X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=loaders.c;h=d187159bcfc87e2e50d7692685f1c1923642e43b;hp=1db122100395edaded6c85f6bb2c5517abbdd71b;hb=4dc3b810a5e3aa5cab68cf62708cdabcfe2993cb;hpb=89f4d632da7483f990e3fa75d4a11aee69c61386 diff --git a/loaders.c b/loaders.c index 1db12210..d187159b 100644 --- a/loaders.c +++ b/loaders.c @@ -1,4 +1,4 @@ -const char loaders_rcs[] = "$Id: loaders.c,v 1.36 2002/01/22 23:46:18 jongfoster Exp $"; +const char loaders_rcs[] = "$Id: loaders.c,v 1.57 2006/12/21 12:22:22 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/loaders.c,v $ @@ -9,7 +9,7 @@ const char loaders_rcs[] = "$Id: loaders.c,v 1.36 2002/01/22 23:46:18 jongfoster * unload files that are no longer in use. * * Copyright : Written by and Copyright (C) 2001 the SourceForge - * IJBSWA team. http://ijbswa.sourceforge.net + * Privoxy team. http://www.privoxy.org/ * * Based on the Internet Junkbuster originally written * by and Copyright (C) 1997 Anonymous Coders and @@ -35,6 +35,112 @@ const char loaders_rcs[] = "$Id: loaders.c,v 1.36 2002/01/22 23:46:18 jongfoster * * Revisions : * $Log: loaders.c,v $ + * Revision 1.57 2006/12/21 12:22:22 fabiankeil + * html_encode filter descriptions. + * + * Have "Ignoring job ..." error messages + * print the filter file name correctly. + * + * Revision 1.56 2006/09/07 10:40:30 fabiankeil + * Turns out trusted referrers above our arbitrary + * limit are downgraded too ordinary trusted URLs. + * Adjusted error message. + * + * Revision 1.55 2006/09/07 10:25:39 fabiankeil + * Fix typo. + * + * Revision 1.54 2006/09/07 10:22:20 fabiankeil + * If too many trusted referrers are used, + * print only one error message instead of logging + * every single trusted referrer above the arbitrary + * limit. + * + * Revision 1.53 2006/08/31 16:25:06 fabiankeil + * Work around a buffer overflow that caused Privoxy to + * segfault if too many trusted referrers were used. Good + * enough for now, but should be replaced with a real + * solution after the next release. + * + * Revision 1.52 2006/07/18 14:48:46 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) + * + * Revision 1.50.2.8 2006/01/30 15:16:25 david__schmidt + * Remove a little residual debugging info + * + * Revision 1.50.2.7 2006/01/29 23:10:56 david__schmidt + * Multiple filter file support + * + * Revision 1.50.2.6 2003/10/24 10:17:54 oes + * Nit: Allowed tabs as separators in filter headings + * + * Revision 1.50.2.5 2003/05/08 15:19:15 oes + * sweep: Made loop structure of sweep step mirror that of mark step + * + * Revision 1.50.2.4 2003/05/06 15:57:12 oes + * Bugfix: Update last_active pointer in sweep() before + * leaving an active client. Closes bugs #724395, #727882 + * + * Revision 1.50.2.3 2002/11/20 17:12:30 oes + * Ooops, forgot one change. + * + * Revision 1.50.2.2 2002/11/20 14:38:15 oes + * Fixed delayed/incomplete freeing of client resources and + * simplified loop structure in sweep. + * Thanks to Oliver Stoeneberg for the hint. + * + * Revision 1.50.2.1 2002/07/26 15:19:24 oes + * - PCRS jobs now chained in order of appearance. Previous + * reverse chaining was counter-intuitive. + * - Changed loglevel of PCRS job compile errors to + * LOG_LEVEL_ERROR + * + * Revision 1.50 2002/04/24 02:12:16 oes + * Jon's multiple AF patch: Sweep now takes care of all AFs + * + * Revision 1.49 2002/04/19 16:53:25 jongfoster + * Optimize away a function call by using an equivalent macro + * + * Revision 1.48 2002/04/05 00:56:09 gliptak + * Correcting