1713 2006-02-15 devnull static int
  6 941e1713 2006-02-15 devnull unhex1(char c)
  7 941e1713 2006-02-15 devnull {
  8 941e1713 2006-02-15 devnull 	if('0' <= c && c <= '9')
  9 941e1713 2006-02-15 devnull 		return c-'0';
 10 941e1713 2006-02-15 devnull 	if('a' <= c && c <= 'f')
 11 941e1713 2006-02-15 devnull 		return c-'a'+10;
 12 941e1713 2006-02-15 devnull 	if('A' <= c && c <= 'F')
 13 941e1713 2006-02-15 devnull 		return c-'A'+10;
 14 941e1713 2006-02-15 devnull 	return 15;
 15 941e1713 2006-02-15 devnull }
 16 941e1713 2006-02-15 devnull 
 17 941e1713 2006-02-15 devnull static int
 18 941e1713 2006-02-15 devnull unhex(char *s)
 19 941e1713 2006-02-15 devnull {
 20 941e1713 2006-02-15 devnull 	return unhex1(s[0])*16+unhex1(s[1]);
 21 941e1713 2006-02-15 devnull }
 22 941e1713 2006-02-15 devnull 
 23 941e1713 2006-02-15 devnull int
 24 6a8add52 2006-02-23 devnull _decqp(uchar *out, int lim, char *in, int n, int underscores)
 25 941e1713 2006-02-15 devnull {
 26 941e1713 2006-02-15 devnull 	char *p, *ep;
 27 941e1713 2006-02-15 devnull 	uchar *eout, *out0;
 28 941e1713 2006-02-15 devnull 	
 29 941e1713 2006-02-15 devnull 	out0 = out;
 30 941e1713 2006-02-15 devnull 	eout = out+lim;
 31 941e1713 2006-02-15 devnull 	for(p=in, ep=in+n; p<ep && out<eout; ){
 32 6a8add52 2006-02-23 devnull 		if(underscores && *p == '_'){
 33 941e1713 2006-02-15 devnull 			*out++ = ' ';
 34 941e1713 2006-02-15 devnull 			p++;
 35 941e1713 2006-02-15 devnull 		}
 36 941e1713 2006-02-15 devnull 		else if(*p == '='){
 37 941e1713 2006-02-15 devnull 			if(p+1 >= ep)
 38 941e1713 2006-02-15 devnull 				break;
 39 941e1713 2006-02-15 devnull 			if(*(p+1) == '\n'){
 40 941e1713 2006-02-15 devnull 				p += 2;
 41 941e1713 2006-02-15 devnull 				continue;
 42 941e1713 2006-02-15 devnull 			}
 43 941e1713 2006-02-15 devnull 			if(p+3 > ep)
 44 941e1713 2006-02-15 devnull 				break;
 45 941e1713 2006-02-15 devnull 			*out++ = unhex(p+1);
 46 941e1713 2006-02-15 devnull 			p += 3;
 47 941e1713 2006-02-15 devnull 		}else
 48 941e1713 2006-02-15 devnull 			*out++ = *p++;
 49 941e1713 2006-02-15 devnull 	}
 50 941e1713 2006-02-15 devnull 	return out-out0;
 51 941e1713 2006-02-15 devnull }
 52 941e1713 2006-02-15 devnull 
 53 6a8add52 2006-02-23 devnull int
 54 6a8add52 2006-02-23 devnull decqp(uchar *out, int lim, char *in, int n)
 55 6a8add52 2006-02-23 devnull {
 56 6a8add52 2006-02-23 devnull 	return _decqp(out, lim, in, n, 0);
 57 6a8add52 2006-02-23 devnull }
 58 6a8add52 2006-02-23 devnull 
 59 941e1713 2006-02-15 devnull char*
 60 941e1713 2006-02-15 devnull decode(int kind, char *s, int *len)
 61 941e1713 2006-02-15 devnull {
 62 941e1713 2006-02-15 devnull 	char *t;
 63 941e1713 2006-02-15 devnull 	int l;
 64 941e1713 2006-02-15 devnull 
 65 941e1713 2006-02-15 devnull 	if(s == nil)
