X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=filters.c;h=48259a8343da84e0ff30ca1b8b3061f22fc50b19;hp=54bd06dcd96a2464ec25c35a622bf01087755a69;hb=c5a40240177ba7f8e30cdac6de6c15a3ddfeca54;hpb=43980e5c0dbe1b2090fa04f69c3989960f1d8975 diff --git a/filters.c b/filters.c index 54bd06dc..48259a83 100644 --- a/filters.c +++ b/filters.c @@ -1,12 +1,11 @@ -const char filters_rcs[] = "$Id: filters.c,v 1.181 2014/06/02 06:19:05 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/filters.c,v $ * * Purpose : Declares functions to parse/crunch headers and pages. * - * Copyright : Written by and Copyright (C) 2001-2014 the - * Privoxy team. http://www.privoxy.org/ + * Copyright : Written by and Copyright (C) 2001-2020 the + * Privoxy team. https://www.privoxy.org/ * * Based on the Internet Junkbuster originally written * by and Copyright (C) 1997 Anonymous Coders and @@ -43,18 +42,12 @@ const char filters_rcs[] = "$Id: filters.c,v 1.181 2014/06/02 06:19:05 fabiankei #include #ifndef _WIN32 -#ifndef __OS2__ #include -#endif /* ndef __OS2__ */ #include #else #include #endif /* ndef _WIN32 */ -#ifdef __OS2__ -#include -#endif /* def __OS2__ */ - #include "project.h" #include "filters.h" #include "encode.h" @@ -70,17 +63,30 @@ const char filters_rcs[] = "$Id: filters.c,v 1.181 2014/06/02 06:19:05 fabiankei #include "deanimate.h" #include "urlmatch.h" #include "loaders.h" +#ifdef FEATURE_CLIENT_TAGS +#include "client-tags.h" +#endif +#ifdef FEATURE_HTTPS_INSPECTION +#include "ssl.h" +#endif #ifdef _WIN32 #include "win32.h" #endif -const char filters_h_rcs[] = FILTERS_H_VERSION; - typedef char *(*filter_function_ptr)(); static filter_function_ptr get_filter_function(const struct client_state *csp); -static jb_err remove_chunked_transfer_coding(char *buffer, size_t *size); static jb_err prepare_for_filtering(struct client_state *csp); +static void apply_url_actions(struct current_action_spec *action, + struct http_request *http, +#ifdef FEATURE_CLIENT_TAGS + const struct list *client_tags, +#endif + struct url_actions *b); + +#ifdef FEATURE_EXTENDED_STATISTICS +static void increment_block_reason_counter(const char *block_reason); +#endif #ifdef FEATURE_ACL #ifdef HAVE_RFC2553 @@ -97,17 +103,15 @@ static jb_err prepare_for_filtering(struct client_state *csp); * 3 : len = length of IP address in octets * 4 : port = port number in network order; * - * Returns : 0 = no errror; -1 otherwise. + * Returns : void * *********************************************************************/ -static int sockaddr_storage_to_ip(const struct sockaddr_storage *addr, - uint8_t **ip, unsigned int *len, - in_port_t **port) +static void sockaddr_storage_to_ip(const struct sockaddr_storage *addr, + uint8_t **ip, unsigned int *len, + in_port_t **port) { - if (NULL == addr) - { - return(-1); - } + assert(NULL != addr); + assert(addr->ss_family == AF_INET || addr->ss_family == AF_INET6); switch (addr->ss_family) { @@ -142,12 +146,7 @@ static int sockaddr_storage_to_ip(const struct sockaddr_storage *addr, } break; - default: - /* Unsupported address family */ - return(-1); } - - return(0); } @@ -208,7 +207,7 @@ static int match_sockaddr(const struct sockaddr_storage *network, return 0; } - /* TODO: Optimize by checking by words insted of octets */ + /* TODO: Optimize by checking by words instead of octets */ for (i = 0; (i < addr_len) && netmask_addr[i]; i++) { if ((network_addr[i] & netmask_addr[i]) != @@ -347,12 +346,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) * Use a temporary acl spec copy so we can log * the unmodified original in case of parse errors. */ - acl_spec = strdup(aspec); - if (acl_spec == NULL) - { - /* XXX: This will be logged as parse error. */ - return(-1); - } + acl_spec = strdup_or_die(aspec); if ((p = strchr(acl_spec, '/')) != NULL) { @@ -391,13 +385,19 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) { p = strchr(acl_spec, ':'); } + if (p != NULL) + { + assert(*p == ':'); + *p = '\0'; + p++; + } #ifdef HAVE_RFC2553 memset(&hints, 0, sizeof(struct addrinfo)); hints.ai_family = AF_UNSPEC; hints.ai_socktype = SOCK_STREAM; - i = getaddrinfo(acl_spec, ((p) ? ++p : NULL), &hints, &result); + i = getaddrinfo(acl_spec, p, &hints, &result); if (i != 0) { @@ -416,7 +416,6 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) { char *endptr; - *p++ = '\0'; port = strtol(p, &endptr, 10); if (port <= 0 || port > 65535 || *endptr != '\0') @@ -447,10 +446,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) } aca->mask.ss_family = aca->addr.ss_family; - if (sockaddr_storage_to_ip(&aca->mask, &mask_data, &addr_len, &mask_port)) - { - return(-1); - } + sockaddr_storage_to_ip(&aca->mask, &mask_data, &addr_len, &mask_port); if (p) { @@ -510,7 +506,7 @@ int acl_addr(const char *aspec, struct access_control_addr *aca) * * Description : Check to see if CONNECT requests to the destination * port of this request are forbidden. The check is - * independend of the actual request method. + * independent of the actual request method. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) @@ -562,6 +558,13 @@ struct http_response *block_url(struct client_state *csp) return cgi_error_memory(); } +#ifdef FEATURE_EXTENDED_STATISTICS + if (csp->action->string[ACTION_STRING_BLOCK] != NULL) + { + increment_block_reason_counter(csp->action->string[ACTION_STRING_BLOCK]); + } +#endif + /* * If it's an image-url, send back an image or redirect * as specified by the relevant +image action @@ -582,12 +585,7 @@ struct http_response *block_url(struct client_state *csp) /* and handle accordingly: */ if ((p == NULL) || (0 == strcmpic(p, "pattern"))) { - rsp->status = strdup("403 Request blocked by Privoxy"); - if (rsp->status == NULL) - { - free_http_response(rsp); - return cgi_error_memory(); - } + rsp->status = strdup_or_die("403 Request blocked by Privoxy"); rsp->body = bindup(image_pattern_data, image_pattern_length); if (rsp->body == NULL) { @@ -604,12 +602,7 @@ struct http_response *block_url(struct client_state *csp) } else if (0 == strcmpic(p, "blank")) { - rsp->status = strdup("403 Request blocked by Privoxy"); - if (rsp->status == NULL) - { - free_http_response(rsp); - return cgi_error_memory(); - } + rsp->status = strdup_or_die("403 Request blocked by Privoxy"); rsp->body = bindup(image_blank_data, image_blank_length); if (rsp->body == NULL) { @@ -626,12 +619,7 @@ struct http_response *block_url(struct client_state *csp) } else { - rsp->status = strdup("302 Local Redirect from Privoxy"); - if (rsp->status == NULL) - { - free_http_response(rsp); - return cgi_error_memory(); - } + rsp->status = strdup_or_die("302 Local Redirect from Privoxy"); if (enlist_unique_header(rsp->headers, "Location", p)) { @@ -651,7 +639,7 @@ struct http_response *block_url(struct client_state *csp) new_content_type = csp->action->string[ACTION_STRING_CONTENT_TYPE]; freez(rsp->body); - rsp->body = strdup(" "); + rsp->body = strdup_or_die(" "); rsp->content_length = 1; if (csp->config->feature_flags & RUNTIME_FEATURE_EMPTY_DOC_RETURNS_OK) @@ -662,18 +650,13 @@ struct http_response *block_url(struct client_state *csp) * Return a 200 OK status for pages blocked with +handle-as-empty-document * if the "handle-as-empty-doc-returns-ok" runtime config option is set. */ - rsp->status = strdup("200 Request blocked by Privoxy"); + rsp->status = strdup_or_die("200 Request blocked by Privoxy"); } else { - rsp->status = strdup("403 Request blocked by Privoxy"); + rsp->status = strdup_or_die("403 Request blocked by Privoxy"); } - if (rsp->status == NULL) - { - free_http_response(rsp); - return cgi_error_memory(); - } if (new_content_type != 0) { log_error(LOG_LEVEL_HEADER, "Overwriting Content-Type with %s", new_content_type); @@ -693,12 +676,7 @@ struct http_response *block_url(struct client_state *csp) jb_err err; struct map * exports; - rsp->status = strdup("403 Request blocked by Privoxy"); - if (rsp->status == NULL) - { - free_http_response(rsp); - return cgi_error_memory(); - } + rsp->status = strdup_or_die("403 Request blocked by Privoxy"); exports = default_exports(csp, NULL); if (exports == NULL) @@ -802,9 +780,9 @@ struct http_response *trust_url(struct client_state *csp) return cgi_error_memory(); } - rsp->status = strdup("403 Request blocked by Privoxy"); + rsp->status = strdup_or_die("403 Request blocked by Privoxy"); exports = default_exports(csp, NULL); - if (exports == NULL || rsp->status == NULL) + if (exports == NULL) { free_http_response(rsp); return cgi_error_memory(); @@ -836,7 +814,7 @@ struct http_response *trust_url(struct client_state *csp) /* * Export the trust list */ - p = strdup(""); + p = strdup_or_die(""); for (tl = csp->config->trust_list; (t = *tl) != NULL ; tl++) { snprintf(buf, sizeof(buf), "
  • %s
  • \n", t->spec); @@ -858,7 +836,7 @@ struct http_response *trust_url(struct client_state *csp) { struct list_entry *l; - p = strdup(""); + p = strdup_or_die(""); for (l = csp->config->trust_info->first; l ; l = l->next) { snprintf(buf, sizeof(buf), "
  • %s
    \n", l->str, l->str); @@ -951,6 +929,7 @@ pcrs_job *compile_dynamic_pcrs_job_list(const struct client_state *csp, const st {"path", csp->http->path, 1}, {"host", csp->http->host, 1}, {"origin", csp->ip_addr_str, 1}, + {"listen-address", csp->listen_addr_str, 1}, {NULL, NULL, 1} }; @@ -1068,7 +1047,7 @@ char *rewrite_url(char *old_url, const char *pcrs_command) * the last URL found. * *********************************************************************/ -char *get_last_url(char *subject, const char *redirect_mode) +static char *get_last_url(char *subject, const char *redirect_mode) { char *new_url = NULL; char *tmp; @@ -1084,7 +1063,7 @@ char *get_last_url(char *subject, const char *redirect_mode) } if (0 == strcmpic(redirect_mode, "check-decoded-url") && strchr(subject, '%')) - { + { char *url_segment = NULL; char **url_segments; size_t max_segments; @@ -1131,14 +1110,8 @@ char *get_last_url(char *subject, const char *redirect_mode) } if (NULL != url_segment) { - url_segment = strdup(url_segment); + url_segment = strdup_or_die(url_segment); freez(dtoken); - if (url_segment == NULL) - { - log_error(LOG_LEVEL_ERROR, - "Out of memory while searching for redirects."); - return NULL; - } break; } freez(dtoken); @@ -1224,7 +1197,6 @@ struct http_response *redirect_url(struct client_state *csp) */ char * redirect_mode; #endif /* def FEATURE_FAST_REDIRECTS */ - char *old_url = NULL; char *new_url = NULL; char *redirection_string; @@ -1238,17 +1210,48 @@ struct http_response *redirect_url(struct client_state *csp) * properly formatted URL and use it for the redirection * directly. * - * According to RFC 2616 section 14.30 the URL - * has to be absolute and if the user tries: - * +redirect{shit/this/will/be/parsed/as/pcrs_command.html} + * According to (the now obsolete) RFC 2616 section 14.30 + * the