1 const char parsers_rcs[] = "$Id: parsers.c,v 1.175 2009/06/05 16:55:16 fabiankeil Exp $";
2 /*********************************************************************
4 * File : $Source: /cvsroot/ijbswa/current/parsers.c,v $
6 * Purpose : Declares functions to parse/crunch headers and pages.
7 * Functions declared include:
8 * `add_to_iob', `client_cookie_adder', `client_from',
9 * `client_referrer', `client_send_cookie', `client_ua',
10 * `client_uagent', `client_x_forwarded',
11 * `client_x_forwarded_adder', `client_xtra_adder',
12 * `content_type', `crumble', `destroy_list', `enlist',
13 * `flush_socket', ``get_header', `sed', `filter_header'
14 * `server_content_encoding', `server_content_disposition',
15 * `server_last_modified', `client_accept_language',
16 * `crunch_client_header', `client_if_modified_since',
17 * `client_if_none_match', `get_destination_from_headers',
18 * `parse_header_time', `decompress_iob' and `server_set_cookie'.
20 * Copyright : Written by and Copyright (C) 2001-2009 the
21 * Privoxy team. http://www.privoxy.org/
23 * Based on the Internet Junkbuster originally written
24 * by and Copyright (C) 1997 Anonymous Coders and
25 * Junkbusters Corporation. http://www.junkbusters.com
27 * This program is free software; you can redistribute it
28 * and/or modify it under the terms of the GNU General
29 * Public License as published by the Free Software
30 * Foundation; either version 2 of the License, or (at
31 * your option) any later version.
33 * This program is distributed in the hope that it will
34 * be useful, but WITHOUT ANY WARRANTY; without even the
35 * implied warranty of MERCHANTABILITY or FITNESS FOR A
36 * PARTICULAR PURPOSE. See the GNU General Public
37 * License for more details.
39 * The GNU General Public License should be included with
40 * this file. If not, you can view it at
41 * http://www.gnu.org/copyleft/gpl.html
42 * or write to the Free Software Foundation, Inc., 59
43 * Temple Place - Suite 330, Boston, MA 02111-1307, USA.
45 *********************************************************************/
52 #include <sys/types.h>
62 * Convince GNU's libc to provide a strptime prototype.
71 #define GZIP_IDENTIFIER_1 0x1f
72 #define GZIP_IDENTIFIER_2 0x8b
74 #define GZIP_FLAG_CHECKSUM 0x02
75 #define GZIP_FLAG_EXTRA_FIELDS 0x04
76 #define GZIP_FLAG_FILE_NAME 0x08
77 #define GZIP_FLAG_COMMENT 0x10
78 #define GZIP_FLAG_RESERVED_BITS 0xe0
81 #if !defined(_WIN32) && !defined(__OS2__)
87 #ifdef FEATURE_PTHREAD
89 /* jcc.h is for mutex semapores only */
90 #endif /* def FEATURE_PTHREAD */
95 #include "jbsockets.h"
101 #ifndef HAVE_STRPTIME
102 #include "strptime.h"
105 const char parsers_h_rcs[] = PARSERS_H_VERSION;
107 /* Fix a problem with Solaris. There should be no effect on other
109 * Solaris's isspace() is a macro which uses its argument directly
110 * as an array index. Therefore we need to make sure that high-bit
111 * characters generate +ve values, and ideally we also want to make
112 * the argument match the declared parameter type of "int".
114 * Why did they write a character function that can't take a simple
115 * "char" argument? Doh!
117 #define ijb_isupper(__X) isupper((int)(unsigned char)(__X))
118 #define ijb_tolower(__X) tolower((int)(unsigned char)(__X))
120 static char *get_header_line(struct iob *iob);
121 static jb_err scan_headers(struct client_state *csp);
122 static jb_err header_tagger(struct client_state *csp, char *header);
123 static jb_err parse_header_time(const char *header_time, time_t *result);
125 static jb_err crumble (struct client_state *csp, char **header);
126 static jb_err filter_header (struct client_state *csp, char **header);
127 static jb_err client_connection (struct client_state *csp, char **header);
128 static jb_err client_referrer (struct client_state *csp, char **header);
129 static jb_err client_uagent (struct client_state *csp, char **header);
130 static jb_err client_ua (struct client_state *csp, char **header);
131 static jb_err client_from (struct client_state *csp, char **header);
132 static jb_err client_send_cookie (struct client_state *csp, char **header);
133 static jb_err client_x_forwarded (struct client_state *csp, char **header);
134 static jb_err client_accept_encoding (struct client_state *csp, char **header);
135 static jb_err client_te (struct client_state *csp, char **header);
136 static jb_err client_max_forwards (struct client_state *csp, char **header);
137 static jb_err client_host (struct client_state *csp, char **header);
138 static jb_err client_if_modified_since (struct client_state *csp, char **header);
139 static jb_err client_accept_language (struct client_state *csp, char **header);
140 static jb_err client_if_none_match (struct client_state *csp, char **header);
141 static jb_err crunch_client_header (struct client_state *csp, char **header);
142 static jb_err client_x_filter (struct client_state *csp, char **header);
143 static jb_err client_range (struct client_state *csp, char **header);
144 static jb_err server_set_cookie (struct client_state *csp, char **header);
145 static jb_err server_connection (struct client_state *csp, char **header);
146 static jb_err server_content_type (struct client_state *csp, char **header);
147 static jb_err server_adjust_content_length(struct client_state *csp, char **header);
148 static jb_err server_content_md5 (struct client_state *csp, char **header);
149 static jb_err server_content_encoding (struct client_state *csp, char **header);
150 static jb_err server_transfer_coding (struct client_state *csp, char **header);
151 static jb_err server_http (struct client_state *csp, char **header);
152 static jb_err crunch_server_header (struct client_state *csp, char **header);
153 static jb_err server_last_modified (struct client_state *csp, char **header);
154 static jb_err server_content_disposition(struct client_state *csp, char **header);
156 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
157 static jb_err server_save_content_length(struct client_state *csp, char **header);
158 static jb_err server_keep_alive(struct client_state *csp, char **header);
159 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
161 static jb_err client_host_adder (struct client_state *csp);
162 static jb_err client_xtra_adder (struct client_state *csp);
163 static jb_err client_x_forwarded_for_adder(struct client_state *csp);
164 static jb_err client_connection_header_adder(struct client_state *csp);
165 static jb_err server_connection_adder(struct client_state *csp);
166 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
167 static jb_err server_proxy_connection_adder(struct client_state *csp);
168 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
170 static jb_err create_forged_referrer(char **header, const char *hostport);
171 static jb_err create_fake_referrer(char **header, const char *fake_referrer);
172 static jb_err handle_conditional_hide_referrer_parameter(char **header,
173 const char *host, const int parameter_conditional_block);
174 static const char *get_appropiate_connection_header(const struct client_state *csp);
177 * List of functions to run on a list of headers.
181 /** The header prefix to match */
184 /** The length of the prefix to match */
187 /** The function to apply to this line */
188 const parser_func_ptr parser;
191 static const struct parsers client_patterns[] = {
192 { "referer:", 8, client_referrer },
193 { "user-agent:", 11, client_uagent },
194 { "ua-", 3, client_ua },
195 { "from:", 5, client_from },
196 { "cookie:", 7, client_send_cookie },
197 { "x-forwarded-for:", 16, client_x_forwarded },
198 { "Accept-Encoding:", 16, client_accept_encoding },
199 { "TE:", 3, client_te },
200 { "Host:", 5, client_host },
201 { "if-modified-since:", 18, client_if_modified_since },
202 #ifndef FEATURE_CONNECTION_KEEP_ALIVE
203 { "Keep-Alive:", 11, crumble },
205 { "connection:", 11, client_connection },
206 { "proxy-connection:", 17, crumble },
207 { "max-forwards:", 13, client_max_forwards },
208 { "Accept-Language:", 16, client_accept_language },
209 { "if-none-match:", 14, client_if_none_match },
210 { "Range:", 6, client_range },
211 { "Request-Range:", 14, client_range },
212 { "If-Range:", 9, client_range },
213 { "X-Filter:", 9, client_x_filter },
214 { "*", 0, crunch_client_header },
215 { "*", 0, filter_header },
219 static const struct parsers server_patterns[] = {
220 { "HTTP/", 5, server_http },
221 { "set-cookie:", 11, server_set_cookie },
222 { "connection:", 11, server_connection },
223 { "Content-Type:", 13, server_content_type },
224 { "Content-MD5:", 12, server_content_md5 },
225 { "Content-Encoding:", 17, server_content_encoding },
226 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
227 { "Content-Length:", 15, server_save_content_length },
228 { "Keep-Alive:", 11, server_keep_alive },
230 { "Keep-Alive:", 11, crumble },
231 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
232 { "Transfer-Encoding:", 18, server_transfer_coding },
233 { "content-disposition:", 20, server_content_disposition },
234 { "Last-Modified:", 14, server_last_modified },
235 { "*", 0, crunch_server_header },
236 { "*", 0, filter_header },
240 static const add_header_func_ptr add_client_headers[] = {
242 client_x_forwarded_for_adder,
244 /* Temporarily disabled: client_accept_encoding_adder, */
245 client_connection_header_adder,
249 static const add_header_func_ptr add_server_headers[] = {
250 server_connection_adder,
251 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
252 server_proxy_connection_adder,
253 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
257 /*********************************************************************
259 * Function : flush_socket
261 * Description : Write any pending "buffered" content.
264 * 1 : fd = file descriptor of the socket to read
265 * 2 : iob = The I/O buffer to flush, usually csp->iob.
267 * Returns : On success, the number of bytes written are returned (zero
268 * indicates nothing was written). On error, -1 is returned,
269 * and errno is set appropriately. If count is zero and the
270 * file descriptor refers to a regular file, 0 will be
271 * returned without causing any other effect. For a special
272 * file, the results are not portable.
274 *********************************************************************/
275 long flush_socket(jb_socket fd, struct iob *iob)
277 long len = iob->eod - iob->cur;
284 if (write_socket(fd, iob->cur, (size_t)len))
288 iob->eod = iob->cur = iob->buf;
294 /*********************************************************************
296 * Function : add_to_iob
298 * Description : Add content to the buffered page, expanding the
299 * buffer if necessary.
302 * 1 : csp = Current client state (buffers, headers, etc...)
303 * 2 : buf = holds the content to be added to the page
304 * 3 : n = number of bytes to be added
306 * Returns : JB_ERR_OK on success, JB_ERR_MEMORY if out-of-memory
307 * or buffer limit reached.
309 *********************************************************************/
310 jb_err add_to_iob(struct client_state *csp, char *buf, long n)
312 struct iob *iob = csp->iob;
313 size_t used, offset, need, want;
316 if (n <= 0) return JB_ERR_OK;
318 used = (size_t)(iob->eod - iob->buf);
319 offset = (size_t)(iob->cur - iob->buf);
320 need = used + (size_t)n + 1;
323 * If the buffer can't hold the new data, extend it first.
324 * Use the next power of two if possible, else use the actual need.
326 if (need > csp->config->buffer_limit)
328 log_error(LOG_LEVEL_INFO,
329 "Buffer limit reached while extending the buffer (iob). Needed: %d. Limit: %d",
330 need, csp->config->buffer_limit);
331 return JB_ERR_MEMORY;
334 if (need > iob->size)
336 for (want = csp->iob->size ? csp->iob->size : 512; want <= need;) want *= 2;
338 if (want <= csp->config->buffer_limit && NULL != (p = (char *)realloc(iob->buf, want)))
342 else if (NULL != (p = (char *)realloc(iob->buf, need)))
348 log_error(LOG_LEVEL_ERROR, "Extending the buffer (iob) failed: %E");
349 return JB_ERR_MEMORY;
352 /* Update the iob pointers */
353 iob->cur = p + offset;
358 /* copy the new data into the iob buffer */
359 memcpy(iob->eod, buf, (size_t)n);
361 /* point to the end of the data */
364 /* null terminate == cheap insurance */
373 /*********************************************************************
375 * Function : decompress_iob
377 * Description : Decompress buffered page, expanding the
378 * buffer as necessary. csp->iob->cur
379 * should point to the the beginning of the
380 * compressed data block.
383 * 1 : csp = Current client state (buffers, headers, etc...)
385 * Returns : JB_ERR_OK on success,
386 * JB_ERR_MEMORY if out-of-memory limit reached, and
387 * JB_ERR_COMPRESS if error decompressing buffer.
389 *********************************************************************/
390 jb_err decompress_iob(struct client_state *csp)
392 char *buf; /* new, uncompressed buffer */
393 char *cur; /* Current iob position (to keep the original
394 * iob->cur unmodified if we return early) */
395 size_t bufsize; /* allocated size of the new buffer */
396 size_t old_size; /* Content size before decompression */
397 size_t skip_size; /* Number of bytes at the beginning of the iob
398 that we should NOT decompress. */
399 int status; /* return status of the inflate() call */
400 z_stream zstr; /* used by calls to zlib */
402 assert(csp->iob->cur - csp->iob->buf > 0);
403 assert(csp->iob->eod - csp->iob->cur > 0);
405 bufsize = csp->iob->size;
406 skip_size = (size_t)(csp->iob->cur - csp->iob->buf);
407 old_size = (size_t)(csp->iob->eod - csp->iob->cur);
411 if (bufsize < (size_t)10)
414 * This is to protect the parsing of gzipped data,
415 * but it should(?) be valid for deflated data also.
