-const char parsers_rcs[] = "$Id: parsers.c,v 1.20 2001/07/30 22:08:36 jongfoster Exp $";
+const char parsers_rcs[] = "$Id: parsers.c,v 1.45 2002/01/09 14:33:03 oes Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/parsers.c,v $
* `client_uagent', `client_x_forwarded',
* `client_x_forwarded_adder', `client_xtra_adder',
* `content_type', `crumble', `destroy_list', `enlist',
- * `flush_socket', `free_http_request', `get_header',
- * `list_to_text', `parse_http_request', `sed',
+ * `flush_socket', ``get_header', `sed',
* and `server_set_cookie'.
*
* Copyright : Written by and Copyright (C) 2001 the SourceForge
* 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
*
* Revisions :
* $Log: parsers.c,v $
+ * Revision 1.45 2002/01/09 14:33:03 oes
+ * Added support for localtime_r.
+ *
+ * Revision 1.44 2001/12/14 01:22:54 steudten
+ * Remove 'user:pass@' from 'proto://user:pass@host' for the
+ * new added header 'Host: ..'. (See Req ID 491818)
+ *
+ * Revision 1.43 2001/11/23 00:26:38 jongfoster
+ * Fixing two really stupid errors in my previous commit
+ *
+ * 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
+ *
+ * Revision 1.30 2001/09/29 12:56:03 joergs
+ * IJB now changes HTTP/1.1 to HTTP/1.0 in requests and answers.
+ *
+ * Revision 1.29 2001/09/24 21:09:24 jongfoster
+ * Fixing 2 memory leaks that Guy spotted, where the paramater to
+ * enlist() was not being free()d.
+ *
+ * Revision 1.28 2001/09/22 16:32:28 jongfoster
+ * Removing unused #includes.
+ *
+ * Revision 1.27 2001/09/20 15:45:25 steudten
+ *
+ * add casting from size_t to int for printf()
+ * remove local variable shadow s2
+ *
+ * Revision 1.26 2001/09/16 17:05:14 jongfoster
+ * Removing unused #include showarg.h
+ *
+ * Revision 1.25 2001/09/16 13:21:27 jongfoster
+ * Changes to use new list functions.
+ *
+ * Revision 1.24 2001/09/13 23:05:50 jongfoster
+ * Changing the string paramater to the header parsers a "const".
+ *
+ * Revision 1.23 2001/09/12 18:08:19 steudten
+ *
+ * In parse_http_request() header rewriting miss the host value, so
+ * from http://www.mydomain.com the result was just " / " not
+ * http://www.mydomain.com/ in case we forward.
+ *
+ * Revision 1.22 2001/09/10 10:58:53 oes
+ * Silenced compiler warnings
+ *
+ * Revision 1.21 2001/07/31 14:46:00 oes
+ * - Persistant connections now suppressed
+ * - sed() no longer appends empty header to csp->headers
+ *
* Revision 1.20 2001/07/30 22:08:36 jongfoster
* Tidying up #defines:
* - All feature #defines are now of the form FEATURE_xxx
* CRLF -> LF
*
* Revision 1.11 2001/05/29 20:11:19 joergs
- * '/* inside comment' warning removed.
+ * '/ * inside comment' warning removed.
*
* Revision 1.10 2001/05/29 09:50:24 jongfoster
* Unified blocklist/imagelist/permissionslist.