URL has to be absolute and if the user tries: + * +redirect{sadly/this/will/be/parsed/as/pcrs_command.html} * she would get undefined results anyway. * + * RFC 7231 7.1.2 actually allows relative references, + * but those start with a leading slash (RFC 3986 4.2) and + * thus can't be mistaken for pcrs commands either. */ if (*redirection_string == 's') { - old_url = csp->http->url; - new_url = rewrite_url(old_url, redirection_string); + char *requested_url; + +#ifdef FEATURE_HTTPS_INSPECTION + if (client_use_ssl(csp)) + { + jb_err err; + + requested_url = strdup_or_die("https://"); + err = string_append(&requested_url, csp->http->hostport); + if (!err) err = string_append(&requested_url, csp->http->path); + if (err) + { + log_error(LOG_LEVEL_FATAL, + "Failed to rebuild URL 'https://%s%s'", + csp->http->hostport, csp->http->path); + } + } + else +#endif + { + requested_url = csp->http->url; + } + new_url = rewrite_url(requested_url, redirection_string); +#ifdef FEATURE_HTTPS_INSPECTION + if (requested_url != csp->http->url) + { + assert(client_use_ssl(csp)); + freez(requested_url); + } +#endif } else { @@ -1262,6 +1265,8 @@ struct http_response *redirect_url(struct client_state *csp) #ifdef FEATURE_FAST_REDIRECTS if ((csp->action->flags & ACTION_FAST_REDIRECTS)) { + char *old_url; + redirect_mode = csp->action->string[ACTION_STRING_FAST_REDIRECTS]; /* @@ -1272,19 +1277,7 @@ struct http_response *redirect_url(struct client_state *csp) new_url = get_last_url(old_url, redirect_mode); freez(old_url); } - - /* - * Disable redirect checkers, so that they - * will be only run more than once if the user - * also enables them through tags. - * - * From a performance point of view - * it doesn't matter, but the duplicated - * log messages are annoying. - */ - csp->action->flags &= ~ACTION_FAST_REDIRECTS; #endif /* def FEATURE_FAST_REDIRECTS */ - csp->action->flags &= ~ACTION_REDIRECT; /* Did any redirect action trigger? */ if (new_url) @@ -1321,8 +1314,8 @@ struct http_response *redirect_url(struct client_state *csp) return cgi_error_memory(); } - if (enlist_unique_header(rsp->headers, "Location", new_url) - || (NULL == (rsp->status = strdup("302 Local Redirect from Privoxy")))) + rsp->status = strdup_or_die("302 Local Redirect from Privoxy"); + if (enlist_unique_header(rsp->headers, "Location", new_url)) { freez(new_url); free_http_response(rsp); @@ -1346,42 +1339,18 @@ struct http_response *redirect_url(struct client_state *csp) * * Function : is_imageurl * - * Description : Given a URL, decide whether it is an image or not, - * using either the info from a previous +image action - * or, #ifdef FEATURE_IMAGE_DETECT_MSIE, and the browser - * is MSIE and not on a Mac, tell from the browser's accept - * header. + * Description : Given a URL, decide whether it should be treated + * as image URL or not. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) * - * Returns : True (nonzero) if URL is an image, false (0) + * Returns : True (nonzero) if URL is an image URL, false (0) * otherwise * *********************************************************************/ int is_imageurl(const struct client_state *csp) { -#ifdef FEATURE_IMAGE_DETECT_MSIE - char *tmp; - - tmp = get_header_value(csp->headers, "User-Agent:"); - if (tmp && strstr(tmp, "MSIE") && !strstr(tmp, "Mac_")) - { - tmp = get_header_value(csp->headers, "Accept:"); - if (tmp && strstr(tmp, "image/gif")) - { - /* Client will accept HTML. If this seems counterintuitive, - * blame Microsoft. - */ - return(0); - } - else - { - return(1); - } - } -#endif /* def FEATURE_IMAGE_DETECT_MSIE */ - return ((csp->action->flags & ACTION_IMAGE) != 0); } @@ -1467,7 +1436,7 @@ int is_untrusted_url(const struct client_state *csp) { char * path; char * path_end; - char * new_entry = strdup("~"); + char * new_entry = strdup_or_die("~"); string_append(&new_entry, csp->http->hostport); @@ -1588,25 +1557,34 @@ struct re_filterfile_spec *get_filter(const struct client_state *csp, /********************************************************************* * - * Function : pcrs_filter_response + * Function : pcrs_filter_impl * * Description : Execute all text substitutions from all applying - * +filter actions on the text buffer that's been - * accumulated in csp->iob->buf. + * (based on filter_response_body value) +filter + * or +client_body_filter actions on the given buffer. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) + * 2 : filter_response_body = when TRUE execute +filter + * actions; execute +client_body_filter actions otherwise + * 3 : data = Target data + * 4 : data_len = Target data len * * Returns : a pointer to the (newly allocated) modified buffer. * or NULL if there were no hits or something went wrong * *********************************************************************/ -static char *pcrs_filter_response(struct client_state *csp) +static char *pcrs_filter_impl(const struct client_state *csp, int filter_response_body, + const char *data, size_t *data_len) { int hits = 0; size_t size, prev_size; + const int filters_idx = + filter_response_body ? ACTION_MULTI_FILTER : ACTION_MULTI_CLIENT_BODY_FILTER; + const enum filter_type filter_type = + filter_response_body ? FT_CONTENT_FILTER : FT_CLIENT_BODY_FILTER; - char *old = NULL; + const char *old = NULL; char *new = NULL; pcrs_job *job; @@ -1616,7 +1594,7 @@ static char *pcrs_filter_response(struct client_state *csp) /* * Sanity