417 log_error(LOG_LEVEL_ERROR, "Buffer too small decompressing iob");
418 return JB_ERR_COMPRESS;
421 if (csp->content_type & CT_GZIP)
424 * Our task is slightly complicated by the facts that data
425 * compressed by gzip does not include a zlib header, and
426 * that there is no easily accessible interface in zlib to
427 * handle a gzip header. We strip off the gzip header by
428 * hand, and later inform zlib not to expect a header.
432 * Strip off the gzip header. Please see RFC 1952 for more
433 * explanation of the appropriate fields.
435 if (((*cur++ & 0xff) != GZIP_IDENTIFIER_1)
436 || ((*cur++ & 0xff) != GZIP_IDENTIFIER_2)
437 || (*cur++ != Z_DEFLATED))
439 log_error(LOG_LEVEL_ERROR, "Invalid gzip header when decompressing");
440 return JB_ERR_COMPRESS;
445 if (flags & GZIP_FLAG_RESERVED_BITS)
447 /* The gzip header has reserved bits set; bail out. */
448 log_error(LOG_LEVEL_ERROR, "Invalid gzip header flags when decompressing");
449 return JB_ERR_COMPRESS;
453 * Skip mtime (4 bytes), extra flags (1 byte)
454 * and OS type (1 byte).
458 /* Skip extra fields if necessary. */
459 if (flags & GZIP_FLAG_EXTRA_FIELDS)
462 * Skip a given number of bytes, specified
463 * as a 16-bit little-endian value.
465 * XXX: this code is untested and should probably be removed.
469 skip_bytes += *cur++ << 8;
472 * The number of bytes to skip should be positive
473 * and we'd like to stay in the buffer.
475 if ((skip_bytes < 0) || (skip_bytes >= (csp->iob->eod - cur)))
477 log_error(LOG_LEVEL_ERROR,
478 "Unreasonable amount of bytes to skip (%d). Stopping decompression",
480 return JB_ERR_COMPRESS;
482 log_error(LOG_LEVEL_INFO,
483 "Skipping %d bytes for gzip compression. Does this sound right?",
488 /* Skip the filename if necessary. */
489 if (flags & GZIP_FLAG_FILE_NAME)
491 /* A null-terminated string is supposed to follow. */
492 while (*cur++ && (cur < csp->iob->eod));
495 /* Skip the comment if necessary. */
496 if (flags & GZIP_FLAG_COMMENT)
498 /* A null-terminated string is supposed to follow. */
499 while (*cur++ && (cur < csp->iob->eod));
502 /* Skip the CRC if necessary. */
503 if (flags & GZIP_FLAG_CHECKSUM)
508 if (cur >= csp->iob->eod)
511 * If the current position pointer reached or passed
512 * the buffer end, we were obviously tricked to skip
515 log_error(LOG_LEVEL_ERROR,
516 "Malformed gzip header detected. Aborting decompression.");
517 return JB_ERR_COMPRESS;
521 else if (csp->content_type & CT_DEFLATE)
524 * XXX: The debug level should be lowered
525 * before the next stable release.
527 log_error(LOG_LEVEL_INFO, "Decompressing deflated iob: %d", *cur);
529 * In theory (that is, according to RFC 1950), deflate-compressed
530 * data should begin with a two-byte zlib header and have an
531 * adler32 checksum at the end. It seems that in practice only
532 * the raw compressed data is sent. Note that this means that
533 * we are not RFC 1950-compliant here, but the advantage is that
534 * this actually works. :)
536 * We add a dummy null byte to tell zlib where the data ends,
537 * and later inform it not to expect a header.
539 * Fortunately, add_to_iob() has thoughtfully null-terminated
540 * the buffer; we can just increment the end pointer to include
547 log_error(LOG_LEVEL_ERROR,
548 "Unable to determine compression format for decompression");
549 return JB_ERR_COMPRESS;
552 /* Set up the fields required by zlib. */
553 zstr.next_in = (Bytef *)cur;
554 zstr.avail_in = (unsigned int)(csp->iob->eod - cur);
555 zstr.zalloc = Z_NULL;
557 zstr.opaque = Z_NULL;
560 * Passing -MAX_WBITS to inflateInit2 tells the library
561 * that there is no zlib header.
563 if (inflateInit2(&zstr, -MAX_WBITS) != Z_OK)
565 log_error(LOG_LEVEL_ERROR, "Error initializing decompression");
566 return JB_ERR_COMPRESS;
570 * Next, we allocate new storage for the inflated data.
571 * We don't modify the existing iob yet, so in case there
572 * is error in decompression we can recover gracefully.
574 buf = zalloc(bufsize);
577 log_error(LOG_LEVEL_ERROR, "Out of memory decompressing iob");
578 return JB_ERR_MEMORY;
581 assert(bufsize >= skip_size);
582 memcpy(buf, csp->iob->buf, skip_size);
583 zstr.avail_out = bufsize - skip_size;
584 zstr.next_out = (Bytef *)buf + skip_size;
586 /* Try to decompress the whole stream in one shot. */
587 while (Z_BUF_ERROR == (status = inflate(&zstr, Z_FINISH)))
589 /* We need to allocate more memory for the output buffer. */
591 char *tmpbuf; /* used for realloc'ing the buffer */
592 size_t oldbufsize = bufsize; /* keep track of the old bufsize */
594 if (0 == zstr.avail_in)
597 * If zlib wants more data then there's a problem, because
598 * the complete compressed file should have been buffered.
600 log_error(LOG_LEVEL_ERROR,
601 "Unexpected end of compressed iob. Using what we got so far.");
606 * If we tried the limit and still didn't have enough
607 * memory, just give up.
609 if (bufsize == csp->config->buffer_limit)
611 log_error(LOG_LEVEL_ERROR, "Buffer limit reached while decompressing iob");
612 return JB_ERR_MEMORY;
615 /* Try doubling the buffer size each time. */
618 /* Don't exceed the buffer limit. */
619 if (bufsize > csp->config->buffer_limit)
621 bufsize = csp->config->buffer_limit;
624 /* Try to allocate the new buffer. */
625 tmpbuf = realloc(buf, bufsize);
628 log_error(LOG_LEVEL_ERROR, "Out of memory decompressing iob");
630 return JB_ERR_MEMORY;
634 char *oldnext_out = (char *)zstr.next_out;
637 * Update the fields for inflate() to use the new
638 * buffer, which may be in a location different from
641 zstr.avail_out += bufsize - oldbufsize;
642 zstr.next_out = (Bytef *)tmpbuf + bufsize - zstr.avail_out;
645 * Compare with an uglier method of calculating these values
646 * that doesn't require the extra oldbufsize variable.
648 assert(zstr.avail_out == tmpbuf + bufsize - (char *)zstr.next_out);
649 assert((char *)zstr.next_out == tmpbuf + ((char *)oldnext_out - buf));
650 assert(zstr.avail_out > 0U);
656 if (Z_STREAM_ERROR == inflateEnd(&zstr))
658 log_error(LOG_LEVEL_ERROR,
659 "Inconsistent stream state after decompression: %s", zstr.msg);
661 * XXX: Intentionally no return.
663 * According to zlib.h, Z_STREAM_ERROR is returned
664 * "if the stream state was inconsistent".
666 * I assume in this case inflate()'s status
667 * would also be something different than Z_STREAM_END
668 * so this check should be redundant, but lets see.
672 if ((status != Z_STREAM_END) && (0 != zstr.avail_in))
675 * We failed to decompress the stream and it's
676 * not simply because of missing data.
678 log_error(LOG_LEVEL_ERROR,
679 "Unexpected error while decompressing to the buffer (iob): %s",
681 return JB_ERR_COMPRESS;
685 * Finally, we can actually update the iob, since the
686 * decompression was successful. First, free the old
689 freez(csp->iob->buf);
691 /* Now, update the iob to use the new buffer. */
693 csp->iob->cur = csp->iob->buf + skip_size;
694 csp->iob->eod = (char *)zstr.next_out;
695 csp->iob->size = bufsize;
698 * Make sure the new uncompressed iob obeys some minimal
699 * consistency conditions.
701 if ((csp->iob->buf < csp->iob->cur)
702 && (csp->iob->cur <= csp->iob->eod)
703 && (csp->iob->eod <= csp->iob->buf + csp->iob->size))
705 const size_t new_size = (size_t)(csp->iob->eod - csp->iob->cur);
706 if (new_size > (size_t)0)
708 log_error(LOG_LEVEL_RE_FILTER,
709 "Decompression successful. Old size: %d, new size: %d.",
714 /* zlib thinks this is OK, so lets do the same. */
715 log_error(LOG_LEVEL_INFO, "Decompression didn't result in any content.");
720 /* It seems that zlib did something weird. */
721 log_error(LOG_LEVEL_ERROR,
722 "Unexpected error decompressing the buffer (iob): %d==%d, %d>%d, %d<%d",
723 csp->iob->cur, csp->iob->buf + skip_size, csp->iob->eod, csp->iob->buf,
724 csp->iob->eod, csp->iob->buf + csp->iob->size);
725 return JB_ERR_COMPRESS;
731 #endif /* defined(FEATURE_ZLIB) */
734 /*********************************************************************
736 * Function : string_move
738 * Description : memmove wrapper to move the last part of a string
739 * towards the beginning, overwriting the part in
740 * the middle. strlcpy() can't be used here as the
744 * 1 : dst = Destination to overwrite
745 * 2 : src = Source to move.
749 *********************************************************************/
750 static void string_move(char *dst, char *src)
754 /* +1 to copy the terminating nul as well. */
755 memmove(dst, src, strlen(src)+1);
759 /*********************************************************************
761 * Function : normalize_lws
763 * Description : Reduces unquoted linear white space in headers
764 * to a single space in accordance with RFC 2616 2.2.
765 * This simplifies parsing and filtering later on.
767 * XXX: Remove log messages before
768 * the next stable release?
771 * 1 : header = A header with linear white space to reduce.
775 *********************************************************************/
776 static void normalize_lws(char *header)
782 if (ijb_isspace(*p) && ijb_isspace(*(p+1)))
786 while (ijb_isspace(*q))
790 log_error(LOG_LEVEL_HEADER, "Reducing white space in '%s'", header);
796 log_error(LOG_LEVEL_HEADER,
797 "Converting tab to space in '%s'", header);
802 char *end_of_token = strstr(p+1, "\"");
804 if (NULL != end_of_token)
806 /* Don't mess with quoted text. */
811 log_error(LOG_LEVEL_HEADER,
812 "Ignoring single quote in '%s'", header);
818 p = strchr(header, ':');
819 if ((p != NULL) && (p != header) && ijb_isspace(*(p-1)))
822 * There's still space before the colon.
830 /*********************************************************************
832 * Function : get_header
834 * Description : This (odd) routine will parse the csp->iob
835 * to get the next complete header.
838 * 1 : iob = The I/O buffer to parse, usually csp->iob.
840 * Returns : Any one of the following:
842 * 1) a pointer to a dynamically allocated string that contains a header line
843 * 2) NULL indicating that the end of the header was reached
844 * 3) "" indicating that the end of the iob was reached before finding
845 * a complete header line.
847 *********************************************************************/
848 char *get_header(struct iob *iob)
852 header = get_header_line(iob);
854 if ((header == NULL) || (*header == '\0'))
857 * No complete header read yet, tell the client.
862 while ((iob->cur[0] == ' ') || (iob->cur[0] == '\t'))
865 * Header spans multiple lines, append the next one.
867 char *continued_header;
869 continued_header = get_header_line(iob);
870 if ((continued_header == NULL) || (*continued_header == '\0'))
873 * No complete header read yet, return what we got.
874 * XXX: Should "unread" header instead.
876 log_error(LOG_LEVEL_INFO,
877 "Failed to read a multi-line header properly: '%s'",
882 if (JB_ERR_OK != string_join(&header, continued_header))
884 log_error(LOG_LEVEL_FATAL,
885 "Out of memory while appending multiple headers.");
889 /* XXX: remove before next stable release. */
890 log_error(LOG_LEVEL_HEADER,
891 "Merged multiple header lines to: '%s'",
896 normalize_lws(header);
903 /*********************************************************************
905 * Function : get_header_line
907 * Description : This (odd) routine will parse the csp->iob
908 * to get the next header line.
911 * 1 : iob = The I/O buffer to parse, usually csp->iob.
913 * Returns : Any one of the following:
915 * 1) a pointer to a dynamically allocated string that contains a header line
916 * 2) NULL indicating that the end of the header was reached
917 * 3) "" indicating that the end of the iob was reached before finding
918 * a complete header line.
920 *********************************************************************/
921 static char *get_header_line(struct iob *iob)
925 if ((iob->cur == NULL)
926 || ((p = strchr(iob->cur, '\n')) == NULL))
928 return(""); /* couldn't find a complete header */
933 ret = strdup(iob->cur);
936 /* FIXME No way to handle error properly */
937 log_error(LOG_LEVEL_FATAL, "Out of memory in get_header_line()");
943 if ((q = strchr(ret, '\r')) != NULL) *q = '\0';
945 /* is this a blank line (i.e. the end of the header) ? */
957 /*********************************************************************
959 * Function : get_header_value
961 * Description : Get the value of a given header from a chained list
962 * of header lines or return NULL if no such header is
963 * present in the list.
