1 /* Licensed to the Apache Software Foundation (ASF) under one or more
2 * contributor license agreements. See the NOTICE file distributed with
3 * this work for additional information regarding copyright ownership.
4 * The ASF licenses this file to You under the Apache License, Version 2.0
5 * (the "License"); you may not use this file except in compliance with
6 * the License. You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
19 #include "apr_strings.h"
21 #define APR_WANT_STRFUNC
28 #include "ap_config.h"
30 #include "http_config.h"
31 #include "http_main.h"
33 #include "http_core.h"
34 #include "http_protocol.h"
35 #include "http_request.h" /* for sub_req_lookup_uri() */
36 #include "util_script.h"
37 #include "apr_date.h" /* For apr_date_parse_http() */
38 #include "util_ebcdic.h"
46 * Various utility functions which are common to a whole lot of
47 * script-type extensions mechanisms, and might as well be gathered
48 * in one place (if only to avoid creating inter-module dependancies
49 * where there don't have to be).
52 #define MALFORMED_MESSAGE "malformed header from script. Bad header="
53 #define MALFORMED_HEADER_LENGTH_TO_SHOW 30
55 APLOG_USE_MODULE(core);
57 static char *http2env(request_rec *r, const char *w)
59 char *res = (char *)apr_palloc(r->pool, sizeof("HTTP_") + strlen(w));
69 while ((c = *w++) != 0) {
71 *cp++ = apr_toupper(c);
77 ap_log_rerror(APLOG_MARK, APLOG_TRACE1, 0, r,
78 "Not exporting header with invalid name as envvar: %s",
79 ap_escape_logitem(r->pool, w));
88 static void add_unless_null(apr_table_t *table, const char *name, const char *val)
91 apr_table_addn(table, name, val);
95 static void env2env(apr_table_t *table, const char *name)
97 add_unless_null(table, name, getenv(name));
100 AP_DECLARE(char **) ap_create_environment(apr_pool_t *p, apr_table_t *t)
102 const apr_array_header_t *env_arr = apr_table_elts(t);
103 const apr_table_entry_t *elts = (const apr_table_entry_t *) env_arr->elts;
104 char **env = (char **) apr_palloc(p, (env_arr->nelts + 2) * sizeof(char *));
110 if (!apr_table_get(t, "TZ")) {
113 env[j++] = apr_pstrcat(p, "TZ=", tz, NULL);
116 for (i = 0; i < env_arr->nelts; ++i) {
120 env[j] = apr_pstrcat(p, elts[i].key, "=", elts[i].val, NULL);
122 if (apr_isdigit(*whack)) {
125 while (*whack != '=') {
126 if (!apr_isalnum(*whack) && *whack != '_') {
138 AP_DECLARE(void) ap_add_common_vars(request_rec *r)
141 server_rec *s = r->server;
142 conn_rec *c = r->connection;
143 const char *env_temp;
144 const apr_array_header_t *hdrs_arr = apr_table_elts(r->headers_in);
145 const apr_table_entry_t *hdrs = (const apr_table_entry_t *) hdrs_arr->elts;
149 /* use a temporary apr_table_t which we'll overlap onto
150 * r->subprocess_env later
151 * (exception: if r->subprocess_env is empty at the start,
152 * write directly into it)
154 if (apr_is_empty_table(r->subprocess_env)) {
155 e = r->subprocess_env;
158 e = apr_table_make(r->pool, 25 + hdrs_arr->nelts);
161 /* First, add environment vars from headers... this is as per
162 * CGI specs, though other sorts of scripting interfaces see
166 for (i = 0; i < hdrs_arr->nelts; ++i) {
171 /* A few headers are special cased --- Authorization to prevent
172 * rogue scripts from capturing passwords; content-type and -length
173 * for no particular reason.
176 if (!strcasecmp(hdrs[i].key, "Content-type")) {
177 apr_table_addn(e, "CONTENT_TYPE", hdrs[i].val);
179 else if (!strcasecmp(hdrs[i].key, "Content-length")) {
180 apr_table_addn(e, "CONTENT_LENGTH", hdrs[i].val);
183 * You really don't want to disable this check, since it leaves you
184 * wide open to CGIs stealing passwords and people viewing them
185 * in the environment with "ps -e". But, if you must...