#include "config.h"
+#ifndef _WIN32
#include <stdio.h>
#include <sys/types.h>
+#endif
+
#include <stdlib.h>
#include <ctype.h>
+#include <assert.h>
#include <string.h>
-#ifndef _WIN32
+#if !defined(_WIN32) && !defined(__OS2__)
#include <unistd.h>
#endif
#include "list.h"
#include "parsers.h"
#include "encode.h"
-#include "filters.h"
-#include "loaders.h"
-#include "showargs.h"
-#include "jcc.h"
#include "ssplit.h"
#include "errlog.h"
#include "jbsockets.h"
#include "miscutil.h"
-#include "cgi.h"
+#include "list.h"
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))
{ "from:", 5, client_from },
{ "cookie:", 7, client_send_cookie },
{ "x-forwarded-for:", 16, client_x_forwarded },
-#ifdef FEATURE_DENY_GZIP
- { "Accept-Encoding: gzip", 21, crumble },
-#endif /* def FEATURE_DENY_GZIP */
-#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 */
+ { "Accept-Encoding:", 16, client_accept_encoding },
+ { "TE:", 3, client_te },
+ { "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", 4, server_http },
{ "set-cookie:", 11, server_set_cookie },
{ "connection:", 11, crumble },
- { "Content-Type:", 13, content_type },
- { "Content-Length:", 15, content_length },
+ { "Content-Type:", 13, server_content_type },
+ { "Content-Length:", 15, server_content_length },
+ { "Content-MD5:", 12, server_content_md5 },
+ { "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,
- connection_close_adder,
+ client_accept_encoding_adder,
+ connection_close_adder,
NULL
};
void (* const add_server_headers[])(struct client_state *) = {
- connection_close_adder,
+ connection_close_adder,
NULL
};
}
+/*********************************************************************
+ *
+ * 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
* 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)
{
- struct list *p;
+ struct list_entry *p;
const struct parsers *v;
char *hdr;
void (* const *f)();
for (v = pats; v->str ; v++)
{
- for (p = csp->headers->next; p ; p = p->next)
+ for (p = csp->headers->first; p ; p = p->next)
{
/* Header crunch()ed in previous run? -> ignore */
if (p->str == NULL) continue;
if (strncmpic(p->str, v->str, v->len) == 0)
{
hdr = v->parser(v, p->str, csp);
- freez(p->str);
+ freez(p->str); /* FIXME: Yuck! patching a list...*/
p->str = hdr;
}
}
hdr = list_to_text(csp->headers);
- return(hdr);
+ return hdr;
}
+/* here begins the family of parser functions that reformat header lines */
+
+
/*********************************************************************
*
- * Function : free_http_request
+ * Function : crumble
*
- * Description : Freez a http_request structure
+ * Description : This is called if a header matches a pattern to "crunch"
*
* Parameters :
- * 1 : http = points to a http_request structure to free
+ * 1 : v = Pointer to parsers structure, which basically holds
+ * headers (client or server) that we want to "crunch"
+ * 2 : s = header (from sed) to "crunch"
+ * 3 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : Always NULL.
*
*********************************************************************/
-void free_http_request(struct http_request *http)
+char *crumble(const struct parsers *v, const char *s, struct client_state *csp)
{
- freez(http->cmd);
- freez(http->gpc);
- freez(http->host);
- freez(http->hostport);
- freez(http->path);
- freez(http->ver);
- freez(http->host_ip_addr_str);
- freez(http->user_agent);
+ log_error(LOG_LEVEL_HEADER, "crunch!");
+ return(NULL);
}
/*********************************************************************
*
- * Function : parse_http_request
+ * Function : server_content_type
*
- * Description : Parse out the host and port from the URL. Find the
- * hostname & path, port (if ':'), and/or password (if '@')
+ * Description : Set the content-type for filterable types (text/.*,
+ * javascript and image/gif) unless filtering has been
+ * forbidden (CT_TABOO) while parsing earlier headers.
*
* Parameters :
- * 1 : req = URL (or is it URI?) to break down
- * 2 : http = pointer to the http structure to hold elements
+ * 1 : v = ignored
+ * 2 : s = header string we are "considering"
* 3 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : N/A
+ * Returns : A duplicate string pointer to this header (ie. pass thru)
*
*********************************************************************/
-void parse_http_request(char *req, struct http_request *http, struct client_state *csp)
+char *server_content_type(const struct parsers *v, const char *s, struct client_state *csp)
{
- char *buf, *v[10], *url, *p;
- int n;
+ if (csp->content_type != CT_TABOO)
+ {
+ if (strstr(s, " text/") || strstr(s, "application/x-javascript"))
+ csp->content_type = CT_TEXT;
+ else if (strstr(s, " image/gif"))
+ csp->content_type = CT_GIF;
+ else
+ csp->content_type = 0;
+ }
- memset(http, '\0', sizeof(*http));
+ return(strdup(s));
- http->cmd = strdup(req);
+}
- buf = strdup(req);
- n = ssplit(buf, " \r\n", v, SZ(v), 1, 1);
- if (n == 3)
+/*********************************************************************
+ *
+ * Function : server_transfer_coding
+ *
+ * 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.