966 * 1 : header_list = pointer to list
967 * 2 : header_name = string with name of header to look for.
968 * Trailing colon required, capitalization
971 * Returns : NULL if not found, else value of header
973 *********************************************************************/
974 char *get_header_value(const struct list *header_list, const char *header_name)
976 struct list_entry *cur_entry;
982 length = strlen(header_name);
984 for (cur_entry = header_list->first; cur_entry ; cur_entry = cur_entry->next)
988 if (!strncmpic(cur_entry->str, header_name, length))
991 * Found: return pointer to start of value
993 ret = cur_entry->str + length;
994 while (*ret && ijb_isspace(*ret)) ret++;
1008 /*********************************************************************
1010 * Function : scan_headers
1012 * Description : Scans headers, applies tags and updates action bits.
1015 * 1 : csp = Current client state (buffers, headers, etc...)
1017 * Returns : JB_ERR_OK
1019 *********************************************************************/
1020 static jb_err scan_headers(struct client_state *csp)
1022 struct list_entry *h; /* Header */
1023 jb_err err = JB_ERR_OK;
1025 for (h = csp->headers->first; (err == JB_ERR_OK) && (h != NULL) ; h = h->next)
1027 /* Header crunch()ed in previous run? -> ignore */
1028 if (h->str == NULL) continue;
1029 log_error(LOG_LEVEL_HEADER, "scan: %s", h->str);
1030 err = header_tagger(csp, h->str);
1037 /*********************************************************************
1041 * Description : add, delete or modify lines in the HTTP header streams.
1042 * On entry, it receives a linked list of headers space
1043 * that was allocated dynamically (both the list nodes
1044 * and the header contents).
1046 * As a side effect it frees the space used by the original
1050 * 1 : csp = Current client state (buffers, headers, etc...)
1051 * 2 : filter_server_headers = Boolean to switch between
1052 * server and header filtering.
1054 * Returns : JB_ERR_OK in case off success, or
1055 * JB_ERR_MEMORY on out-of-memory error.
1057 *********************************************************************/
1058 jb_err sed(struct client_state *csp, int filter_server_headers)
1060 /* XXX: use more descriptive names. */
1061 struct list_entry *p;
1062 const struct parsers *v;
1063 const add_header_func_ptr *f;
1064 jb_err err = JB_ERR_OK;
1066 if (filter_server_headers)
1068 v = server_patterns;
1069 f = add_server_headers;
1073 v = client_patterns;
1074 f = add_client_headers;
1079 while ((err == JB_ERR_OK) && (v->str != NULL))
1081 for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL); p = p->next)
1083 /* Header crunch()ed in previous run? -> ignore */
1084 if (p->str == NULL) continue;
1086 /* Does the current parser handle this header? */
1087 if ((strncmpic(p->str, v->str, v->len) == 0) ||
1088 (v->len == CHECK_EVERY_HEADER_REMAINING))
1090 err = v->parser(csp, &(p->str));
1096 /* place additional headers on the csp->headers list */
1097 while ((err == JB_ERR_OK) && (*f))
1107 /*********************************************************************
1109 * Function : update_server_headers
1111 * Description : Updates server headers after the body has been modified.
1114 * 1 : csp = Current client state (buffers, headers, etc...)
1116 * Returns : JB_ERR_OK in case off success, or
1117 * JB_ERR_MEMORY on out-of-memory error.
1119 *********************************************************************/
1120 jb_err update_server_headers(struct client_state *csp)
1122 jb_err err = JB_ERR_OK;
1124 static const struct parsers server_patterns_light[] = {
1125 { "Content-Length:", 15, server_adjust_content_length },
1126 { "Transfer-Encoding:", 18, server_transfer_coding },
1128 { "Content-Encoding:", 17, server_content_encoding },
1129 #endif /* def FEATURE_ZLIB */
1133 if (strncmpic(csp->http->cmd, "HEAD", 4))
1135 const struct parsers *v;
1136 struct list_entry *p;
1138 for (v = server_patterns_light; (err == JB_ERR_OK) && (v->str != NULL); v++)
1140 for (p = csp->headers->first; (err == JB_ERR_OK) && (p != NULL); p = p->next)
1142 /* Header crunch()ed in previous run? -> ignore */
1143 if (p->str == NULL) continue;
1145 /* Does the current parser handle this header? */
1146 if (strncmpic(p->str, v->str, v->len) == 0)
1148 err = v->parser(csp, (char **)&(p->str));
1154 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
1155 if ((JB_ERR_OK == err)
1156 && (csp->flags & CSP_FLAG_MODIFIED)
1157 && (csp->flags & CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE)
1158 && !(csp->flags & CSP_FLAG_SERVER_CONTENT_LENGTH_SET))
1161 * XXX: lame. should factor the regeneration code out
1162 * of server_adjust_content_length instead.
1164 log_error(LOG_LEVEL_HEADER,
1165 "Content modified with no Content-Length header set. "
1166 "Creating a fake one for adjustment later on.");
1167 err = enlist(csp->headers, "Content-Length: 0");
1168 if (JB_ERR_OK == err)
1170 err = server_adjust_content_length(csp, &(csp->headers->last->str));
1173 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
1179 /*********************************************************************
1181 * Function : header_tagger
1183 * Description : Executes all text substitutions from applying
1184 * tag actions and saves the result as tag.
1186 * XXX: Shares enough code with filter_header() and
1187 * pcrs_filter_response() to warrant some helper functions.
1190 * 1 : csp = Current client state (buffers, headers, etc...)
1191 * 2 : header = Header that is used as tagger input
1193 * Returns : JB_ERR_OK on success and always succeeds
1195 *********************************************************************/
1196 static jb_err header_tagger(struct client_state *csp, char *header)
1198 int wanted_filter_type;
1199 int multi_action_index;
1203 struct file_list *fl;
1204 struct re_filterfile_spec *b;
1205 struct list_entry *tag_name;
1207 int found_filters = 0;
1208 const size_t header_length = strlen(header);
1210 if (csp->flags & CSP_FLAG_CLIENT_HEADER_PARSING_DONE)
1212 wanted_filter_type = FT_SERVER_HEADER_TAGGER;
1213 multi_action_index = ACTION_MULTI_SERVER_HEADER_TAGGER;
1217 wanted_filter_type = FT_CLIENT_HEADER_TAGGER;
1218 multi_action_index = ACTION_MULTI_CLIENT_HEADER_TAGGER;
1221 /* Check if there are any filters */
1222 for (i = 0; i < MAX_AF_FILES; i++)
1235 if (0 == found_filters)
1237 log_error(LOG_LEVEL_ERROR, "Inconsistent configuration: "
1238 "tagging enabled, but no taggers available.");
1242 for (i = 0; i < MAX_AF_FILES; i++)
1245 if ((NULL == fl) || (NULL == fl->f))
1248 * Either there are no filter files
1249 * left, or this filter file just
1250 * contains no valid filters.
1252 * Continue to be sure we don't miss
1253 * valid filter files that are chained
1254 * after empty or invalid ones.
1259 /* For all filters, */
1260 for (b = fl->f; b; b = b->next)
1262 if (b->type != wanted_filter_type)
1264 /* skip the ones we don't care about, */
1267 /* leaving only taggers that could apply, of which we use the ones, */
1268 for (tag_name = csp->action->multi[multi_action_index]->first;
1269 NULL != tag_name; tag_name = tag_name->next)
1271 /* that do apply, and */
1272 if (strcmp(b->name, tag_name->str) == 0)
1274 char *modified_tag = NULL;
1276 size_t size = header_length;
1277 pcrs_job *joblist = b->joblist;
1279 if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
1281 if (NULL == joblist)
1283 log_error(LOG_LEVEL_RE_FILTER,
1284 "Tagger %s has empty joblist. Nothing to do.", b->name);
1288 /* execute their pcrs_joblist on the header. */
1289 for (job = joblist; NULL != job; job = job->next)
1291 const int hits = pcrs_execute(job, tag, size, &modified_tag, &size);
1295 /* Success, continue with the modified version. */
1304 /* Tagger doesn't match */
1307 /* Regex failure, log it but continue anyway. */
1308 assert(NULL != header);
1309 log_error(LOG_LEVEL_ERROR,
1310 "Problems with tagger \'%s\' and header \'%s\': %s",
1311 b->name, *header, pcrs_strerror(hits));
1313 freez(modified_tag);
1317 if (b->dynamic) pcrs_free_joblist(joblist);
1319 /* If this tagger matched */
1325 * There is to technical limitation which makes
1326 * it impossible to use empty tags, but I assume
1327 * no one would do it intentionally.
1330 log_error(LOG_LEVEL_INFO,
1331 "Tagger \'%s\' created an empty tag. Ignored.",
1336 if (!list_contains_item(csp->tags, tag))
1338 if (JB_ERR_OK != enlist(csp->tags, tag))
1340 log_error(LOG_LEVEL_ERROR,
1341 "Insufficient memory to add tag \'%s\', "
1342 "based on tagger \'%s\' and header \'%s\'",
1343 tag, b->name, *header);
1347 char *action_message;
1349 * update the action bits right away, to make
1350 * tagging based on tags set by earlier taggers
1351 * of the same kind possible.
1353 if (update_action_bits_for_tag(csp, tag))
1355 action_message = "Action bits updated accordingly.";
1359 action_message = "No action bits update necessary.";
1362 log_error(LOG_LEVEL_HEADER,
1363 "Tagger \'%s\' added tag \'%s\'. %s",
1364 b->name, tag, action_message);
1369 /* XXX: Is this log-worthy? */
1370 log_error(LOG_LEVEL_HEADER,
1371 "Tagger \'%s\' didn't add tag \'%s\'. "
1372 "Tag already present", b->name, tag);
1375 } /* if the tagger matched */
1376 } /* if the tagger applies */
1377 } /* for every tagger that could apply */
1378 } /* for all filters */
1379 } /* for all filter files */
1384 /* here begins the family of parser functions that reformat header lines */
1386 /*********************************************************************
1388 * Function : filter_header
1390 * Description : Executes all text substitutions from all applying
1391 * +(server|client)-header-filter actions on the header.
1392 * Most of the code was copied from pcrs_filter_response,
1393 * including the rather short variable names
1396 * 1 : csp = Current client state (buffers, headers, etc...)
1397 * 2 : header = On input, pointer to header to modify.
1398 * On output, pointer to the modified header, or NULL
1399 * to remove the header. This function frees the
1400 * original string if necessary.
1402 * Returns : JB_ERR_OK on success and always succeeds
1404 *********************************************************************/
1405 static jb_err filter_header(struct client_state *csp, char **header)
1409 size_t size = strlen(*header);
1411 char *newheader = NULL;
1414 struct file_list *fl;
1415 struct re_filterfile_spec *b;
1416 struct list_entry *filtername;
1418 int i, found_filters = 0;
1419 int wanted_filter_type;
1420 int multi_action_index;
1422 if (csp->flags & CSP_FLAG_NO_FILTERING)
1427 if (csp->flags & CSP_FLAG_CLIENT_HEADER_PARSING_DONE)
1429 wanted_filter_type = FT_SERVER_HEADER_FILTER;
1430 multi_action_index = ACTION_MULTI_SERVER_HEADER_FILTER;
1434 wanted_filter_type = FT_CLIENT_HEADER_FILTER;
1435 multi_action_index = ACTION_MULTI_CLIENT_HEADER_FILTER;
1439 * Need to check the set of re_filterfiles...
1441 for (i = 0; i < MAX_AF_FILES; i++)
1454 if (0 == found_filters)
1456 log_error(LOG_LEVEL_ERROR, "Inconsistent configuration: "
1457 "header filtering enabled, but no matching filters available.");
1461 for (i = 0; i < MAX_AF_FILES; i++)
1464 if ((NULL == fl) || (NULL == fl->f))
1467 * Either there are no filter files
1468 * left, or this filter file just
1469 * contains no valid filters.
1471 * Continue to be sure we don't miss
1472 * valid filter files that are chained
1473 * after empty or invalid ones.
1478 * For all applying +filter actions, look if a filter by that
1479 * name exists and if yes, execute its pcrs_joblist on the
1482 for (b = fl->f; b; b = b->next)
1484 if (b->type != wanted_filter_type)
1486 /* Skip other filter types */
1490 for (filtername = csp->action->multi[multi_action_index]->first;
1491 filtername ; filtername = filtername->next)
1493 if (strcmp(b->name, filtername->str) == 0)
1495 int current_hits = 0;
1496 pcrs_job *joblist = b->joblist;
1498 if (b->dynamic) joblist = compile_dynamic_pcrs_job_list(csp, b);
1500 if (NULL == joblist)
1502 log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->name);
1506 log_error(LOG_LEVEL_RE_FILTER, "filtering \'%s\' (size %d) with \'%s\' ...",
1507 *header, size, b->name);
1509 /* Apply all jobs from the joblist */
1510 for (job = joblist; NULL != job; job = job->next)
1512 matches = pcrs_execute(job, *header, size, &newheader, &size);
1515 current_hits += matches;
1516 log_error(LOG_LEVEL_HEADER, "Transforming \"%s\" to \"%s\"", *header, newheader);
1518 *header = newheader;
1520 else if ( 0 == matches )
1522 /* Filter doesn't change header */
1528 log_error(LOG_LEVEL_ERROR, "Filtering \'%s\' with \'%s\' didn't work out: %s",
1529 *header, b->name, pcrs_strerror(matches));
1530 if (newheader != NULL)
1532 log_error(LOG_LEVEL_ERROR, "Freeing what's left: %s", newheader);
1538 if (b->dynamic) pcrs_free_joblist(joblist);
1540 log_error(LOG_LEVEL_RE_FILTER, "... produced %d hits (new size %d).", current_hits, size);
1541 hits += current_hits;
1548 * Additionally checking for hits is important because if
1549 * the continue hack is triggered, server headers can
1550 * arrive empty to separate multiple heads from each other.
