X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=parsers.c;h=6bf55ea00b21617bd2f718714062e5967292cd83;hp=48f927566cadeb801f63e80fabe9d1b78e9d8ce4;hb=54a73604bb9b22a8237eaf4078bcdf2a72767a78;hpb=d8341fe14a165d5ba6cd360060647bea02271da7 diff --git a/parsers.c b/parsers.c index 48f92756..6bf55ea0 100644 --- a/parsers.c +++ b/parsers.c @@ -1,4 +1,4 @@ -const char parsers_rcs[] = "$Id: parsers.c,v 1.31 2001/10/05 14:25:02 oes Exp $"; +const char parsers_rcs[] = "$Id: parsers.c,v 1.42 2001/11/22 21:59:30 jongfoster Exp $"; /********************************************************************* * * File : $Source: /cvsroot/ijbswa/current/parsers.c,v $ @@ -18,10 +18,10 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.31 2001/10/05 14:25:02 oes Exp $" * IJBSWA team. http://ijbswa.sourceforge.net * * Based on the Internet Junkbuster originally written - * by and Copyright (C) 1997 Anonymous Coders and + * by and Copyright (C) 1997 Anonymous Coders and * Junkbusters Corporation. http://www.junkbusters.com * - * This program is free software; you can redistribute it + * This program is free software; you can redistribute it * and/or modify it under the terms of the GNU General * Public License as published by the Free Software * Foundation; either version 2 of the License, or (at @@ -41,6 +41,76 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.31 2001/10/05 14:25:02 oes Exp $" * * Revisions : * $Log: parsers.c,v $ + * Revision 1.42 2001/11/22 21:59:30 jongfoster + * Adding code to handle +no-cookies-keep + * + * Revision 1.41 2001/11/05 23:43:05 steudten + * Add time+date to log files. + * + * Revision 1.40 2001/10/26 20:13:09 jongfoster + * ctype.h is needed in Windows, too. + * + * Revision 1.39 2001/10/26 17:40:04 oes + * Introduced get_header_value() + * Removed http->user_agent, csp->referrer and csp->accept_types + * Removed client_accept() + * + * Revision 1.38 2001/10/25 03:40:48 david__schmidt + * Change in porting tactics: OS/2's EMX porting layer doesn't allow multiple + * threads to call select() simultaneously. So, it's time to do a real, live, + * native OS/2 port. See defines for __EMX__ (the porting layer) vs. __OS2__ + * (native). Both versions will work, but using __OS2__ offers multi-threading. + * + * Revision 1.37 2001/10/23 21:36:02 jongfoster + * Documenting sed()'s error behaviou (doc change only) + * + * Revision 1.36 2001/10/13 12:51:51 joergs + * Removed client_host, (was only required for the old 2.0.2-11 http://noijb. + * force-load), instead crumble Host: and add it (again) in client_host_adder + * (in case we get a HTTP/1.0 request without Host: header and forward it to + * a HTTP/1.1 server/proxy). + * + * Revision 1.35 2001/10/09 22:39:21 jongfoster + * assert.h is also required under Win32, so moving out of #ifndef _WIN32 + * block. + * + * Revision 1.34 2001/10/07 18:50:55 oes + * Added server_content_encoding, renamed server_transfer_encoding + * + * Revision 1.33 2001/10/07 18:04:49 oes + * Changed server_http11 to server_http and its pattern to "HTTP". + * Additional functionality: it now saves the HTTP status into + * csp->http->status and sets CT_TABOO for Status 206 (partial range) + * + * Revision 1.32 2001/10/07 15:43:28 oes + * Removed FEATURE_DENY_GZIP and replaced it with client_accept_encoding, + * client_te and client_accept_encoding_adder, triggered by the new + * +no-compression action. For HTTP/1.1 the Accept-Encoding header is + * changed to allow only identity and chunked, and the TE header is + * crunched. For HTTP/1.0, Accept-Encoding is crunched. + * + * parse_http_request no longer does anything than parsing. The rewriting + * of http->cmd and version mangling are gone. It now also recognizes + * the put and delete methods and saves the url in http->url. Removed + * unused variable. + * + * renamed content_type and content_length to have the server_ prefix + * + * server_content_type now only works if csp->content_type != CT_TABOO + * + * added server_transfer_encoding, which + * - Sets CT_TABOO to prohibit filtering if encoding compresses + * - Raises the CSP_FLAG_CHUNKED flag if Encoding is "chunked" + * - Change