typo to clean up on realloc failure + * + * Revision 1.47 2002/03/26 22:29:55 swa + * we have a new homepage! + * + * Revision 1.46 2002/03/24 13:25:43 swa + * name change related issues + * + * Revision 1.45 2002/03/16 23:54:06 jongfoster + * Adding graceful termination feature, to help look for memory leaks. + * If you enable this (which, by design, has to be done by hand + * editing config.h) and then go to http://i.j.b/die, then the program + * will exit cleanly after the *next* request. It should free all the + * memory that was used. + * + * Revision 1.44 2002/03/16 21:51:00 jongfoster + * Fixing free(NULL). + * + * Revision 1.43 2002/03/16 20:28:34 oes + * Added descriptions to the filters so users will know what they select in the cgi editor + * + * Revision 1.42 2002/03/13 00:27:05 jongfoster + * Killing warnings + * + * Revision 1.41 2002/03/12 01:42:50 oes + * Introduced modular filters + * + * Revision 1.40 2002/03/08 17:46:04 jongfoster + * Fixing int/size_t warnings + * + * Revision 1.39 2002/03/07 03:46:17 oes + * Fixed compiler warnings + * + * Revision 1.38 2002/03/06 22:54:35 jongfoster + * Automated function-comment nitpicking. + * + * Revision 1.37 2002/03/03 15:07:49 oes + * Re-enabled automatic config reloading + * * Revision 1.36 2002/01/22 23:46:18 jongfoster * Moving edit_read_line() and simple_read_line() to loaders.c, and * extending them to support reading MS-DOS, Mac and UNIX style files @@ -247,6 +353,7 @@ const char loaders_rcs[] = "$Id: loaders.c,v 1.36 2002/01/22 23:46:18 jongfoster #include "errlog.h" #include "actions.h" #include "urlmatch.h" +#include "encode.h" const char loaders_h_rcs[] = LOADERS_H_VERSION; @@ -259,7 +366,12 @@ const char loaders_h_rcs[] = LOADERS_H_VERSION; static struct file_list *current_trustfile = NULL; #endif /* def FEATURE_TRUST */ -static struct file_list *current_re_filterfile = NULL; +static int load_one_re_filterfile(struct client_state *csp, int fileid); + +static struct file_list *current_re_filterfile[MAX_AF_FILES] = { + NULL, NULL, NULL, NULL, NULL, + NULL, NULL, NULL, NULL, NULL +}; @@ -290,7 +402,8 @@ static struct file_list *current_re_filterfile = NULL; void sweep(void) { struct file_list *fl, *nfl; - struct client_state *csp, *ncsp; + struct client_state *csp, *last_active; + int i; /* clear all of the file's active flags */ for ( fl = files->next; NULL != fl; fl = fl->next ) @@ -298,89 +411,112 @@ void sweep(void) fl->active = 0; } - for (csp = clients; csp && (ncsp = csp->next) ; csp = csp->next) + last_active = clients; + csp = clients->next; + + while (NULL != csp) { - if (ncsp->flags & CSP_FLAG_ACTIVE) + if (csp->flags & CSP_FLAG_ACTIVE) { - /* mark this client's files as active */ + /* Mark this client's files as active */ /* * Always have a configuration file. * (Also note the slightly non-standard extra * indirection here.) */ - ncsp->config->config_file_list->active = 1; + csp->config->config_file_list->active = 1; - if (ncsp->actions_list) /* actions files */ + /* + * Actions files + */ + for (i = 0; i < MAX_AF_FILES; i++) { - ncsp->actions_list->active = 1; + if (csp->actions_list[i]) + { + csp->actions_list[i]->active = 1; + } } - if (ncsp->rlist) /* pcrsjob files */ + /* + * Filter files + */ + for (i = 0; i < MAX_AF_FILES; i++) { - ncsp->rlist->active = 1; + if (csp->rlist[i]) + { + csp->rlist[i]->active = 1; + } } + /* + * Trust file + */ #ifdef FEATURE_TRUST - if (ncsp->tlist) /* trust files */ + if (csp->tlist) { - ncsp->tlist->active = 1; + csp->tlist->active = 1; } #endif /* def FEATURE_TRUST */ + + last_active = csp; + csp = csp->next; } - else + else /* - * this client is not active, release its resources - * and the ones of all inactive clients that might - * follow it + * This client is not active. Free its resources. */ { - while (!