X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=parsers.c;h=b65474ce3c3b3a7725c854c75e9410c4abfe1f54;hp=630537a38b735672b115465ff37b93a0856f6b72;hb=4490d451f9b61baada414233897a83ec8d9908aa;hpb=3eaf270b84f4baf1b84499f005749c0406ea3892 diff --git a/parsers.c b/parsers.c index 630537a3..b65474ce 100644 --- a/parsers.c +++ b/parsers.c @@ -1,12 +1,11 @@ -const char parsers_rcs[] = "$Id: parsers.c,v 1.298 2015/01/24 16:41:51 fabiankeil Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/parsers.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 @@ -90,8 +89,6 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.298 2015/01/24 16:41:51 fabiankei #include "strptime.h" #endif -const char parsers_h_rcs[] = PARSERS_H_VERSION; - static char *get_header_line(struct iob *iob); static jb_err scan_headers(struct client_state *csp); static jb_err header_tagger(struct client_state *csp, char *header); @@ -252,13 +249,14 @@ static const add_header_func_ptr add_server_headers[] = { /********************************************************************* * - * Function : flush_socket + * Function : flush_iob * * Description : Write any pending "buffered" content. * * Parameters : * 1 : fd = file descriptor of the socket to read * 2 : iob = The I/O buffer to flush, usually csp->iob. + * 3 : delay = Number of milliseconds to delay the writes * * Returns : On success, the number of bytes written are returned (zero * indicates nothing was written). On error, -1 is returned, @@ -268,7 +266,7 @@ static const add_header_func_ptr add_server_headers[] = { * file, the results are not portable. * *********************************************************************/ -long flush_socket(jb_socket fd, struct iob *iob) +long flush_iob(jb_socket fd, struct iob *iob, unsigned int delay) { long len = iob->eod - iob->cur; @@ -277,7 +275,7 @@ long flush_socket(jb_socket fd, struct iob *iob) return(0); } - if (write_socket(fd, iob->cur, (size_t)len)) + if (write_socket_delayed(fd, iob->cur, (size_t)len, delay)) { return(-1); } @@ -387,7 +385,7 @@ jb_err add_to_iob(struct iob *iob, const size_t buffer_limit, char *src, long n) void clear_iob(struct iob *iob) { free(iob->buf); - memset(iob, '\0', sizeof(*iob));; + memset(iob, '\0', sizeof(*iob)); } @@ -421,8 +419,13 @@ jb_err decompress_iob(struct client_state *csp) int status; /* return status of the inflate() call */ z_stream zstr; /* used by calls to zlib */ +#ifdef FUZZ + assert(csp->iob->cur - csp->iob->buf >= 0); + assert(csp->iob->eod - csp->iob->cur >= 0); +#else assert(csp->iob->cur - csp->iob->buf > 0); assert(csp->iob->eod - csp->iob->cur > 0); +#endif bufsize = csp->iob->size; skip_size = (size_t)(csp->iob->cur - csp->iob->buf); @@ -630,6 +633,7 @@ jb_err decompress_iob(struct client_state *csp) if (bufsize >= csp->config->buffer_limit) { log_error(LOG_LEVEL_ERROR, "Buffer limit reached while decompressing iob"); + freez(buf); return JB_ERR_MEMORY; } @@ -718,7 +722,7 @@ jb_err decompress_iob(struct client_state *csp) * Make sure the new uncompressed iob obeys some minimal * consistency conditions. */ - if ((csp->iob->buf < csp->iob->cur) + if ((csp->iob->buf <= csp->iob->cur) && (csp->iob->cur <= csp->iob->eod) && (csp->iob->eod <= csp->iob->buf + csp->iob->size)) { @@ -1182,6 +1186,59 @@ jb_err sed(struct client_state *csp, int filter_server_headers) } +#ifdef FEATURE_HTTPS_INSPECTION +/********************************************************************* + * + * Function : sed_https + * + * Description : add, delete or modify lines in the HTTPS client + * header streams. Wrapper around sed(). + * + * Parameters : + * 1 : csp = Current client state (buffers, headers, etc...) + * + * Returns : JB_ERR_OK in case off success, or + * JB_ERR_MEMORY on some out-of-memory errors, or + * JB_ERR_PARSE in case of fatal parse errors. + * + *********************************************************************/ +jb_err sed_https(struct