| 1 | /* |
| 2 | ashd - A Sane HTTP Daemon |
| 3 | Copyright (C) 2008 Fredrik Tolf <fredrik@dolda2000.com> |
| 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 | |
| 19 | #include <stdlib.h> |
| 20 | #include <stdio.h> |
| 21 | #include <string.h> |
| 22 | #include <unistd.h> |
| 23 | #include <signal.h> |
| 24 | #include <errno.h> |
| 25 | #include <ctype.h> |
| 26 | #include <regex.h> |
| 27 | |
| 28 | #ifdef HAVE_CONFIG_H |
| 29 | #include <config.h> |
| 30 | #endif |
| 31 | #include <utils.h> |
| 32 | #include <log.h> |
| 33 | #include <req.h> |
| 34 | #include <proc.h> |
| 35 | #include <resp.h> |
| 36 | #include <cf.h> |
| 37 | |
| 38 | #define PAT_REST 0 |
| 39 | #define PAT_URL 1 |
| 40 | #define PAT_METHOD 2 |
| 41 | #define PAT_HEADER 3 |
| 42 | #define PAT_ALL 4 |
| 43 | #define PAT_DEFAULT 5 |
| 44 | |
| 45 | #define PATFL_MSS 1 |
| 46 | |
| 47 | struct config { |
| 48 | struct child *children; |
| 49 | struct pattern *patterns; |
| 50 | }; |
| 51 | |
| 52 | struct rule { |
| 53 | int type; |
| 54 | int fl; |
| 55 | char *header; |
| 56 | regex_t *pattern; |
| 57 | }; |
| 58 | |
| 59 | struct pattern { |
| 60 | struct pattern *next; |
| 61 | char *childnm; |
| 62 | struct rule **rules; |
| 63 | char *restpat; |
| 64 | }; |
| 65 | |
| 66 | static struct config *config; |
| 67 | |
| 68 | static void freepattern(struct pattern *pat) |
| 69 | { |
| 70 | struct rule **rule; |
| 71 | |
| 72 | for(rule = pat->rules; *rule; rule++) { |
| 73 | if((*rule)->header != NULL) |
| 74 | free((*rule)->header); |
| 75 | if((*rule)->pattern != NULL) { |
| 76 | regfree((*rule)->pattern); |
| 77 | free((*rule)->pattern); |
| 78 | } |
| 79 | free(*rule); |
| 80 | } |
| 81 | if(pat->childnm != NULL) |
| 82 | free(pat->childnm); |
| 83 | free(pat); |
| 84 | } |
| 85 | |
| 86 | static struct child *getchild(struct config *cf, char *name) |
| 87 | { |
| 88 | struct child *ch; |
| 89 | |
| 90 | for(ch = cf->children; ch; ch = ch->next) { |
| 91 | if(!strcmp(ch->name, name)) |
| 92 | break; |
| 93 | } |
| 94 | return(ch); |
| 95 | } |
| 96 | |
| 97 | static struct rule *newrule(struct pattern *pat) |
| 98 | { |
| 99 | int i; |
| 100 | struct rule *rule; |
| 101 | |
| 102 | for(i = 0; pat->rules[i]; i++); |
| 103 | pat->rules = srealloc(pat->rules, sizeof(*pat->rules) * (i + 2)); |
| 104 | rule = pat->rules[i] = szmalloc(sizeof(*rule)); |
| 105 | pat->rules[i + 1] = NULL; |
| 106 | return(rule); |
| 107 | } |
| 108 | |
| 109 | static struct pattern *newpattern(void) |
| 110 | { |
| 111 | struct pattern *pat; |
| 112 | |
| 113 | omalloc(pat); |
| 114 | pat->rules = szmalloc(sizeof(*pat->rules)); |
| 115 | return(pat); |
| 116 | } |
| 117 | |
| 118 | static regex_t *regalloc(char *regex, int flags) |
| 119 | { |
| 120 | regex_t *ret; |
| 121 | int res; |
| 122 | char errbuf[256]; |
| 123 | |
| 124 | omalloc(ret); |
| 125 | if((res = regcomp(ret, regex, flags | REG_EXTENDED)) != 0) { |
| 126 | regerror(res, ret, errbuf, sizeof(errbuf)); |
| 127 | flog(LOG_WARNING, "%s: %s", regex, errbuf); |
| 128 | free(ret); |
| 129 | return(NULL); |
| 130 | } |
| 131 | return(ret); |
| 132 | } |
| 133 | |
| 134 | static struct pattern *parsepattern(struct cfstate *s) |