1552 if ((0 == size) && hits)
1554 log_error(LOG_LEVEL_HEADER, "Removing empty header %s", *header);
1562 /*********************************************************************
1564 * Function : server_connection
1566 * Description : Makes sure a proper "Connection:" header is
1567 * set and signals connection_header_adder to
1571 * 1 : csp = Current client state (buffers, headers, etc...)
1572 * 2 : header = On input, pointer to header to modify.
1573 * On output, pointer to the modified header, or NULL
1574 * to remove the header. This function frees the
1575 * original string if necessary.
1577 * Returns : JB_ERR_OK on success, or
1578 * JB_ERR_MEMORY on out-of-memory error.
1580 *********************************************************************/
1581 static jb_err server_connection(struct client_state *csp, char **header)
1583 if (!strcmpic(*header, "Connection: keep-alive"))
1585 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
1586 if ((csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE))
1588 csp->flags |= CSP_FLAG_SERVER_CONNECTION_KEEP_ALIVE;
1591 if ((csp->flags & CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE))
1593 log_error(LOG_LEVEL_HEADER,
1594 "Keeping the server header '%s' around.", *header);
1597 #endif /* FEATURE_CONNECTION_KEEP_ALIVE */
1599 char *old_header = *header;
1601 *header = strdup("Connection: close");
1604 return JB_ERR_MEMORY;
1606 log_error(LOG_LEVEL_HEADER, "Replaced: \'%s\' with \'%s\'", old_header, *header);
1611 /* Signal server_connection_adder() to return early. */
1612 csp->flags |= CSP_FLAG_SERVER_CONNECTION_HEADER_SET;
1618 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
1619 /*********************************************************************
1621 * Function : server_keep_alive
1623 * Description : Stores the servers keep alive timeout.
1626 * 1 : csp = Current client state (buffers, headers, etc...)
1627 * 2 : header = On input, pointer to header to modify.
1628 * On output, pointer to the modified header, or NULL
1629 * to remove the header. This function frees the
1630 * original string if necessary.
1632 * Returns : JB_ERR_OK.
1634 *********************************************************************/
1635 static jb_err server_keep_alive(struct client_state *csp, char **header)
1637 unsigned int keep_alive_timeout;
1638 const char *timeout_position = strstr(*header, "timeout=");
1640 if ((NULL == timeout_position)
1641 || (1 != sscanf(timeout_position, "timeout=%u", &keep_alive_timeout)))
1643 log_error(LOG_LEVEL_ERROR, "Couldn't parse: %s", *header);
1647 if (keep_alive_timeout < csp->server_connection.keep_alive_timeout)
1649 log_error(LOG_LEVEL_HEADER,
1650 "Reducing keep-alive timeout from %u to %u.",
1651 csp->server_connection.keep_alive_timeout, keep_alive_timeout);
1652 csp->server_connection.keep_alive_timeout = keep_alive_timeout;
1656 /* XXX: Is this log worthy? */
1657 log_error(LOG_LEVEL_HEADER,
1658 "Server keep-alive timeout is %u. Sticking with %u.",
1659 keep_alive_timeout, csp->server_connection.keep_alive_timeout);
1665 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
1669 /*********************************************************************
1671 * Function : client_connection
1673 * Description : Makes sure a proper "Connection:" header is
1674 * set and signals connection_header_adder
1678 * 1 : csp = Current client state (buffers, headers, etc...)
1679 * 2 : header = On input, pointer to header to modify.
1680 * On output, pointer to the modified header, or NULL
1681 * to remove the header. This function frees the
1682 * original string if necessary.
1684 * Returns : JB_ERR_OK on success, or
1685 * JB_ERR_MEMORY on out-of-memory error.
1687 *********************************************************************/
1688 static jb_err client_connection(struct client_state *csp, char **header)
1690 const char *wanted_header = get_appropiate_connection_header(csp);
1692 if (strcmpic(*header, wanted_header))
1694 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
1695 if (!(csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_SHARING))
1697 log_error(LOG_LEVEL_HEADER,
1698 "Keeping the client header '%s' around. "
1699 "The connection will not be kept alive.",
1703 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
1705 char *old_header = *header;
1707 *header = strdup(wanted_header);
1710 return JB_ERR_MEMORY;
1712 log_error(LOG_LEVEL_HEADER,
1713 "Replaced: \'%s\' with \'%s\'", old_header, *header);
1717 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
1720 log_error(LOG_LEVEL_HEADER,
1721 "Keeping the client header '%s' around. "
1722 "The server connection will be kept alive if possible.",
1724 csp->flags |= CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE;
1726 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
1728 /* Signal client_connection_adder() to return early. */
1729 csp->flags |= CSP_FLAG_CLIENT_CONNECTION_HEADER_SET;
1735 /*********************************************************************
1737 * Function : crumble
1739 * Description : This is called if a header matches a pattern to "crunch"
1742 * 1 : csp = Current client state (buffers, headers, etc...)
1743 * 2 : header = On input, pointer to header to modify.
1744 * On output, pointer to the modified header, or NULL
1745 * to remove the header. This function frees the
1746 * original string if necessary.
1748 * Returns : JB_ERR_OK on success, or
1749 * JB_ERR_MEMORY on out-of-memory error.
1751 *********************************************************************/
1752 static jb_err crumble(struct client_state *csp, char **header)
1755 log_error(LOG_LEVEL_HEADER, "crumble crunched: %s!", *header);
1761 /*********************************************************************
1763 * Function : crunch_server_header
1765 * Description : Crunch server header if it matches a string supplied by the
1766 * user. Called from `sed'.
1769 * 1 : csp = Current client state (buffers, headers, etc...)
1770 * 2 : header = On input, pointer to header to modify.
1771 * On output, pointer to the modified header, or NULL
1772 * to remove the header. This function frees the
1773 * original string if necessary.
1775 * Returns : JB_ERR_OK on success and always succeeds
1777 *********************************************************************/
1778 static jb_err crunch_server_header(struct client_state *csp, char **header)
1780 const char *crunch_pattern;
1782 /* Do we feel like crunching? */
1783 if ((csp->action->flags & ACTION_CRUNCH_SERVER_HEADER))
1785 crunch_pattern = csp->action->string[ACTION_STRING_SERVER_HEADER];
1787 /* Is the current header the lucky one? */
1788 if (strstr(*header, crunch_pattern))
1790 log_error(LOG_LEVEL_HEADER, "Crunching server header: %s (contains: %s)", *header, crunch_pattern);
1799 /*********************************************************************
1801 * Function : server_content_type
1803 * Description : Set the content-type for filterable types (text/.*,
1804 * .*xml.*, javascript and image/gif) unless filtering has been
1805 * forbidden (CT_TABOO) while parsing earlier headers.
1806 * NOTE: Since text/plain is commonly used by web servers
1807 * for files whose correct type is unknown, we don't
1808 * set CT_TEXT for it.
1811 * 1 : csp = Current client state (buffers, headers, etc...)
1812 * 2 : header = On input, pointer to header to modify.
1813 * On output, pointer to the modified header, or NULL
1814 * to remove the header. This function frees the
1815 * original string if necessary.
1817 * Returns : JB_ERR_OK on success, or
1818 * JB_ERR_MEMORY on out-of-memory error.
1820 *********************************************************************/
1821 static jb_err server_content_type(struct client_state *csp, char **header)
1823 /* Remove header if it isn't the first Content-Type header */
1824 if ((csp->content_type & CT_DECLARED))
1827 * Another, slightly slower, way to see if
1828 * we already parsed another Content-Type header.
1830 assert(NULL != get_header_value(csp->headers, "Content-Type:"));
1832 log_error(LOG_LEVEL_ERROR,
1833 "Multiple Content-Type headers. Removing and ignoring: \'%s\'",
1841 * Signal that the Content-Type has been set.
1843 csp->content_type |= CT_DECLARED;
1845 if (!(csp->content_type & CT_TABOO))
1848 * XXX: The assumption that text/plain is a sign of
1849 * binary data seems to be somewhat unreasonable nowadays
1850 * and should be dropped after 3.0.8 is out.
1852 if ((strstr(*header, "text/") && !strstr(*header, "plain"))
1853 || strstr(*header, "xml")
1854 || strstr(*header, "application/x-javascript"))
1856 csp->content_type |= CT_TEXT;
1858 else if (strstr(*header, "image/gif"))
1860 csp->content_type |= CT_GIF;
1865 * Are we messing with the content type?
1867 if (csp->action->flags & ACTION_CONTENT_TYPE_OVERWRITE)
1870 * Make sure the user doesn't accidently
1871 * change the content type of binary documents.
1873 if ((csp->content_type & CT_TEXT) || (csp->action->flags & ACTION_FORCE_TEXT_MODE))
1876 *header = strdup("Content-Type: ");
1877 string_append(header, csp->action->string[ACTION_STRING_CONTENT_TYPE]);
1881 log_error(LOG_LEVEL_HEADER, "Insufficient memory to replace Content-Type!");
1882 return JB_ERR_MEMORY;
1884 log_error(LOG_LEVEL_HEADER, "Modified: %s!", *header);
1888 log_error(LOG_LEVEL_HEADER, "%s not replaced. "
1889 "It doesn't look like a content type that should be filtered. "
1890 "Enable force-text-mode if you know what you're doing.", *header);
1898 /*********************************************************************
1900 * Function : server_transfer_coding
1902 * Description : - Prohibit filtering (CT_TABOO) if transfer coding compresses
1903 * - Raise the CSP_FLAG_CHUNKED flag if coding is "chunked"
1904 * - Remove header if body was chunked but has been
1905 * de-chunked for filtering.
1908 * 1 : csp = Current client state (buffers, headers, etc...)
1909 * 2 : header = On input, pointer to header to modify.
1910 * On output, pointer to the modified header, or NULL
1911 * to remove the header. This function frees the
1912 * original string if necessary.
1914 * Returns : JB_ERR_OK on success, or
1915 * JB_ERR_MEMORY on out-of-memory error.
1917 *********************************************************************/
1918 static jb_err server_transfer_coding(struct client_state *csp, char **header)
1921 * Turn off pcrs and gif filtering if body compressed
1923 if (strstr(*header, "gzip") || strstr(*header, "compress") || strstr(*header, "deflate"))
1927 * XXX: Added to test if we could use CT_GZIP and CT_DEFLATE here.
1929 log_error(LOG_LEVEL_INFO, "Marking content type for %s as CT_TABOO because of %s.",
1930 csp->http->cmd, *header);
1931 #endif /* def FEATURE_ZLIB */
1932 csp->content_type = CT_TABOO;
1936 * Raise flag if body chunked
1938 if (strstr(*header, "chunked"))
1940 csp->flags |= CSP_FLAG_CHUNKED;
1943 * If the body was modified, it has been de-chunked first
1944 * and the header must be removed.
1946 * FIXME: If there is more than one transfer encoding,
1947 * only the "chunked" part should be removed here.
1949 if (csp->flags & CSP_FLAG_MODIFIED)
1951 log_error(LOG_LEVEL_HEADER, "Removing: %s", *header);
1960 /*********************************************************************
1962 * Function : server_content_encoding
1964 * Description : This function is run twice for each request,
1965 * unless FEATURE_ZLIB and filtering are disabled.
1967 * The first run is used to check if the content
1968 * is compressed, if FEATURE_ZLIB is disabled
1969 * filtering is then disabled as well, if FEATURE_ZLIB
1970 * is enabled the content is marked for decompression.
1972 * The second run is used to remove the Content-Encoding
1973 * header if the decompression was successful.
1976 * 1 : csp = Current client state (buffers, headers, etc...)
1977 * 2 : header = On input, pointer to header to modify.
1978 * On output, pointer to the modified header, or NULL
1979 * to remove the header. This function frees the
1980 * original string if necessary.
1982 * Returns : JB_ERR_OK on success, or
1983 * JB_ERR_MEMORY on out-of-memory error.
1985 *********************************************************************/
1986 static jb_err server_content_encoding(struct client_state *csp, char **header)
1989 if ((csp->flags & CSP_FLAG_MODIFIED)
1990 && (csp->content_type & (CT_GZIP | CT_DEFLATE)))
1993 * We successfully decompressed the content,
1994 * and have to clean the header now, so the
1995 * client no longer expects compressed data..
1997 * XXX: There is a difference between cleaning
1998 * and removing it completely.
