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-1251",	"cp1251",
108 	"koi8-r",			"koi8"
109 };
110 
111 char*
112 tcs(char *charset, char *s)
113 {
114 	static char buf[4096];
115 	int i, n;
116 	int fd[3], p[2], pp[2];
117 	uchar *us;
118 	char *t, *u;
119 	char *argv[4];
120 	Rune r;
121 	
122 	if(s == nil || charset == nil || *s == 0)
123 		return s;
124 
125 	if(cistrcmp(charset, "utf-8") == 0)
126 		return s;
127 	if(cistrcmp(charset, "iso-8859-1") == 0 || cistrcmp(charset, "us-ascii") == 0){
128 latin1:
129 		n = 0;
130 		for(us=(uchar*)s; *us; us++)
131 			n += runelen(*us);
132 		n++;
133 		t = emalloc(n);
134 		for(us=(uchar*)s, u=t; *us; us++){
135 			r = *us;
136 			u += runetochar(u, &r);
137 		}
138 		*u = 0;
139 		free(s);
140 		return t;
141 	}
142 	for(i=0; i<nelem(tcstab); i++)
143 		if(cistrcmp(charset, tcstab[i].mime) == 0)
144 			goto tcs;
145 	goto latin1;
146 
147 tcs:
148 	argv[0] = "tcs";
149 	argv[1] = "-f";
150 	argv[2] = charset;
151 	argv[3] = nil;
152 	
153 	if(pipe(p) < 0 || pipe(pp) < 0)
154 		sysfatal("pipe: %r");
155 	fd[0] = p[0];
156 	fd[1] = pp[0];
157 	fd[2] = dup(2, -1);
158 	if(threadspawnl(fd, "tcs", "tcs", "-f", tcstab[i].tcs, nil) < 0){
159 		close(p[0]);
160 		close(p[1]);
161 		close(pp[0]);
162 		close(pp[1]);
163 		close(fd[2]);
164 		goto latin1;
165 	}
166 	close(p[0]);
167 	close(pp[0]);
168 	write(p[1], s, strlen(s));
169 	close(p[1]);
170 	n = readn(pp[1], buf, sizeof buf-1);
171 	close(pp[1]);
172 	if(n <= 0)
173 		goto latin1;
174 	free(s);
175 	buf[n] = 0;
176 	return estrdup(buf);
177 }
178 
179 char*
180 unrfc2047(char *s)
181 {
182 	char *p, *q, *t, *u, *v;
183 	int len;
184 	Rune r;
185 	Fmt fmt;
186 	
187 	if(s == nil)
188 		return nil;
189 
190 	if(strstr(s, "=?") == nil)
191 		return s;
192 	
193 	fmtstrinit(&fmt);
194 	for(p=s; *p; ){
195 		/* =?charset?e?text?= */
196 		if(*p=='=' && *(p+1)=='?'){
197 			p += 2;
198 			q = strchr(p, '?');
199 			if(q == nil)
200 				goto emit;
201 			q++;
202 			if(*q == '?' || *(q+1) != '?')
203 				goto emit;
204 			t = q+2;
205 			u = strchr(t, '?');
206 			if(u == nil || *(u+1) != '=')
207 				goto emit;
208 			switch(*q){
209 			case 'q':
210 			case 'Q':
211 				*u = 0;
212 				v = decode(QuotedPrintableU, t, &len);
213 				break;
214 			case 'b':
215 			case 'B':
216 				*u = 0;
217 				v = decode(Base64, t, &len);
218 				break;
219 			default:
220 				goto emit;
221 			}
222 			*(q-1) = 0;
223 			v = tcs(p, v);
224 			fmtstrcpy(&fmt, v);
225 			free(v);
226 			p = u+2;
227 		}
228 	emit:
229 		p += chartorune(&r, p);
230 		fmtrune(&fmt, r);
231 	}
232 	p = fmtstrflush(&fmt);
233 	if(p == nil)
234 		sysfatal("out of memory");
235 	free(s);
236 	return p;
237 }
238 
239 #ifdef TEST
240 char *test[] = 
241 {
242 	"hello world",
243 	"hello =?iso-8859-1?q?this is some text?=",
244 	"=?US-ASCII?Q?Keith_Moore?=",
245 	"=?ISO-8859-1?Q?Keld_J=F8rn_Simonsen?=",
246 	"=?ISO-8859-1?Q?Andr=E9?= Pirard",
247 	"=?ISO-8859-1?B?SWYgeW91IGNhbiByZWFkIHRoaXMgeW8=?=",
248 	"=?ISO-8859-2?B?dSB1bmRlcnN0YW5kIHRoZSBleGFtcGxlLg==?=",
249 	"=?ISO-8859-1?Q?Olle_J=E4rnefors?=",
250 	"=?iso-2022-jp?B?GyRCTTVKISRKP006SiRyS34kPyQ3JEZKcz03JCIkahsoQg==?=",
251 	"=?UTF-8?B?Ik5pbHMgTy4gU2Vsw6VzZGFsIg==?="
252 };
253 
254 void
255 threadmain(int argc, char **argv)
256 {
257 	int i;
258 	
259 	for(i=0; i<nelem(test); i++)
260 		print("%s\n\t%s\n", test[i], unrfc2047(estrdup(test[i])));
261 	threadexitsall(0);
262 }
263 
264 #endif