187 #ifndef SECURITY_HOLE_PASS_AUTHORIZATION
188 else if (!strcasecmp(hdrs[i].key, "Authorization")
189 || !strcasecmp(hdrs[i].key, "Proxy-Authorization")) {
194 add_unless_null(e, http2env(r, hdrs[i].key), hdrs[i].val);
197 env_temp = apr_table_get(r->subprocess_env, "PATH");
198 if (env_temp == NULL) {
199 env_temp = getenv("PATH");
201 if (env_temp == NULL) {
202 env_temp = DEFAULT_PATH;
204 apr_table_addn(e, "PATH", apr_pstrdup(r->pool, env_temp));
207 env2env(e, "SystemRoot");
208 env2env(e, "COMSPEC");
209 env2env(e, "PATHEXT");
210 env2env(e, "WINDIR");
212 env2env(e, "COMSPEC");
215 env2env(e, "PERLLIB_PREFIX");
217 env2env(e, "LIBRARY_PATH");
218 #elif defined(DARWIN)
219 env2env(e, "DYLD_LIBRARY_PATH");
221 env2env(e, "LIBPATH");
222 #elif defined(__HPUX__)
223 /* HPUX PARISC 2.0W knows both, otherwise redundancy is harmless */
224 env2env(e, "SHLIB_PATH");
225 env2env(e, "LD_LIBRARY_PATH");
226 #else /* Some Unix */
227 env2env(e, "LD_LIBRARY_PATH");
230 apr_table_addn(e, "SERVER_SIGNATURE", ap_psignature("", r));
231 apr_table_addn(e, "SERVER_SOFTWARE", ap_get_server_banner());
232 apr_table_addn(e, "SERVER_NAME",
233 ap_escape_html(r->pool, ap_get_server_name(r)));
234 apr_table_addn(e, "SERVER_ADDR", r->connection->local_ip); /* Apache */
235 apr_table_addn(e, "SERVER_PORT",
236 apr_psprintf(r->pool, "%u", ap_get_server_port(r)));
237 add_unless_null(e, "REMOTE_HOST",
238 ap_get_remote_host(c, r->per_dir_config, REMOTE_HOST, NULL));
239 apr_table_addn(e, "REMOTE_ADDR", c->remote_ip);
240 apr_table_addn(e, "DOCUMENT_ROOT", ap_document_root(r)); /* Apache */
241 apr_table_setn(e, "REQUEST_SCHEME", ap_http_scheme(r));
242 apr_table_addn(e, "CONTEXT_PREFIX", ap_context_prefix(r));
243 apr_table_addn(e, "CONTEXT_DOCUMENT_ROOT", ap_context_document_root(r));
244 apr_table_addn(e, "SERVER_ADMIN", s->server_admin); /* Apache */
245 apr_table_addn(e, "SCRIPT_FILENAME", r->filename); /* Apache */
247 rport = c->remote_addr->port;
248 apr_table_addn(e, "REMOTE_PORT", apr_itoa(r->pool, rport));
251 apr_table_addn(e, "REMOTE_USER", r->user);
254 request_rec *back = r->prev;
258 apr_table_addn(e, "REDIRECT_REMOTE_USER", back->user);
264 add_unless_null(e, "AUTH_TYPE", r->ap_auth_type);
265 env_temp = ap_get_remote_logname(r);
267 apr_table_addn(e, "REMOTE_IDENT", apr_pstrdup(r->pool, env_temp));
270 /* Apache custom error responses. If we have redirected set two new vars */
273 add_unless_null(e, "REDIRECT_QUERY_STRING", r->prev->args);
274 add_unless_null(e, "REDIRECT_URL", r->prev->uri);
277 if (e != r->subprocess_env) {
278 apr_table_overlap(r->subprocess_env, e, APR_OVERLAP_TABLES_SET);
282 /* This "cute" little function comes about because the path info on
283 * filenames and URLs aren't always the same. So we take the two,
284 * and find as much of the two that match as possible.
287 AP_DECLARE(int) ap_find_path_info(const char *uri, const char *path_info)
289 int lu = strlen(uri);
290 int lp = strlen(path_info);
292 while (lu-- && lp-- && uri[lu] == path_info[lp]) {
293 if (path_info[lp] == '/') {
294 while (lu && uri[lu-1] == '/') lu--;
302 while (uri[lu] != '\0' && uri[lu] != '/') {
308 /* Obtain the Request-URI from the original request-line, returning
309 * a new string from the request pool containing the URI or "".