| 135 | { |
| 136 | struct pattern *pat; |
| 137 | int sl; |
| 138 | struct rule *rule; |
| 139 | regex_t *regex; |
| 140 | int rxfl; |
| 141 | |
| 142 | if(!strcmp(s->argv[0], "match")) { |
| 143 | s->expstart = 1; |
| 144 | pat = newpattern(); |
| 145 | } else { |
| 146 | return(NULL); |
| 147 | } |
| 148 | |
| 149 | sl = s->lno; |
| 150 | while(1) { |
| 151 | getcfline(s); |
| 152 | if(!strcmp(s->argv[0], "point") || |
| 153 | !strcmp(s->argv[0], "url") || |
| 154 | !strcmp(s->argv[0], "method")) { |
| 155 | if(s->argc < 2) { |
| 156 | flog(LOG_WARNING, "%s:%i: missing pattern for `%s' match", s->file, s->lno, s->argv[0]); |
| 157 | continue; |
| 158 | } |
| 159 | if(s->argc >= 3) { |
| 160 | if(strchr(s->argv[2], 'i')) |
| 161 | rxfl |= REG_ICASE; |
| 162 | } |
| 163 | if((regex = regalloc(s->argv[1], rxfl)) == NULL) { |
| 164 | flog(LOG_WARNING, "%s:%i: invalid regex for `%s' match", s->file, s->lno, s->argv[0]); |
| 165 | continue; |
| 166 | } |
| 167 | rule = newrule(pat); |
| 168 | if(!strcmp(s->argv[0], "point")) |
| 169 | rule->type = PAT_REST; |
| 170 | else if(!strcmp(s->argv[0], "url")) |
| 171 | rule->type = PAT_URL; |
| 172 | else if(!strcmp(s->argv[0], "method")) |
| 173 | rule->type = PAT_METHOD; |
| 174 | rule->pattern = regex; |
| 175 | if(s->argc >= 3) { |
| 176 | if(strchr(s->argv[2], 's')) |
| 177 | rule->fl |= PATFL_MSS; |
| 178 | } |
| 179 | } else if(!strcmp(s->argv[0], "header")) { |
| 180 | if(s->argc < 3) { |
| 181 | flog(LOG_WARNING, "%s:%i: missing header name or pattern for `header' match", s->file, s->lno); |
| 182 | continue; |
| 183 | } |
| 184 | if(s->argc >= 4) { |
| 185 | if(strchr(s->argv[3], 'i')) |
| 186 | rxfl |= REG_ICASE; |
| 187 | } |
| 188 | if((regex = regalloc(s->argv[2], rxfl)) == NULL) { |
| 189 | flog(LOG_WARNING, "%s:%i: invalid regex for `header' match", s->file, s->lno); |
| 190 | continue; |
| 191 | } |
| 192 | rule = newrule(pat); |
| 193 | rule->type = PAT_HEADER; |
| 194 | rule->header = sstrdup(s->argv[1]); |
| 195 | rule->pattern = regex; |
| 196 | if(s->argc >= 4) { |
| 197 | if(strchr(s->argv[3], 's')) |
| 198 | rule->fl |= PATFL_MSS; |
| 199 | } |
| 200 | } else if(!strcmp(s->argv[0], "all")) { |
| 201 | newrule(pat)->type = PAT_ALL; |
| 202 | } else if(!strcmp(s->argv[0], "default")) { |
| 203 | newrule(pat)->type = PAT_DEFAULT; |
| 204 | } else if(!strcmp(s->argv[0], "handler")) { |
| 205 | if(s->argc < 2) { |
| 206 | flog(LOG_WARNING, "%s:%i: missing child name for `handler' directive", s->file, s->lno); |
| 207 | continue; |
| 208 | } |
| 209 | if(pat->childnm != NULL) |
| 210 | free(pat->childnm); |
| 211 | pat->childnm = sstrdup(s->argv[1]); |
| 212 | } else if(!strcmp(s->argv[0], "restpat")) { |
| 213 | if(s->argc < 2) { |
| 214 | flog(LOG_WARNING, "%s:%i: missing pattern for `restpat' directive", s->file, s->lno); |
| 215 | continue; |
| 216 | } |
| 217 | if(pat->restpat != NULL) |
| 218 | free(pat->restpat); |
| 219 | pat->restpat = sstrdup(s->argv[1]); |
| 220 | } else if(!strcmp(s->argv[0], "end") || !strcmp(s->argv[0], "eof")) { |
| 221 | break; |
| 222 | } else { |
| 223 | flog(LOG_WARNING, "%s:%i: unknown directive `%s' in pattern declaration", s->file, s->lno, s->argv[0]); |