first */ - if (csp->iob->cur >= csp->iob->eod) + if (*data_len == 0) { return(NULL); } @@ -1628,15 +1606,15 @@ static char *pcrs_filter_response(struct client_state *csp) return(NULL); } - size = (size_t)(csp->iob->eod - csp->iob->cur); - old = csp->iob->cur; + size = *data_len; + old = data; /* - * For all applying +filter actions, look if a filter by that + * For all applying actions, look if a filter by that * name exists and if yes, execute it's pcrs_joblist on the * buffer. */ - for (filtername = csp->action->multi[ACTION_MULTI_FILTER]->first; + for (filtername = csp->action->multi[filters_idx]->first; filtername != NULL; filtername = filtername->next) { int current_hits = 0; /* Number of hits caused by this filter */ @@ -1644,7 +1622,7 @@ static char *pcrs_filter_response(struct client_state *csp) int job_hits = 0; /* How many hits the current job caused */ pcrs_job *joblist; - b = get_filter(csp, filtername->str, FT_CONTENT_FILTER); + b = get_filter(csp, filtername->str, filter_type); if (b == NULL) { continue; @@ -1675,7 +1653,7 @@ static char *pcrs_filter_response(struct client_state *csp) * input for the next one. */ current_hits += job_hits; - if (old != csp->iob->cur) + if (old != data) { freez(old); } @@ -1707,29 +1685,82 @@ static char *pcrs_filter_response(struct client_state *csp) if (b->dynamic) pcrs_free_joblist(joblist); - log_error(LOG_LEVEL_RE_FILTER, - "filtering %s%s (size %d) with \'%s\' produced %d hits (new size %d).", - csp->http->hostport, csp->http->path, prev_size, b->name, current_hits, size); - + if (filter_response_body) + { + log_error(LOG_LEVEL_RE_FILTER, + "filtering %s%s (size %lu) with \'%s\' produced %d hits (new size %lu).", + csp->http->hostport, csp->http->path, prev_size, b->name, current_hits, size); + } + else + { + log_error(LOG_LEVEL_RE_FILTER, "filtering request body from client %s " + "(size %lu) with \'%s\' produced %d hits (new size %lu).", + csp->ip_addr_str, prev_size, b->name, current_hits, size); + } +#ifdef FEATURE_EXTENDED_STATISTICS + update_filter_statistics(b->name, current_hits); +#endif hits += current_hits; } /* * If there were no hits, destroy our copy and let - * chat() use the original in csp->iob + * chat() use the original content */ if (!hits) { + if (old != data && old != new) + { + freez(old); + } freez(new); return(NULL); } - csp->flags |= CSP_FLAG_MODIFIED; - csp->content_length = size; - clear_iob(csp->iob); - + *data_len = size; return(new); +} + + +/********************************************************************* + * + * Function : pcrs_filter_response_body + * + * Description : Execute all text substitutions from all applying + * +filter actions on the text buffer that's been + * accumulated in csp->iob->buf. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * + * Returns : a pointer to the (newly allocated) modified buffer. + * or NULL if there were no hits or something went wrong + * + *********************************************************************/ +static char *pcrs_filter_response_body(struct client_state *csp) +{ + size_t size = (size_t)(csp->iob->eod - csp->iob->cur); + char *new = NULL; + + /* + * Sanity first + */ + if (csp->iob->cur >= csp->iob->eod) + { + return NULL; + } + + new = pcrs_filter_impl(csp, TRUE, csp->iob->cur, &size); + + if (new != NULL) + { + csp->flags |= CSP_FLAG_MODIFIED; + csp->content_length = size; + clear_iob(csp->iob); + } + + return new; } @@ -1790,6 +1821,7 @@ static void set_privoxy_variables(const struct client_state *csp) { "PRIVOXY_PATH", csp->http->path }, { "PRIVOXY_HOST", csp->http->host }, { "PRIVOXY_ORIGIN", csp->ip_addr_str }, + { "PRIVOXY_LISTEN_ADDRESS", csp->listen_addr_str }, }; for (i = 0; i < SZ(env); i++) @@ -1871,8 +1903,9 @@ static char *execute_external_filter(const struct client_state *csp, */ if ((*size != 0) && fwrite(content, *size, 1, fp) != 1) { - log_error(LOG_LEVEL_ERROR, "fwrite(..., %d, 1, ..) failed: %E", *size); + log_error(LOG_LEVEL_ERROR, "fwrite(..., %lu, 1, ..) failed: %E", *size); unlink(file_name); + fclose(fp); return NULL; } fclose(fp); @@ -1903,7 +1936,8 @@ static char *execute_external_filter(const struct client_state *csp, return NULL; } - filter_output = malloc_or_die(*size); + /* Allocate at least one byte */ + filter_output = malloc_or_die(*size + 1); new_size = 0; while (!feof(fp) && !ferror(fp)) @@ -1917,7 +1951,7 @@ static char *execute_external_filter(const struct client_state *csp, char *p; /* Could be considered wasteful if the content is 'large'. */ - *size = (*size != 0) ? *size * 2 : READ_LENGTH; + *size += (*size >= READ_LENGTH) ? *size : READ_LENGTH; p = realloc(filter_output, *size); if (p == NULL) @@ -1928,6 +1962,7 @@ static char *execute_external_filter(const struct client_state *csp, } filter_output = p; } + assert(new_size + READ_LENGTH < *size); len = fread(&filter_output[new_size], 1, READ_LENGTH, fp); if (len > 0) { @@ -1944,7 +1979,7 @@ static char *execute_external_filter(const struct client_state *csp, { log_error(LOG_LEVEL_RE_FILTER, "Executing '%s' resulted in return value %d. " - "Read %d of up to %d bytes.", name, (ret >> 8), new_size, *size); + "Read %lu of up to %lu bytes.", name, (ret >> 8), new_size, *size); } unlink(file_name); @@ -1956,6 +1991,28 @@ static char *execute_external_filter(const struct client_state *csp, #endif /* def FEATURE_EXTERNAL_FILTERS */ +/********************************************************************* + * + * Function : pcrs_filter_request_body + * + * Description : Execute all text substitutions from all applying + * +client_body_filter actions on the given text buffer. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * 2 : data = Target data + * 3 : data_len = Target data len + * + * Returns : a pointer to the (newly allocated) modified buffer. + * or NULL if there were no hits or something went wrong + * + *********************************************************************/ +static char *pcrs_filter_request_body(const struct client_state *csp, const char *data, size_t *data_len) +{ + return pcrs_filter_impl(csp, FALSE, data, data_len); +} + + /********************************************************************* * * Function : gif_deanimate_response @@ -1971,7 +2028,11 @@ static char *execute_external_filter(const struct client_state *csp, * or NULL in case something went wrong. * *********************************************************************/ +#ifdef FUZZ +char *gif_deanimate_response(struct client_state *csp) +#else static char *gif_deanimate_response(struct client_state *csp) +#endif { struct binbuffer *in, *out; char *p; @@ -1979,12 +2040,8 @@ static char *gif_deanimate_response(struct client_state *csp) size = (size_t)(csp->iob->eod - csp->iob->cur); - if ( (NULL == (in = (struct binbuffer *)zalloc(sizeof *in ))) - || (NULL == (out = (struct binbuffer *)zalloc(sizeof *out))) ) - { - log_error(LOG_LEVEL_DEANIMATE, "failed! (no mem)"); - return NULL; - } + in = zalloc_or_die(sizeof(*in)); + out = zalloc_or_die(sizeof(*out)); in->buffer = csp->iob->cur; in->size = size; @@ -2004,7 +2061,8 @@ static char *gif_deanimate_response(struct client_state *csp) } else { - log_error(LOG_LEVEL_DEANIMATE, "Success! GIF shrunk from %d bytes to %d.", size, out->offset); + log_error(LOG_LEVEL_DEANIMATE, + "Success! GIF shrunk from %lu bytes to %lu.", size, out->offset); } csp->content_length = out->offset; csp->flags |= CSP_FLAG_MODIFIED; @@ -2041,12 +2099,11 @@ static filter_function_ptr get_filter_function(const struct client_state *csp) * the content type and action settings. */ if ((csp->content_type & CT_TEXT) && - (csp->rlist != NULL) && (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) { - filter_function = pcrs_filter_response; + filter_function = pcrs_filter_response_body; } - else if ((csp->content_type & CT_GIF) && + else if ((csp->content_type & CT_GIF) && (csp->action->flags & ACTION_DEANIMATE)) { filter_function = gif_deanimate_response; @@ -2061,7 +2118,8 @@ static filter_function_ptr get_filter_function(const struct client_state *csp) * Function : remove_chunked_transfer_coding * * Description : In-situ remove the "chunked" transfer coding as defined - * in rfc2616 from a buffer. + * in RFC 7230 4.1 from a buffer. XXX: The implementation + * is neither complete nor compliant (TODO #129). * * Parameters : * 1 : buffer = Pointer to the text buffer @@ -2073,11 +2131,22 @@ static filter_function_ptr get_filter_function(const struct client_state *csp) * JB_ERR_PARSE otherwise * *********************************************************************/ +#ifdef FUZZ +extern jb_err remove_chunked_transfer_coding(char *buffer, size_t *size) +#else static jb_err remove_chunked_transfer_coding(char *buffer, size_t *size) +#endif { size_t newsize = 0; unsigned int chunksize = 0; char *from_p, *to_p; + const char *end_of_buffer = buffer + *size; + + if (*size == 0) + { + log_error(LOG_LEVEL_FATAL, "Invalid chunked input. Buffer is empty."); + return JB_ERR_PARSE; + } assert(buffer); from_p = to_p = buffer; @@ -2090,27 +2159,65 @@ static jb_err remove_chunked_transfer_coding(char *buffer, size_t *size) while (chunksize > 0U) { - if (NULL == (from_p = strstr(from_p, "\r\n"))) + /* + * If the chunk-size is valid, we should have at least + * chunk-size bytes of chunk-data and five bytes of + * meta data (chunk-size, CRLF, CRLF) left in the buffer. + */ + if (chunksize + 5 >= *size - newsize) { - log_error(LOG_LEVEL_ERROR, "Parse error while stripping \"chunked\" transfer coding"); + log_error(LOG_LEVEL_ERROR, + "Chunk size %u exceeds buffered data left. " + "Already digested %lu of %lu buffered bytes.", + chunksize, newsize, *size); return JB_ERR_PARSE; } - if (chunksize >= *size - newsize) + /* + * Skip the chunk-size, the optional chunk-ext and the CRLF + * that is supposed to be located directly before the start + * of chunk-data. + */ + if (NULL == (from_p = strstr(from_p, "\r\n"))) { log_error(LOG_LEVEL_ERROR, - "Chunk size %u exceeds buffered data left. " - "Already digested %u of %u buffered bytes.", - chunksize, (unsigned int)newsize, (unsigned int)*size); + "Failed to strip \"chunked\" transfer coding. " + "Line with chunk size doesn't seem to end properly."); return JB_ERR_PARSE; } - newsize += chunksize; from_p += 2; + /* + * The previous strstr() does not enforce chunk-validity + * and is sattisfied as long a CRLF is left in the buffer. + * + * Make sure the bytes we consider chunk-data are within + * the valid range. + */ + if (from_p + chunksize >= end_of_buffer) + { + log_error(LOG_LEVEL_ERROR, + "Failed to decode content for filtering. " + "One chunk end is beyond the end of the buffer."); + return JB_ERR_PARSE; + } + memmove(to_p, from_p, (size_t) chunksize); + newsize += chunksize; to_p = buffer + newsize; - from_p += chunksize + 2; + from_p += chunksize; + /* + * Not