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