 66 941e1713 2006-02-15 devnull 		return s;
 67 941e1713 2006-02-15 devnull 	switch(kind){
 68 384d563d 2006-02-23 devnull 	case QuotedPrintable:
 69 6a8add52 2006-02-23 devnull 	case QuotedPrintableU:
 70 941e1713 2006-02-15 devnull 		l = strlen(s)+1;
 71 941e1713 2006-02-15 devnull 		t = emalloc(l);
 72 1ff2c2b9 2006-02-23 devnull 		l = _decqp((uchar*)t, l, s, l-1, kind==QuotedPrintableU);
 73 941e1713 2006-02-15 devnull 		*len = l;
 74 941e1713 2006-02-15 devnull 		t[l] = 0;
 75 941e1713 2006-02-15 devnull 		return t;
 76 941e1713 2006-02-15 devnull 	
 77 941e1713 2006-02-15 devnull 	case Base64:
 78 941e1713 2006-02-15 devnull 		l = strlen(s)+1;
 79 941e1713 2006-02-15 devnull 		t = emalloc(l);
 80 941e1713 2006-02-15 devnull 		l = dec64((uchar*)t, l, s, l-1);
 81 941e1713 2006-02-15 devnull 		*len = l;
 82 941e1713 2006-02-15 devnull 		t[l] = 0;
 83 941e1713 2006-02-15 devnull 		return t;
 84 941e1713 2006-02-15 devnull 
 85 941e1713 2006-02-15 devnull 	default:
 86 941e1713 2006-02-15 devnull 		*len = strlen(s);
 87 941e1713 2006-02-15 devnull 		return estrdup(s);
 88 941e1713 2006-02-15 devnull 	}
 89 941e1713 2006-02-15 devnull }
 90 941e1713 2006-02-15 devnull 
 91 941e1713 2006-02-15 devnull struct {
 92 941e1713 2006-02-15 devnull 	char *mime;
 93 941e1713 2006-02-15 devnull 	char *tcs;
 94 941e1713 2006-02-15 devnull } tcstab[] = {
 95 941e1713 2006-02-15 devnull 	"iso-8859-2",		"8859-2",
 96 941e1713 2006-02-15 devnull 	"iso-8859-3",		"8859-3",
 97 941e1713 2006-02-15 devnull 	"iso-8859-4",		"8859-4",
 98 941e1713 2006-02-15 devnull 	"iso-8859-5",		"8859-5",
 99 941e1713 2006-02-15 devnull 	"iso-8859-6",		"8859-6",
100 941e1713 2006-02-15 devnull 	"iso-8859-7",		"8859-7",
101 941e1713 2006-02-15 devnull 	"iso-8859-8",		"8859-8",
102 941e1713 2006-02-15 devnull 	"iso-8859-9",		"8859-9",
103 941e1713 2006-02-15 devnull 	"iso-8859-10",	"8859-10",
104 941e1713 2006-02-15 devnull 	"iso-8859-15",	"8859-15",
105 941e1713 2006-02-15 devnull 	"big5",			"big5",
106 941e1713 2006-02-15 devnull 	"iso-2022-jp",	"jis-kanji",
107 865e8b0f 2008-05-06 rsc 	"windows-1250",	"windows-1250",
108 865e8b0f 2008-05-06 rsc 	"windows-1251",	"windows-1251",
109 865e8b0f 2008-05-06 rsc 	"windows-1252",	"windows-1252",
110 865e8b0f 2008-05-06 rsc 	"windows-1253",	"windows-1253",
111 865e8b0f 2008-05-06 rsc 	"windows-1254",	"windows-1254",
112 865e8b0f 2008-05-06 rsc 	"windows-1255",	"windows-1255",
113 865e8b0f 2008-05-06 rsc 	"windows-1256",	"windows-1256",
114 865e8b0f 2008-05-06 rsc 	"windows-1257",	"windows-1257",
115 865e8b0f 2008-05-06 rsc 	"windows-1258",	"windows-1258",
116 cbeb0b26 2006-04-01 devnull 	"koi8-r",			"koi8"
117 941e1713 2006-02-15 devnull };
118 941e1713 2006-02-15 devnull 
119 c5f88424 2008-07-14 rsc typedef struct Writeargs Writeargs;
120 c5f88424 2008-07-14 rsc struct Writeargs