311 static char *original_uri(request_rec *r)
315 if (r->the_request == NULL) {
316 return (char *) apr_pcalloc(r->pool, 1);
319 first = r->the_request; /* use the request-line */
321 while (*first && !apr_isspace(*first)) {
322 ++first; /* skip over the method */
324 while (apr_isspace(*first)) {
325 ++first; /* and the space(s) */
329 while (*last && !apr_isspace(*last)) {
330 ++last; /* end at next whitespace */
333 return apr_pstrmemdup(r->pool, first, last - first);
336 AP_DECLARE(void) ap_add_cgi_vars(request_rec *r)
338 apr_table_t *e = r->subprocess_env;
340 apr_table_setn(e, "GATEWAY_INTERFACE", "CGI/1.1");
341 apr_table_setn(e, "SERVER_PROTOCOL", r->protocol);
342 apr_table_setn(e, "REQUEST_METHOD", r->method);
343 apr_table_setn(e, "QUERY_STRING", r->args ? r->args : "");
344 apr_table_setn(e, "REQUEST_URI", original_uri(r));
346 /* Note that the code below special-cases scripts run from includes,
347 * because it "knows" that the sub_request has been hacked to have the
348 * args and path_info of the original request, and not any that may have
349 * come with the script URI in the include command. Ugh.
352 if (!strcmp(r->protocol, "INCLUDED")) {
353 apr_table_setn(e, "SCRIPT_NAME", r->uri);
354 if (r->path_info && *r->path_info) {
355 apr_table_setn(e, "PATH_INFO", r->path_info);
358 else if (!r->path_info || !*r->path_info) {
359 apr_table_setn(e, "SCRIPT_NAME", r->uri);
362 int path_info_start = ap_find_path_info(r->uri, r->path_info);
364 apr_table_setn(e, "SCRIPT_NAME",
365 apr_pstrndup(r->pool, r->uri, path_info_start));
367 apr_table_setn(e, "PATH_INFO", r->path_info);
370 if (r->path_info && r->path_info[0]) {
372 * To get PATH_TRANSLATED, treat PATH_INFO as a URI path.
373 * Need to re-escape it for this, since the entire URI was
374 * un-escaped before we determined where the PATH_INFO began.
378 pa_req = ap_sub_req_lookup_uri(ap_escape_uri(r->pool, r->path_info), r,
381 if (pa_req->filename) {
382 char *pt = apr_pstrcat(r->pool, pa_req->filename, pa_req->path_info,
385 /* We need to make this a real Windows path name */
386 apr_filepath_merge(&pt, "", pt, APR_FILEPATH_NATIVE, r->pool);
388 apr_table_setn(e, "PATH_TRANSLATED", pt);
390 ap_destroy_sub_req(pa_req);
395 static int set_cookie_doo_doo(void *v, const char *key, const char *val)
397 apr_table_addn(v, key, val);
401 #define HTTP_UNSET (-HTTP_OK)
403 AP_DECLARE(int) ap_scan_script_header_err_core(request_rec *r, char *buffer,
404 int (*getsfunc) (char *, int, void *),
407 char x[MAX_STRING_LEN];
410 int cgi_status = HTTP_UNSET;
412 apr_table_t *cookie_table;
417 w = buffer ? buffer : x;
419 /* temporary place to hold headers to merge in later */
420 merge = apr_table_make(r->pool, 10);
422 /* The HTTP specification says that it is legal to merge duplicate
423 * headers into one. Some browsers that support Cookies don't like
424 * merged headers and prefer that each Set-Cookie header is sent
425 * separately. Lets humour those browsers by not merging.
426 * Oh what a pain it is.
428 cookie_table = apr_table_make(r->pool, 2);
429 apr_table_do(set_cookie_doo_doo, cookie_table, r->err_headers_out, "Set-Cookie", NULL);
433 int rv = (*getsfunc) (w, MAX_STRING_LEN - 1, getsfunc_data);
435 ap_log_rerror(APLOG_MARK, APLOG_ERR|APLOG_TOCLIENT, 0, r,
436 "Premature end of script headers: %s",
437 apr_filepath_name_get(r->filename));
438 return HTTP_INTERNAL_SERVER_ERROR;
441 ap_log_rerror(APLOG_MARK, APLOG_ERR|APLOG_TOCLIENT, 0, r,
442 "Script timed out before returning headers: %s",
443 apr_filepath_name_get(r->filename));
444 return HTTP_GATEWAY_TIME_OUT;
447 /* Delete terminal (CR?)LF */
450 /* Indeed, the host's '\n':
451 '\012' for UNIX; '\015' for MacOS; '\025' for OS/390
452 -- whatever the script generates.
