X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=parsers.c;h=d56fde8bf1636f6da292016b1407934a6088712c;hp=ac5bfef8e70e00797e946c4bde1f625bc921cd37;hb=d9ab2562dfeadc4f1d2bcae73e9fae62208eaa8e;hpb=1b65660a5bb4e9d8511906406ebcc8a008c7baee diff --git a/parsers.c b/parsers.c index ac5bfef8..d56fde8b 100644 --- a/parsers.c +++ b/parsers.c @@ -1,4 +1,4 @@ -const char parsers_rcs[] = "$Id: parsers.c,v 1.58 2006/07/18 14:48:47 david__schmidt Exp $"; +const char parsers_rcs[] = "$Id: parsers.c,v 1.74 2006/10/02 16:59:12 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/parsers.c,v $ @@ -40,6 +40,89 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.58 2006/07/18 14:48:47 david__sch * * Revisions : * $Log: parsers.c,v $ + * Revision 1.74 2006/10/02 16:59:12 fabiankeil + * The special header "X-Filter: No" now disables + * header filtering as well. + * + * Revision 1.73 2006/09/23 13:26:38 roro + * Replace TABs by spaces in source code. + * + * Revision 1.72 2006/09/23 12:37:21 fabiankeil + * Don't print a log message every time filter_headers is + * entered or left. It only creates noise without any real + * information. + * + * Revision 1.71 2006/09/21 19:55:17 fabiankeil + * Fix +hide-if-modified-since{-n}. + * + * Revision 1.70 2006/09/08 12:06:34 fabiankeil + * Have hide-if-modified-since interpret the random + * range value as minutes instead of hours. Allows + * more fine-grained configuration. + * + * Revision 1.69 2006/09/06 16:25:51 fabiankeil + * Always have parse_header_time return a pointer + * that actual makes sense, even though we currently + * only need it to detect problems. + * + * Revision 1.68 2006/09/06 10:43:32 fabiankeil + * Added config option enable-remote-http-toggle + * to specify if Privoxy should recognize special + * headers (currently only X-Filter) to change its + * behaviour. Disabled by default. + * + * Revision 1.67 2006/09/04 11:01:26 fabiankeil + * After filtering de-chunked instances, remove + * "Transfer-Encoding" header entirely instead of changing + * it to "Transfer-Encoding: identity", which is invalid. + * Thanks Michael Shields . Fixes PR 1318658. + * + * Don't use localtime in parse_header_time. An empty time struct + * is good enough, it gets overwritten by strptime anyway. + * + * Revision 1.66 2006/09/03 19:38:28 fabiankeil + * Use gmtime_r if available, fallback to gmtime with mutex + * protection for MacOSX and use vanilla gmtime for the rest. + * + * Revision 1.65 2006/08/22 10:55:56 fabiankeil + * Changed client_referrer to use the right type (size_t) for + * hostlenght and to shorten the temporary referrer string with + * '\0' instead of adding a useless line break. + * + * Revision 1.64 2006/08/17 17:15:10 fabiankeil + * - Back to timegm() using GnuPG's replacement if necessary. + * Using mktime() and localtime() could add a on hour offset if + * the randomize factor was big enough to lead to a summer/wintertime + * switch. + * + * - Removed now-useless Privoxy 3.0.3 compatibility glue. + * + * - Moved randomization code into pick_from_range(). + * + * - Changed parse_header_time definition. + * time_t isn't guaranteed to be signed and + * if it isn't, -1 isn't available as error code. + * Changed some variable types in client_if_modified_since() + * because of the same reason. + * + * Revision 1.63 2006/08/14 13:18:08 david__schmidt + * OS/2 compilation compatibility fixups + * + * Revision 1.62 2006/08/14 08:58:42 fabiankeil + * Changed include from strptime.c to strptime.h + * + * Revision 1.61 2006/08/14 08:25:19 fabiankeil + * Split filter-headers{} into filter-client-headers{} + * and filter-server-headers{}. + * Added parse_header_time() to share some code. + * Replaced timegm() with mktime(). + * + * Revision 1.60 2006/08/12 03:54:37 david__schmidt + * Windows service integration + * + * Revision 1.59 2006/08/03 02:46:41 david__schmidt + * Incorporate Fabian Keil's patch work: http://www.fabiankeil.de/sourcecode/privoxy/ + * * Revision 1.58 2006/07/18 14:48:47 david__schmidt * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch) * with what was really the latest development (the v_3_0_branch branch) @@ -456,11 +539,10 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.58 2006/07/18 14:48:47 david__sch #include "project.h" -#ifdef OSX_DARWIN -#include +#ifdef FEATURE_PTHREAD #include "jcc.h" /* jcc.h is for mutex semapores only */ -#endif /* def OSX_DARWIN */ +#endif /* def FEATURE_PTHREAD */ #include "list.h" #include "parsers.h" #include "encode.h" @@ -470,6 +552,10 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.58 2006/07/18 14:48:47 david__sch #include "miscutil.h" #include "list.h" +#ifndef HAVE_STRPTIME +#include "strptime.h" +#endif + const char parsers_h_rcs[] = PARSERS_H_VERSION; /* Fix a problem with Solaris. There should be no effect on other @@ -505,7 +591,7 @@ const struct parsers client_patterns[] = { { "if-none-match:", 14, client_if_none_match }, { "X-Filter:", 9, client_x_filter }, { "*", 0, crunch_client_header }, - { "*", 0, filter_header }, + { "*", 0, filter_client_header }, { NULL, 0, NULL } }; @@ -522,7 +608,7 @@ const struct parsers server_patterns[] = { { "content-disposition:", 20, server_content_disposition }, { "Last-Modified:", 14, server_last_modified }, { "*", 0, crunch_server_header }, - { "*", 0, filter_header }, + { "*", 0, filter_server_header }, { NULL, 0, NULL } }; @@ -865,6 +951,57 @@ char *sed(const struct parsers pats[], /* here begins the family of parser functions that reformat header lines */ +/********************************************************************* + * + * Function : filter_server_header + * + * Description : Checks if server header filtering is enabled. + * If it is, filter_header is called to do the work. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * 2 : header = On input, pointer to header to modify. + * On output, pointer to the modified header, or NULL + * to remove the header. This function frees the + * original string if necessary. + * + * Returns : JB_ERR_OK on success and always succeeds + * + *********************************************************************/ +jb_err filter_server_header(struct client_state *csp, char **header) +{ + if (csp->action->flags & ACTION_FILTER_SERVER_HEADERS) + { + filter_header(csp, header); + } + return(JB_ERR_OK); +} + +/********************************************************************* + * + * Function : filter_client_header + * + * Description : Checks if client header filtering is enabled. + * If it is, filter_header is called to do the work. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * 2 : header = On input, pointer to header to modify. + * On output, pointer to the modified header, or NULL + * to remove the header. This function frees the + * original string if necessary. + * + * Returns : JB_ERR_OK on success and always succeeds + * + *********************************************************************/ +jb_err filter_client_header(struct client_state *csp, char **header) +{ + if (csp->action->flags & ACTION_FILTER_CLIENT_HEADERS) + { + filter_header(csp, header); + } + return(JB_ERR_OK); +} /********************************************************************* * @@ -900,24 +1037,12 @@ jb_err filter_header(struct client_state *csp, char **header) int i, found_filters = 0; -#ifndef MAX_AF_FILES -# define MAX_AF_FILES 1 -# define INDEX_OR_NOT -#else -# define INDEX_OR_NOT [i] -#endif - - if (!(csp->action->flags & ACTION_FILTER_HEADERS)) - { - return(JB_ERR_OK); - } - log_error(LOG_LEVEL_RE_FILTER, "Entered filter_headers"); /* * Need to check the set of re_filterfiles... */ for (i = 0; i < MAX_AF_FILES; i++) { - fl = csp->rlist INDEX_OR_NOT; + fl = csp->rlist[i]; if (NULL != fl) { if (NULL != fl->f) @@ -936,12 +1061,12 @@ jb_err filter_header(struct client_state *csp, char **header) for (i = 0; i < MAX_AF_FILES; i++) { - fl = csp->rlist INDEX_OR_NOT; + fl = csp->rlist[i]; if ((NULL == fl) || (NULL == fl->f)) break; /* * For all applying +filter actions, look if a filter by that - * name exists and if yes, execute it's pcrs_joblist on the + * name exists and if yes, execute its pcrs_joblist on the * buffer. */ for (b = fl->f; b; b = b->next) @@ -999,12 +1124,11 @@ jb_err filter_header(struct client_state *csp, char **header) if ( 0 == size ) { - log_error(LOG_LEVEL_HEADER, "Removing empty header %s", *header); + log_error(LOG_LEVEL_HEADER, "Removing empty header %s", *header); freez(*header); } - log_error(LOG_LEVEL_RE_FILTER, "Leaving filter headers"); - return(JB_ERR_OK); + return(JB_ERR_OK); } @@ -1165,8 +1289,8 @@ jb_err server_content_type(struct client_state *csp, char **header) * * Description : - Prohibit filtering (CT_TABOO) if transfer coding compresses * - Raise the CSP_FLAG_CHUNKED flag if coding is "chunked" - * - Change from "chunked" to "identity" if body was chunked - * but has been de-chunked for filtering. + * - Remove header if body was chunked but has been + * de-chunked for filtering. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) @@ -1197,15 +1321,13 @@ jb_err server_transfer_coding(struct client_state *csp, char **header) csp->flags |= CSP_FLAG_CHUNKED; /* - * If the body was modified, it has been - * de-chunked first, so adjust the header: + * If the body was modified, it has been de-chunked first + * and the header must be removed. */ if (csp->flags & CSP_FLAG_MODIFIED) { + log_error(LOG_LEVEL_HEADER, "Removing: %s", *header); freez(*header); - *header = strdup("Transfer-Encoding: identity"); - log_error(LOG_LEVEL_HEADER, "Set: %s", *header); - return (header == NULL) ? JB_ERR_MEMORY : JB_ERR_OK; } } @@ -1404,7 +1526,10 @@ jb_err server_last_modified(struct client_state *csp, char **header) char buf[BUFFER_SIZE]; char newheader[50]; - struct tm *timeptr; +#ifdef HAVE_GMTIME_R + struct tm gmt; +#endif + struct tm *timeptr = NULL; time_t now, last_modified; long int rtime; long int days, hours, minutes, seconds; @@ -1452,29 +1577,44 @@ jb_err server_last_modified(struct client_state *csp, char **header) { log_error(LOG_LEVEL_HEADER, "Randomizing: %s", *header); now = time(NULL); +#ifdef HAVE_GMTIME_R + timeptr = gmtime_r(&now, &gmt); +#elif FEATURE_PTHREAD + pthread_mutex_lock(&gmtime_mutex); + timeptr = gmtime(&now); + pthread_mutex_unlock(&gmtime_mutex); +#else timeptr = gmtime(&now); - if (strptime(*header, "Last-Modified: %a, %d %b %Y %T", timeptr) == NULL) +#endif + if ((timeptr = parse_header_time(*header, &last_modified)) == NULL) { - log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s (crunching!)", *header); - freez(*header); + log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s (crunching!)", *header); + freez(*header); } else { - last_modified = timegm(timeptr); rtime = difftime(now, last_modified); if (rtime) { - rtime = random() % rtime + 1; + rtime = pick_from_range(rtime); last_modified += rtime; +#ifdef HAVE_GMTIME_R + timeptr = gmtime_r(&last_modified, &gmt); +#elif FEATURE_PTHREAD + pthread_mutex_lock(&gmtime_mutex); + timeptr = gmtime(&last_modified); + pthread_mutex_unlock(&gmtime_mutex); +#else timeptr = gmtime(&last_modified); - strftime(newheader, sizeof(newheader), "%a, %d %b %Y %T GMT", timeptr); +#endif + strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr); freez(*header); *header = strdup("Last-Modified: "); string_append(header, newheader); if (*header == NULL) { - log_error(LOG_LEVEL_ERROR, " Insufficent memory, header crunched without replacement."); + log_error(LOG_LEVEL_ERROR, "Insufficent memory, header crunched without replacement."); return JB_ERR_MEMORY; } @@ -1598,7 +1738,7 @@ jb_err client_referrer(struct client_state *csp, char **header) const char *newval; const char *host; char *referer; - int hostlenght; + size_t hostlenght; #ifdef FEATURE_FORCE_LOAD /* Since the referrer can include the prefix even @@ -1656,7 +1796,7 @@ jb_err client_referrer(struct client_state *csp, char **header) *if www.example.org/www.example.com-shall-see-the-referer/ *links to www.example.com/ */ - referer[hostlenght+17] = '\n'; + referer[hostlenght+17] = '\0'; } if ( 0 == strstr(referer, host)) /*Host has changed*/ { @@ -2147,11 +2287,14 @@ jb_err client_host(struct client_state *csp, char **header) jb_err client_if_modified_since(struct client_state *csp, char **header) { char newheader[50]; - struct tm *timeptr; +#ifdef HAVE_GMTIME_R + struct tm gmt; +#endif + struct tm *timeptr = NULL; time_t tm = 0; const char *newval; - time_t rtime; - time_t hours, minutes, seconds; + long int rtime; + long int hours, minutes, seconds; int negative = 0; char * endptr; @@ -2178,11 +2321,7 @@ jb_err client_if_modified_since(struct client_state *csp, char **header) } else /* add random value */ { - /* - * tm must be initinalized to prevent segmentation faults. - */ - timeptr = gmtime(&tm); - if (strptime(*header, "If-Modified-Since: %a, %d %b %Y %T", timeptr) == NULL) + if ((timeptr = parse_header_time(*header, &tm)) == NULL) { log_error(LOG_LEVEL_HEADER, "Couldn't