client_state *csp) +{ + jb_err err; + struct list headers; + + /* + * Temporarly replace csp->headers with csp->https_headers + * to trick sed() into filtering the https headers. + */ + headers.first = csp->headers->first; + headers.last = csp->headers->last; + csp->headers->first = csp->https_headers->first; + csp->headers->last = csp->https_headers->last; + + /* + * Start with fresh tags. Already exising tags may + * be set again. This is necessary to overrule + * URL-based patterns. + */ + destroy_list(csp->tags); + + /* + * We want client header filters and taggers + * so temporarly remove the flag. + */ + csp->flags &= ~CSP_FLAG_CLIENT_HEADER_PARSING_DONE; + err = sed(csp, FILTER_CLIENT_HEADERS); + csp->flags |= CSP_FLAG_CLIENT_HEADER_PARSING_DONE; + + csp->headers->first = headers.first; + csp->headers->last = headers.last; + + return err; +} +#endif /* def FEATURE_HTTPS_INSPECTION */ + + /********************************************************************* * * Function : update_server_headers @@ -1811,7 +1868,9 @@ static jb_err client_keep_alive(struct client_state *csp, char **header) static jb_err get_content_length(const char *header_value, unsigned long long *length) { #ifdef _WIN32 - assert(sizeof(unsigned long long) > 4); +#if SIZEOF_LONG_LONG < 8 +#error sizeof(unsigned long long) too small +#endif if (1 != sscanf(header_value, "%I64u", length)) #else if (1 != sscanf(header_value, "%llu", length)) @@ -2372,8 +2431,7 @@ static jb_err server_content_encoding(struct client_state *csp, char **header) /* * Log a warning if the user expects the content to be filtered. */ - if ((csp->rlist != NULL) && - (!list_is_empty(csp->action->multi[ACTION_MULTI_FILTER]))) + if (content_filters_enabled(csp->action)) { log_error(LOG_LEVEL_INFO, "Compressed content detected, content filtering disabled. " @@ -3287,6 +3345,13 @@ static jb_err client_host(struct client_state *csp, char **header) { char *p, *q; + if (strlen(*header) < 7) + { + log_error(LOG_LEVEL_HEADER, "Removing empty Host header"); + freez(*header); + return JB_ERR_OK; + } + if (!csp->http->hostport || (*csp->http->hostport == '*') || *csp->http->hostport == ' ' || *csp->http->hostport == '\0') { @@ -3600,9 +3665,8 @@ static jb_err client_host_adder(struct client_state *csp) if (!csp->http->hostport || !*(csp->http->hostport)) { - /* XXX: When does this happen and why is it OK? */ - log_error(LOG_LEVEL_INFO, "Weirdness in client_host_adder detected and ignored."); - return JB_ERR_OK; + log_error(LOG_LEVEL_ERROR, "Destination host unknown."); + return JB_ERR_PARSE; } /* @@ -3795,7 +3859,8 @@ static jb_err server_proxy_connection_adder(struct client_state *csp) * Function : client_connection_header_adder * * Description : Adds a proper "Connection:" header to csp->headers - * unless the header was already present. Called from `sed'. + * unless the header was already present or it's a + * CONNECT request. Called from `sed'. * * Parameters : * 1 : csp = Current client state (buffers, headers, etc...) @@ -3814,10 +3879,20 @@ static jb_err client_connection_header_adder(struct client_state *csp) return JB_ERR_OK; } + /* + * In case of CONNECT requests "Connection: close" is implied, + * but actually setting the header has been reported to cause + * problems with some forwarding proxies that close the + * connection prematurely. + */ + if (csp->http->ssl != 0) + { + return JB_ERR_OK; + } + #ifdef FEATURE_CONNECTION_KEEP_ALIVE if ((csp->config->feature_flags & RUNTIME_FEATURE_CONNECTION_KEEP_ALIVE) && !(csp->flags & CSP_FLAG_SERVER_SOCKET_TAINTED) - && (csp->http->ssl == 0) && !strcmpic(csp->http->ver, "HTTP/1.1")) { csp->flags |= CSP_FLAG_CLIENT_CONNECTION_KEEP_ALIVE; @@ -4308,7 +4383,13 @@ static jb_err parse_header_time(const char *header_time, time_t *result) time_t result2; tm = gmtime(result); - strftime(recreated_date, sizeof(recreated_date), time_formats[i], tm); + if (!strftime(recreated_date, sizeof(recreated_date), + time_formats[i], tm)) + { + log_error(LOG_LEVEL_ERROR, "Failed to recreate date '%s' with '%s'.", + header_time, time_formats[i]); + continue; + } memset(&gmt, 0, sizeof(gmt)); if (NULL == strptime(recreated_date, time_formats[i], &gmt)) { @@ -4427,12 +4508,12 @@ jb_err get_destination_from_headers(const struct list *headers, struct http_requ } else { - http->port = http->ssl ? 