2000 log_error(LOG_LEVEL_HEADER, "Crunching: %s", *header);
2003 else if (strstr(*header, "gzip"))
2005 /* Mark for gzip decompression */
2006 csp->content_type |= CT_GZIP;
2008 else if (strstr(*header, "deflate"))
2010 /* Mark for zlib decompression */
2011 csp->content_type |= CT_DEFLATE;
2013 else if (strstr(*header, "compress"))
2016 * We can't decompress this; therefore we can't filter
2019 csp->content_type |= CT_TABOO;
2021 #else /* !defined(FEATURE_ZLIB) */
2022 if (strstr(*header, "gzip") || strstr(*header, "compress") || strstr(*header, "deflate"))
2025 * Body is compressed, turn off pcrs and gif filtering.
2027 csp->content_type |= CT_TABOO;
2030 * Log a warning if the user expects the content to be filtered.
2032 if ((csp->rlist != NULL) &&
2033 (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER])))
2035 log_error(LOG_LEVEL_INFO,
2036 "Compressed content detected, content filtering disabled. "
2037 "Consider recompiling Privoxy with zlib support or "
2038 "enable the prevent-compression action.");
2041 #endif /* defined(FEATURE_ZLIB) */
2048 /*********************************************************************
2050 * Function : server_adjust_content_length
2052 * Description : Adjust Content-Length header if we modified
2056 * 1 : csp = Current client state (buffers, headers, etc...)
2057 * 2 : header = On input, pointer to header to modify.
2058 * On output, pointer to the modified header, or NULL
2059 * to remove the header. This function frees the
2060 * original string if necessary.
2062 * Returns : JB_ERR_OK on success, or
2063 * JB_ERR_MEMORY on out-of-memory error.
2065 *********************************************************************/
2066 static jb_err server_adjust_content_length(struct client_state *csp, char **header)
2068 const size_t max_header_length = 80;
2070 /* Regenerate header if the content was modified. */
2071 if (csp->flags & CSP_FLAG_MODIFIED)
2074 *header = (char *) zalloc(max_header_length);
2075 if (*header == NULL)
2077 return JB_ERR_MEMORY;
2080 snprintf(*header, max_header_length, "Content-Length: %d",
2081 (int)csp->content_length);
2082 log_error(LOG_LEVEL_HEADER, "Adjusted Content-Length to %d",
2083 (int)csp->content_length);
2090 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
2091 /*********************************************************************
2093 * Function : server_save_content_length
2095 * Description : Save the Content-Length sent by the server.
2098 * 1 : csp = Current client state (buffers, headers, etc...)
2099 * 2 : header = On input, pointer to header to modify.
2100 * On output, pointer to the modified header, or NULL
2101 * to remove the header. This function frees the
2102 * original string if necessary.
2104 * Returns : JB_ERR_OK on success, or
2105 * JB_ERR_MEMORY on out-of-memory error.
2107 *********************************************************************/
2108 static jb_err server_save_content_length(struct client_state *csp, char **header)
2110 unsigned long long content_length = 0;
2112 assert(*(*header+14) == ':');
2114 if (1 != sscanf(*header+14, ": %llu", &content_length))
2116 log_error(LOG_LEVEL_ERROR, "Crunching invalid header: %s", *header);
2121 csp->expected_content_length = content_length;
2122 csp->flags |= CSP_FLAG_SERVER_CONTENT_LENGTH_SET;
2123 csp->flags |= CSP_FLAG_CONTENT_LENGTH_SET;
2128 #endif /* def FEATURE_CONNECTION_KEEP_ALIVE */
2131 /*********************************************************************
2133 * Function : server_content_md5
2135 * Description : Crumble any Content-MD5 headers if the document was
2136 * modified. FIXME: Should we re-compute instead?
2139 * 1 : csp = Current client state (buffers, headers, etc...)
2140 * 2 : header = On input, pointer to header to modify.
2141 * On output, pointer to the modified header, or NULL
2142 * to remove the header. This function frees the
2143 * original string if necessary.
2145 * Returns : JB_ERR_OK on success, or
2146 * JB_ERR_MEMORY on out-of-memory error.
2148 *********************************************************************/
2149 static jb_err server_content_md5(struct client_state *csp, char **header)
2151 if (csp->flags & CSP_FLAG_MODIFIED)
2153 log_error(LOG_LEVEL_HEADER, "Crunching Content-MD5");
2161 /*********************************************************************
2163 * Function : server_content_disposition
2165 * Description : If enabled, blocks or modifies the "Content-Disposition" header.
2166 * Called from `sed'.
2169 * 1 : csp = Current client state (buffers, headers, etc...)
2170 * 2 : header = On input, pointer to header to modify.
2171 * On output, pointer to the modified header, or NULL
2172 * to remove the header. This function frees the
2173 * original string if necessary.
2175 * Returns : JB_ERR_OK on success, or
2176 * JB_ERR_MEMORY on out-of-memory error.
2178 *********************************************************************/
2179 static jb_err server_content_disposition(struct client_state *csp, char **header)
2184 * Are we messing with the Content-Disposition header?
2186 if ((csp->action->flags & ACTION_HIDE_CONTENT_DISPOSITION) == 0)
2192 newval = csp->action->string[ACTION_STRING_CONTENT_DISPOSITION];
2194 if ((newval == NULL) || (0 == strcmpic(newval, "block")))
2197 * Blocking content-disposition header
2199 log_error(LOG_LEVEL_HEADER, "Crunching %s!", *header);
2206 * Replacing Content-Disposition header
2209 *header = strdup("Content-Disposition: ");
2210 string_append(header, newval);
2212 if (*header != NULL)
2214 log_error(LOG_LEVEL_HEADER,
2215 "Content-Disposition header crunched and replaced with: %s", *header);
2218 return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
2222 /*********************************************************************
2224 * Function : server_last_modified
2226 * Description : Changes Last-Modified header to the actual date
2227 * to help hide-if-modified-since.
2228 * Called from `sed'.
2231 * 1 : csp = Current client state (buffers, headers, etc...)
2232 * 2 : header = On input, pointer to header to modify.
2233 * On output, pointer to the modified header, or NULL
2234 * to remove the header. This function frees the
2235 * original string if necessary.
2237 * Returns : JB_ERR_OK on success, or
2238 * JB_ERR_MEMORY on out-of-memory error.
2240 *********************************************************************/
2241 static jb_err server_last_modified(struct client_state *csp, char **header)
2244 char buf[BUFFER_SIZE];
2247 #ifdef HAVE_GMTIME_R
2250 struct tm *timeptr = NULL;
2251 time_t now, last_modified;
2252 long int days, hours, minutes, seconds;
2255 * Are we messing with the Last-Modified header?
2257 if ((csp->action->flags & ACTION_OVERWRITE_LAST_MODIFIED) == 0)
2263 newval = csp->action->string[ACTION_STRING_LAST_MODIFIED];
2265 if (0 == strcmpic(newval, "block") )
2268 * Blocking Last-Modified header. Useless but why not.
2270 log_error(LOG_LEVEL_HEADER, "Crunching %s!", *header);
2274 else if (0 == strcmpic(newval, "reset-to-request-time"))
2277 * Setting Last-Modified Header to now.
2279 get_http_time(0, buf, sizeof(buf));
2281 *header = strdup("Last-Modified: ");
2282 string_append(header, buf);
2284 if (*header == NULL)
2286 log_error(LOG_LEVEL_HEADER, "Insufficient memory. Last-Modified header got lost, boohoo.");
2290 log_error(LOG_LEVEL_HEADER, "Reset to present time: %s", *header);
2293 else if (0 == strcmpic(newval, "randomize"))
2295 const char *header_time = *header + sizeof("Last-Modified:");
2297 log_error(LOG_LEVEL_HEADER, "Randomizing: %s", *header);
2299 #ifdef HAVE_GMTIME_R
2300 gmtime_r(&now, &gmt);
2301 #elif defined(MUTEX_LOCKS_AVAILABLE)
2302 privoxy_mutex_lock(&gmtime_mutex);
2304 privoxy_mutex_unlock(&gmtime_mutex);
2308 if (JB_ERR_OK != parse_header_time(header_time, &last_modified))
2310 log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s in %s (crunching!)", header_time, *header);
2315 long int rtime = (long int)difftime(now, last_modified);
2318 const int negative_delta = (rtime < 0);
2323 log_error(LOG_LEVEL_HEADER, "Server time in the future.");
2325 rtime = pick_from_range(rtime);
2330 last_modified += rtime;
2331 #ifdef HAVE_GMTIME_R
2332 timeptr = gmtime_r(&last_modified, &gmt);
2333 #elif defined(MUTEX_LOCKS_AVAILABLE)
2334 privoxy_mutex_lock(&gmtime_mutex);
2335 timeptr = gmtime(&last_modified);
2336 privoxy_mutex_unlock(&gmtime_mutex);
2338 timeptr = gmtime(&last_modified);
2340 strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr);
2342 *header = strdup("Last-Modified: ");
2343 string_append(header, newheader);
2345 if (*header == NULL)
2347 log_error(LOG_LEVEL_ERROR, "Insufficient memory, header crunched without replacement.");
2348 return JB_ERR_MEMORY;
2351 days = rtime / (3600 * 24);
2352 hours = rtime / 3600 % 24;
2353 minutes = rtime / 60 % 60;
2354 seconds = rtime % 60;
2356 log_error(LOG_LEVEL_HEADER,
2357 "Randomized: %s (added %d da%s %d hou%s %d minut%s %d second%s",
2358 *header, days, (days == 1) ? "y" : "ys", hours, (hours == 1) ? "r" : "rs",
2359 minutes, (minutes == 1) ? "e" : "es", seconds, (seconds == 1) ? ")" : "s)");
2363 log_error(LOG_LEVEL_HEADER, "Randomized ... or not. No time difference to work with.");
2372 /*********************************************************************
2374 * Function : client_accept_encoding
2376 * Description : Rewrite the client's Accept-Encoding header so that
2377 * if doesn't allow compression, if the action applies.
2378 * Note: For HTTP/1.0 the absence of the header is enough.
2381 * 1 : csp = Current client state (buffers, headers, etc...)
2382 * 2 : header = On input, pointer to header to modify.
2383 * On output, pointer to the modified header, or NULL
2384 * to remove the header. This function frees the
2385 * original string if necessary.
2387 * Returns : JB_ERR_OK on success, or
2388 * JB_ERR_MEMORY on out-of-memory error.
2390 *********************************************************************/
2391 static jb_err client_accept_encoding(struct client_state *csp, char **header)
2393 if ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
2395 log_error(LOG_LEVEL_HEADER, "Suppressed offer to compress content");
2399 /* Temporarily disable the correct behaviour to
2400 * work around a PHP bug.
2402 * if (!strcmpic(csp->http->ver, "HTTP/1.1"))
2404 * *header = strdup("Accept-Encoding: identity;q=1.0, *;q=0");
2405 * if (*header == NULL)
2407 * return JB_ERR_MEMORY;
2418 /*********************************************************************
2420 * Function : client_te
2422 * Description : Rewrite the client's TE header so that
2423 * if doesn't allow compression, if the action applies.
2426 * 1 : csp = Current client state (buffers, headers, etc...)
2427 * 2 : header = On input, pointer to header to modify.
2428 * On output, pointer to the modified header, or NULL
2429 * to remove the header. This function frees the
2430 * original string if necessary.
2432 * Returns : JB_ERR_OK on success, or
2433 * JB_ERR_MEMORY on out-of-memory error.
2435 *********************************************************************/
2436 static jb_err client_te(struct client_state *csp, char **header)
2438 if ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
2441 log_error(LOG_LEVEL_HEADER, "Suppressed offer to compress transfer");
2448 /*********************************************************************
2450 * Function : client_referrer
2452 * Description : Handle the "referer" config setting properly.
2453 * Called from `sed'.
2456 * 1 : csp = Current client state (buffers, headers, etc...)
2457 * 2 : header = On input, pointer to header to modify.
2458 * On output, pointer to the modified header, or NULL
2459 * to remove the header. This function frees the
2460 * original string if necessary.
2462 * Returns : JB_ERR_OK on success, or
2463 * JB_ERR_MEMORY on out-of-memory error.
2465 *********************************************************************/
2466 static jb_err client_referrer(struct client_state *csp, char **header)
2468 const char *parameter;
2469 /* booleans for parameters we have to check multiple times */
2470 int parameter_conditional_block;
2471 int parameter_conditional_forge;
2473 #ifdef FEATURE_FORCE_LOAD
2475 * Since the referrer can include the prefix even
2476 * if the request itself is non-forced, we must
2477 * clean it unconditionally.
2479 * XXX: strclean is too broad
2481 strclean(*header, FORCE_PREFIX);
2482 #endif /* def FEATURE_FORCE_LOAD */
2484 if ((csp->action->flags & ACTION_HIDE_REFERER) == 0)
2486 /* Nothing left to do */
2490 parameter = csp->action->string[ACTION_STRING_REFERER];
2491 assert(parameter != NULL);
2492 parameter_conditional_block = (0 == strcmpic(parameter, "conditional-block"));
2493 parameter_conditional_forge = (0 == strcmpic(parameter, "conditional-forge"));
2495 if (!parameter_conditional_block && !parameter_conditional_forge)
2498 * As conditional-block and conditional-forge are the only
2499 * parameters that rely on the original referrer, we can
2500 * remove it now for all the others.