+ *
+ * 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_transfer_coding(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"))
{
- /* this could be a CONNECT request */
- if (strcmpic(v[0], "connect") == 0)
- {
- http->ssl = 1;
- http->gpc = strdup(v[0]);
- http->hostport = strdup(v[1]);
- http->ver = strdup(v[2]);
- }
-
- /* or it could be a GET or a POST (possibly webDAV too) */
- if ((0 == strcmpic(v[0], "get"))
- || (0 == strcmpic(v[0], "head"))
- || (0 == strcmpic(v[0], "post"))
- /* These are the headers as defined in RFC2518 to add webDAV support: */
- || (0 == strcmpic(v[0], "propfind"))
- || (0 == strcmpic(v[0], "proppatch"))
- || (0 == strcmpic(v[0], "move"))
- || (0 == strcmpic(v[0], "copy"))
- || (0 == strcmpic(v[0], "mkcol"))
- || (0 == strcmpic(v[0], "lock"))
- || (0 == strcmpic(v[0], "unlock"))
- )
- {
- http->ssl = 0;
- http->gpc = strdup(v[0]);
- url = v[1];
- http->ver = strdup(v[2]);
+ csp->content_type = CT_TABOO;
+ }
- if (strncmpic(url, "http://", 7) == 0)
- {
- url += 7;
- }
- else if (strncmpic(url, "https://", 8) == 0)
- {
- url += 8;
- }
- else
- {
- url = NULL;
- }
+ /*
+ * Raise flag if body chunked
+ */
+ if (strstr(s, "chunked"))
+ {
+ csp->flags |= CSP_FLAG_CHUNKED;
- if (url)
- {
- if (p = strchr(url, '/'))
- {
- http->path = strdup(p);
- *p = '\0';
- http->hostport = strdup(url);
- }
- /*
- * Repair broken HTTP requests that don't contain a path
- */
- else
- {
- /* Repair hostport & path */
- http->path = strdup("/");
- http->hostport = strdup(url);
-
- /* Even repair cmd in case we're just forwarding. Boy are we nice ;-) */
- freez(http->cmd);
- http->cmd = strsav(http->cmd, http->gpc);
- http->cmd = strsav(http->cmd, " / ");
- http->cmd = strsav(http->cmd, http->ver);
- }
- }
+ /*
+ * If the body was modified, it has been
+ * de-chunked first, so adjust the header:
+ */
+ if (csp->flags & CSP_FLAG_MODIFIED)
+ {
+ return(strdup("Transfer-Encoding: identity"));
}
}
- freez(buf);
-
-
- if (http->hostport == NULL)
- {
- free_http_request(http);
- return;
- }
+ return(strdup(s));
- buf = strdup(http->hostport);
+}
- /* check if url contains password */
- n = ssplit(buf, "@", v, SZ(v), 1, 1);
- if (n == 2)
+/*********************************************************************
+ *
+ * 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"))
{
- char * newbuf = NULL;
- newbuf = strdup(v[1]);
- freez(buf);
- buf = newbuf;
+ csp->content_type = CT_TABOO;
}
- n = ssplit(buf, ":", v, SZ(v), 1, 1);
-
- if (n == 1)
- {
- http->host = strdup(v[0]);
- http->port = 80;
- }
+ return(strdup(s));
- if (n == 2)
- {
- http->host = strdup(v[0]);
- http->port = atoi(v[1]);
- }
+}
- freez(buf);
- if (http->host == NULL)
+/*********************************************************************
+ *
+ * Function : server_content_length
+ *
+ * Description : Adjust Content-Length header if we modified
+ * the body.
+ *
+ * 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_length(const struct parsers *v, const char *s, struct client_state *csp)
+{
+ if (csp->content_length != 0) /* Content length has been modified */
{
- free_http_request(http);
- }
+ char * s2 = (char *) zalloc(100);
+ sprintf(s2, "Content-Length: %d", (int) csp->content_length);
- if (http->path == NULL)
+ log_error(LOG_LEVEL_HEADER, "Adjust Content-Length to %d", (int) csp->content_length);
+ return(s2);
+ }
+ else
{
- http->path = strdup("");
+ return(strdup(s));
}
}
-/* here begins the family of parser functions that reformat header lines */
-
-
/*********************************************************************
*
- * Function : crumble
+ * Function : server_content_md5
*
- * Description : This is called if a header matches a pattern to "crunch"
+ * Description : Crumble any Content-MD5 headers if the document was
+ * modified. FIXME: Should we re-compute instead?