121 c5f88424 2008-07-14 rsc {
122 c5f88424 2008-07-14 rsc 	int fd;
123 c5f88424 2008-07-14 rsc 	char *s;
124 c5f88424 2008-07-14 rsc };
125 c5f88424 2008-07-14 rsc 
126 c5f88424 2008-07-14 rsc static void
127 c5f88424 2008-07-14 rsc twriter(void *v)
128 c5f88424 2008-07-14 rsc {
129 c5f88424 2008-07-14 rsc 	Writeargs *w;
130 c5f88424 2008-07-14 rsc 	
131 c5f88424 2008-07-14 rsc 	w = v;
132 c5f88424 2008-07-14 rsc 	write(w->fd, w->s, strlen(w->s));
133 c5f88424 2008-07-14 rsc 	close(w->fd);
134 c5f88424 2008-07-14 rsc 	free(w->s);
135 c5f88424 2008-07-14 rsc 	free(w);
136 c5f88424 2008-07-14 rsc }
137 c5f88424 2008-07-14 rsc 
138 941e1713 2006-02-15 devnull char*
139 941e1713 2006-02-15 devnull tcs(char *charset, char *s)
140 941e1713 2006-02-15 devnull {
141 c5f88424 2008-07-14 rsc 	char *buf;
142 c5f88424 2008-07-14 rsc 	int i, n, nbuf;
143 941e1713 2006-02-15 devnull 	int fd[3], p[2], pp[2];
144 941e1713 2006-02-15 devnull 	uchar *us;
145 941e1713 2006-02-15 devnull 	char *t, *u;
146 941e1713 2006-02-15 devnull 	char *argv[4];
147 941e1713 2006-02-15 devnull 	Rune r;
148 c5f88424 2008-07-14 rsc 	Writeargs *w;
149 c5f88424 2008-07-14 rsc 
150 941e1713 2006-02-15 devnull 	if(s == nil || charset == nil || *s == 0)
151 941e1713 2006-02-15 devnull 		return s;
152 941e1713 2006-02-15 devnull 
153 941e1713 2006-02-15 devnull 	if(cistrcmp(charset, "utf-8") == 0)
154 941e1713 2006-02-15 devnull 		return s;
155 941e1713 2006-02-15 devnull 	if(cistrcmp(charset, "iso-8859-1") == 0 || cistrcmp(charset, "us-ascii") == 0){
156 941e1713 2006-02-15 devnull latin1:
157 941e1713 2006-02-15 devnull 		n = 0;
158 941e1713 2006-02-15 devnull 		for(us=(uchar*)s; *us; us++)
159 941e1713 2006-02-15 devnull 			n += runelen(*us);
160 941e1713 2006-02-15 devnull 		n++;
161 941e1713 2006-02-15 devnull 		t = emalloc(n);
162 941e1713 2006-02-15 devnull 		for(us=(uchar*)s, u=t; *us; us++){
163 941e1713 2006-02-15 devnull 			r = *us;
164 941e1713 2006-02-15 devnull 			u += runetochar(u, &r);
165 941e1713 2006-02-15 devnull 		}
166 941e1713 2006-02-15 devnull 		*u = 0;
167 941e1713 2006-02-15 devnull 		free(s);
168 941e1713 2006-02-15 devnull 		return t;
169 941e1713 2006-02-15 devnull 	}
170 941e1713 2006-02-15 devnull 	for(i=0; i<nelem(tcstab); i++)
171 941e1713 2006-02-15 devnull 		if(cistrcmp(charset, tcstab[i].mime) == 0)
172 941e1713 2006-02-15 devnull 			goto tcs;
173 941e1713 2006-02-15 devnull 	goto latin1;
174 941e1713 2006-02-15 devnull 
175 941e1713 2006-02-15 devnull tcs:
176 941e1713 2006-02-15 devnull 	argv[0] = "tcs";
177 941e1713 2006-02-15 devnull 	argv[1] = "-f";
178 941e1713 2006-02-15 devnull 	argv[2] = charset;
179 941e1713 2006-02-15 devnull 	argv[3] = nil;
180 941e1713 2006-02-15 devnull 	
181 941e1713 2006-02-15 devnull 	if(pipe(p) < 0 || pipe(pp) < 0)
