blob: 8e7d455f608c4a08d67a5f509035c4afbaffa94a
1 | /* vi: set sw=4 ts=4: */ |
2 | /* |
3 | * reformime: parse MIME-encoded message |
4 | * |
5 | * Copyright (C) 2008 by Vladimir Dronnikov <dronnikov@gmail.com> |
6 | * |
7 | * Licensed under GPLv2, see file LICENSE in this source tree. |
8 | */ |
9 | |
10 | //kbuild:lib-$(CONFIG_REFORMIME) += reformime.o mail.o |
11 | |
12 | #include "libbb.h" |
13 | #include "mail.h" |
14 | |
15 | #if 0 |
16 | # define dbg_error_msg(...) bb_error_msg(__VA_ARGS__) |
17 | #else |
18 | # define dbg_error_msg(...) ((void)0) |
19 | #endif |
20 | |
21 | static const char *find_token(const char *const string_array[], const char *key, const char *defvalue) |
22 | { |
23 | const char *r = NULL; |
24 | int i; |
25 | for (i = 0; string_array[i] != NULL; i++) { |
26 | if (strcasecmp(string_array[i], key) == 0) { |
27 | r = (char *)string_array[i+1]; |
28 | break; |
29 | } |
30 | } |
31 | return (r) ? r : defvalue; |
32 | } |
33 | |
34 | static const char *xfind_token(const char *const string_array[], const char *key) |
35 | { |
36 | const char *r = find_token(string_array, key, NULL); |
37 | if (r) |
38 | return r; |
39 | bb_error_msg_and_die("not found: '%s'", key); |
40 | } |
41 | |
42 | enum { |
43 | OPT_x = 1 << 0, |
44 | OPT_X = 1 << 1, |
45 | #if ENABLE_FEATURE_REFORMIME_COMPAT |
46 | OPT_d = 1 << 2, |
47 | OPT_e = 1 << 3, |
48 | OPT_i = 1 << 4, |
49 | OPT_s = 1 << 5, |
50 | OPT_r = 1 << 6, |
51 | OPT_c = 1 << 7, |
52 | OPT_m = 1 << 8, |
53 | OPT_h = 1 << 9, |
54 | OPT_o = 1 << 10, |
55 | OPT_O = 1 << 11, |
56 | #endif |
57 | }; |
58 | |
59 | static int parse(const char *boundary, char **argv) |
60 | { |
61 | int boundary_len = strlen(boundary); |
62 | char uniq[sizeof("%%llu.%u") + sizeof(int)*3]; |
63 | |
64 | dbg_error_msg("BOUNDARY[%s]", boundary); |
65 | |
66 | // prepare unique string pattern |
67 | sprintf(uniq, "%%llu.%u", (unsigned)getpid()); |
68 | dbg_error_msg("UNIQ[%s]", uniq); |
69 | |
70 | while (1) { |
71 | char *header; |
72 | const char *tokens[32]; /* 32 is enough */ |
73 | const char *type; |
74 | |
75 | /* Read the header (everything up to two \n) */ |
76 | { |
77 | unsigned header_idx = 0; |
78 | int last_ch = 0; |
79 | header = NULL; |
80 | while (1) { |
81 | int ch = fgetc(stdin); |
82 | if (ch == '\r') /* Support both line endings */ |
83 | continue; |
84 | if (ch == EOF) |
85 | break; |
86 | if (ch == '\n' && last_ch == ch) |
87 | break; |
88 | if (!(header_idx & 0xff)) |
89 | header = xrealloc(header, header_idx + 0x101); |
90 | header[header_idx++] = last_ch = ch; |
91 | } |
92 | if (!header) { |
93 | dbg_error_msg("EOF"); |
94 | break; |
95 | } |
96 | header[header_idx] = '\0'; |
97 | dbg_error_msg("H:'%s'", p); |
98 | } |
99 | |
100 | /* Split to tokens */ |
101 | { |
102 | char *s, *p; |
103 | unsigned ntokens; |
104 | const char *delims = ";=\" \t\n"; |
105 | |
106 | /* Skip to last Content-Type: */ |
107 | s = p = header; |