(ncsp->flags & CSP_FLAG_ACTIVE)) - { - csp->next = ncsp->next; + last_active->next = csp->next; - freez(ncsp->ip_addr_str); - freez(ncsp->my_ip_addr_str); - freez(ncsp->my_hostname); - freez(ncsp->x_forwarded); - freez(ncsp->iob->buf); + freez(csp->ip_addr_str); + freez(csp->my_ip_addr_str); + freez(csp->my_hostname); + freez(csp->x_forwarded); + freez(csp->iob->buf); - free_http_request(ncsp->http); + free_http_request(csp->http); - destroy_list(ncsp->headers); - destroy_list(ncsp->cookie_list); + destroy_list(csp->headers); + destroy_list(csp->cookie_list); - free_current_action(ncsp->action); + free_current_action(csp->action); #ifdef FEATURE_STATISTICS - urls_read++; - if (ncsp->flags & CSP_FLAG_REJECTED) - { - urls_rejected++; - } + urls_read++; + if (csp->flags & CSP_FLAG_REJECTED) + { + urls_rejected++; + } #endif /* def FEATURE_STATISTICS */ - freez(ncsp); - - /* are there any more in sequence after it? */ - if( !(ncsp = csp->next) ) - break; - } + freez(csp); + + csp = last_active->next; } } - for (fl = files; fl && (nfl = fl->next) ; fl = fl->next) + nfl = files; + fl = files->next; + + while (fl != NULL) { - if ( ( 0 == nfl->active ) && ( NULL != nfl->unloader ) ) + if ( ( 0 == fl->active ) && ( NULL != fl->unloader ) ) { - fl->next = nfl->next; + nfl->next = fl->next; - (nfl->unloader)(nfl->f); + (fl->unloader)(fl->f); - freez(nfl->filename); + freez(fl->filename); + freez(fl); - freez(nfl); + fl = nfl->next; + } + else + { + nfl = fl; + fl = fl->next; } } @@ -442,6 +578,7 @@ int check_file_changed(const struct file_list * current, return 1; } + fs->filename = strdup(filename); fs->lastmodified = statbuf->st_mtime; @@ -486,8 +623,8 @@ int check_file_changed(const struct file_list * current, *********************************************************************/ jb_err simple_read_line(FILE *fp, char **dest, int *newline) { - int len = 0; - int buflen = BUFFER_SIZE; + size_t len = 0; + size_t buflen = BUFFER_SIZE; char * buf; char * p; int ch; @@ -507,9 +644,9 @@ jb_err simple_read_line(FILE *fp, char **dest, int *newline) #define CHAR_CR '\r' /* ASCII 13 */ #define CHAR_LF '\n' /* ASCII 10 */ - while (FOREVER) + for (;;) { - ch = fgetc(fp); + ch = getc(fp); if (ch == EOF) { if (len > 0) @@ -571,12 +708,12 @@ jb_err simple_read_line(FILE *fp, char **dest, int *newline) return JB_ERR_OK; } - *p++ = ch; + *p++ = (char)ch; if (++len >= buflen) { buflen += BUFFER_SIZE; - if (NULL == (p = realloc(buf, buflen))); + if (NULL == (p = realloc(buf, buflen))) { free(buf); return JB_ERR_MEMORY; @@ -827,9 +964,9 @@ jb_err edit_read_line(FILE *fp, * the caller cares about "raw" or just "data"). */ - free(raw); - free(prefix); - free(data); + freez(raw); + freez(prefix); + freez(data); return JB_ERR_FILE; } @@ -846,7 +983,7 @@ jb_err edit_read_line(FILE *fp, } else { - free(raw); + freez(raw); } if (prefix_out) { @@ -854,7 +991,7 @@ jb_err edit_read_line(FILE *fp, } else { - free(prefix); + freez(prefix); } if (data_out) { @@ -862,7 +999,7 @@ jb_err edit_read_line(FILE *fp, } else { - free(data); + freez(data); } return JB_ERR_OK; } @@ -881,13 +1018,13 @@ jb_err edit_read_line(FILE *fp, * 1 : buf = Buffer to use. * 2 : buflen = Size of buffer