2505 if (0 == strcmpic(parameter, "block"))
2507 log_error(LOG_LEVEL_HEADER, "Referer crunched!");
2510 else if (parameter_conditional_block || parameter_conditional_forge)
2512 return handle_conditional_hide_referrer_parameter(header,
2513 csp->http->hostport, parameter_conditional_block);
2515 else if (0 == strcmpic(parameter, "forge"))
2517 return create_forged_referrer(header, csp->http->hostport);
2521 /* interpret parameter as user-supplied referer to fake */
2522 return create_fake_referrer(header, parameter);
2527 /*********************************************************************
2529 * Function : client_accept_language
2531 * Description : Handle the "Accept-Language" config setting properly.
2532 * Called from `sed'.
2535 * 1 : csp = Current client state (buffers, headers, etc...)
2536 * 2 : header = On input, pointer to header to modify.
2537 * On output, pointer to the modified header, or NULL
2538 * to remove the header. This function frees the
2539 * original string if necessary.
2541 * Returns : JB_ERR_OK on success, or
2542 * JB_ERR_MEMORY on out-of-memory error.
2544 *********************************************************************/
2545 static jb_err client_accept_language(struct client_state *csp, char **header)
2550 * Are we messing with the Accept-Language?
2552 if ((csp->action->flags & ACTION_HIDE_ACCEPT_LANGUAGE) == 0)
2554 /*I don't think so*/
2558 newval = csp->action->string[ACTION_STRING_LANGUAGE];
2560 if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
2563 * Blocking Accept-Language header
2565 log_error(LOG_LEVEL_HEADER, "Crunching Accept-Language!");
2572 * Replacing Accept-Language header
2575 *header = strdup("Accept-Language: ");
2576 string_append(header, newval);
2578 if (*header == NULL)
2580 log_error(LOG_LEVEL_ERROR,
2581 "Insufficient memory. Accept-Language header crunched without replacement.");
2585 log_error(LOG_LEVEL_HEADER,
2586 "Accept-Language header crunched and replaced with: %s", *header);
2589 return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
2593 /*********************************************************************
2595 * Function : crunch_client_header
2597 * Description : Crunch client header if it matches a string supplied by the
2598 * user. Called from `sed'.
2601 * 1 : csp = Current client state (buffers, headers, etc...)
2602 * 2 : header = On input, pointer to header to modify.
2603 * On output, pointer to the modified header, or NULL
2604 * to remove the header. This function frees the
2605 * original string if necessary.
2607 * Returns : JB_ERR_OK on success and always succeeds
2609 *********************************************************************/
2610 static jb_err crunch_client_header(struct client_state *csp, char **header)
2612 const char *crunch_pattern;
2614 /* Do we feel like crunching? */
2615 if ((csp->action->flags & ACTION_CRUNCH_CLIENT_HEADER))
2617 crunch_pattern = csp->action->string[ACTION_STRING_CLIENT_HEADER];
2619 /* Is the current header the lucky one? */
2620 if (strstr(*header, crunch_pattern))
2622 log_error(LOG_LEVEL_HEADER, "Crunching client header: %s (contains: %s)", *header, crunch_pattern);
2630 /*********************************************************************
2632 * Function : client_uagent
2634 * Description : Handle the "user-agent" config setting properly
2635 * and remember its original value to enable browser
2636 * bug workarounds. Called from `sed'.
2639 * 1 : csp = Current client state (buffers, headers, etc...)
2640 * 2 : header = On input, pointer to header to modify.
2641 * On output, pointer to the modified header, or NULL
2642 * to remove the header. This function frees the
2643 * original string if necessary.
2645 * Returns : JB_ERR_OK on success, or
2646 * JB_ERR_MEMORY on out-of-memory error.
2648 *********************************************************************/
2649 static jb_err client_uagent(struct client_state *csp, char **header)
2653 if ((csp->action->flags & ACTION_HIDE_USER_AGENT) == 0)
2658 newval = csp->action->string[ACTION_STRING_USER_AGENT];
2665 *header = strdup("User-Agent: ");
2666 string_append(header, newval);
2668 log_error(LOG_LEVEL_HEADER, "Modified: %s", *header);
2670 return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
2674 /*********************************************************************
2676 * Function : client_ua
2678 * Description : Handle "ua-" headers properly. Called from `sed'.
2681 * 1 : csp = Current client state (buffers, headers, etc...)
2682 * 2 : header = On input, pointer to header to modify.
2683 * On output, pointer to the modified header, or NULL
2684 * to remove the header. This function frees the
2685 * original string if necessary.
2687 * Returns : JB_ERR_OK on success, or
2688 * JB_ERR_MEMORY on out-of-memory error.
2690 *********************************************************************/
2691 static jb_err client_ua(struct client_state *csp, char **header)
2693 if ((csp->action->flags & ACTION_HIDE_USER_AGENT) != 0)
2695 log_error(LOG_LEVEL_HEADER, "crunched User-Agent!");
2703 /*********************************************************************
2705 * Function : client_from
2707 * Description : Handle the "from" config setting properly.
2708 * Called from `sed'.
2711 * 1 : csp = Current client state (buffers, headers, etc...)
2712 * 2 : header = On input, pointer to header to modify.
2713 * On output, pointer to the modified header, or NULL
2714 * to remove the header. This function frees the
2715 * original string if necessary.
2717 * Returns : JB_ERR_OK on success, or
2718 * JB_ERR_MEMORY on out-of-memory error.
2720 *********************************************************************/
2721 static jb_err client_from(struct client_state *csp, char **header)
2725 if ((csp->action->flags & ACTION_HIDE_FROM) == 0)
2732 newval = csp->action->string[ACTION_STRING_FROM];
2735 * Are we blocking the e-mail address?
2737 if ((newval == NULL) || (0 == strcmpic(newval, "block")) )
2739 log_error(LOG_LEVEL_HEADER, "crunched From!");
2743 log_error(LOG_LEVEL_HEADER, " modified");
2745 *header = strdup("From: ");
2746 string_append(header, newval);
2748 return (*header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK;
2752 /*********************************************************************
2754 * Function : client_send_cookie
2756 * Description : Crunches the "cookie" header if necessary.
2757 * Called from `sed'.
2759 * XXX: Stupid name, doesn't send squat.
2762 * 1 : csp = Current client state (buffers, headers, etc...)
2763 * 2 : header = On input, pointer to header to modify.
2764 * On output, pointer to the modified header, or NULL
2765 * to remove the header. This function frees the
2766 * original string if necessary.
2768 * Returns : JB_ERR_OK on success, or
2769 * JB_ERR_MEMORY on out-of-memory error.
2771 *********************************************************************/
2772 static jb_err client_send_cookie(struct client_state *csp, char **header)
2774 if (csp->action->flags & ACTION_NO_COOKIE_READ)
2776 log_error(LOG_LEVEL_HEADER, "Crunched outgoing cookie: %s", *header);
2784 /*********************************************************************
2786 * Function : client_x_forwarded
2788 * Description : Handle the "x-forwarded-for" config setting properly,
2789 * also used in the add_client_headers list. Called from `sed'.
2792 * 1 : csp = Current client state (buffers, headers, etc...)
2793 * 2 : header = On input, pointer to header to modify.
2794 * On output, pointer to the modified header, or NULL
2795 * to remove the header. This function frees the
2796 * original string if necessary.
2798 * Returns : JB_ERR_OK on success, or
2799 * JB_ERR_MEMORY on out-of-memory error.
2801 *********************************************************************/
2802 jb_err client_x_forwarded(struct client_state *csp, char **header)
2804 if (0 != (csp->action->flags & ACTION_CHANGE_X_FORWARDED_FOR))
2806 const char *parameter = csp->action->string[ACTION_STRING_CHANGE_X_FORWARDED_FOR];
2808 if (0 == strcmpic(parameter, "block"))
2811 log_error(LOG_LEVEL_HEADER, "crunched x-forwarded-for!");
2813 else if (0 == strcmpic(parameter, "add"))
2815 string_append(header, ", ");
2816 string_append(header, csp->ip_addr_str);
2818 if (*header == NULL)
2820 return JB_ERR_MEMORY;
2822 log_error(LOG_LEVEL_HEADER,
2823 "Appended client IP address to %s", *header);
2824 csp->flags |= CSP_FLAG_X_FORWARDED_FOR_APPENDED;
2828 log_error(LOG_LEVEL_FATAL,
2829 "Invalid change-x-forwarded-for parameter: '%s'", parameter);
2837 /*********************************************************************
2839 * Function : client_max_forwards
2841 * Description : If the HTTP method is OPTIONS or TRACE, subtract one
2842 * from the value of the Max-Forwards header field.
2845 * 1 : csp = Current client state (buffers, headers, etc...)
2846 * 2 : header = On input, pointer to header to modify.
2847 * On output, pointer to the modified header, or NULL
2848 * to remove the header. This function frees the
2849 * original string if necessary.
2851 * Returns : JB_ERR_OK on success, or
2852 * JB_ERR_MEMORY on out-of-memory error.
2854 *********************************************************************/
2855 static jb_err client_max_forwards(struct client_state *csp, char **header)
2859 if ((0 == strcmpic(csp->http->gpc, "trace")) ||
2860 (0 == strcmpic(csp->http->gpc, "options")))
2862 assert(*(*header+12) == ':');
2863 if (1 == sscanf(*header+12, ": %d", &max_forwards))
2865 if (max_forwards > 0)
2867 snprintf(*header, strlen(*header)+1, "Max-Forwards: %d", --max_forwards);
2868 log_error(LOG_LEVEL_HEADER,
2869 "Max-Forwards value for %s request reduced to %d.",
2870 csp->http->gpc, max_forwards);
2872 else if (max_forwards < 0)
2874 log_error(LOG_LEVEL_ERROR, "Crunching invalid header: %s", *header);
2880 log_error(LOG_LEVEL_ERROR, "Crunching invalid header: %s", *header);
2889 /*********************************************************************
2891 * Function : client_host
2893 * Description : If the request URI did not contain host and
2894 * port information, parse and evaluate the Host
2897 * Also, kill ill-formed HOST: headers as sent by
2898 * Apple's iTunes software when used with a proxy.
2901 * 1 : csp = Current client state (buffers, headers, etc...)
2902 * 2 : header = On input, pointer to header to modify.
2903 * On output, pointer to the modified header, or NULL
2904 * to remove the header. This function frees the
2905 * original string if necessary.
2907 * Returns : JB_ERR_OK on success, or
2908 * JB_ERR_MEMORY on out-of-memory error.
2910 *********************************************************************/
2911 static jb_err client_host(struct client_state *csp, char **header)
2916 * If the header field name is all upper-case, chances are that it's
2917 * an ill-formed one from iTunes. BTW, killing innocent headers here is
2918 * not a problem -- they are regenerated later.
2920 if ((*header)[1] == 'O')
2922 log_error(LOG_LEVEL_HEADER, "Killed all-caps Host header line: %s", *header);
2927 if (!csp->http->hostport || (*csp->http->hostport == '*') ||
2928 *csp->http->hostport == ' ' || *csp->http->hostport == '\0')
2931 if (NULL == (p = strdup((*header)+6)))
2933 return JB_ERR_MEMORY;
2936 if (NULL == (q = strdup(p)))
2939 return JB_ERR_MEMORY;
2942 freez(csp->http->hostport);
2943 csp->http->hostport = p;
2944 freez(csp->http->host);
2945 csp->http->host = q;
2946 q = strchr(csp->http->host, ':');
2949 /* Terminate hostname and evaluate port string */
2951 csp->http->port = atoi(q);
2955 csp->http->port = csp->http->ssl ? 443 : 80;
2958 log_error(LOG_LEVEL_HEADER, "New host and port from Host field: %s = %s:%d",
2959 csp->http->hostport, csp->http->host, csp->http->port);
2962 /* Signal client_host_adder() to return right away */
2963 csp->flags |= CSP_FLAG_HOST_HEADER_IS_SET;
2969 /*********************************************************************
2971 * Function : client_if_modified_since
2973 * Description : Remove or modify the If-Modified-Since header.
2976 * 1 : csp = Current client state (buffers, headers, etc...)
2977 * 2 : header = On input, pointer to header to modify.
2978 * On output, pointer to the modified header, or NULL
2979 * to remove the header. This function frees the
2980 * original string if necessary.
2982 * Returns : JB_ERR_OK on success, or
2983 * JB_ERR_MEMORY on out-of-memory error.
2985 *********************************************************************/
2986 static jb_err client_if_modified_since(struct client_state *csp, char **header)
2989 #ifdef HAVE_GMTIME_R
2992 struct tm *timeptr = NULL;
2995 long int hours, minutes, seconds;
2998 if ( 0 == strcmpic(*header, "If-Modified-Since: Wed, 08 Jun 1955 12:00:00 GMT"))
3001 * The client got an error message because of a temporary problem,
3002 * the problem is gone and the client now tries to revalidate our
3003 * error message on the real server. The revalidation would always
3004 * end with the transmission of the whole document and there is
3005 * no need to expose the bogus If-Modified-Since header.