454 if (p > 0 && w[p - 1] == '\n') {
455 if (p > 1 && w[p - 2] == CR) {
464 * If we've finished reading the headers, check to make sure any
465 * HTTP/1.1 conditions are met. If so, we're done; normal processing
466 * will handle the script's output. If not, just return the error.
467 * The appropriate thing to do would be to send the script process a
468 * SIGPIPE to let it know we're ignoring it, close the channel to the
469 * script process, and *then* return the failed-to-meet-condition
470 * error. Otherwise we'd be waiting for the script to finish
471 * blithering before telling the client the output was no good.
472 * However, we don't have the information to do that, so we have to
473 * leave it to an upper layer.
476 int cond_status = OK;
478 /* PR#38070: This fails because it gets confused when a
479 * CGI Status header overrides ap_meets_conditions.
481 * We can fix that by dropping ap_meets_conditions when
482 * Status has been set. Since this is the only place
483 * cgi_status gets used, let's test it explicitly.
485 * The alternative would be to ignore CGI Status when
486 * ap_meets_conditions returns anything interesting.
487 * That would be safer wrt HTTP, but would break CGI.
489 if ((cgi_status == HTTP_UNSET) && (r->method_number == M_GET)) {
490 cond_status = ap_meets_conditions(r);
492 apr_table_overlap(r->err_headers_out, merge,
493 APR_OVERLAP_TABLES_MERGE);
494 if (!apr_is_empty_table(cookie_table)) {
495 /* the cookies have already been copied to the cookie_table */
496 apr_table_unset(r->err_headers_out, "Set-Cookie");
497 r->err_headers_out = apr_table_overlay(r->pool,
498 r->err_headers_out, cookie_table);
503 /* if we see a bogus header don't ignore it. Shout and scream */
505 #if APR_CHARSET_EBCDIC
506 /* Chances are that we received an ASCII header text instead of
507 * the expected EBCDIC header lines. Try to auto-detect:
509 if (!(l = strchr(w, ':'))) {
510 int maybeASCII = 0, maybeEBCDIC = 0;
511 unsigned char *cp, native;
512 apr_size_t inbytes_left, outbytes_left;
514 for (cp = w; *cp != '\0'; ++cp) {
515 native = apr_xlate_conv_byte(ap_hdrs_from_ascii, *cp);
516 if (apr_isprint(*cp) && !apr_isprint(native))
518 if (!apr_isprint(*cp) && apr_isprint(native))
521 if (maybeASCII > maybeEBCDIC) {
522 ap_log_error(APLOG_MARK, APLOG_ERR, 0, r->server,
523 "CGI Interface Error: Script headers apparently ASCII: (CGI = %s)",
525 inbytes_left = outbytes_left = cp - w;
526 apr_xlate_conv_buffer(ap_hdrs_from_ascii,
527 w, &inbytes_left, w, &outbytes_left);
530 #endif /*APR_CHARSET_EBCDIC*/
531 if (!(l = strchr(w, ':'))) {
532 char malformed[(sizeof MALFORMED_MESSAGE) + 1
533 + MALFORMED_HEADER_LENGTH_TO_SHOW];
535 strcpy(malformed, MALFORMED_MESSAGE);
536 strncat(malformed, w, MALFORMED_HEADER_LENGTH_TO_SHOW);
539 /* Soak up all the script output - may save an outright kill */
540 while ((*getsfunc) (w, MAX_STRING_LEN - 1, getsfunc_data)) {
545 ap_log_rerror(APLOG_MARK, APLOG_ERR|APLOG_TOCLIENT, 0, r,
547 apr_filepath_name_get(r->filename));
548 return HTTP_INTERNAL_SERVER_ERROR;
552 while (*l && apr_isspace(*l)) {
556 if (!strcasecmp(w, "Content-type")) {
559 /* Nuke trailing whitespace */
561 char *endp = l + strlen(l) - 1;
562 while (endp > l && apr_isspace(*endp)) {
566 tmp = apr_pstrdup(r->pool, l);
567 ap_content_type_tolower(tmp);
568 ap_set_content_type(r, tmp);
571 * If the script returned a specific status, that's what
572 * we'll use - otherwise we assume 200 OK.
574 else if (!strcasecmp(w, "Status")) {
575 r->status = cgi_status = atoi(l);
576 r->status_line = apr_pstrdup(r->pool, l);
578 else if (!strcasecmp(w, "Location")) {
579 apr_table_set(r->headers_out, w, l);
581 else if (!strcasecmp(w, "Content-Length")) {
582 apr_table_set(r->headers_out, w, l);
584 else if (!strcasecmp(w, "Content-Range")) {
585 apr_table_set(r->headers_out, w, l);
587 else if (!strcasecmp(w, "Transfer-Encoding")) {
588 apr_table_set(r->headers_out, w, l);
590 else if (!strcasecmp(w, "ETag")) {
591 apr_table_set(r->headers_out, w, l);
594 * If the script gave us a Last-Modified header, we can't just
595 * pass it on blindly because of restrictions on future values.
597 else if (!strcasecmp(w, "Last-Modified")) {
598 ap_update_mtime(r, apr_date_parse_http(l));
599 ap_set_last_modified(r);
601 else if (!strcasecmp(w, "Set-Cookie")) {
602 apr_table_add(cookie_table, w, l);
605 apr_table_add(merge, w, l);
608 /* never reached - we leave this function within the while loop above */
612 static int getsfunc_FILE(char *buf, int len, void *f)
614 return apr_file_gets(buf, len, (apr_file_t *) f) == APR_SUCCESS;
617 AP_DECLARE(int) ap_scan_script_header_err(request_rec *r, apr_file_t *f,
620 return ap_scan_script_header_err_core(r, buffer, getsfunc_FILE, f);
623 static int getsfunc_BRIGADE(char *buf, int len, void *arg)
625 apr_bucket_brigade *bb = (apr_bucket_brigade *)arg;
626 const char *dst_end = buf + len - 1; /* leave room for terminating null */
628 apr_bucket *e = APR_BRIGADE_FIRST(bb);
632 while ((dst < dst_end) && !done && !APR_BUCKET_IS_EOS(e)) {
633 const char *bucket_data;
634 apr_size_t bucket_data_len;
639 rv = apr_bucket_read(e, &bucket_data, &bucket_data_len,
641 if (rv != APR_SUCCESS || (bucket_data_len == 0)) {
642 return APR_STATUS_IS_TIMEUP(rv) ? -1 : 0;
645 src_end = bucket_data + bucket_data_len;
646 while ((src < src_end) && (dst < dst_end) && !done) {
650 else if (*src != '\r') {
657 apr_bucket_split(e, src - bucket_data);
659 next = APR_BUCKET_NEXT(e);
660 APR_BUCKET_REMOVE(e);
661 apr_bucket_destroy(e);
668 AP_DECLARE(int) ap_scan_script_header_err_brigade(request_rec *r,
669 apr_bucket_brigade *bb,
672 return ap_scan_script_header_err_core(r, buffer, getsfunc_BRIGADE, bb);
681 static int getsfunc_STRING(char *w, int len, void *pvastrs)
683 struct vastrs *strs = (struct vastrs*) pvastrs;
687 if (!strs->curpos || !*strs->curpos)
689 p = ap_strchr_c(strs->curpos, '\n');
693 p = ap_strchr_c(strs->curpos, '\0');
694 t = p - strs->curpos;
697 strncpy (w, strs->curpos, t);
699 if (!strs->curpos[t]) {
701 strs->curpos = va_arg(strs->args, const char *);
708 /* ap_scan_script_header_err_strs() accepts additional const char* args...
709 * each is treated as one or more header lines, and the first non-header
710 * character is returned to **arg, **data. (The first optional arg is
713 AP_DECLARE_NONSTD(int) ap_scan_script_header_err_strs(request_rec *r,
721 va_start(strs.args, termarg);
723 strs.curpos = va_arg(strs.args, char*);
724 res = ap_scan_script_header_err_core(r, buffer, getsfunc_STRING, (void *) &strs);
726 *termch = strs.curpos;
735 argstr_to_table(char *str, apr_table_t *parms)
745 key = apr_strtok(str, "&", &strtok_state);
747 value = strchr(key, '=');
749 *value = '\0'; /* Split the string in two */
750 value++; /* Skip passed the = */
755 ap_unescape_url(key);
756 ap_unescape_url(value);
757 apr_table_set(parms, key, value);
758 key = apr_strtok(NULL, "&", &strtok_state);
762 AP_DECLARE(void) ap_args_to_table(request_rec *r, apr_table_t **table)
764 apr_table_t *t = apr_table_make(r->pool, 10);
765 argstr_to_table(apr_pstrdup(r->pool, r->args), t);