443 : 80; + http->port = 80; } /* Rebuild request URL */ freez(http->url); - http->url = strdup(http->ssl ? "https://" : "http://"); + http->url = strdup("http://"); string_append(&http->url, http->hostport); string_append(&http->url, http->path); if (http->url == NULL) @@ -4440,14 +4521,113 @@ jb_err get_destination_from_headers(const struct list *headers, struct http_requ return JB_ERR_MEMORY; } - log_error(LOG_LEVEL_HEADER, "Destination extracted from \"Host:\" header. New request URL: %s", + log_error(LOG_LEVEL_HEADER, + "Destination extracted from \"Host\" header. New request URL: %s", http->url); + /* + * Regenerate request line in "proxy format" + * to make rewrites more convenient. + */ + assert(http->cmd != NULL); + freez(http->cmd); + http->cmd = strdup_or_die(http->gpc); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->url); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->ver); + if (http->cmd == NULL) + { + return JB_ERR_MEMORY; + } + return JB_ERR_OK; } +#ifdef FEATURE_HTTPS_INSPECTION +/********************************************************************* + * + * Function : get_destination_from_https_headers + * + * Description : Parse the previously encrypted "Host:" header to + * get the request's destination. + * + * Parameters : + * 1 : headers = List of headers (one of them hopefully being + * the "Host:" header) + * 2 : http = storage for the result (host, port and hostport). + * + * Returns : JB_ERR_MEMORY (or terminates) in case of memory problems, + * JB_ERR_PARSE if the host header couldn't be found, + * JB_ERR_OK otherwise. + * + *********************************************************************/ +jb_err get_destination_from_https_headers(const struct list *headers, struct http_request *http) +{ + char *q; + char *p; + char *host; + + host = get_header_value(headers, "Host:"); + + if (NULL == host) + { + log_error(LOG_LEVEL_ERROR, "No \"Host:\" header found."); + return JB_ERR_PARSE; + } + + p = strdup_or_die(host); + chomp(p); + q = strdup_or_die(p); + + freez(http->hostport); + http->hostport = p; + freez(http->host); + http->host = q; + q = strchr(http->host, ':'); + if (q != NULL) + { + /* Terminate hostname and evaluate port string */ + *q++ = '\0'; + http->port = atoi(q); + } + else + { + http->port = 443; + } + + /* Rebuild request URL */ + freez(http->url); + http->url = strdup_or_die(http->path); + + log_error(LOG_LEVEL_HEADER, + "Destination extracted from \"Host\" header. New request URL: %s", + http->url); + + /* + * Regenerate request line in "proxy format" + * to make rewrites more convenient. + */ + assert(http->cmd != NULL); + freez(http->cmd); + http->cmd = strdup_or_die(http->gpc); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->url); + string_append(&http->cmd, " "); + string_append(&http->cmd, http->ver); + if (http->cmd == NULL) + { + return JB_ERR_MEMORY; + } + + return JB_ERR_OK; + +} +#endif /* def FEATURE_HTTPS_INSPECTION */ + + /********************************************************************* * * Function : create_forged_referrer @@ -4598,7 +4778,14 @@ static jb_err handle_conditional_hide_referrer_parameter(char **header, static void create_content_length_header(unsigned long long content_length, char *header, size_t buffer_length) { +#ifdef _WIN32 +#if SIZEOF_LONG_LONG < 8 +#error sizeof(unsigned long long) too small +#endif + snprintf(header, buffer_length, "Content-Length: %I64u", content_length); +#else snprintf(header, buffer_length, "Content-Length: %llu", content_length); +#endif }