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 	Rune r;
147 	Writeargs *w;
148 
149 	if(s == nil || charset == nil || *s == 0)
150 		return s;
151 
152 	if(cistrcmp(charset, "utf-8") == 0)
153 		return s;
154 	if(cistrcmp(charset, "iso-8859-1") == 0 || cistrcmp(charset, "us-ascii") == 0){
155 latin1:
156 		n = 0;
157 		for(us=(uchar*)s; *us; us++)
158 			n += runelen(*us);
159 		n++;
160 		t = emalloc(n);
161 		for(us=(uchar*)s, u=t; *us; us++){
162 			r = *us;
163 			u += runetochar(u, &r);
164 		}
165 		*u = 0;
166 		free(s);
167 		return t;
168 	}
169 	for(i=0; i<nelem(tcstab); i++)
170 		if(cistrcmp(charset, tcstab[i].mime) == 0)
171 			goto tcs;
172 	goto latin1;
173 
174 tcs:
175 	if(pipe(p) < 0 || pipe(pp) < 0)
176 		sysfatal("pipe: %r");
177 	fd[0] = p[0];
178 	fd[1] = pp[0];
179 	fd[2] = dup(2, -1);
180 	if(threadspawnl(fd, "tcs", "tcs", "-f", tcstab[i].tcs, nil) < 0){
181 		close(p[0]);
182 		close(p[1]);
183 		close(pp[0]);
184 		close(pp[1]);
185 		close(fd[2]);
186 		goto latin1;
187 	}
188 	close(p[0]);
189 	close(pp[0]);
190 
191 	nbuf = UTFmax*strlen(s)+100;	/* just a guess at worst case */
192 	buf = emalloc(nbuf);
193 
194 	w = emalloc(sizeof *w);
195 	w->fd = p[1];
196 	w->s = estrdup(s);
197 	proccreate(twriter, w, STACK);
198 
199 	n = readn(pp[1], buf, nbuf-1);
200 	close(pp[1]);
201 	if(n <= 0){
202 		free(buf);
203 		goto latin1;
204 	}
205 	buf[n] = 0;
206 	free(s);
207 	s = estrdup(buf);
208 	free(buf);
209 	return s;
210 }
211 
212 char*
213 unrfc2047(char *s)
214 {
215 	char *p, *q, *t, *u, *v;
216 	int len;
217 	Rune r;
218 	Fmt fmt;
219 
220 	if(s == nil)
221 		return nil;
222 
223 	if(strstr(s, "=?") == nil)
224 		return s;
225 
226 	fmtstrinit(&fmt);
227 	for(p=s; *p; ){
228 		/* =?charset?e?text?= */
229 		if(*p=='=' && *(p+1)=='?'){
230 			p += 2;
231 			q = strchr(p, '?');
232 			if(q == nil)
233 				goto emit;
234 			q++;
235 			if(*q == '?' || *(q+1) != '?')
236 				goto emit;
237 			t = q+2;
238 			u = strchr(t, '?');
239 			if(u == nil || *(u+1) != '=')
240 				goto emit;
241 			switch(*q){
242 			case 'q':
243 			case 'Q':
244 				*u = 0;
245 				v = decode(QuotedPrintableU, t, &len);
246 				break;
247 			case 'b':
248 			case 'B':
249 				*u = 0;
250 				v = decode(Base64, t, &len);
251 				break;
252 			default:
253 				goto emit;
254 			}
255 			*(q-1) = 0;
256 			v = tcs(p, v);
257 			fmtstrcpy(&fmt, v);
258 			free(v);
259 			p = u+2;
260 		}
261 	emit:
262 		p += chartorune(&r, p);
263 		fmtrune(&fmt, r);
264 	}
265 	p = fmtstrflush(&fmt);
266 	if(p == nil)
267 		sysfatal("out of memory");
268 	free(s);
269 	return p;
270 }
271 
272 #ifdef TEST
273 char *test[] =
274 {
275 	"hello world",
276 	"hello =?iso-8859-1?q?this is some text?=",
277 	"=?US-ASCII?Q?Keith_Moore?=",
278 	"=?ISO-8859-1?Q?Keld_J=F8rn_Simonsen?=",
279 	"=?ISO-8859-1?Q?Andr=E9?= Pirard",
280 	"=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=",
281 	"=?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?=",
282 	"=?ISO-8859-1?Q?Olle_J=E4rnefors?=",
283 	"=?iso-2022-jp?B?GyRCTTVKISRKP006SiRyS34kPyQ3JEZKcz03JCIkahsoQg==?=",
284 	"=?UTF-8?B?Ik5pbHMgTy4gU2Vsw6VzZGFsIg==?="
285 };
286 
287 void
288 threadmain(int argc, char **argv)
289 {
290 	int i;
291 
292 	for(i=0; i<nelem(test); i++)
293 		print("%s\n\t%s\n", test[i], unrfc2047(estrdup(test[i])));
294 	threadexitsall(0);
295 }
296 
297 #endif