from "chunked" to "identity" if body was chunked + * but has been de-chunked for filtering. + * + * added server_content_md5 which crunches any Content-MD5 headers + * if the body was modified. + * + * made server_http11 conditional on +downgrade action + * + * Replaced 6 boolean members of csp with one bitmap (csp->flags) + * * Revision 1.31 2001/10/05 14:25:02 oes * Crumble Keep-Alive from Server * @@ -252,14 +322,14 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.31 2001/10/05 14:25:02 oes Exp $" #ifndef _WIN32 #include #include -#include -#include -#include #endif +#include +#include +#include #include -#ifndef _WIN32 +#if !defined(_WIN32) && !defined(__OS2__) #include #endif @@ -271,6 +341,7 @@ const char parsers_rcs[] = "$Id: parsers.c,v 1.31 2001/10/05 14:25:02 oes Exp $" #include "errlog.h" #include "jbsockets.h" #include "miscutil.h" +#include "list.h" const char parsers_h_rcs[] = PARSERS_H_VERSION; @@ -280,8 +351,8 @@ const char parsers_h_rcs[] = PARSERS_H_VERSION; * as an array index. Therefore we need to make sure that high-bit * characters generate +ve values, and ideally we also want to make * the argument match the declared parameter type of "int". - * - * Why did they write a character function that can't take a simple + * + * Why did they write a character function that can't take a simple * "char" argument? Doh! */ #define ijb_isupper(__X) isupper((int)(unsigned char)(__X)) @@ -297,45 +368,42 @@ const struct parsers client_patterns[] = { { "x-forwarded-for:", 16, client_x_forwarded }, { "Accept-Encoding:", 16, client_accept_encoding }, { "TE:", 3, client_te }, -#if defined(FEATURE_IMAGE_DETECT_MSIE) - { "Accept:", 7, client_accept }, -#endif /* defined(FEATURE_IMAGE_DETECT_MSIE) */ -#ifdef FEATURE_FORCE_LOAD - { "Host:", 5, client_host }, -#endif /* def FEATURE_FORCE_LOAD */ + { "Host:", 5, crumble }, /* { "if-modified-since:", 18, crumble }, */ { "Keep-Alive:", 11, crumble }, { "connection:", 11, crumble }, - { "proxy-connection:", 17, crumble }, + { "proxy-connection:", 17, crumble }, { NULL, 0, NULL } }; const struct parsers server_patterns[] = { - { "HTTP/1.1 ", 9, server_http11 }, + { "HTTP", 4, server_http }, { "set-cookie:", 11, server_set_cookie }, { "connection:", 11, crumble }, { "Content-Type:", 13, server_content_type }, { "Content-Length:", 15, server_content_length }, { "Content-MD5:", 12, server_content_md5 }, - { "Transfer-Encoding:", 18, server_transfer_encoding }, + { "Content-Encoding:", 17, server_content_encoding }, + { "Transfer-Encoding:", 18, server_transfer_coding }, { "Keep-Alive:", 11, crumble }, { NULL, 0, NULL } }; void (* const add_client_headers[])(struct client_state *) = { + client_host_adder, client_cookie_adder, client_x_forwarded_adder, client_xtra_adder, client_accept_encoding_adder, - connection_close_adder, + connection_close_adder, NULL }; void (* const add_server_headers[])(struct client_state *) = { - connection_close_adder, + connection_close_adder, NULL }; @@ -496,6 +564,56 @@ char *get_header(struct client_state *csp) } +/********************************************************************* + * + * Function : get_header_value + * + * Description : Get the value of a given header from a chained list + * of header lines or return NULL if no such header is + * present in the list. + * + * Parameters : + * 1 : header_list = pointer to list + * 2 : header_name = string with name of header to look for. + * Trailing colon required, capitalization + * doesn't matter. + * + * Returns : NULL if not found, else value of header + * + *********************************************************************/ +char *get_header_value(const struct list *header_list, const char *header_name) +{ + struct list_entry *cur_entry; + char *ret = NULL; + size_t length = 0; + + assert(header_list); + assert(header_name); + length = strlen(header_name); + + for (cur_entry = header_list->first; cur_entry ; cur_entry = cur_entry->next) + { + if (cur_entry->str) + { + if (!strncmpic(cur_entry->str, header_name, length)) + { + /* + * Found: return pointer to start of value + */ + ret = (char *) (cur_entry->str + length); + while (*ret && ijb_isspace(*ret)) ret++; + return(ret); + } + } + } + + /* + * Not found + */ + return NULL; + +} + /********************************************************************* * * Function : sed @@ -514,7 +632,8 @@ char *get_header(struct client_state *csp) * headers (client or server) * 3 : csp = Current client state (buffers, headers, etc...) * - * Returns : Single pointer to a fully formed header. + * Returns : Single pointer to a fully formed header, or NULL + * on out-of-memory error. * *********************************************************************/ char *sed(const struct parsers pats[], void (* const more_headers[])(struct client_state *), struct client_state *csp) @@ -536,7 +655,7 @@ char *sed(const struct parsers pats[], void (* const more_headers[])(struct clie if (strncmpic(p->str, v->str, v->len) == 0) { hdr = v->parser(v, p->str, csp); - freez((char *)p->str); /* FIXME: Yuck! patching a list...*/ + freez(p->str); /* FIXME: Yuck! patching a list...*/ p->str = hdr; } } @@ -550,7 +669,7 @@ char *sed(const struct parsers pats[], void (* const more_headers[])(struct clie hdr = list_to_text(csp->headers); - return(hdr); + return hdr; } @@ -579,7 +698,6 @@ void free_http_request(struct http_request *http) freez(http->path); freez(http->ver); freez(http->host_ip_addr_str); - freez(http->user_agent); } @@ -605,7 +723,7 @@ void parse_http_request(char *req, struct http_request *http, struct client_stat int n; memset(http, '\0', sizeof(*http)); - http->cmd = strdup(req); + http->cmd = strdup(req); buf = strdup(req); n = ssplit(buf, " \r\n", v, SZ(v), 1, 1); @@ -665,7 +783,7 @@ void parse_http_request(char *req, struct http_request *http, struct client_stat *p = '\0'; http->hostport = strdup(url); } - /* + /* * Repair broken HTTP requests that don't contain a path */ else @@ -789,10 +907,10 @@ char *server_content_type(const struct parsers *v, const char *s, struct client_ /********************************************************************* * - * Function : server_transfer_encoding + * Function : server_transfer_coding * - * Description : - Prohibit filtering (CT_TABOO) if encoding compresses - * - Raise the CSP_FLAG_CHUNKED flag if Encoding is "chunked" + * 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. * @@ -804,7 +922,7 @@ char *server_content_type(const struct parsers *v, const char *s, struct client_ * Returns : A duplicate string pointer to this header (ie. pass thru) * *********************************************************************/ -char *server_transfer_encoding(const struct parsers *v, const char *s, struct client_state *csp) +char *server_transfer_coding(const struct parsers *v, const char *s, struct client_state *csp) { /* * Turn off pcrs and gif filtering if body compressed @@ -814,7 +932,7 @@ char *server_transfer_encoding(const struct parsers *v, const char *s, struct cl csp->content_type = CT_TABOO; } - /* + /* * Raise flag if body chunked */ if (strstr(s, "chunked")) @@ -822,7 +940,7 @@ char *server_transfer_encoding(const struct parsers *v, const char *s, struct cl csp->flags |= CSP_FLAG_CHUNKED; /* - * If the body was modified, it has been + * If the body was modified, it has been * de-chunked first, so adjust the header: */ if (csp->flags & CSP_FLAG_MODIFIED) @@ -836,6 +954,35 @@ char *server_transfer_encoding(const struct parsers *v, const char *s, struct cl } +/********************************************************************* + * + * Function : server_content_encoding + * + * Description : Prohibit filtering (CT_TABOO) if content encoding compresses + * + * Parameters : + * 1 : v = ignored + * 2 : s = header string we are "considering" + * 3 : csp = Current client state (buffers, headers, etc...) + * + * Returns : A duplicate string pointer to this header (ie. pass thru) + * + *********************************************************************/ +char *server_content_encoding(const struct parsers *v, const char *s, struct client_state *csp) +{ + /* + * Turn off pcrs and gif filtering if body compressed + */ + if (strstr(s, "gzip") || strstr(s, "compress") || strstr(s, "deflate")) + { + csp->content_type = CT_TABOO; + } + + return(strdup(s)); + +} + + /********************************************************************* * * Function : server_content_length @@ -990,15 +1137,11 @@ char *client_referrer(const struct parsers *v, const char *s, struct client_stat #ifdef FEATURE_FORCE_LOAD /* Since the referrer can include the prefix even * even if the request itself is non-forced, we must - * clean it unconditionally + * clean it unconditionally */ strclean(s, FORCE_PREFIX); #endif /* def FEATURE_FORCE_LOAD */ -#ifdef FEATURE_TRUST - csp->referrer = strdup(s); -#endif /* def FEATURE_TRUST */ - /* * Are we sending referer? */ @@ -1092,22 +1235,6 @@ char *client_uagent(const struct parsers *v, const char *s, struct client_state const char * newval; char * s2; - /* Save the client's User-Agent: value */ - if (strlen(s) >= 12) - { - csp->http->user_agent = strdup(s + 12); - } - -#ifdef FEATURE_IMAGE_DETECT_MSIE - if (strstr (s, "MSIE ")) - { - /* This is Microsoft Internet Explorer. - * Enable auto-detect. - */ - csp->accept_types |= ACCEPT_TYPE_IS_MSIE; - } -#endif /* def FEATURE_IMAGE_DETECT_MSIE */ - if ((csp->action->flags & ACTION_HIDE_USER_AGENT) == 0) { return(strdup(s)); @@ -1270,48 +1397,32 @@ char *client_x_forwarded(const struct parsers *v, const char *s, struct client_s } -#if defined(FEATURE_IMAGE_DETECT_MSIE) +/* the following functions add headers directly to the header list */ + /********************************************************************* * - * Function : client_accept + * Function : client_host_adder * - * Description : Detect whether the client wants HTML or an image. - * Clients do not always make this information available - * in a sane way. Always passes the header through - * the proxy unchanged. + * Description : (re)adds the host header. Called from `sed'. * * Parameters : - * 1 : v = Ignored. - * 2 : s = Header string. Null terminated. - * 3 : csp = Current client state (buffers, headers, etc...) + * 1 : csp = Current client state (buffers, headers, etc...) * - * Returns : Duplicate of argument s. + * Returns : N/A * *********************************************************************/ -char *client_accept(const struct parsers *v, const char *s, struct client_state *csp) +void client_host_adder(struct client_state *csp) { -#ifdef FEATURE_IMAGE_DETECT_MSIE - if (strstr (s, "image/gif")) - { - /* Client will accept HTML. If this seems counterintuitive, - * blame Microsoft. - */ - csp->accept_types |= ACCEPT_TYPE_MSIE_HTML; - } - else - { - csp->accept_types |= ACCEPT_TYPE_MSIE_IMAGE; - } -#endif /* def FEATURE_IMAGE_DETECT_MSIE */ - - return(strdup(s)); - -} -#endif /* defined(FEATURE_IMAGE_DETECT_MSIE) */ + char *p = NULL; + p = strsav(p, "Host: "); + p = strsav(p, csp->http->hostport); + log_error(LOG_LEVEL_HEADER, "addh: %s", p); + enlist(csp->headers, p); -/* the following functions add headers directly to the header list */ + freez(p); +} /********************************************************************* @@ -1485,10 +1596,13 @@ void connection_close_adder(struct client_state *csp) /********************************************************************* * - * Function : server_http11 + * Function : server_http * - * Description : Rewrite HTTP/1.1 answers to HTTP/1.0 if +downgrade - * action applies. + * Description : - Save the HTTP Status into csp->http->status + * - Set CT_TABOO to prevent filtering if the answer + * is a partial range (HTTP status 206) + * - Rewrite HTTP/1.1 answers to HTTP/1.0 if +downgrade + * action applies. * * Parameters : * 1 : v = parser pattern that matched this header @@ -1498,22 +1612,22 @@ void connection_close_adder(struct client_state *csp) * Returns : Copy of changed or original answer. * *********************************************************************/ -char *server_http11(const struct parsers *v, const char *s, struct client_state *csp) +char *server_http(const struct parsers *v, const char *s, struct client_state *csp) { - char *ret; + char *ret = strdup(s); - if ((csp->action->flags & ACTION_DOWNGRADE) != 0) + sscanf(ret, "HTTP/%*d.