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