108 | while ((p = strchr(p, '\n')) != NULL) { |
109 | p++; |
110 | if (strncasecmp(p, "Content-Type:", sizeof("Content-Type:")-1) == 0) |
111 | s = p; |
112 | } |
113 | dbg_error_msg("L:'%s'", p); |
114 | ntokens = 0; |
115 | s = strtok(s, delims); |
116 | while (s) { |
117 | tokens[ntokens] = s; |
118 | if (ntokens < ARRAY_SIZE(tokens) - 1) |
119 | ntokens++; |
120 | dbg_error_msg("L[%d]='%s'", ntokens, s); |
121 | s = strtok(NULL, delims); |
122 | } |
123 | tokens[ntokens] = NULL; |
124 | dbg_error_msg("EMPTYLINE, ntokens:%d", ntokens); |
125 | if (ntokens == 0) |
126 | break; |
127 | } |
128 | |
129 | /* Is it multipart? */ |
130 | type = find_token(tokens, "Content-Type:", "text/plain"); |
131 | dbg_error_msg("TYPE:'%s'", type); |
132 | if (0 == strncasecmp(type, "multipart/", 10)) { |
133 | /* Yes, recurse */ |
134 | if (strcasecmp(type + 10, "mixed") != 0) |
135 | bb_error_msg_and_die("no support of content type '%s'", type); |
136 | parse(xfind_token(tokens, "boundary"), argv); |
137 | |
138 | } else { |
139 | /* No, process one non-multipart section */ |
140 | char *end; |
141 | pid_t pid = pid; |
142 | FILE *fp; |
143 | |
144 | const char *charset = find_token(tokens, "charset", CONFIG_FEATURE_MIME_CHARSET); |
145 | const char *encoding = find_token(tokens, "Content-Transfer-Encoding:", "7bit"); |
146 | |
147 | /* Compose target filename */ |
148 | char *filename = (char *)find_token(tokens, "filename", NULL); |
149 | if (!filename) |
150 | filename = xasprintf(uniq, monotonic_us()); |
151 | else |
152 | filename = bb_get_last_path_component_strip(xstrdup(filename)); |
153 | |
154 | if (opts & OPT_X) { |
155 | int fd[2]; |
156 | |
157 | /* start external helper */ |
158 | xpipe(fd); |
159 | pid = vfork(); |
160 | if (0 == pid) { |
161 | /* child reads from fd[0] */ |
162 | close(fd[1]); |
163 | xmove_fd(fd[0], STDIN_FILENO); |
164 | xsetenv("CONTENT_TYPE", type); |
165 | xsetenv("CHARSET", charset); |
166 | xsetenv("ENCODING", encoding); |
167 | xsetenv("FILENAME", filename); |
168 | BB_EXECVP_or_die(argv); |
169 | } |
170 | /* parent will write to fd[1] */ |
171 | close(fd[0]); |
172 | fp = xfdopen_for_write(fd[1]); |
173 | signal(SIGPIPE, SIG_IGN); |
174 | } else { |
175 | /* write to file */ |
176 | char *fname = xasprintf("%s%s", *argv, filename); |
177 | fp = xfopen_for_write(fname); |
178 | free(fname); |
179 | } |
180 | free(filename); |
181 | |
182 | /* write to fp */ |
183 | end = NULL; |
184 | if (0 == strcasecmp(encoding, "base64")) { |
185 | read_base64(stdin, fp, '-'); |
186 | } else |
187 | if (0 != strcasecmp(encoding, "7bit") |
188 | && 0 != strcasecmp(encoding, "8bit") |
189 | ) { |
190 | /* quoted-printable, binary, user-defined are unsupported so far */ |
191 | bb_error_msg_and_die("encoding '%s' not supported", encoding); |
192 | } else { |
193 | /* plain 7bit or 8bit */ |
194 | while ((end = xmalloc_fgets(stdin)) != NULL) { |