in bytes. * 3 : fp = File to read from - * 4 : linenum = linenumber in file + * 4 : linenum = linenumber in file * * Returns : NULL on EOF or error * Otherwise, returns buf. * *********************************************************************/ -char *read_config_line(char *buf, int buflen, FILE *fp, unsigned long *linenum) +char *read_config_line(char *buf, size_t buflen, FILE *fp, unsigned long *linenum) { jb_err err; char *buf2 = NULL; @@ -903,7 +1040,7 @@ char *read_config_line(char *buf, int buflen, FILE *fp, unsigned long *linenum) else { assert(buf2); - assert(strlen(buf2) + 1U < (unsigned)buflen); + assert(strlen(buf2) + 1U < buflen); strncpy(buf, buf2, buflen - 1); free(buf2); buf[buflen - 1] = '\0'; @@ -927,18 +1064,46 @@ char *read_config_line(char *buf, int buflen, FILE *fp, unsigned long *linenum) *********************************************************************/ static void unload_trustfile(void *f) { - struct block_spec *b = (struct block_spec *)f; - if (b == NULL) return; + struct block_spec *cur = (struct block_spec *)f; + struct block_spec *next; - unload_trustfile(b->next); /* Stack is cheap, isn't it? */ + while (cur != NULL) + { + next = cur->next; - free_url_spec(b->url); + free_url_spec(cur->url); + free(cur); - freez(b); + cur = next; + } } +#ifdef FEATURE_GRACEFUL_TERMINATION +/********************************************************************* + * + * Function : unload_current_trust_file + * + * Description : Unloads current trust file - reset to state at + * beginning of program. + * + * Parameters : None + * + * Returns : N/A + * + *********************************************************************/ +void unload_current_trust_file(void) +{ + if (current_trustfile) + { + current_trustfile->unloader = unload_trustfile; + current_trustfile = NULL; + } +} +#endif /* FEATURE_GRACEFUL_TERMINATION */ + + /********************************************************************* * * Function : load_trustfile @@ -962,6 +1127,7 @@ int load_trustfile(struct client_state *csp) int reject, trusted; struct file_list *fs; unsigned long linenum = 0; + int trusted_referrers = 0; if (!check_file_changed(current_trustfile, csp->config->trustfile, &fs)) { @@ -1006,7 +1172,7 @@ int load_trustfile(struct client_state *csp) reject = 0; p = buf; q = p+1; - while ((*p++ = *q++)) + while ((*p++ = *q++) != '\0') { /* nop */ } @@ -1043,11 +1209,24 @@ int load_trustfile(struct client_state *csp) */ if (trusted) { - *tl++ = b->url; - /* FIXME BUFFER OVERFLOW if >=64 entries */ + if(++trusted_referrers < MAX_TRUSTED_REFERRERS) + { + *tl++ = b->url; + } } } + if(trusted_referrers >= MAX_TRUSTED_REFERRERS) + { + /* + * FIXME: ... after Privoxy 3.0.4 is out. + */ + log_error(LOG_LEVEL_ERROR, "Too many trusted referrers. Current limit is %d, you are using %d.\n" + " Additional trusted referrers are treated like ordinary trusted URLs.\n" + " (You can increase this limit by changing MAX_TRUSTED_REFERRERS in project.h and recompiling).", + MAX_TRUSTED_REFERRERS, trusted_referrers); + } + *tl = NULL; fclose(fp); @@ -1082,7 +1261,8 @@ load_trustfile_error: * * Function : unload_re_filterfile * - * Description : Unload the re_filter list. + * Description : Unload the re_filter list by freeing all chained + * re_filterfile specs and their data. * * Parameters : * 1 : f = the data structure associated with the filterfile. @@ -1092,27 +1272,62 @@ load_trustfile_error: *********************************************************************/ static