| 224 | } |
| 225 | } |
| 226 | |
| 227 | if(pat->rules[0] == NULL) { |
| 228 | flog(LOG_WARNING, "%s:%i: missing rules in match declaration", s->file, sl); |
| 229 | freepattern(pat); |
| 230 | return(NULL); |
| 231 | } |
| 232 | if(pat->childnm == NULL) { |
| 233 | flog(LOG_WARNING, "%s:%i: missing handler in match declaration", s->file, sl); |
| 234 | freepattern(pat); |
| 235 | return(NULL); |
| 236 | } |
| 237 | return(pat); |
| 238 | } |
| 239 | |
| 240 | static struct config *readconfig(char *filename) |
| 241 | { |
| 242 | struct cfstate *s; |
| 243 | struct config *cf; |
| 244 | struct child *ch; |
| 245 | struct pattern *pat; |
| 246 | FILE *in; |
| 247 | |
| 248 | if((in = fopen(filename, "r")) == NULL) { |
| 249 | flog(LOG_WARNING, "%s: %s", filename, strerror(errno)); |
| 250 | return(NULL); |
| 251 | } |
| 252 | s = mkcfparser(in, filename); |
| 253 | omalloc(cf); |
| 254 | |
| 255 | while(1) { |
| 256 | getcfline(s); |
| 257 | if((ch = parsechild(s)) != NULL) { |
| 258 | ch->next = cf->children; |
| 259 | cf->children = ch; |
| 260 | } else if((pat = parsepattern(s)) != NULL) { |
| 261 | pat->next = cf->patterns; |
| 262 | cf->patterns = pat; |
| 263 | } else if(!strcmp(s->argv[0], "eof")) { |
| 264 | break; |
| 265 | } else { |
| 266 | flog(LOG_WARNING, "%s:%i: unknown directive `%s'", s->file, s->lno, s->argv[0]); |
| 267 | } |
| 268 | } |
| 269 | |
| 270 | freecfparser(s); |
| 271 | fclose(in); |
| 272 | return(cf); |
| 273 | } |
| 274 | |
| 275 | static void exprestpat(struct hthead *req, struct pattern *pat, char **mstr) |
| 276 | { |
| 277 | char *p, *p2, *hdr; |
| 278 | int mc; |
| 279 | struct charbuf buf; |
| 280 | |
| 281 | if(mstr == NULL) |
| 282 | mc = 0; |
| 283 | else |
| 284 | for(mc = 0; mstr[mc]; mc++); |
| 285 | bufinit(buf); |
| 286 | for(p = pat->restpat; *p; ) { |
| 287 | if(*p == '$') { |
| 288 | p++; |
| 289 | if((*p >= '0') && (*p <= '9')) { |
| 290 | if(*p - '0' < mc) |
| 291 | bufcatstr(buf, mstr[*p - '0']); |
| 292 | p++; |
| 293 | } else if(*p == '_') { |
| 294 | bufcatstr(buf, req->rest); |
| 295 | p++; |
| 296 | } else if(*p == '$') { |
| 297 | bufadd(buf, '$'); |
| 298 | p++; |
| 299 | } else if(*p == '{') { |
| 300 | if((p2 = strchr(p, '}')) == NULL) { |
| 301 | p++; |
| 302 | } else { |
| 303 | hdr = getheader(req, sprintf3("%.*s", p2 - p - 1, p + 1)); |
| 304 | if(hdr) |
| 305 | bufcatstr(buf, hdr); |
| 306 | } |
| 307 | } else if(!*p) { |
| 308 | } |
| 309 | } else { |
| 310 | bufadd(buf, *(p++)); |
| 311 | } |
| 312 | } |
| 313 | bufadd(buf, 0); |
| 314 | replrest(req, buf.b); |
| 315 | buffree(buf); |
| 316 | } |
| 317 | |
| 318 | static char *findmatch(struct config *cf, struct hthead *req, int trydefault) |
| 319 | { |
| 320 | int i, o; |
| 321 | struct pattern *pat; |
| 322 | struct rule *rule; |
| 323 | int rmo, matched; |
| 324 | char *pstr; |
| 325 | char **mstr; |
| 326 | regmatch_t gr[10]; |
| 327 | |
| 328 | mstr = NULL; |
| 329 | for(pat = cf->patterns; pat != NULL; pat = pat->next) { |
| 330 | rmo = -1; |
| 331 | for(i = 0; (rule = pat->rules[i]) != NULL; i++) { |
| 332 | matched = 0; |
| 333 | if(rule->type == PAT_REST) { |
| 334 | if((matched = !regexec(rule->pattern, pstr = req->rest, 10, gr, 0))) |
| 335 | rmo = gr[0].rm_eo; |
| 336 | else |
| 337 | break; |
| 338 | } else if(rule->type == PAT_URL) { |
| 339 | if(!(matched = !regexec(rule->pattern, pstr = req->url, 10, gr, 0))) |
| 340 | break; |
| 341 | } else if(rule->type == PAT_METHOD) { |
| 342 | if(!(matched = !regexec(rule->pattern, pstr = req->method, 10, gr, 0))) |
| 343 | break; |
| 344 | } else if(rule->type == PAT_HEADER) { |
| 345 | if(!(pstr = getheader(req, rule->header))) |
| 346 | break; |
| 347 | if(!(matched = !regexec(rule->pattern, pstr, 10, gr, 0))) |
| 348 | break; |
| 349 | } else if(rule->type == PAT_ALL) { |
| 350 | } else if(rule->type == PAT_DEFAULT) { |
| 351 | if(!trydefault) |
| 352 | break; |
| 353 | } |
| 354 | if(matched && (rule->fl & PATFL_MSS)) { |
| 355 | if(mstr) { |
| 356 | flog(LOG_WARNING, "two pattern rules marked with `s' flag found (for handler %s)", pat->childnm); |
| 357 | freeca(mstr); |
| 358 | } |
| 359 | for(o = 0; o < 10; o++) { |
| 360 | if(gr[o].rm_so < 0) |
| 361 | break; |
| 362 | } |
| 363 | mstr = szmalloc((o + 1) * sizeof(*mstr)); |
| 364 | for(o = 0; o < 10; o++) { |
| 365 | if(gr[o].rm_so < 0) |
| 366 | break; |
| 367 | mstr[o] = smalloc(gr[o].rm_eo - gr[o].rm_so + 1); |
| 368 | memcpy(mstr[o], pstr + gr[o].rm_so, gr[o].rm_eo - gr[o].rm_so); |
| 369 | mstr[o][gr[o].rm_eo - gr[o].rm_so] = 0; |
| 370 | } |
| 371 | } |
| 372 | } |
| 373 | if(!rule) { |
| 374 | if(pat->restpat) { |
| 375 | exprestpat(req, pat, mstr); |
| 376 | } else if(rmo != -1) { |
| 377 | replrest(req, req->rest + rmo); |
| 378 | } |
| 379 | if(mstr) |
| 380 | freeca(mstr); |
| 381 | return(pat->childnm); |
| 382 | } |
| 383 | if(mstr) { |
| 384 | freeca(mstr); |
| 385 | mstr = NULL; |
| 386 | } |
| 387 | } |
| 388 | return(NULL); |
| 389 | } |
| 390 | |
| 391 | static void serve(struct hthead *req, int fd) |
| 392 | { |
| 393 | char *chnm; |
| 394 | struct child *ch; |
| 395 | |
| 396 | if(((chnm = findmatch(config, req, 0)) == NULL) && ((chnm = findmatch(config, req, 1)) == NULL)) { |
| 397 | simpleerror(fd, 404, "Not Found", "The requested resource could not be found on this server."); |
| 398 | return; |
| 399 | } |
| 400 | if((ch = getchild(config, chnm)) == NULL) { |
| 401 | flog(LOG_ERR, "child %s requested, but was not declared", chnm); |
| 402 | simpleerror(fd, 500, "Configuration Error", "The server is erroneously configured. Handler %s was requested, but not declared.", chnm); |
| 403 | return; |
| 404 | } |
| 405 | |
| 406 | if(childhandle(ch, req, fd)) |
| 407 | simpleerror(fd, 500, "Server Error", "The request handler crashed."); |
| 408 | } |
| 409 | |
| 410 | int main(int argc, char **argv) |
| 411 | { |
| 412 | struct hthead *req; |
| 413 | int fd; |
| 414 | |
| 415 | if(argc < 2) { |
| 416 | flog(LOG_ERR, "usage: patplex CONFIGFILE"); |
| 417 | exit(1); |
| 418 | } |
| 419 | config = readconfig(argv[1]); |
| 420 | signal(SIGCHLD, SIG_IGN); |
| 421 | while(1) { |
| 422 | if((fd = recvreq(0, &req)) < 0) { |
| 423 | if(errno != 0) |
| 424 | flog(LOG_ERR, "recvreq: %s", strerror(errno)); |
| 425 | break; |
| 426 | } |
| 427 | serve(req, fd); |
| 428 | freehthead(req); |
| 429 | close(fd); |
| 430 | } |
| 431 | return(0); |
| 432 | } |