*
* Parameters :
- * 1 : v = Pointer to parsers structure, which basically holds
- * headers (client or server) that we want to "crunch"
- * 2 : s = header (from sed) to "crunch"
+ * 1 : v = ignored
+ * 2 : s = header string we are "considering"
* 3 : csp = Current client state (buffers, headers, etc...)
*
- * Returns : Always NULL.
+ * Returns : A duplicate string pointer to this header (ie. pass thru)
*
*********************************************************************/
-char *crumble(const struct parsers *v, char *s, struct client_state *csp)
+char *server_content_md5(const struct parsers *v, const char *s, struct client_state *csp)
{
- log_error(LOG_LEVEL_HEADER, "crunch!");
- return(NULL);
+ if (csp->flags & CSP_FLAG_MODIFIED)
+ {
+ log_error(LOG_LEVEL_HEADER, "Crunching Content-MD5");
+ return(NULL);
+ }
+ else
+ {
+ return(strdup(s));
+ }
}
/*********************************************************************
*
- * Function : content_type
+ * Function : client_accept_encoding
*
- * Description : Is this a text/.* or javascript MIME Type?
+ * Description : Rewrite the client's Accept-Encoding header so that
+ * if doesn't allow compression, if the action applies.
+ * Note: For HTTP/1.0 the absence of the header is enough.
*
* 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)
+ * Returns : A copy of the client's original or the modified header.
*
*********************************************************************/
-char *content_type(const struct parsers *v, char *s, struct client_state *csp)
+char *client_accept_encoding(const struct parsers *v, const char *s, struct client_state *csp)
{
- if (strstr(s, " text/") || strstr(s, "application/x-javascript"))
- csp->content_type = CT_TEXT;
- else if (strstr(s, " image/gif"))
- csp->content_type = CT_GIF;
+ if ((csp->action->flags & ACTION_NO_COMPRESSION) == 0)
+ {
+ return(strdup(s));
+ }
else
- csp->content_type = 0;
+ {
+ log_error(LOG_LEVEL_HEADER, "Supressed offer to compress content");
- return(strdup(s));
+ if (!strcmpic(csp->http->ver, "HTTP/1.1"))
+ {
+ return(strdup("Accept-Encoding: identity;q=1.0, *;q=0"));
+ }
+ else
+ {
+ return(NULL);
+ }
+ }
}
/*********************************************************************
*
- * Function : content_length
+ * Function : client_te
*
- * Description : Adjust Content-Length header if we modified
- * the body.
+ * Description : Rewrite the client's TE header so that
+ * if doesn't allow compression, if the action applies.
*
* 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)
+ * Returns : A copy of the client's original or the modified header.
*
*********************************************************************/
-char *content_length(const struct parsers *v, char *s, struct client_state *csp)
+char *client_te(const struct parsers *v, const char *s, struct client_state *csp)
{
- if (csp->content_length != 0) /* Content has been modified */
+ if ((csp->action->flags & ACTION_NO_COMPRESSION) == 0)
{
- s = (char *) zalloc(100);
- sprintf(s, "Content-Length: %d", csp->content_length);
-
- log_error(LOG_LEVEL_HEADER, "Adjust Content-Length to %d", csp->content_length);
- return(s);
+ return(strdup(s));
}
else
{
- return(strdup(s));
+ log_error(LOG_LEVEL_HEADER, "Supressed offer to compress transfer");
+ return(NULL);
}
}
-
/*********************************************************************
*
* Function : client_referrer
* or modified header
*
*********************************************************************/
-char *client_referrer(const struct parsers *v, char *s, struct client_state *csp)
+char *client_referrer(const struct parsers *v, const char *s, struct client_state *csp)
{
const char * newval;
+ char * s2;
#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?
*/
* to fool stupid checks for in-site links
*/
log_error(LOG_LEVEL_HEADER, "crunch+forge!");
- s = strsav(NULL, "Referer: ");
- s = strsav(s, "http://");
- s = strsav(s, csp->http->hostport);
- s = strsav(s, "/");
- return(s);
+ s2 = strsav(NULL, "Referer: ");
+ s2 = strsav(s2, "http://");
+ s2 = strsav(s2, csp->http->hostport);
+ s2 = strsav(s2, "/");
+ return(s2);
}
/*
/*
* We have a specific (fixed) referer we want to send.