void unload_re_filterfile(void *f) { - struct re_filterfile_spec *b = (struct re_filterfile_spec *)f; + struct re_filterfile_spec *a, *b = (struct re_filterfile_spec *)f; - if (b == NULL) + while (b != NULL) { - return; - } + a = b->next; + + destroy_list(b->patterns); + pcrs_free_joblist(b->joblist); + freez(b->name); + freez(b->description); + freez(b); - destroy_list(b->patterns); - pcrs_free_joblist(b->joblist); - freez(b); + b = a; + } return; } + +#ifdef FEATURE_GRACEFUL_TERMINATION +/********************************************************************* + * + * Function : unload_current_re_filterfile + * + * Description : Unloads current re_filter file - reset to state at + * beginning of program. + * + * Parameters : None + * + * Returns : N/A + * + *********************************************************************/ +void unload_current_re_filterfile(void) +{ + int i; + + for (i = 0; i < MAX_AF_FILES; i++) + { + if (current_re_filterfile[i]) + { + current_re_filterfile[i]->unloader = unload_re_filterfile; + current_re_filterfile[i] = NULL; + } + } +} +#endif + + /********************************************************************* * * Function : load_re_filterfile * - * Description : Load the re_filterfile. Each non-comment, non-empty - * line is instantly added to the joblist, which is - * a chained list of pcrs_job structs. + * Description : Load the re_filterfile. + * Generate a chained list of re_filterfile_spec's from + * the "FILTER: " blocks, compiling all their substitutions + * into chained lists of pcrs_job structs. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) @@ -1121,23 +1336,65 @@ static void unload_re_filterfile(void *f) * *********************************************************************/ int load_re_filterfile(struct client_state *csp) +{ + int i; + int result; + + for (i = 0; i < MAX_AF_FILES; i++) + { + if (csp->config->re_filterfile[i]) + { + result = load_one_re_filterfile(csp, i); + if (result) + { + return result; + } + } + else if (current_re_filterfile[i]) + { + current_re_filterfile[i]->unloader = unload_re_filterfile; + current_re_filterfile[i] = NULL; + } + } + + return 0; +} + +/********************************************************************* + * + * Function : load_one_re_filterfile + * + * Description : Load a re_filterfile. + * Generate a chained list of re_filterfile_spec's from + * the "FILTER: " blocks, compiling all their substitutions + * into chained lists of pcrs_job structs. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * + * Returns : 0 => Ok, everything else is an error. + * + *********************************************************************/ +int load_one_re_filterfile(struct client_state *csp, int fileid) { FILE *fp; - struct re_filterfile_spec *bl; + struct re_filterfile_spec *new_bl, *bl = NULL; struct file_list *fs; char buf[BUFFER_SIZE]; int error; unsigned long linenum = 0; - pcrs_job *dummy; + pcrs_job *dummy, *lastjob = NULL; - if (!check_file_changed(current_re_filterfile, csp->config->re_filterfile, &fs)) + /* + * No need to reload if unchanged + */ + if (!check_file_changed(current_re_filterfile[fileid], csp->config->re_filterfile[fileid], &fs)) { - /* No need to load */ if (csp) { - csp->rlist = current_re_filterfile; + csp->rlist[fileid] = current_re_filterfile[fileid]; } return(0); } @@ -1146,60 +1403,135 @@ int load_re_filterfile(struct client_state *csp) goto load_re_filterfile_error; } - fs->f = bl = (struct re_filterfile_spec *)zalloc(sizeof(*bl)); - if (bl == NULL) - { - goto load_re_filterfile_error; - } - - /* Open the file or fail */ - if ((fp = fopen(csp->config->re_filterfile, "r")) == NULL) + /* + * Open the file or fail + */ + if ((fp = fopen(csp->config->re_filterfile[fileid], "r")) == NULL) { goto load_re_filterfile_error; } - /* Read line by line */ + /* + * Read line by line + */ while (read_config_line(buf, sizeof(buf), fp, &linenum) != NULL) { - enlist( bl->patterns, buf ); - - /* We have a meaningful line -> make it a job */ - if ((dummy = pcrs_compile_command(buf, &error)) == NULL) + /* + * If this is the head of a new filter block, make it a + * re_filterfile spec of its own and chain it to the list: + */ + if (strncmp(buf, "FILTER:", 7) == 0) { - log_error(LOG_LEVEL_RE_FILTER, - "Adding re_filter job %s failed with error %d.", buf, error); + new_bl = (struct re_filterfile_spec *)zalloc(sizeof(*bl)); + if (new_bl == NULL) + { + goto load_re_filterfile_error; + } + + new_bl->name = chomp(buf + 7); + + /* + * If a filter description is available, + * encode it to HTML and save it. + */ + if (NULL != (new_bl->description = strpbrk(new_bl->name, " \t"))) + { + *new_bl->description++ = '\0'; + new_bl->description = html_encode(chomp(new_bl->description)); + if (NULL == new_bl->description) + { + new_bl->description = strdup("Out of memory while encoding this filter's description to HTML"); + } + } + else + { + new_bl->description = strdup("No description available for this filter"); + } + + new_bl->name = strdup(chomp(new_bl->name)); + + /* + * If this is the first filter block, chain it + * to the file_list rather than its (nonexistant) + * predecessor + */ + if (fs->f == NULL) + { + fs->f = new_bl; + } + else + { + bl->next = new_bl; + } + bl = new_bl; + + log_error(LOG_LEVEL_RE_FILTER, "Reading in filter \"%s\" (\"%s\")", bl->name, bl->description); + continue; } + + /* + * Else, save the expression, make it a pcrs_job + * and chain it into the current filter's joblist + */ + if (bl != NULL) + { + enlist(bl->patterns, buf); + + if ((dummy = pcrs_compile_command(buf, &error)) == NULL) + { + log_error(LOG_LEVEL_ERROR, + "Adding re_filter job %s to filter %s failed with error %d.", buf, bl->name, error); + continue; + } + else + { + if (bl->joblist == NULL) + { + bl->joblist = dummy; + } + else + { + lastjob->next = dummy; + } + lastjob = dummy; + log_error(LOG_LEVEL_RE_FILTER, "Adding re_filter job %s to filter %s succeeded.", buf, bl->name); + } + } else { - dummy->next = bl->joblist; - bl->joblist = dummy; - log_error(LOG_LEVEL_RE_FILTER, "Adding re_filter job %s succeeded.", buf); + log_error(LOG_LEVEL_ERROR, "Ignoring job %s outside filter block in %s, line %d", + buf, csp->config->re_filterfile[fileid], linenum); } } fclose(fp); - /* the old one is now obsolete */ - if ( NULL != current_re_filterfile ) + /* + * Schedule the now-obsolete old data for unloading + */ + if ( NULL != current_re_filterfile[fileid] ) { - current_re_filterfile->unloader = unload_re_filterfile; + current_re_filterfile[fileid]->unloader = unload_re_filterfile; } + /* + * Chain this file into the global list of loaded files + */ fs->next = files->next; files->next = fs; - current_re_filterfile = fs; + current_re_filterfile[fileid] = fs; if (csp) { - csp->rlist = fs; + csp->rlist[fileid] = fs; } return( 0 ); load_re_filterfile_error: log_error(LOG_LEVEL_FATAL, "can't load re_filterfile '%s': %E", - csp->config->re_filterfile); + csp->config->re_filterfile[fileid]); return(-1); }