3007 log_error(LOG_LEVEL_HEADER, "Crunching useless If-Modified-Since header.");
3010 else if (csp->action->flags & ACTION_HIDE_IF_MODIFIED_SINCE)
3012 newval = csp->action->string[ACTION_STRING_IF_MODIFIED_SINCE];
3014 if ((0 == strcmpic(newval, "block")))
3016 log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
3019 else /* add random value */
3021 const char *header_time = *header + sizeof("If-Modified-Since:");
3023 if (JB_ERR_OK != parse_header_time(header_time, &tm))
3025 log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s in %s (crunching!)", header_time, *header);
3030 long int rtime = strtol(newval, &endptr, 0);
3031 const int negative_range = (rtime < 0);
3035 log_error(LOG_LEVEL_HEADER, "Randomizing: %s (random range: %d minut%s)",
3036 *header, rtime, (rtime == 1 || rtime == -1) ? "e": "es");
3042 rtime = pick_from_range(rtime);
3046 log_error(LOG_LEVEL_ERROR, "Random range is 0. Assuming time transformation test.",
3049 tm += rtime * (negative_range ? -1 : 1);
3050 #ifdef HAVE_GMTIME_R
3051 timeptr = gmtime_r(&tm, &gmt);
3052 #elif defined(MUTEX_LOCKS_AVAILABLE)
3053 privoxy_mutex_lock(&gmtime_mutex);
3054 timeptr = gmtime(&tm);
3055 privoxy_mutex_unlock(&gmtime_mutex);
3057 timeptr = gmtime(&tm);
3059 strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr);
3062 *header = strdup("If-Modified-Since: ");
3063 string_append(header, newheader);
3065 if (*header == NULL)
3067 log_error(LOG_LEVEL_HEADER, "Insufficient memory, header crunched without replacement.");
3068 return JB_ERR_MEMORY;
3071 hours = rtime / 3600;
3072 minutes = rtime / 60 % 60;
3073 seconds = rtime % 60;
3075 log_error(LOG_LEVEL_HEADER,
3076 "Randomized: %s (%s %d hou%s %d minut%s %d second%s",
3077 *header, (negative_range) ? "subtracted" : "added", hours,
3078 (hours == 1) ? "r" : "rs", minutes, (minutes == 1) ? "e" : "es",
3079 seconds, (seconds == 1) ? ")" : "s)");
3088 /*********************************************************************
3090 * Function : client_if_none_match
3092 * Description : Remove the If-None-Match header.
3095 * 1 : csp = Current client state (buffers, headers, etc...)
3096 * 2 : header = On input, pointer to header to modify.
3097 * On output, pointer to the modified header, or NULL
3098 * to remove the header. This function frees the
3099 * original string if necessary.
3101 * Returns : JB_ERR_OK on success, or
3102 * JB_ERR_MEMORY on out-of-memory error.
3104 *********************************************************************/
3105 static jb_err client_if_none_match(struct client_state *csp, char **header)
3107 if (csp->action->flags & ACTION_CRUNCH_IF_NONE_MATCH)
3109 log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
3117 /*********************************************************************
3119 * Function : client_x_filter
3121 * Description : Disables filtering if the client set "X-Filter: No".
3122 * Called from `sed'.
3125 * 1 : csp = Current client state (buffers, headers, etc...)
3126 * 2 : header = On input, pointer to header to modify.
3127 * On output, pointer to the modified header, or NULL
3128 * to remove the header. This function frees the
3129 * original string if necessary.
3131 * Returns : JB_ERR_OK on success
3133 *********************************************************************/
3134 jb_err client_x_filter(struct client_state *csp, char **header)
3136 if ( 0 == strcmpic(*header, "X-Filter: No"))
3138 if (!(csp->config->feature_flags & RUNTIME_FEATURE_HTTP_TOGGLE))
3140 log_error(LOG_LEVEL_INFO, "Ignored the client's request to fetch without filtering.");
3144 if (csp->action->flags & ACTION_FORCE_TEXT_MODE)
3146 log_error(LOG_LEVEL_HEADER,
3147 "force-text-mode overruled the client's request to fetch without filtering!");
3151 csp->content_type = CT_TABOO; /* XXX: This hack shouldn't be necessary */
3152 csp->flags |= CSP_FLAG_NO_FILTERING;
3153 log_error(LOG_LEVEL_HEADER, "Accepted the client's request to fetch without filtering.");
3155 log_error(LOG_LEVEL_HEADER, "Crunching %s", *header);
3163 /*********************************************************************
3165 * Function : client_range
3167 * Description : Removes Range, Request-Range and If-Range headers if
3168 * content filtering is enabled. If the client's version
3169 * of the document has been altered by Privoxy, the server
3170 * could interpret the range differently than the client
3171 * intended in which case the user could end up with
3172 * corrupted content.
3175 * 1 : csp = Current client state (buffers, headers, etc...)
3176 * 2 : header = On input, pointer to header to modify.
3177 * On output, pointer to the modified header, or NULL
3178 * to remove the header. This function frees the
3179 * original string if necessary.
3181 * Returns : JB_ERR_OK
3183 *********************************************************************/
3184 static jb_err client_range(struct client_state *csp, char **header)
3186 if (content_filters_enabled(csp->action))
3188 log_error(LOG_LEVEL_HEADER, "Content filtering is enabled."
3189 " Crunching: \'%s\' to prevent range-mismatch problems.", *header);
3196 /* the following functions add headers directly to the header list */
3198 /*********************************************************************
3200 * Function : client_host_adder
3202 * Description : Adds the Host: header field if it is missing.
3203 * Called from `sed'.
3206 * 1 : csp = Current client state (buffers, headers, etc...)
3208 * Returns : JB_ERR_OK on success, or
3209 * JB_ERR_MEMORY on out-of-memory error.
3211 *********************************************************************/
3212 static jb_err client_host_adder(struct client_state *csp)
3217 if (csp->flags & CSP_FLAG_HOST_HEADER_IS_SET)
3219 /* Header already set by the client, nothing to do. */
3223 if ( !csp->http->hostport || !*(csp->http->hostport))
3225 /* XXX: When does this happen and why is it OK? */
3226 log_error(LOG_LEVEL_INFO, "Weirdness in client_host_adder detected and ignored.");
3231 * remove 'user:pass@' from 'proto://user:pass@host'
3233 if ( (p = strchr( csp->http->hostport, '@')) != NULL )
3239 p = csp->http->hostport;
3242 /* XXX: Just add it, we already made sure that it will be unique */
3243 log_error(LOG_LEVEL_HEADER, "addh-unique: Host: %s", p);
3244 err = enlist_unique_header(csp->headers, "Host", p);
3251 /*********************************************************************
3253 * Function : client_accept_encoding_adder
3255 * Description : Add an Accept-Encoding header to the client's request
3256 * that disables compression if the action applies, and
3257 * the header is not already there. Called from `sed'.
3258 * Note: For HTTP/1.0, the absence of the header is enough.
3261 * 1 : csp = Current client state (buffers, headers, etc...)
3263 * Returns : JB_ERR_OK on success, or
3264 * JB_ERR_MEMORY on out-of-memory error.
3266 *********************************************************************/
3267 static jb_err client_accept_encoding_adder(struct client_state *csp)
3269 if ( ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
3270 && (!strcmpic(csp->http->ver, "HTTP/1.1")) )
3272 return enlist_unique(csp->headers, "Accept-Encoding: identity;q=1.0, *;q=0", 16);
3280 /*********************************************************************
3282 * Function : client_xtra_adder
3284 * Description : Used in the add_client_headers list. Called from `sed'.
3287 * 1 : csp = Current client state (buffers, headers, etc...)
3289 * Returns : JB_ERR_OK on success, or
3290 * JB_ERR_MEMORY on out-of-memory error.
3292 *********************************************************************/
3293 static jb_err client_xtra_adder(struct client_state *csp)
3295 struct list_entry *lst;
3298 for (lst = csp->action->multi[ACTION_MULTI_ADD_HEADER]->first;
3299 lst ; lst = lst->next)
3301 log_error(LOG_LEVEL_HEADER, "addh: %s", lst->str);
3302 err = enlist(csp->headers, lst->str);
3314 /*********************************************************************
3316 * Function : client_x_forwarded_for_adder
3318 * Description : Used in the add_client_headers list. Called from `sed'.
3321 * 1 : csp = Current client state (buffers, headers, etc...)
3323 * Returns : JB_ERR_OK on success, or
3324 * JB_ERR_MEMORY on out-of-memory error.
3326 *********************************************************************/
3327 static jb_err client_x_forwarded_for_adder(struct client_state *csp)
3329 char *header = NULL;
3332 if (!((csp->action->flags & ACTION_CHANGE_X_FORWARDED_FOR)
3333 && (0 == strcmpic(csp->action->string[ACTION_STRING_CHANGE_X_FORWARDED_FOR], "add")))
3334 || (csp->flags & CSP_FLAG_X_FORWARDED_FOR_APPENDED))
3337 * If we aren't adding X-Forwarded-For headers,
3338 * or we already appended an existing X-Forwarded-For
3339 * header, there's nothing left to do here.
3344 header = strdup("X-Forwarded-For: ");
3345 string_append(&header, csp->ip_addr_str);
3349 return JB_ERR_MEMORY;
3352 log_error(LOG_LEVEL_HEADER, "addh: %s", header);
3353 err = enlist(csp->headers, header);
3360 /*********************************************************************
3362 * Function : server_connection_adder
3364 * Description : Adds an appropiate "Connection:" header to csp->headers
3365 * unless the header was already present. Called from `sed'.
3368 * 1 : csp = Current client state (buffers, headers, etc...)
3370 * Returns : JB_ERR_OK on success, or
3371 * JB_ERR_MEMORY on out-of-memory error.
3373 *********************************************************************/
3374 static jb_err server_connection_adder(struct client_state *csp)
3376 const unsigned int flags = csp->flags;
3377 const char *response_status_line = csp->headers->first->str;
3378 const char *wanted_header = get_appropiate_connection_header(csp);
3380 if ((flags & CSP_FLAG_CLIENT_HEADER_PARSING_DONE)
3381 && (flags & CSP_FLAG_SERVER_CONNECTION_HEADER_SET))
3387 * XXX: if we downgraded the response, this check will fail.
3389 if ((csp->config->feature_flags &
3390 RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE)
3391 && (NULL != response_status_line)
3392 && !strncmpic(response_status_line, "HTTP/1.1", 8))
3394 log_error(LOG_LEVEL_HEADER, "A HTTP/1.1 response "
3395 "without Connection header implies keep-alive.");
3396 csp->flags |= CSP_FLAG_SERVER_CONNECTION_KEEP_ALIVE;
3399 log_error(LOG_LEVEL_HEADER, "Adding: %s", wanted_header);
3401 return enlist(csp->headers, wanted_header);
3405 #ifdef FEATURE_CONNECTION_KEEP_ALIVE
3406 /*********************************************************************
3408 * Function : server_proxy_connection_adder
3410 * Description : Adds a "Proxy-Connection: keep-alive" header to
3411 * csp->headers if the client asked for keep-alive.
3412 * XXX: We should reuse existant ones.
3415 * 1 : csp = Current client state (buffers, headers, etc...)
3417 * Returns : JB_ERR_OK on success, or
3418 * JB_ERR_MEMORY on out-of-memory error.
3420 *********************************************************************/
3421 static jb_err server_proxy_connection_adder(struct client_state *csp)
3423 static const char proxy_connection_header[] = "Proxy-Connection: keep-alive";
3424 jb_err err = JB_ERR_OK;
3426 if ((csp->flags & CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE))
3428 log_error(LOG_LEVEL_HEADER, "Adding: %s", proxy_connection_header);
3429 err = enlist(csp->headers, proxy_connection_header);
3434 #endif /* FEATURE_CONNECTION_KEEP_ALIVE */
3437 /*********************************************************************
3439 * Function : client_connection_header_adder
3441 * Description : Adds a proper "Connection:" header to csp->headers
3442 * unless the header was already present. Called from `sed'.
3445 * 1 : csp = Current client state (buffers, headers, etc...)
3447 * Returns : JB_ERR_OK on success, or
3448 * JB_ERR_MEMORY on out-of-memory error.
3450 *********************************************************************/
3451 static jb_err client_connection_header_adder(struct client_state *csp)
3453 const char *wanted_header = get_appropiate_connection_header(csp);
3455 if (!(csp->flags & CSP_FLAG_CLIENT_HEADER_PARSING_DONE)
3456 && (csp->flags & CSP_FLAG_CLIENT_CONNECTION_HEADER_SET))
3461 if ((csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE)
3462 && (csp->http->ssl == 0))
3464 csp->flags |= CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE;
3467 log_error(LOG_LEVEL_HEADER, "Adding: %s", wanted_header);
3469 return enlist(csp->headers, wanted_header);
3473 /*********************************************************************
3475 * Function : server_http
3477 * Description : - Save the HTTP Status into csp->http->status
3478 * - Set CT_TABOO to prevent filtering if the answer
3479 * is a partial range (HTTP status 206)
3480 * - Rewrite HTTP/1.1 answers to HTTP/1.0 if +downgrade
3484 * 1 : csp = Current client state (buffers, headers, etc...)
3485 * 2 : header = On input, pointer to header to modify.
3486 * On output, pointer to the modified header, or NULL
3487 * to remove the header. This function frees the
3488 * original string if necessary.
3490 * Returns : JB_ERR_OK on success, or
3491 * JB_ERR_MEMORY on out-of-memory error.