%*d %d", &(csp->http->status)); + if (csp->http->status == 206) { - /* "HTTP/1.1 ..." -> "HTTP/1.0 ..." */ - ret = strdup(s); - ret[7] = '0'; - - return(ret); + csp->content_type = CT_TABOO; } - else + + if ((csp->action->flags & ACTION_DOWNGRADE) != 0) { - return(strdup(s)); + ret[7] = '0'; + log_error(LOG_LEVEL_HEADER, "Downgraded answer to HTTP/1.0"); } + return(ret); } @@ -1539,7 +1653,20 @@ char *server_set_cookie(const struct parsers *v, const char *s, struct client_st #ifdef FEATURE_COOKIE_JAR if (csp->config->jar) { - fprintf(csp->config->jar, "%s\t%s\n", csp->http->host, (s + v->len + 1)); + /* + * Write timestamp into outbuf. + * + * Complex because not all OSs have tm_gmtoff or + * the %z field in strftime() + */ + char tempbuf[ BUFFER_SIZE ]; + time_t now; + struct tm *tm_now; + time (&now); + tm_now = localtime (&now); + strftime (tempbuf, BUFFER_SIZE-6, "%b %d %H:%M:%S ", tm_now); + + fprintf(csp->config->jar, "%s %s\t%s\n", tempbuf, csp->http->host, (s + v->len + 1)); } #endif /* def FEATURE_COOKIE_JAR */ @@ -1547,48 +1674,95 @@ char *server_set_cookie(const struct parsers *v, const char *s, struct client_st { return(crumble(v, s, csp)); } + else if ((csp->action->flags & ACTION_NO_COOKIE_KEEP) != 0) + { + /* Flag whether or not to log a message */ + int changed = 0; - return(strdup(s)); + /* A variable to store the tag we're working on */ + char * cur_tag; -} + /* Make a copy of the header we can write to */ + char * result = strdup(s); + if (result == NULL) + { + /* FIXME: This error handling is incorrect */ + return NULL; + } + /* Skip "Set-Cookie:" (11 characters) in header */ + cur_tag = result + 11; -#ifdef FEATURE_FORCE_LOAD -/********************************************************************* - * - * Function : client_host - * - * Description : Clean the FORCE_PREFIX out of the 'host' http - * header, if we use force - * - * Parameters : - * 1 : v = ignored - * 2 : s = header (from sed) to clean - * 3 : csp = Current client state (buffers, headers, etc...) - * - * Returns : A malloc'ed pointer to the cleaned host header - * - *********************************************************************/ -char *client_host(const struct parsers *v, const char *s, struct client_state *csp) -{ - char *cleanhost = strdup(s); - - if(csp->flags & CSP_FLAG_FORCED) - { - strclean(cleanhost, FORCE_PREFIX); + /* skip whitespace between "Set-Cookie:" and value */ + while (*cur_tag && ijb_isspace(*cur_tag)) + { + cur_tag++; + } + + /* Loop through each tag in the cookie */ + while (*cur_tag) + { + /* Find next tag */ + char * next_tag = strchr(cur_tag, ';'); + if (next_tag != NULL) + { + /* Skip the ';' character itself */ + next_tag++; + + /* skip whitespace ";" and start of tag */ + while (*next_tag && ijb_isspace(*next_tag)) + { + next_tag++; + } + } + else + { + /* "Next tag" is the end of the string */ + next_tag = cur_tag + strlen(cur_tag); + } + + /* Is this the "Expires" tag? */ + if (strncmpic(cur_tag, "expires=", 8) == 0) + { + /* Delete the tag by copying the rest of the string over it. + * (Note that we cannot just use "strcpy(cur_tag, next_tag)", + * since the behaviour of strcpy is undefined for overlapping + * strings.) + */ + memmove(cur_tag, next_tag, strlen(next_tag) + 1); + + /* That changed the header, need to issue a log message */ + changed = 1; + + /* Note that the next tag has now been moved to *cur_tag, + * so we do not need to update the cur_tag pointer. + */ + } + else + { + /* Move on to next cookie tag */ + cur_tag = next_tag; + } + } + + if (changed) + { + log_error(LOG_LEVEL_HEADER, "Changed cookie to a temporary one."); + } + + return result; } - - return(cleanhost); + + return(strdup(s)); } -#endif /* def FEATURE_FORCE_LOAD */ - - -#ifdef FEATURE_FORCE_LOAD + + +#ifdef FEATURE_FORCE_LOAD /********************************************************************* * * Function : strclean * - * Description : In-Situ-Eliminate all occurances of substring in + * Description : In-Situ-Eliminate all occurances of substring in * string * * Parameters : @@ -1608,7 +1782,7 @@ int strclean(const char *string, const char *substring) p = pos + len; do { - *(p - len) = *p; + *(p - len) = *p; } while (*p++ != '\0');