182 941e1713 2006-02-15 devnull 		sysfatal("pipe: %r");
183 941e1713 2006-02-15 devnull 	fd[0] = p[0];
184 941e1713 2006-02-15 devnull 	fd[1] = pp[0];
185 941e1713 2006-02-15 devnull 	fd[2] = dup(2, -1);
186 941e1713 2006-02-15 devnull 	if(threadspawnl(fd, "tcs", "tcs", "-f", tcstab[i].tcs, nil) < 0){
187 941e1713 2006-02-15 devnull 		close(p[0]);
188 941e1713 2006-02-15 devnull 		close(p[1]);
189 941e1713 2006-02-15 devnull 		close(pp[0]);
190 941e1713 2006-02-15 devnull 		close(pp[1]);
191 941e1713 2006-02-15 devnull 		close(fd[2]);
192 941e1713 2006-02-15 devnull 		goto latin1;
193 941e1713 2006-02-15 devnull 	}
194 941e1713 2006-02-15 devnull 	close(p[0]);
195 941e1713 2006-02-15 devnull 	close(pp[0]);
196 c5f88424 2008-07-14 rsc 
197 c5f88424 2008-07-14 rsc 	nbuf = UTFmax*strlen(s)+100;	/* just a guess at worst case */
198 c5f88424 2008-07-14 rsc 	buf = emalloc(nbuf);
199 c5f88424 2008-07-14 rsc 
200 c5f88424 2008-07-14 rsc 	w = emalloc(sizeof *w);
201 c5f88424 2008-07-14 rsc 	w->fd = p[1];
202 c5f88424 2008-07-14 rsc 	w->s = estrdup(s);
203 c5f88424 2008-07-14 rsc 	proccreate(twriter, w, STACK);
204 c5f88424 2008-07-14 rsc 
205 c5f88424 2008-07-14 rsc 	n = readn(pp[1], buf, nbuf-1);
206 941e1713 2006-02-15 devnull 	close(pp[1]);
207 c5f88424 2008-07-14 rsc 	if(n <= 0){
208 c5f88424 2008-07-14 rsc 		free(buf);
209 941e1713 2006-02-15 devnull 		goto latin1;
210 c5f88424 2008-07-14 rsc 	}
211 941e1713 2006-02-15 devnull 	buf[n] = 0;
212 c5f88424 2008-07-14 rsc 	free(s);
213 c5f88424 2008-07-14 rsc 	s = estrdup(buf);
214 c5f88424 2008-07-14 rsc 	free(buf);
215 c5f88424 2008-07-14 rsc 	return s;
216 941e1713 2006-02-15 devnull }
217 941e1713 2006-02-15 devnull 
218 941e1713 2006-02-15 devnull char*
219 941e1713 2006-02-15 devnull unrfc2047(char *s)
220 941e1713 2006-02-15 devnull {
221 941e1713 2006-02-15 devnull 	char *p, *q, *t, *u, *v;
222 941e1713 2006-02-15 devnull 	int len;
223 941e1713 2006-02-15 devnull 	Rune r;
224 941e1713 2006-02-15 devnull 	Fmt fmt;
225 941e1713 2006-02-15 devnull 	
226 941e1713 2006-02-15 devnull 	if(s == nil)
227 941e1713 2006-02-15 devnull 		return nil;
228 941e1713 2006-02-15 devnull 
229 941e1713 2006-02-15 devnull 	if(strstr(s, "=?") == nil)
230 941e1713 2006-02-15 devnull 		return s;
231 941e1713 2006-02-15 devnull 	
232 941e1713 2006-02-15 devnull 	fmtstrinit(&fmt);
233 941e1713 2006-02-15 devnull 	for(p=s; *p; ){
234 941e1713 2006-02-15 devnull 		/* =?charset?e?text?= */
235 941e1713 2006-02-15 devnull 		if(*p=='=' && *(p+1)=='?'){
236 941e1713 2006-02-15 devnull 			p += 2;
237 941e1713 2006-02-15 devnull 			q = strchr(p, '?');
238 941e1713 2006-02-15 devnull 			if(q == nil)
239 941e1713 2006-02-15 devnull 				goto emit;
240 941e1713 2006-02-15 devnull 			q++;
241 941e1713 2006-02-15 devnull 			if(*q == '?' || *(q+1) != '?')