*/
+ char * s3;
log_error(LOG_LEVEL_HEADER, "modified");
- s = strsav( NULL, "Referer: " );
- s = strsav( s, newval );
- return(s);
+ s3 = strsav( NULL, "Referer: " );
+ s3 = strsav( s3, newval );
+ return(s3);
}
/* Should never get here! */
* to fool stupid checks for in-site links
*/
log_error(LOG_LEVEL_HEADER, "crunch+forge!");
- s = strsav(NULL, "Referer: ");
- s = strsav(s, "http://");
- s = strsav(s, csp->http->hostport);
- s = strsav(s, "/");
- return(s);
+ s2 = strsav(NULL, "Referer: ");
+ s2 = strsav(s2, "http://");
+ s2 = strsav(s2, csp->http->hostport);
+ s2 = strsav(s2, "/");
+ return(s2);
}
* a malloc'ed string pointer to this header (ie. pass thru).
*
*********************************************************************/
-char *client_uagent(const struct parsers *v, char *s, struct client_state *csp)
+char *client_uagent(const struct parsers *v, const char *s, struct client_state *csp)
{
const char * newval;
-
- /* 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 */
+ char * s2;
if ((csp->action->flags & ACTION_HIDE_USER_AGENT) == 0)
{
log_error(LOG_LEVEL_HEADER, "modified");
- s = strsav( NULL, "User-Agent: " );
- s = strsav( s, newval );
- return(s);
+ s2 = strsav( NULL, "User-Agent: " );
+ s2 = strsav( s2, newval );
+ return(s2);
}
* Returns : NULL if crunched, or a malloc'ed string to original header
*
*********************************************************************/
-char *client_ua(const struct parsers *v, char *s, struct client_state *csp)
+char *client_ua(const struct parsers *v, const char *s, struct client_state *csp)
{
if ((csp->action->flags & ACTION_HIDE_USER_AGENT) == 0)
{
* modified/original header.
*
*********************************************************************/
-char *client_from(const struct parsers *v, char *s, struct client_state *csp)
+char *client_from(const struct parsers *v, const char *s, struct client_state *csp)
{
const char * newval;
+ char * s2;
if ((csp->action->flags & ACTION_HIDE_FROM) == 0)
{
log_error(LOG_LEVEL_HEADER, " modified");
- s = strsav( NULL, "From: " );
- s = strsav( s, newval );
- return(s);
+ s2 = strsav( NULL, "From: " );
+ s2 = strsav( s2, newval );
+ return(s2);
}
* Returns : Always NULL.
*
*********************************************************************/
-char *client_send_cookie(const struct parsers *v, char *s, struct client_state *csp)
+char *client_send_cookie(const struct parsers *v, const char *s, struct client_state *csp)
{
if ((csp->action->flags & ACTION_NO_COOKIE_READ) == 0)
{
* Returns : Always NULL.
*
*********************************************************************/
-char *client_x_forwarded(const struct parsers *v, char *s, struct client_state *csp)
+char *client_x_forwarded(const struct parsers *v, const char *s, struct client_state *csp)
{
if ((csp->action->flags & ACTION_HIDE_FORWARDED) == 0)
{
}
-#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, char *s, struct client_state *csp)
+void client_host_adder(struct client_state *csp)
{
-#ifdef FEATURE_IMAGE_DETECT_MSIE
- if (strstr (s, "image/gif"))
+ char *p = NULL,
+ *pos = NULL;
+
+ if ( !csp->http->hostport || !*(csp->http->hostport)) return;
+ p = strsav(p, "Host: ");
+ /*
+ ** remove 'user:pass@' from 'proto://user:pass@host'
+ */
+ if ( (pos = strchr( csp->http->hostport, '@')) != NULL )
{
- /* Client will accept HTML. If this seems counterintuitive,
- * blame Microsoft.