parse: %s (crunching!)", *header); freez(*header); @@ -2190,20 +2329,34 @@ jb_err client_if_modified_since(struct client_state *csp, char **header) else { rtime = strtol(newval, &endptr, 0); - - log_error(LOG_LEVEL_HEADER, "Randomizing: %s (random range: %d hou%s)", - *header, rtime, (rtime == 1 || rtime == -1) ? "r": "rs"); - - rtime *= 3600; - rtime = random() % rtime; - - if(newval[0] == '-') + if(rtime) { - rtime *= -1; + log_error(LOG_LEVEL_HEADER, "Randomizing: %s (random range: %d minut%s)", + *header, rtime, (rtime == 1 || rtime == -1) ? "e": "es"); + if(rtime < 0) + { + rtime *= -1; + negative = 1; + } + rtime *= 60; + rtime = pick_from_range(rtime); + } + else + { + log_error(LOG_LEVEL_ERROR, "Random range is 0. Assuming time transformation test.", + *header); } - tm = timegm(timeptr) + rtime; + tm += rtime * (negative ? -1 : 1); +#ifdef HAVE_GMTIME_R + timeptr = gmtime_r(&tm, &gmt); +#elif FEATURE_PTHREAD + pthread_mutex_lock(&gmtime_mutex); timeptr = gmtime(&tm); - strftime(newheader, sizeof(newheader), "%a, %d %b %Y %T GMT", timeptr); + pthread_mutex_unlock(&gmtime_mutex); +#else + timeptr = gmtime(&tm); +#endif + strftime(newheader, sizeof(newheader), "%a, %d %b %Y %H:%M:%S GMT", timeptr); freez(*header); *header = strdup("If-Modified-Since: "); @@ -2211,18 +2364,13 @@ jb_err client_if_modified_since(struct client_state *csp, char **header) if (*header == NULL) { - log_error(LOG_LEVEL_HEADER, " Insufficent memory, header crunched without replacement."); + log_error(LOG_LEVEL_HEADER, "Insufficent memory, header crunched without replacement."); return JB_ERR_MEMORY; } if(LOG_LEVEL_HEADER & debug) /* Save cycles if the user isn't interested. */ { - if(rtime < 0) - { - rtime *= -1; - negative = 1; - } - hours = rtime / 3600 % 24; + hours = rtime / 3600; minutes = rtime / 60 % 60; seconds = rtime % 60; @@ -2286,17 +2434,26 @@ jb_err client_x_filter(struct client_state *csp, char **header) { if ( 0 == strcmpic(*header, "X-Filter: No")) { - if (csp->action->flags & ACTION_FORCE_TEXT_MODE) + if (!(csp->config->feature_flags & RUNTIME_FEATURE_HTTP_TOGGLE)) { - log_error(LOG_LEVEL_HEADER, "force-text-mode overruled the client's request to disable filtering!"); + log_error(LOG_LEVEL_INFO, "Ignored the client's request to fetch without filtering."); } else - { - csp->content_type = CT_TABOO; - log_error(LOG_LEVEL_HEADER, "Disabled filter mode on behalf of the client."); + { + if (csp->action->flags & ACTION_FORCE_TEXT_MODE) + { + log_error(LOG_LEVEL_HEADER, "force-text-mode overruled the client's request to fetch without filtering!"); + } + else + { + csp->content_type = CT_TABOO; + csp->action->flags &= ~ACTION_FILTER_SERVER_HEADERS; + csp->action->flags &= ~ACTION_FILTER_CLIENT_HEADERS; + log_error(LOG_LEVEL_HEADER, "Accepted the client's request to fetch without filtering."); + } + log_error(LOG_LEVEL_HEADER, "Crunching %s", *header); + freez(*header); } - log_error(LOG_LEVEL_HEADER, "Crunching %s", *header); - freez(*header); } return JB_ERR_OK; } @@ -2621,7 +2778,7 @@ jb_err server_set_cookie(struct client_state *csp, char **header) time (&now); #ifdef HAVE_LOCALTIME_R tm_now = *localtime_r(&now, &tm_now); -#elif OSX_DARWIN +#elif FEATURE_PTHREAD pthread_mutex_lock(&localtime_mutex); tm_now = *localtime (&now); pthread_mutex_unlock(&localtime_mutex); @@ -2749,6 +2906,41 @@ int strclean(const char *string, const char *substring) } #endif /* def FEATURE_FORCE_LOAD */ +/********************************************************************* + * + * Function : parse_header_time + * + * Description : Transforms time inside a HTTP header into + * the usual time format. + * + * Parameters : + * 1 : header = header to parse + * 2 : tm = storage for the resulting time in seconds + * + * Returns : Time struct containing the header time, or + * NULL in case of a parsing problem. + * + *********************************************************************/ +struct tm *parse_header_time(char *header, time_t *tm) { + + char * timestring; + struct tm gmt; + struct tm * timeptr; + + /* Skipping header name */ + timestring = strstr(header, ": "); + if (strptime(timestring, ": %a, %d %b %Y %H:%M:%S", &gmt) == NULL) + { + timeptr = NULL; + } + else + { + *tm = timegm(&gmt); + timeptr=&gmt; + } + return(timeptr); +} + /* Local Variables: