Commit | Line | Data |
---|---|---|
5b708e0c RK |
1 | /* |
2 | * This file is part of DisOrder. | |
3 | * Copyright (C) 2004, 2005, 2007, 2008 Richard Kettlewell | |
4 | * | |
5 | * This program is free software; you can redistribute it and/or modify | |
6 | * it under the terms of the GNU General Public License as published by | |
7 | * the Free Software Foundation; either version 2 of the License, or | |
8 | * (at your option) any later version. | |
9 | * | |
10 | * This program is distributed in the hope that it will be useful, but | |
11 | * WITHOUT ANY WARRANTY; without even the implied warranty of | |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
13 | * General Public License for more details. | |
14 | * | |
15 | * You should have received a copy of the GNU General Public License | |
16 | * along with this program; if not, write to the Free Software | |
17 | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 | |
18 | * USA | |
19 | */ | |
20 | /** @file lib/cgi.c | |
21 | * @brief CGI tools | |
22 | */ | |
23 | ||
24 | #include <config.h> | |
25 | #include "types.h" | |
26 | ||
27 | #include <stdlib.h> | |
28 | #include <string.h> | |
29 | #include <assert.h> | |
30 | #include <unistd.h> | |
31 | #include <errno.h> | |
32 | #include <stdio.h> | |
33 | ||
34 | #include "cgi.h" | |
35 | #include "mem.h" | |
36 | #include "log.h" | |
37 | #include "vector.h" | |
38 | #include "hash.h" | |
39 | #include "kvp.h" | |
40 | #include "mime.h" | |
41 | #include "unicode.h" | |
42 | #include "sink.h" | |
43 | ||
44 | /** @brief Hash of arguments */ | |
45 | static hash *cgi_args; | |
46 | ||
47 | /** @brief Get CGI arguments from a GET request's query string */ | |
48 | static struct kvp *cgi__init_get(void) { | |
49 | const char *q; | |
50 | ||
51 | if((q = getenv("QUERY_STRING"))) | |
52 | return kvp_urldecode(q, strlen(q)); | |
53 | error(0, "QUERY_STRING not set, assuming empty"); | |
54 | return NULL; | |
55 | } | |
56 | ||
57 | /** @brief Read the HTTP request body */ | |
58 | static void cgi__input(char **ptrp, size_t *np) { | |
59 | const char *cl; | |
60 | char *q; | |
61 | size_t n, m = 0; | |
62 | int r; | |
63 | ||
64 | if(!(cl = getenv("CONTENT_LENGTH"))) | |
65 | fatal(0, "CONTENT_LENGTH not set"); | |
66 | n = atol(cl); | |
67 | /* We check for overflow and also limit the input to 16MB. Lower | |
68 | * would probably do. */ | |
69 | if(!(n+1) || n > 16 * 1024 * 1024) | |
70 | fatal(0, "input is much too large"); | |
71 | q = xmalloc_noptr(n + 1); | |
72 | while(m < n) { | |
73 | r = read(0, q + m, n - m); | |
74 | if(r > 0) | |
75 | m += r; | |
76 | else if(r == 0) | |
77 | fatal(0, "unexpected end of file reading request body"); | |
78 | else switch(errno) { | |
79 | case EINTR: break; | |
80 | default: fatal(errno, "error reading request body"); | |
81 | } | |
82 | } | |
83 | if(memchr(q, 0, n)) | |
84 | fatal(0, "null character in request body"); | |
85 | q[n + 1] = 0; | |
86 | *ptrp = q; | |
87 | if(np) | |
88 | *np = n; | |
89 | } | |
90 | ||
91 | /** @brief Called for each part header field (see cgi__part_callback()) */ | |
92 | static int cgi__field_callback(const char *name, const char *value, | |
93 | void *u) { | |
94 | char *disposition, *pname, *pvalue; | |
95 | char **namep = u; | |
96 | ||
97 | if(!strcmp(name, "content-disposition")) { | |
98 | if(mime_rfc2388_content_disposition(value, | |
99 | &disposition, | |
100 | &pname, | |
101 | &pvalue)) | |
102 | fatal(0, "error parsing Content-Disposition field"); | |
103 | if(!strcmp(disposition, "form-data") | |
104 | && pname | |
105 | && !strcmp(pname, "name")) { | |
106 | if(*namep) | |
107 | fatal(0, "duplicate Content-Disposition field"); | |
108 | *namep = pvalue; | |
109 | } | |
110 | } | |
111 | return 0; | |
112 | } | |
113 | ||
114 | /** @brief Called for each part (see cgi__init_multipart()) */ | |
115 | static int cgi__part_callback(const char *s, | |
116 | void *u) { | |
117 | char *name = 0; | |
118 | struct kvp *k, **head = u; | |
119 | ||
120 | if(!(s = mime_parse(s, cgi__field_callback, &name))) | |
121 | fatal(0, "error parsing part header"); | |
122 | if(!name) | |
123 | fatal(0, "no name found"); | |
124 | k = xmalloc(sizeof *k); | |
125 | k->next = *head; | |
126 | k->name = name; | |
127 | k->value = s; | |
128 | *head = k; | |
129 | return 0; | |
130 | } | |
131 | ||
132 | /** @brief Initialize CGI arguments from a multipart/form-data request body */ | |
133 | static struct kvp *cgi__init_multipart(const char *boundary) { | |
134 | char *q; | |
135 | struct kvp *head = 0; | |
136 | ||
137 | cgi__input(&q, 0); | |
138 | if(mime_multipart(q, cgi__part_callback, boundary, &head)) | |
139 | fatal(0, "invalid multipart object"); | |
140 | return head; | |
141 | } | |
142 | ||
143 | /** @brief Initialize CGI arguments from a POST request */ | |
144 | static struct kvp *cgi__init_post(void) { | |
145 | const char *ct, *boundary; | |
146 | char *q, *type; | |
147 | size_t n; | |
148 | struct kvp *k; | |
149 | ||
150 | if(!(ct = getenv("CONTENT_TYPE"))) | |
151 | ct = "application/x-www-form-urlencoded"; | |
152 | if(mime_content_type(ct, &type, &k)) | |
153 | fatal(0, "invalid content type '%s'", ct); | |
154 | if(!strcmp(type, "application/x-www-form-urlencoded")) { | |
155 | cgi__input(&q, &n); | |
156 | return kvp_urldecode(q, n); | |
157 | } | |
158 | if(!strcmp(type, "multipart/form-data")) { | |
159 | if(!(boundary = kvp_get(k, "boundary"))) | |
160 | fatal(0, "no boundary parameter found"); | |
161 | return cgi__init_multipart(boundary); | |
162 | } | |
163 | fatal(0, "unrecognized content type '%s'", type); | |
164 | } | |
165 | ||
166 | /** @brief Initialize CGI arguments | |
167 | * | |
168 | * Must be called before other cgi_ functions are used. | |
169 | * | |
170 | * This function can be called more than once, in which case it | |
171 | * revisits the environment and (perhaps) standard input. This is | |
172 | * only intended to be used for testing, actual CGI applications | |
173 | * should call it exactly once. | |
174 | */ | |
175 | void cgi_init(void) { | |
176 | const char *p; | |
177 | struct kvp *k; | |
178 | ||
179 | cgi_args = hash_new(sizeof (char *)); | |
180 | if(!(p = getenv("REQUEST_METHOD"))) | |
181 | error(0, "REQUEST_METHOD not set, assuming GET"); | |
182 | if(!p || !strcmp(p, "GET")) | |
183 | k = cgi__init_get(); | |
184 | else if(!strcmp(p, "POST")) | |
185 | k = cgi__init_post(); | |
186 | else | |
187 | fatal(0, "unknown request method %s", p); | |
188 | /* Validate the arguments and put them in a hash */ | |
189 | for(; k; k = k->next) { | |
190 | if(!utf8_valid(k->name, strlen(k->name)) | |
191 | || !utf8_valid(k->value, strlen(k->value))) | |
192 | error(0, "invalid UTF-8 sequence in cgi argument %s", k->name); | |
193 | else | |
194 | hash_add(cgi_args, k->name, &k->value, HASH_INSERT_OR_REPLACE); | |
195 | /* We just drop bogus arguments. */ | |
196 | } | |
197 | } | |
198 | ||
199 | /** @brief Get a CGI argument by name | |
200 | * | |
201 | * cgi_init() must be called first. Names and values are all valid | |
202 | * UTF-8 strings (and this is enforced at initialization time). | |
203 | */ | |
204 | const char *cgi_get(const char *name) { | |
205 | const char **v = hash_find(cgi_args, name); | |
206 | ||
207 | return v ? *v : NULL; | |
208 | } | |
209 | ||
210 | /** @brief Add SGML-style quoting | |
211 | * @param src String to quote (UTF-8) | |
212 | * @return Quoted string | |
213 | * | |
214 | * Quotes characters for insertion into HTML output. Anything that is | |
215 | * not a printable ASCII character will be converted to a numeric | |
216 | * character references, as will '"', '&', '<' and '>' (since those | |
217 | * have special meanings). | |
218 | * | |
219 | * Quoting everything down to ASCII means we don't care what the | |
220 | * content encoding really is (as long as it's not anything insane | |
221 | * like EBCDIC). | |
222 | */ | |
223 | char *cgi_sgmlquote(const char *src) { | |
224 | uint32_t *ucs, c; | |
225 | int n; | |
226 | struct dynstr d[1]; | |
227 | struct sink *s; | |
228 | ||
229 | if(!(ucs = utf8_to_utf32(src, strlen(src), 0))) | |
230 | exit(1); | |
231 | dynstr_init(d); | |
232 | s = sink_dynstr(d); | |
233 | n = 1; | |