- */
- csp->accept_types |= ACCEPT_TYPE_MSIE_HTML;
+ p = strsav(p, pos+1);
}
else
{
- csp->accept_types |= ACCEPT_TYPE_MSIE_IMAGE;
+ p = strsav(p, csp->http->hostport);
}
-#endif /* def FEATURE_IMAGE_DETECT_MSIE */
-
- return(strdup(s));
+ log_error(LOG_LEVEL_HEADER, "addh: %s", p);
+ enlist(csp->headers, p);
+ freez(p);
}
-#endif /* defined(FEATURE_IMAGE_DETECT_MSIE) */
-
-
-
-/* the following functions add headers directly to the header list */
/*********************************************************************
*********************************************************************/
void client_cookie_adder(struct client_state *csp)
{
- struct list *lst;
+ struct list_entry *lst;
char *tmp = NULL;
char *e;
- for (lst = csp->cookie_list->next; lst ; lst = lst->next)
+ for (lst = csp->cookie_list->first; lst ; lst = lst->next)
{
if (tmp)
{
tmp = strsav(tmp, lst->str);
}
- for (lst = csp->action->multi[ACTION_MULTI_WAFER]->next; lst ; lst = lst->next)
+ for (lst = csp->action->multi[ACTION_MULTI_WAFER]->first; lst ; lst = lst->next)
{
if (tmp)
{
}
+/*********************************************************************
+ *
+ * Function : client_accept_encoding_adder
+ *
+ * Description : Add an Accept-Encoding header to the client's request
+ * that disables compression if the action applies, and
+ * the header is not already there. Called from `sed'.
+ * Note: For HTTP/1.0, the absence of the header is enough.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : N/A
+ *
+ *********************************************************************/
+void client_accept_encoding_adder(struct client_state *csp)
+{
+ if ( ((csp->action->flags & ACTION_NO_COMPRESSION) != 0)
+ && (!strcmpic(csp->http->ver, "HTTP/1.1")) )
+ {
+ enlist_unique(csp->headers, "Accept-Encoding: identity;q=1.0, *;q=0", 16);
+ }
+
+}
+
+
/*********************************************************************
*
* Function : client_xtra_adder
*********************************************************************/
void client_xtra_adder(struct client_state *csp)
{
- struct list *lst = csp->action->multi[ACTION_MULTI_ADD_HEADER];
+ struct list_entry *lst;
- for (lst = lst->next; lst ; lst = lst->next)
+ for (lst = csp->action->multi[ACTION_MULTI_ADD_HEADER]->first;
+ lst ; lst = lst->next)
{
log_error(LOG_LEVEL_HEADER, "addh: %s", lst->str);
enlist(csp->headers, lst->str);
log_error(LOG_LEVEL_HEADER, "addh: %s", p);
enlist(csp->headers, p);
+ freez(p);
}
*********************************************************************/
void connection_close_adder(struct client_state *csp)
{
- enlist(csp->headers, strdup("Connection: close"));
+ enlist(csp->headers, "Connection: close");
+}
+
+
+/*********************************************************************
+ *
+ * Function : server_http
+ *
+ * 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
+ * 2 : s = header that matched this pattern
+ * 3 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : Copy of changed or original answer.
+ *
+ *********************************************************************/
+char *server_http(const struct parsers *v, const char *s, struct client_state *csp)
+{
+ char *ret = strdup(s);
+
+ sscanf(ret, "HTTP/%*d.%*d %d", &(csp->http->status));
+ if (csp->http->status == 206)
+ {
+ csp->content_type = CT_TABOO;
+ }
+
+ if ((csp->action->flags & ACTION_DOWNGRADE) != 0)
+ {
+ ret[7] = '0';
+ log_error(LOG_LEVEL_HEADER, "Downgraded answer to HTTP/1.0");
+ }
+ return(ret);
}
* Returns : `crumble' or a newly malloc'ed string.
*
*********************************************************************/
-char *server_set_cookie(const struct parsers *v, char *s, struct client_state *csp)
+char *server_set_cookie(const struct parsers *v, const char *s, struct client_state *csp)
{
#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);
+#ifdef HAVE_LOCALTIME_R
+ tm_now = *localtime_r(&now, &tm_now);
+#else
+ tm_now = *localtime (&now);
+#endif
+ 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 */
{
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;
+
+ /* 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(strdup(s));
}
#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, char *s, struct client_state *csp)
-{
- char *cleanhost = strdup(s);
-
- if(csp->force)
- strclean(cleanhost, FORCE_PREFIX);
-
- return(cleanhost);
-}
-#endif /* def 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 :
p = pos + len;
do
{
- *(p - len) = *p;
+ *(p - len) = *p;
}
while (*p++ != '\0');