merging this check with the previous one allows us + * to keep chunks without trailing CRLF. It's not clear + * if we actually have to care about those, though. + */ + if (from_p + 2 >= end_of_buffer) + { + log_error(LOG_LEVEL_ERROR, "Not enough room for trailing CRLF."); + return JB_ERR_PARSE; + } + from_p += 2; if (sscanf(from_p, "%x", &chunksize) != 1) { log_error(LOG_LEVEL_INFO, "Invalid \"chunked\" transfer encoding detected and ignored."); @@ -2119,7 +2226,8 @@ static jb_err remove_chunked_transfer_coding(char *buffer, size_t *size) } /* XXX: Should get its own loglevel. */ - log_error(LOG_LEVEL_RE_FILTER, "De-chunking successful. Shrunk from %d to %d", *size, newsize); + log_error(LOG_LEVEL_RE_FILTER, + "De-chunking successful. Shrunk from %lu to %lu", *size, newsize); *size = newsize; @@ -2172,7 +2280,11 @@ static jb_err prepare_for_filtering(struct client_state *csp) * If the body has a supported transfer-encoding, * decompress it, adjusting size and iob->eod. */ - if (csp->content_type & (CT_GZIP|CT_DEFLATE)) + if ((csp->content_type & (CT_GZIP|CT_DEFLATE)) +#ifdef FEATURE_BROTLI + || (csp->content_type & CT_BROTLI) +#endif + ) { if (0 == csp->iob->eod - csp->iob->cur) { @@ -2190,11 +2302,14 @@ static jb_err prepare_for_filtering(struct client_state *csp) else { /* - * Unset CT_GZIP and CT_DEFLATE to remember not - * to modify the Content-Encoding header later. + * Unset content types to remember not to + * modify the Content-Encoding header later. */ csp->content_type &= ~CT_GZIP; csp->content_type &= ~CT_DEFLATE; +#ifdef FEATURE_BROTLI + csp->content_type &= ~CT_BROTLI; +#endif } } #endif @@ -2252,7 +2367,8 @@ char *execute_content_filters(struct client_state *csp) content = (content_filter != NULL) ? (*content_filter)(csp) : NULL; #ifdef FEATURE_EXTERNAL_FILTERS - if (!list_is_empty(csp->action->multi[ACTION_MULTI_EXTERNAL_FILTER])) + if ((csp->content_type & CT_TEXT) && + !list_is_empty(csp->action->multi[ACTION_MULTI_EXTERNAL_FILTER])) { struct list_entry *filtername; size_t size = (size_t)csp->content_length; @@ -2266,7 +2382,15 @@ char *execute_content_filters(struct client_state *csp) for (filtername = csp->action->multi[ACTION_MULTI_EXTERNAL_FILTER]->first; filtername ; filtername = filtername->next) { - content = execute_external_filter(csp, filtername->str, content, &size); + char *result = execute_external_filter(csp, filtername->str, content, &size); + if (result != NULL) + { + if (content != csp->iob->cur) + { + free(content); + } + content = result; + } } csp->flags |= CSP_FLAG_MODIFIED; csp->content_length = size; @@ -2278,6 +2402,46 @@ char *execute_content_filters(struct client_state *csp) } +/********************************************************************* + * + * Function : execute_client_body_filters + * + * Description : Executes client body filters for the request that is buffered + * in the client_iob. Upon success moves client_iob cur pointer + * to the end of the processed data. + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * 2 : content_length = content length. Upon successful filtering + * the passed value is updated with the new content length. + * + * Returns : Pointer to the modified buffer, or + * NULL if filtering failed or wasn't necessary. + * + *********************************************************************/ +char *execute_client_body_filters(struct client_state *csp, size_t *content_length) +{ + char *ret; + + assert(client_body_filters_enabled(csp->action)); + + if (content_length == 0) + { + /* + * No content, no filtering necessary. + */ + return NULL; + } + + ret = pcrs_filter_request_body(csp, csp->client_iob->cur, content_length); + if (ret != NULL) + { + csp->client_iob->cur = csp->client_iob->eod; + } + return ret; +} + + /********************************************************************* * * Function : get_url_actions @@ -2306,13 +2470,16 @@ void get_url_actions(struct client_state *csp, struct http_request *http) return; } +#ifdef FEATURE_CLIENT_TAGS + apply_url_actions(csp->action, http, csp->client_tags, b); +#else apply_url_actions(csp->action, http, b); +#endif } return; } - /********************************************************************* * * Function : apply_url_actions @@ -2322,14 +2489,18 @@ void get_url_actions(struct client_state *csp, struct http_request *http) * Parameters : * 1 : action = Destination. * 2 : http = Current URL - * 3 : b = list of URL actions to apply + * 3 : client_tags = list of client tags + * 4 : b = list of URL actions to apply * * Returns : N/A * *********************************************************************/ -void apply_url_actions(struct current_action_spec *action, - struct http_request *http, - struct url_actions *b) +static void apply_url_actions(struct current_action_spec *action, + struct http_request *http, +#ifdef FEATURE_CLIENT_TAGS + const struct list *client_tags, +#endif + struct url_actions *b) { if (b == NULL) { @@ -2343,6 +2514,12 @@ void apply_url_actions(struct current_action_spec *action, { merge_current_action(action, b->action); } +#ifdef FEATURE_CLIENT_TAGS + if (client_tag_match(b->url, client_tags)) + { + merge_current_action(action, b->action); + } +#endif } } @@ -2373,7 +2550,7 @@ void apply_url_actions(struct current_action_spec *action, * Invalid syntax is fatal. * *********************************************************************/ -const static