234 | /* format the string */ | |
235 | while((c = *ucs++)) { | |
236 | switch(c) { | |
237 | default: | |
238 | if(c > 126 || c < 32) { | |
239 | case '"': | |
240 | case '&': | |
241 | case '<': | |
242 | case '>': | |
243 | /* For simplicity we always use numeric character references | |
244 | * even if a named reference is available. */ | |
245 | sink_printf(s, "&#%"PRIu32";", c); | |
246 | break; | |
247 | } else | |
248 | sink_writec(s, (char)c); | |
249 | } | |
250 | } | |
251 | dynstr_terminate(d); | |
252 | return d->vec; | |
253 | } | |
254 | ||
255 | /** @brief Write a CGI attribute | |
256 | * @param output Where to send output | |
257 | * @param name Attribute name | |
258 | * @param value Attribute value | |
259 | */ | |
260 | void cgi_attr(struct sink *output, const char *name, const char *value) { | |
261 | /* Try to avoid needless quoting */ | |
262 | if(!value[strspn(value, "abcdefghijklmnopqrstuvwxyz" | |
263 | "ABCDEFGHIJKLMNOPQRSTUVWXYZ" | |
264 | "0123456789")]) | |
265 | sink_printf(output, "%s=%s", name, value); | |
266 | else | |
267 | sink_printf(output, "%s=\"%s\"", name, cgi_sgmlquote(value)); | |
268 | } | |
269 | ||
270 | /** @brief Write an open tag | |
271 | * @param output Where to send output | |
272 | * @param name Element name | |
273 | * @param ... Attribute name/value pairs | |
274 | * | |
275 | * The name/value pair list is terminated by a single (char *)0. | |
276 | */ | |
277 | void cgi_opentag(struct sink *output, const char *name, ...) { | |
278 | va_list ap; | |
279 | const char *n, *v; | |
280 | ||
281 | sink_printf(output, "<%s", name); | |
282 | va_start(ap, name); | |
283 | while((n = va_arg(ap, const char *))) { | |
284 | sink_printf(output, " "); | |
285 | v = va_arg(ap, const char *); | |
286 | if(v) | |
287 | cgi_attr(output, n, v); | |
288 | else | |
289 | sink_printf(output, n); | |
290 | } | |
291 | va_end(ap); | |
292 | sink_printf(output, ">"); | |
293 | } | |
294 | ||
295 | /** @brief Write a close tag | |
296 | * @param output Where to send output | |
297 | * @param name Element name | |
298 | */ | |
299 | void cgi_closetag(struct sink *output, const char *name) { | |
300 | sink_printf(output, "</%s>", name); | |
301 | } | |
302 | ||
303 | /** @brief Construct a URL | |
304 | * @param url Base URL | |
305 | * @param ... Name/value pairs for constructed query string | |
306 | * @return Constructed URL | |
307 | * | |
308 | * The name/value pair list is terminated by a single (char *)0. | |
309 | */ | |
310 | char *cgi_makeurl(const char *url, ...) { | |
311 | va_list ap; | |
312 | struct kvp *kvp, *k, **kk = &kvp; | |
313 | struct dynstr d; | |
314 | const char *n, *v; | |
315 | ||
316 | dynstr_init(&d); | |
317 | dynstr_append_string(&d, url); | |
318 | va_start(ap, url); | |
319 | while((n = va_arg(ap, const char *))) { | |
320 | v = va_arg(ap, const char *); | |
321 | *kk = k = xmalloc(sizeof *k); | |
322 | kk = &k->next; | |
323 | k->name = n; | |
324 | k->value = v; | |
325 | } | |
326 | va_end(ap); | |
327 | *kk = 0; | |
328 | if(kvp) { | |
329 | dynstr_append(&d, '?'); | |
330 | dynstr_append_string(&d, kvp_urlencode(kvp, 0)); | |
331 | } | |
332 | dynstr_terminate(&d); | |
333 | return d.vec; | |
334 | } | |
335 | ||
9faa7a88 RK |
336 | /** @brief Construct a URL from current parameters |
337 | * @param url Base URL | |
338 | * @return Constructed URL | |
339 | */ | |
340 | char *cgi_thisurl(const char *url) { | |
341 | struct dynstr d[1]; | |
342 | char **keys = hash_keys(cgi_args); | |
343 | int n; | |
344 | ||
345 | dynstr_init(d); | |
346 | dynstr_append_string(d, url); | |
347 | if(*keys) { | |
348 | dynstr_append(d, '?'); | |
349 | for(n = 0; keys[n]; ++n) { | |
350 | dynstr_append_string(d, urlencodestring(keys[n])); | |
351 | dynstr_append(d, '='); | |
352 | dynstr_append_string(d, cgi_get(keys[n])); | |
353 | } | |
354 | } | |
355 | dynstr_terminate(d); | |
356 | return d->vec; | |
357 | } | |
358 | ||
5b708e0c RK |
359 | /* |
360 | Local Variables: | |
361 | c-basic-offset:2 | |
362 | comment-column:40 | |
363 | fill-column:79 | |
364 | indent-tabs-mode:nil | |
365 | End: | |
366 | */ |