242 941e1713 2006-02-15 devnull 				goto emit;
243 941e1713 2006-02-15 devnull 			t = q+2;
244 941e1713 2006-02-15 devnull 			u = strchr(t, '?');
245 941e1713 2006-02-15 devnull 			if(u == nil || *(u+1) != '=')
246 941e1713 2006-02-15 devnull 				goto emit;
247 941e1713 2006-02-15 devnull 			switch(*q){
248 941e1713 2006-02-15 devnull 			case 'q':
249 941e1713 2006-02-15 devnull 			case 'Q':
250 941e1713 2006-02-15 devnull 				*u = 0;
251 6a8add52 2006-02-23 devnull 				v = decode(QuotedPrintableU, t, &len);
252 941e1713 2006-02-15 devnull 				break;
253 941e1713 2006-02-15 devnull 			case 'b':
254 941e1713 2006-02-15 devnull 			case 'B':
255 941e1713 2006-02-15 devnull 				*u = 0;
256 941e1713 2006-02-15 devnull 				v = decode(Base64, t, &len);
257 941e1713 2006-02-15 devnull 				break;
258 941e1713 2006-02-15 devnull 			default:
259 941e1713 2006-02-15 devnull 				goto emit;
260 941e1713 2006-02-15 devnull 			}
261 941e1713 2006-02-15 devnull 			*(q-1) = 0;
262 941e1713 2006-02-15 devnull 			v = tcs(p, v);
263 941e1713 2006-02-15 devnull 			fmtstrcpy(&fmt, v);
264 941e1713 2006-02-15 devnull 			free(v);
265 941e1713 2006-02-15 devnull 			p = u+2;
266 941e1713 2006-02-15 devnull 		}
267 941e1713 2006-02-15 devnull 	emit:
268 941e1713 2006-02-15 devnull 		p += chartorune(&r, p);
269 941e1713 2006-02-15 devnull 		fmtrune(&fmt, r);
270 941e1713 2006-02-15 devnull 	}
271 941e1713 2006-02-15 devnull 	p = fmtstrflush(&fmt);
272 941e1713 2006-02-15 devnull 	if(p == nil)
273 941e1713 2006-02-15 devnull 		sysfatal("out of memory");
274 941e1713 2006-02-15 devnull 	free(s);
275 941e1713 2006-02-15 devnull 	return p;
276 941e1713 2006-02-15 devnull }
277 941e1713 2006-02-15 devnull 
278 941e1713 2006-02-15 devnull #ifdef TEST
279 941e1713 2006-02-15 devnull char *test[] = 
280 941e1713 2006-02-15 devnull {
281 941e1713 2006-02-15 devnull 	"hello world",
282 941e1713 2006-02-15 devnull 	"hello =?iso-8859-1?q?this is some text?=",
283 941e1713 2006-02-15 devnull 	"=?US-ASCII?Q?Keith_Moore?=",
284 941e1713 2006-02-15 devnull 	"=?ISO-8859-1?Q?Keld_J=F8rn_Simonsen?=",
285 941e1713 2006-02-15 devnull 	"=?ISO-8859-1?Q?Andr=E9?= Pirard",
286 941e1713 2006-02-15 devnull 	"=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=",
287 941e1713 2006-02-15 devnull 	"=?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?=",
288 941e1713 2006-02-15 devnull 	"=?ISO-8859-1?Q?Olle_J=E4rnefors?=",
289 941e1713 2006-02-15 devnull 	"=?iso-2022-jp?B?GyRCTTVKISRKP006SiRyS34kPyQ3JEZKcz03JCIkahsoQg==?=",
290 941e1713 2006-02-15 devnull 	"=?UTF-8?B?Ik5pbHMgTy4gU2Vsw6VzZGFsIg==?="
291 941e1713 2006-02-15 devnull };
292 941e1713 2006-02-15 devnull 
293 941e1713 2006-02-15 devnull void
294 941e1713 2006-02-15 devnull threadmain(int argc, char **argv)
295 941e1713 2006-02-15 devnull {
296 941e1713 2006-02-15 devnull 	int i;
297 941e1713 2006-02-15 devnull 	
298 941e1713 2006-02-15 devnull 	for(i=0; i<nelem(test); i++)
299 941e1713 2006-02-15 devnull 		print("%s\n\t%s\n", test[i], unrfc2047(estrdup(test[i])));
300 941e1713 2006-02-15 devnull 	threadexitsall(0);
301 941e1713 2006-02-15 devnull }
302 941e1713 2006-02-15 devnull 
303 941e1713 2006-02-15 devnull #endif