struct forward_spec *get_forward_override_settings(struct client_state *csp) +static const struct forward_spec *get_forward_override_settings(struct client_state *csp) { const char *forward_override_line = csp->action->string[ACTION_STRING_FORWARD_OVERRIDE]; char forward_settings[BUFFER_SIZE]; @@ -2404,14 +2581,7 @@ const static struct forward_spec *get_forward_override_settings(struct client_st * the lifetime of this request. Save its location * in csp as well, so sweep() can free it later on. */ - fwd = csp->fwd = zalloc(sizeof(*fwd)); - if (NULL == fwd) - { - log_error(LOG_LEVEL_FATAL, - "can't allocate memory for forward-override{%s}", forward_override_line); - /* Never get here - LOG_LEVEL_FATAL causes program exit */ - return NULL; - } + fwd = csp->fwd = zalloc_or_die(sizeof(*fwd)); vec_count = ssplit(forward_settings, " \t", vec, SZ(vec)); if ((vec_count == 2) && !strcasecmp(vec[0], "forward")) @@ -2421,6 +2591,14 @@ const static struct forward_spec *get_forward_override_settings(struct client_st /* Parse the parent HTTP proxy host:port */ http_parent = vec[1]; + } + else if ((vec_count == 2) && !strcasecmp(vec[0], "forward-webserver")) + { + fwd->type = FORWARD_WEBSERVER; + + /* Parse the parent HTTP server host:port */ + http_parent = vec[1]; + } else if (vec_count == 3) { @@ -2449,10 +2627,11 @@ const static struct forward_spec *get_forward_override_settings(struct client_st if (NULL != socks_proxy) { - /* Parse the SOCKS proxy host[:port] */ + /* Parse the SOCKS proxy [user:pass@]host[:port] */ fwd->gateway_port = 1080; parse_forwarder_address(socks_proxy, - &fwd->gateway_host, &fwd->gateway_port); + &fwd->gateway_host, &fwd->gateway_port, + &fwd->auth_username, &fwd->auth_password); http_parent = vec[2]; } @@ -2470,7 +2649,8 @@ const static struct forward_spec *get_forward_override_settings(struct client_st { fwd->forward_port = 8000; parse_forwarder_address(http_parent, - &fwd->forward_host, &fwd->forward_port); + &fwd->forward_host, &fwd->forward_port, + NULL, NULL); } assert (NULL != fwd); @@ -2576,12 +2756,7 @@ struct http_response *direct_response(struct client_state *csp) return cgi_error_memory(); } - if (NULL == (rsp->status = strdup("501 Not Implemented"))) - { - free_http_response(rsp); - return cgi_error_memory(); - } - + rsp->status = strdup_or_die("501 Not Implemented"); rsp->is_static = 1; rsp->crunch_reason = UNSUPPORTED; @@ -2641,7 +2816,7 @@ int content_requires_filtering(struct client_state *csp) * The server didn't bother to declare a MIME-Type. * Assume it's text that can be filtered. * - * This also regulary happens with 304 responses, + * This also regularly happens with 304 responses, * therefore logging anything here would cause * too much noise. */ @@ -2653,8 +2828,8 @@ int content_requires_filtering(struct client_state *csp) * the content type and action settings. */ if ((csp->content_type & CT_TEXT) && - (csp->rlist != NULL) && - (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) + (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]) || + !list_is_empty(csp->action->multi[ACTION_MULTI_EXTERNAL_FILTER]))) { return TRUE; } @@ -2664,7 +2839,7 @@ int content_requires_filtering(struct client_state *csp) return TRUE; } - return (!list_is_empty(csp->action->multi[ACTION_MULTI_EXTERNAL_FILTER])); + return FALSE; } @@ -2690,6 +2865,25 @@ int content_filters_enabled(const struct current_action_spec *action) } +/********************************************************************* + * + * Function : client_body_filters_enabled + * + * Description : Checks whether there are any client body filters + * enabled for the current request. + * + * Parameters : + * 1 : action = Action spec to check. + * + * Returns : TRUE for yes, FALSE otherwise + * + *********************************************************************/ +int client_body_filters_enabled(const struct current_action_spec *action) +{ + return !list_is_empty(action->multi[ACTION_MULTI_CLIENT_BODY_FILTER]); +} + + /********************************************************************* * * Function : filters_available @@ -2716,6 +2910,283 @@ int filters_available(const struct client_state *csp) return FALSE; } +#ifdef FEATURE_EXTENDED_STATISTICS + +struct filter_statistics_entry +{ + char *filter; + unsigned long long executions; + unsigned long long response_bodies_modified; + unsigned long long hits; + + struct filter_statistics_entry *next; +}; + +static struct filter_statistics_entry *filter_statistics = NULL; + + +/********************************************************************* + * + * Function : register_filter_for_statistics + * + * Description : Registers a filter so we can gather statistics for + * it unless the filter has already been registered + * before. + * + * Parameters : + * 1 : filter = Name of the filter to register + * + * Returns : void + * + *********************************************************************/ +void register_filter_for_statistics(const char *filter) +{ + struct filter_statistics_entry *entry; + + privoxy_mutex_lock(&filter_statistics_mutex); + + if (filter_statistics == NULL) + { + filter_statistics = zalloc_or_die(sizeof(struct filter_statistics_entry)); + entry = filter_statistics; + entry->filter = strdup_or_die(filter); + privoxy_mutex_unlock(&filter_statistics_mutex); + return; + } + entry = filter_statistics; + while (entry != NULL) + { + if (!strcmp(entry->filter, filter)) + { + /* Already