3493 *********************************************************************/
3494 static jb_err server_http(struct client_state *csp, char **header)
3496 sscanf(*header, "HTTP/%*d.%*d %d", &(csp->http->status));
3497 if (csp->http->status == 206)
3499 csp->content_type = CT_TABOO;
3502 if ((csp->action->flags & ACTION_DOWNGRADE) != 0)
3504 /* XXX: Should we do a real validity check here? */
3505 if (strlen(*header) > 8)
3508 log_error(LOG_LEVEL_HEADER, "Downgraded answer to HTTP/1.0");
3513 * XXX: Should we block the request or
3514 * enlist a valid status code line here?
3516 log_error(LOG_LEVEL_INFO, "Malformed server response detected. "
3517 "Downgrading to HTTP/1.0 impossible.");
3525 /*********************************************************************
3527 * Function : server_set_cookie
3529 * Description : Handle the server "cookie" header properly.
3530 * Log cookie to the jar file. Then "crunch",
3531 * accept or rewrite it to a session cookie.
3532 * Called from `sed'.
3534 * TODO: Allow the user to specify a new expiration
3535 * time to cause the cookie to expire even before the
3536 * browser is closed.
3539 * 1 : csp = Current client state (buffers, headers, etc...)
3540 * 2 : header = On input, pointer to header to modify.
3541 * On output, pointer to the modified header, or NULL
3542 * to remove the header. This function frees the
3543 * original string if necessary.
3545 * Returns : JB_ERR_OK on success, or
3546 * JB_ERR_MEMORY on out-of-memory error.
3548 *********************************************************************/
3549 static jb_err server_set_cookie(struct client_state *csp, char **header)
3556 if ((csp->action->flags & ACTION_NO_COOKIE_SET) != 0)
3558 log_error(LOG_LEVEL_HEADER, "Crunching incoming cookie: %s", *header);
3561 else if ((csp->action->flags & ACTION_NO_COOKIE_KEEP) != 0)
3563 /* Flag whether or not to log a message */
3566 /* A variable to store the tag we're working on */
3569 /* Skip "Set-Cookie:" (11 characters) in header */
3570 cur_tag = *header + 11;
3572 /* skip whitespace between "Set-Cookie:" and value */
3573 while (*cur_tag && ijb_isspace(*cur_tag))
3578 /* Loop through each tag in the cookie */
3582 char *next_tag = strchr(cur_tag, ';');
3583 if (next_tag != NULL)
3585 /* Skip the ';' character itself */
3588 /* skip whitespace ";" and start of tag */
3589 while (*next_tag && ijb_isspace(*next_tag))
3596 /* "Next tag" is the end of the string */
3597 next_tag = cur_tag + strlen(cur_tag);
3601 * Check the expiration date to see
3602 * if the cookie is still valid, if yes,
3603 * rewrite it to a session cookie.
3605 if ((strncmpic(cur_tag, "expires=", 8) == 0) && *(cur_tag + 8))
3607 char *expiration_date = cur_tag + 8; /* Skip "[Ee]xpires=" */
3609 /* Did we detect the date properly? */
3610 if (JB_ERR_OK != parse_header_time(expiration_date, &cookie_time))
3613 * Nope, treat it as if it was still valid.
3615 * XXX: Should we remove the whole cookie instead?
3617 log_error(LOG_LEVEL_ERROR,
3618 "Can't parse \'%s\', send by %s. Unsupported time format?", cur_tag, csp->http->url);
3619 string_move(cur_tag, next_tag);
3625 * Yes. Check if the cookie is still valid.
3627 * If the cookie is already expired it's probably
3628 * a delete cookie and even if it isn't, the browser
3629 * will discard it anyway.
3633 * XXX: timegm() isn't available on some AmigaOS
3634 * versions and our replacement doesn't work.
3636 * Our options are to either:
3638 * - disable session-cookies-only completely if timegm
3641 * - to simply remove all expired tags, like it has
3642 * been done until Privoxy 3.0.6 and to live with
3643 * the consequence that it can cause login/logout
3644 * problems on servers that don't validate their
3645 * input properly, or
3647 * - to replace it with mktime in which
3648 * case there is a slight chance of valid cookies
3649 * passing as already expired.
3651 * This is the way it's currently done and it's not
3652 * as bad as it sounds. If the missing GMT offset is
3653 * enough to change the result of the expiration check
3654 * the cookie will be only valid for a few hours
3655 * anyway, which in many cases will be shorter
3656 * than a browser session.
3658 if (cookie_time - now < 0)
3660 log_error(LOG_LEVEL_HEADER,
3661 "Cookie \'%s\' is already expired and can pass unmodified.", *header);
3662 /* Just in case some clown sets more then one expiration date */
3668 * Still valid, delete expiration date by copying
3669 * the rest of the string over it.
3671 string_move(cur_tag, next_tag);
3673 /* That changed the header, need to issue a log message */
3677 * Note that the next tag has now been moved to *cur_tag,
3678 * so we do not need to update the cur_tag pointer.
3686 /* Move on to next cookie tag */
3693 assert(NULL != *header);
3694 log_error(LOG_LEVEL_HEADER, "Cookie rewritten to a temporary one: %s",
3703 #ifdef FEATURE_FORCE_LOAD
3704 /*********************************************************************
3706 * Function : strclean
3708 * Description : In-Situ-Eliminate all occurances of substring in
3712 * 1 : string = string to clean
3713 * 2 : substring = substring to eliminate
3715 * Returns : Number of eliminations
3717 *********************************************************************/
3718 int strclean(char *string, const char *substring)
3724 len = strlen(substring);
3726 while((pos = strstr(string, substring)) != NULL)
3733 while (*p++ != '\0');
3740 #endif /* def FEATURE_FORCE_LOAD */
3743 /*********************************************************************
3745 * Function : parse_header_time
3747 * Description : Parses time formats used in HTTP header strings
3748 * to get the numerical respresentation.
3751 * 1 : header_time = HTTP header time as string.
3752 * 2 : result = storage for header_time in seconds
3754 * Returns : JB_ERR_OK if the time format was recognized, or
3755 * JB_ERR_PARSE otherwise.
3757 *********************************************************************/
3758 static jb_err parse_header_time(const char *header_time, time_t *result)
3763 * Zero out gmt to prevent time zone offsets.
3765 * While this is only necessary on some platforms
3766 * (mingw32 for example), I don't know how to
3767 * detect these automatically and doing it everywhere
3770 memset(&gmt, 0, sizeof(gmt));
3772 /* Tue, 02 Jun 2037 20:00:00 */
3773 if ((NULL == strptime(header_time, "%a, %d %b %Y %H:%M:%S", &gmt))
3774 /* Tue, 02-Jun-2037 20:00:00 */
3775 && (NULL == strptime(header_time, "%a, %d-%b-%Y %H:%M:%S", &gmt))
3776 /* Tue, 02-Jun-37 20:00:00 */
3777 && (NULL == strptime(header_time, "%a, %d-%b-%y %H:%M:%S", &gmt))
3778 /* Tuesday, 02-Jun-2037 20:00:00 */
3779 && (NULL == strptime(header_time, "%A, %d-%b-%Y %H:%M:%S", &gmt))
3780 /* Tuesday Jun 02 20:00:00 2037 */
3781 && (NULL == strptime(header_time, "%A %b %d %H:%M:%S %Y", &gmt)))
3783 return JB_ERR_PARSE;
3786 *result = timegm(&gmt);
3793 /*********************************************************************
3795 * Function : get_destination_from_headers
3797 * Description : Parse the "Host:" header to get the request's destination.
3798 * Only needed if the client's request was forcefully
3799 * redirected into Privoxy.
3801 * Code mainly copied from client_host() which is currently
3802 * run too late for this purpose.
3805 * 1 : headers = List of headers (one of them hopefully being
3806 * the "Host:" header)
3807 * 2 : http = storage for the result (host, port and hostport).
3809 * Returns : JB_ERR_MEMORY in case of memory problems,
3810 * JB_ERR_PARSE if the host header couldn't be found,
3811 * JB_ERR_OK otherwise.
3813 *********************************************************************/
3814 jb_err get_destination_from_headers(const struct list *headers, struct http_request *http)
3820 host = get_header_value(headers, "Host:");
3824 log_error(LOG_LEVEL_ERROR, "No \"Host:\" header found.");
3825 return JB_ERR_PARSE;
3828 if (NULL == (p = strdup((host))))
3830 log_error(LOG_LEVEL_ERROR, "Out of memory while parsing \"Host:\" header");
3831 return JB_ERR_MEMORY;
3834 if (NULL == (q = strdup(p)))
3837 log_error(LOG_LEVEL_ERROR, "Out of memory while parsing \"Host:\" header");
3838 return JB_ERR_MEMORY;
3841 freez(http->hostport);
3845 q = strchr(http->host, ':');
3848 /* Terminate hostname and evaluate port string */
3850 http->port = atoi(q);
3854 http->port = http->ssl ? 443 : 80;
3857 /* Rebuild request URL */
3859 http->url = strdup(http->ssl ? "https://" : "http://");
3860 string_append(&http->url, http->hostport);
3861 string_append(&http->url, http->path);
3862 if (http->url == NULL)
3864 return JB_ERR_MEMORY;
3867 log_error(LOG_LEVEL_HEADER, "Destination extracted from \"Host:\" header. New request URL: %s",
3875 /*********************************************************************
3877 * Function : create_forged_referrer
3879 * Description : Helper for client_referrer to forge a referer as
3880 * 'http://[hostname:port/' to fool stupid
3881 * checks for in-site links
3884 * 1 : header = Pointer to header pointer
3885 * 2 : hostport = Host and optionally port as string
3887 * Returns : JB_ERR_OK in case of success, or
3888 * JB_ERR_MEMORY in case of memory problems.
3890 *********************************************************************/
3891 static jb_err create_forged_referrer(char **header, const char *hostport)
3893 assert(NULL == *header);
3895 *header = strdup("Referer: http://");
3896 string_append(header, hostport);
3897 string_append(header, "/");
3899 if (NULL == *header)
3901 return JB_ERR_MEMORY;
3904 log_error(LOG_LEVEL_HEADER, "Referer forged to: %s", *header);
3911 /*********************************************************************
3913 * Function : create_fake_referrer
3915 * Description : Helper for client_referrer to create a fake referrer
3916 * based on a string supplied by the user.
3919 * 1 : header = Pointer to header pointer
3920 * 2 : hosthost = Referrer to fake
3922 * Returns : JB_ERR_OK in case of success, or
3923 * JB_ERR_MEMORY in case of memory problems.
3925 *********************************************************************/
3926 static jb_err create_fake_referrer(char **header, const char *fake_referrer)
3928 assert(NULL == *header);
3930 if ((0 != strncmpic(fake_referrer, "http://", 7)) && (0 != strncmpic(fake_referrer, "https://", 8)))
3932 log_error(LOG_LEVEL_HEADER,
3933 "Parameter: +hide-referrer{%s} is a bad idea, but I don't care.", fake_referrer);
3935 *header = strdup("Referer: ");
3936 string_append(header, fake_referrer);
3938 if (NULL == *header)
3940 return JB_ERR_MEMORY;
3943 log_error(LOG_LEVEL_HEADER, "Referer replaced with: %s", *header);
3950 /*********************************************************************
3952 * Function : handle_conditional_hide_referrer_parameter
3954 * Description : Helper for client_referrer to crunch or forge
3955 * the referrer header if the host has changed.
3958 * 1 : header = Pointer to header pointer
3959 * 2 : host = The target host (may include the port)
3960 * 3 : parameter_conditional_block = Boolean to signal
3961 * if we're in conditional-block mode. If not set,
3962 * we're in conditional-forge mode.
3964 * Returns : JB_ERR_OK in case of success, or
3965 * JB_ERR_MEMORY in case of memory problems.
3967 *********************************************************************/
3968 static jb_err handle_conditional_hide_referrer_parameter(char **header,
3969 const char *host, const int parameter_conditional_block)
3971 char *referer = strdup(*header);
3972 const size_t hostlenght = strlen(host);
3973 const char *referer_url = NULL;
3975 if (NULL == referer)
3978 return JB_ERR_MEMORY;
3981 /* referer begins with 'Referer: http[s]://' */
3982 if ((hostlenght+17) < strlen(referer))
3985 * Shorten referer to make sure the referer is blocked
3986 * if www.example.org/www.example.com-shall-see-the-referer/
3987 * links to www.example.com/
3989 referer[hostlenght+17] = '\0';
3991 referer_url = strstr(referer, "http://");
3992 if ((NULL == referer_url) || (NULL == strstr(referer_url, host)))
3994 /* Host has changed, Referer is invalid or a https URL. */
3995 if (parameter_conditional_block)
3997 log_error(LOG_LEVEL_HEADER, "New host is: %s. Crunching %s!", host, *header);
4004 return create_forged_referrer(header, host);
4014 /*********************************************************************
4016 * Function : get_appropiate_connection_header
4018 * Description : Returns an appropiate Connection header
4019 * depending on whether or not we try to keep
4020 * the connection to the server alive.
4023 * 1 : csp = Current client state (buffers, headers, etc...)
4025 * Returns : Pointer to statically allocated header buffer.
4027 *********************************************************************/
4028 static const char *get_appropiate_connection_header(const struct client_state *csp)
4030 static const char connection_keep_alive[] = "Connection: keep-alive";
4031 static const char connection_close[] = "Connection: close";
4033 if ((csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE)
4034 && (csp->http->ssl == 0))
4036 return connection_keep_alive;
4038 return connection_close;