195 | if ('-' == end[0] |
196 | && '-' == end[1] |
197 | && strncmp(end + 2, boundary, boundary_len) == 0 |
198 | ) { |
199 | break; |
200 | } |
201 | fputs(end, fp); |
202 | } |
203 | } |
204 | fclose(fp); |
205 | |
206 | /* Wait for child */ |
207 | if (opts & OPT_X) { |
208 | int rc; |
209 | signal(SIGPIPE, SIG_DFL); |
210 | rc = (wait4pid(pid) & 0xff); |
211 | if (rc != 0) |
212 | return rc + 20; |
213 | } |
214 | |
215 | /* Multipart ended? */ |
216 | if (end && '-' == end[2 + boundary_len] && '-' == end[2 + boundary_len + 1]) { |
217 | dbg_error_msg("FINISHED MPART:'%s'", end); |
218 | break; |
219 | } |
220 | dbg_error_msg("FINISHED:'%s'", end); |
221 | free(end); |
222 | } /* end of "handle one non-multipart block" */ |
223 | |
224 | free(header); |
225 | } /* while (1) */ |
226 | |
227 | dbg_error_msg("ENDPARSE[%s]", boundary); |
228 | |
229 | return EXIT_SUCCESS; |
230 | } |
231 | |
232 | //usage:#define reformime_trivial_usage |
233 | //usage: "[OPTIONS]" |
234 | //usage:#define reformime_full_usage "\n\n" |
235 | //usage: "Parse MIME-encoded message on stdin\n" |
236 | //usage: "\n -x PREFIX Extract content of MIME sections to files" |
237 | //usage: "\n -X PROG ARGS Filter content of MIME sections through PROG" |
238 | //usage: "\n Must be the last option" |
239 | //usage: "\n" |
240 | //usage: "\nOther options are silently ignored" |
241 | |
242 | /* |
243 | Usage: reformime [options] |
244 | -d - parse a delivery status notification. |
245 | -e - extract contents of MIME section. |
246 | -x - extract MIME section to a file. |
247 | -X - pipe MIME section to a program. |
248 | -i - show MIME info. |
249 | -s n.n.n.n - specify MIME section. |
250 | -r - rewrite message, filling in missing MIME headers. |
251 | -r7 - also convert 8bit/raw encoding to quoted-printable, if possible. |
252 | -r8 - also convert quoted-printable encoding to 8bit, if possible. |
253 | -c charset - default charset for rewriting, -o, and -O. |
254 | -m [file] [file]... - create a MIME message digest. |
255 | -h "header" - decode RFC 2047-encoded header. |
256 | -o "header" - encode unstructured header using RFC 2047. |
257 | -O "header" - encode address list header using RFC 2047. |
258 | */ |
259 | |
260 | int reformime_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE; |
261 | int reformime_main(int argc UNUSED_PARAM, char **argv) |
262 | { |
263 | const char *opt_prefix = ""; |
264 | |
265 | INIT_G(); |
266 | |
267 | // parse options |
268 | // N.B. only -x and -X are supported so far |
269 | opt_complementary = "x--X:X--x" IF_FEATURE_REFORMIME_COMPAT(":m::"); |
270 | opts = getopt32(argv, |
271 | "x:X" IF_FEATURE_REFORMIME_COMPAT("deis:r:c:m:h:o:O:"), |
272 | &opt_prefix |
273 | IF_FEATURE_REFORMIME_COMPAT(, NULL, NULL, &G.opt_charset, NULL, NULL, NULL, NULL) |
274 | ); |
275 | argv += optind; |
276 | |
277 | return parse("", (opts & OPT_X) ? argv : (char **)&opt_prefix); |
278 | } |
279 |