registered, nothing to do. */ + break; + } + if (entry->next == NULL) + { + entry->next = zalloc_or_die(sizeof(struct filter_statistics_entry)); + entry->next->filter = strdup_or_die(filter); + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&filter_statistics_mutex); + +} + + +/********************************************************************* + * + * Function : update_filter_statistics + * + * Description : Updates the statistics for a filter. + * + * Parameters : + * 1 : filter = Name of the filter to update + * 2 : hits = Hit count. + * + * Returns : void + * + *********************************************************************/ +void update_filter_statistics(const char *filter, int hits) +{ + struct filter_statistics_entry *entry; + + privoxy_mutex_lock(&filter_statistics_mutex); + + entry = filter_statistics; + while (entry != NULL) + { + if (!strcmp(entry->filter, filter)) + { + entry->executions++; + if (hits != 0) + { + entry->response_bodies_modified++; + entry->hits += (unsigned)hits; + } + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&filter_statistics_mutex); + +} + + +/********************************************************************* + * + * Function : get_filter_statistics + * + * Description : Gets the statistics for a filter. + * + * Parameters : + * 1 : filter = Name of the filter to get statistics for. + * 2 : executions = Storage for the execution count. + * 3 : response_bodies_modified = Storage for the number + * of modified response bodies. + * 4 : hits = Storage for the number of hits. + * + * Returns : void + * + *********************************************************************/ +void get_filter_statistics(const char *filter, unsigned long long *executions, + unsigned long long *response_bodies_modified, + unsigned long long *hits) +{ + struct filter_statistics_entry *entry; + + privoxy_mutex_lock(&filter_statistics_mutex); + + entry = filter_statistics; + while (entry != NULL) + { + if (!strcmp(entry->filter, filter)) + { + *executions = entry->executions; + *response_bodies_modified = entry->response_bodies_modified; + *hits = entry->hits; + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&filter_statistics_mutex); + +} + + +struct block_statistics_entry +{ + char *block_reason; + unsigned long long count; + + struct block_statistics_entry *next; +}; + +static struct block_statistics_entry *block_statistics = NULL; + +/********************************************************************* + * + * Function : register_block_reason_for_statistics + * + * Description : Registers a block reason so we can gather statistics + * for it unless the block reason has already been + * registered before. + * + * Parameters : + * 1 : block_reason = Block reason to register + * + * Returns : void + * + *********************************************************************/ +void register_block_reason_for_statistics(const char *block_reason) +{ + struct block_statistics_entry *entry; + + privoxy_mutex_lock(&block_reason_statistics_mutex); + + if (block_statistics == NULL) + { + block_statistics = zalloc_or_die(sizeof(struct block_statistics_entry)); + entry = block_statistics; + entry->block_reason = strdup_or_die(block_reason); + privoxy_mutex_unlock(&block_reason_statistics_mutex); + return; + } + entry = block_statistics; + while (entry != NULL) + { + if (!strcmp(entry->block_reason, block_reason)) + { + /* Already registered, nothing to do. */ + break; + } + if (entry->next == NULL) + { + entry->next = zalloc_or_die(sizeof(struct block_statistics_entry)); + entry->next->block_reason = strdup_or_die(block_reason); + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&block_reason_statistics_mutex); + +} + + +/********************************************************************* + * + * Function : increment_block_reason_counter + * + * Description : Updates the counter for a block reason. + * + * Parameters : + * 1 : block_reason = Block reason to count + * + * Returns : void + * + *********************************************************************/ +static void increment_block_reason_counter(const char *block_reason) +{ + struct block_statistics_entry *entry; + + privoxy_mutex_lock(&block_reason_statistics_mutex); + + entry = block_statistics; + while (entry != NULL) + { + if (!strcmp(entry->block_reason, block_reason)) + { + entry->count++; + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&block_reason_statistics_mutex); + +} + + +/********************************************************************* + * + * Function : get_block_reason_count + * + * Description : Gets number of times a block reason was used. + * + * Parameters : + * 1 : block_reason = Block reason to get statistics for. + * 2 : count = Storage for the number of times the block + * reason was used. + * + * Returns : void + * + *********************************************************************/ +void get_block_reason_count(const char *block_reason, unsigned long long *count) +{ + struct block_statistics_entry *entry; + + privoxy_mutex_lock(&block_reason_statistics_mutex); + + entry = block_statistics; + while (entry != NULL) + { + if (!strcmp(entry->block_reason, block_reason)) + { + *count = entry->count; + break; + } + entry = entry->next; + } + + privoxy_mutex_unlock(&block_reason_statistics_mutex); + +} + +#endif /* def FEATURE_EXTENDED_STATISTICS */ /* Local Variables: