X-Git-Url: http://www.privoxy.org/gitweb/?p=privoxy.git;a=blobdiff_plain;f=filters.c;h=1a1da409823f7a6ed0df34740e5f3da4222b71b4;hp=cff05c4a9bf24d3dc8aacc0531c4b31721aba4e9;hb=7537bf8433682b12ccf1099e1e30c978f4bd52a2;hpb=91f8d1cfa59b145296b7b5dc4ee07bdb32517b8b
diff --git a/filters.c b/filters.c
index cff05c4a..1a1da409 100644
--- a/filters.c
+++ b/filters.c
@@ -1,4 +1,4 @@
-const char filters_rcs[] = "$Id: filters.c,v 1.53 2002/03/26 22:29:54 swa Exp $";
+const char filters_rcs[] = "$Id: filters.c,v 1.63 2006/08/31 10:11:28 fabiankeil Exp $";
/*********************************************************************
*
* File : $Source: /cvsroot/ijbswa/current/filters.c,v $
@@ -9,9 +9,10 @@ const char filters_rcs[] = "$Id: filters.c,v 1.53 2002/03/26 22:29:54 swa Exp $"
* `block_url', `url_actions', `domain_split',
* `filter_popups', `forward_url', 'redirect_url',
* `ij_untrusted_url', `intercept_url', `pcrs_filter_respose',
- * 'ijb_send_banner', and `trust_url'
+ * `ijb_send_banner', `trust_url', `gif_deanimate_response',
+ * `jpeg_inspect_response'
*
- * Copyright : Written by and Copyright (C) 2001 the SourceForge
+ * Copyright : Written by and Copyright (C) 2001, 2004 the SourceForge
* Privoxy team. http://www.privoxy.org/
*
* Based on the Internet Junkbuster originally written
@@ -38,6 +39,76 @@ const char filters_rcs[] = "$Id: filters.c,v 1.53 2002/03/26 22:29:54 swa Exp $"
*
* Revisions :
* $Log: filters.c,v $
+ * Revision 1.63 2006/08/31 10:11:28 fabiankeil
+ * Don't free p which is still in use and will be later
+ * freed by free_map(). Don't claim the referrer is unknown
+ * when the client didn't set one.
+ *
+ * Revision 1.62 2006/08/14 00:27:47 david__schmidt
+ * Feature request 595948: Re-Filter logging in single line
+ *
+ * Revision 1.61 2006/08/03 02:46:41 david__schmidt
+ * Incorporate Fabian Keil's patch work:
http://www.fabiankeil.de/sourcecode/privoxy/
+ *
+ * Revision 1.60 2006/07/18 14:48:46 david__schmidt
+ * Reorganizing the repository: swapping out what was HEAD (the old 3.1 branch)
+ * with what was really the latest development (the v_3_0_branch branch)
+ *
+ * Revision 1.58.2.9 2006/01/29 23:10:56 david__schmidt
+ * Multiple filter file support
+ *
+ * Revision 1.58.2.8 2005/05/07 21:50:55 david__schmidt
+ * A few memory leaks plugged (mostly on error paths)
+ *
+ * Revision 1.58.2.7 2004/10/03 12:53:32 david__schmidt
+ * Add the ability to check jpeg images for invalid
+ * lengths of comment blocks. Defensive strategy
+ * against the exploit:
+ * Microsoft Security Bulletin MS04-028
+ * Buffer Overrun in JPEG Processing (GDI+) Could
+ * Allow Code Execution (833987)
+ * Enabled with +inspect-jpegs in actions files.
+ *
+ * Revision 1.58.2.6 2003/12/06 22:18:27 gliptak
+ * Correcting compile problem with FEATURE_IMAGE_BLOCKING
+ *
+ * Revision 1.58.2.5 2003/11/11 13:10:31 oes
+ * Fixed bug #839859: "See why" link URL now gets url-encoded.
+ *
+ * Revision 1.58.2.4 2003/02/28 12:52:45 oes
+ * Fixed a typo
+ *
+ * Revision 1.58.2.3 2002/09/25 14:51:51 oes
+ * Added basic support for OPTIONS and TRACE HTTP methods:
+ * New function direct_response which handles OPTIONS and
+ * TRACE requests whose Max-Forwards header field is zero.
+ *
+ * Revision 1.58.2.2 2002/08/01 17:18:28 oes
+ * Fixed BR 537651 / SR 579724 (MSIE image detect improper for IE/Mac)
+ *
+ * Revision 1.58.2.1 2002/07/26 15:18:53 oes
+ * - Bugfix: Executing a filters without jobs no longer results in
+ * turing off *all* filters.
+ * - Security fix: Malicious web servers can't cause a seg fault
+ * through bogus chunk sizes anymore
+ *
+ * Revision 1.58 2002/04/24 02:11:17 oes
+ * Jon's multiple AF patch: url_actions now evaluates rules
+ * from all AFs.
+ *
+ * Revision 1.57 2002/04/08 20:38:34 swa
+ * fixed JB spelling
+ *
+ * Revision 1.56 2002/04/05 15:51:24 oes
+ * - bugfix: error-pages now get correct request protocol
+ * - fix for invalid HTML in trust info
+ *
+ * Revision 1.55 2002/04/02 16:13:51 oes
+ * Fix: No "Go there anyway" for SSL
+ *
+ * Revision 1.54 2002/04/02 14:55:56 oes
+ * Bugfix: is_untrusted_url() now depends on FEATURE_TRUST, not FEATURE_COOKIE_JAR
+ *
* Revision 1.53 2002/03/26 22:29:54 swa
* we have a new homepage!
*
@@ -360,7 +431,7 @@ const char filters_rcs[] = "$Id: filters.c,v 1.53 2002/03/26 22:29:54 swa Exp $"
* a lot! ;-)
*
* Revision 1.3 2001/05/20 16:44:47 jongfoster
- * Removing last hardcoded JunkBusters.com URLs.
+ * Removing last hardcoded Junkbusters.com URLs.
*
* Revision 1.2 2001/05/20 01:21:20 jongfoster
* Version 2.9.4 checkin.
@@ -664,10 +735,8 @@ int match_portlist(const char *portlist, int port)
*********************************************************************/
struct http_response *block_url(struct client_state *csp)
{
-#ifdef FEATURE_IMAGE_BLOCKING
- char *p;
-#endif /* def FEATURE_IMAGE_BLOCKING */
struct http_response *rsp;
+ const char *new_content_type = NULL;
/*
* If it's not blocked, don't block it ;-)
@@ -676,7 +745,10 @@ struct http_response *block_url(struct client_state *csp)
{
return NULL;
}
-
+ if (csp->action->flags & ACTION_REDIRECT)
+ {
+ log_error(LOG_LEVEL_ERROR, "redirect{} overruled by block.");
+ }
/*
* Else, prepare a response
*/
@@ -693,14 +765,25 @@ struct http_response *block_url(struct client_state *csp)
if (((csp->action->flags & ACTION_IMAGE_BLOCKER) != 0)
&& is_imageurl(csp))
{
+ char *p;
/* determine HOW images should be blocked */
p = csp->action->string[ACTION_STRING_IMAGE_BLOCKER];
+ if(csp->action->flags & ACTION_HANDLE_AS_EMPTY_DOCUMENT)
+ {
+ log_error(LOG_LEVEL_ERROR, "handle-as-empty-document overruled by handle-as-image.");
+ }
#if 1 /* Two alternative strategies, use this one for now: */
/* 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->body = bindup(image_pattern_data, image_pattern_length);
if (rsp->body == NULL)
{
@@ -718,6 +801,12 @@ 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->body = bindup(image_blank_data, image_blank_length);
if (rsp->body == NULL)
{
@@ -773,6 +862,34 @@ struct http_response *block_url(struct client_state *csp)
return cgi_error_memory();
}
#endif /* Preceeding code is disabled for now */
+ }
+ else if(csp->action->flags & ACTION_HANDLE_AS_EMPTY_DOCUMENT)
+ {
+ /*
+ * Send empty document.
+ */
+ new_content_type = csp->action->string[ACTION_STRING_CONTENT_TYPE];
+
+ freez(rsp->body);
+ rsp->body = strdup(" ");
+ rsp->content_length = 1;
+
+ rsp->status = strdup("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);
+ if (enlist_unique_header(rsp->headers, "Content-Type", new_content_type))
+ {
+ free_http_response(rsp);
+ return cgi_error_memory();
+ }
+ }
+
}
else
#endif /* def FEATURE_IMAGE_BLOCKING */
@@ -783,6 +900,7 @@ struct http_response *block_url(struct client_state *csp)
{
jb_err err;
struct map * exports;
+ char *p;
/*
* Workaround for stupid Netscape bug which prevents
@@ -818,12 +936,16 @@ struct http_response *block_url(struct client_state *csp)
#ifdef FEATURE_FORCE_LOAD
err = map(exports, "force-prefix", 1, FORCE_PREFIX, 1);
-#else /* ifndef FEATURE_FORCE_LOAD */
- err = map_block_killer(exports, "force-support");
+ if (csp->http->ssl != 0)
#endif /* ndef FEATURE_FORCE_LOAD */
+ {
+ err = map_block_killer(exports, "force-support");
+ }
+ if (!err) err = map(exports, "protocol", 1, csp->http->ssl ? "https://" : "http://", 1);
if (!err) err = map(exports, "hostport", 1, html_encode(csp->http->hostport), 0);
if (!err) err = map(exports, "path", 1, html_encode(csp->http->path), 0);
+ if (!err) err = map(exports, "path-ue", 1, url_encode(csp->http->path), 0);
if (err)
{
@@ -851,7 +973,7 @@ struct http_response *block_url(struct client_state *csp)
* Function : trust_url FIXME: I should be called distrust_url
*
* Description : Calls is_untrusted_url to determine if the URL is trusted
- * and if not, returns a HTTP 304 response with a reject message.
+ * and if not, returns a HTTP 403 response with a reject message.
*
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
@@ -885,17 +1007,19 @@ struct http_response *trust_url(struct client_state *csp)
return cgi_error_memory();
}
+ rsp->status = strdup("403 Request blocked by Privoxy");
exports = default_exports(csp, NULL);
- if (exports == NULL)
+ if (exports == NULL || rsp->status == NULL)
{
free_http_response(rsp);
return cgi_error_memory();
}
/*
- * Export the host, port, and referrer information
+ * Export the protocol, host, port, and referrer information
*/
err = map(exports, "hostport", 1, csp->http->hostport, 1);
+ if (!err) err = map(exports, "protocol", 1, csp->http->ssl ? "https://" : "http://", 1);
if (!err) err = map(exports, "path", 1, csp->http->path, 1);
if (NULL != (p = get_header_value(csp->headers, "Referer:")))
@@ -904,7 +1028,7 @@ struct http_response *trust_url(struct client_state *csp)
}
else
{
- if (!err) err = map(exports, "referrer", 1, "unknown", 1);
+ if (!err) err = map(exports, "referrer", 1, "none set", 1);
}
if (err)
@@ -942,7 +1066,7 @@ struct http_response *trust_url(struct client_state *csp)
p = strdup("");
for (l = csp->config->trust_info->first; l ; l = l->next)
{
- sprintf(buf, " %s
\n",l->str, l->str);
+ sprintf(buf, " %s
\n",l->str, l->str);
string_append(&p, buf);
}
err = map(exports, "trust-info", 1, p, 0);
@@ -1008,18 +1132,55 @@ struct http_response *redirect_url(struct client_state *csp)
{
char *p, *q;
struct http_response *rsp;
+ char *redirect_mode = NULL;
+ int x, y;
- p = q = csp->http->path;
- log_error(LOG_LEVEL_REDIRECTS, "checking path for redirects: %s", p);
-
- /*
- * find the last URL encoded in the request
- */
- while ((p = strstr(p, "http://")) != NULL)
+ if ((csp->action->flags & ACTION_REDIRECT))
{
- q = p++;
+ q = csp->action->string[ACTION_STRING_REDIRECT];
}
+ else
+ {
+ redirect_mode = csp->action->string[ACTION_STRING_FAST_REDIRECTS];
+ if (0 == strcmpic(redirect_mode, "check-decoded-url"))
+ {
+ p = q = csp->http->path;
+ log_error(LOG_LEVEL_REDIRECTS, "Decoding path: %s if necessary.", p);
+ while (*p)
+ {
+ if (*p == '%') /* Escape sequence? */
+ {
+ /* Yes, translate from hexadecimal to decimal */
+ p++;
+ /* First byte */
+ x=((int)*p++)-48;
+ if (x>9) x-=7;
+ x<<=4;
+ /* Second byte */
+ y=((int)*p++)-48;
+ if (y>9)y-=7;
+ /* Merge */
+ *q++=(char)(x|y);
+ }
+ else
+ {
+ /* No, forward character. */
+ *q++=*p++;
+ }
+ }
+ *q='\0';
+ }
+ p = q = csp->http->path;
+ log_error(LOG_LEVEL_REDIRECTS, "Checking path for redirects: %s", p);
+ /*
+ * find the last URL encoded in the request
+ */
+ while ((p = strstr(p, "http://")) != NULL)
+ {
+ q = p++;
+ }
+ }
/*
* if there was any, generate and return a HTTP redirect
*/
@@ -1057,8 +1218,9 @@ struct http_response *redirect_url(struct client_state *csp)
*
* 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, the info from
- * the browser's accept header.
+ * or, #ifdef FEATURE_IMAGE_DETECT_MSIE, and the browser
+ * is MSIE and not on a Mac, tell from the browser's accept
+ * header.
*
* Parameters :
* 1 : csp = Current client state (buffers, headers, etc...)
@@ -1073,7 +1235,7 @@ int is_imageurl(struct client_state *csp)
char *tmp;
tmp = get_header_value(csp->headers, "User-Agent:");
- if (tmp && strstr(tmp, "MSIE"))
+ if (tmp && strstr(tmp, "MSIE") && !strstr(tmp, "Mac_"))
{
tmp = get_header_value(csp->headers, "Accept:");
if (tmp && strstr(tmp, "image/gif"))
@@ -1222,7 +1384,7 @@ int is_untrusted_url(struct client_state *csp)
*
* Function : pcrs_filter_response
*
- * Description : Ecexute all text substitutions from all applying
+ * Description : Execute all text substitutions from all applying
* +filter actions on the text buffer that's been accumulated
* in csp->iob->buf. If this changes the contents, set
* csp->content_length to the modified size and raise the
@@ -1238,7 +1400,7 @@ int is_untrusted_url(struct client_state *csp)
char *pcrs_filter_response(struct client_state *csp)
{
int hits=0;
- size_t size;
+ size_t size, prev_size;
char *old = csp->iob->cur, *new = NULL;
pcrs_job *job;
@@ -1247,6 +1409,8 @@ char *pcrs_filter_response(struct client_state *csp)
struct re_filterfile_spec *b;
struct list_entry *filtername;
+ int i, found_filters = 0;
+
/*
* Sanity first
*/
@@ -1256,10 +1420,26 @@ char *pcrs_filter_response(struct client_state *csp)
}
size = csp->iob->eod - csp->iob->cur;
- if ( ( NULL == (fl = csp->rlist) ) || ( NULL == fl->f) )
+ /*
+ * Need to check the set of re_filterfiles...
+ */
+ for (i = 0; i < MAX_AF_FILES; i++)
+ {
+ fl = csp->rlist[i];
+ if (NULL != fl)
+ {
+ if (NULL != fl->f)
+ {
+ found_filters = 1;
+ break;
+ }
+ }
+ }
+
+ if (0 == found_filters)
{
log_error(LOG_LEVEL_ERROR, "Unable to get current state of regexp filtering.");
- return(NULL);
+ return(NULL);
}
/*
@@ -1277,6 +1457,11 @@ char *pcrs_filter_response(struct client_state *csp)
csp->flags |= CSP_FLAG_MODIFIED;
}
+ for (i = 0; i < MAX_AF_FILES; i++)
+ {
+ fl = csp->rlist[i];
+ if ((NULL == fl) || (NULL == fl->f))
+ break;
/*
* For all applying +filter actions, look if a filter by that
* name exists and if yes, execute it's pcrs_joblist on the
@@ -1294,12 +1479,10 @@ char *pcrs_filter_response(struct client_state *csp)
if ( NULL == b->joblist )
{
log_error(LOG_LEVEL_RE_FILTER, "Filter %s has empty joblist. Nothing to do.", b->name);
- return(NULL);
+ continue;
}
- log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) with filter %s...",
- csp->http->hostport, csp->http->path, size, b->name);
-
+ prev_size = size;
/* Apply all jobs from the joblist */
for (job = b->joblist; NULL != job; job = job->next)
{
@@ -1308,11 +1491,14 @@ char *pcrs_filter_response(struct client_state *csp)
old=new;
}
- log_error(LOG_LEVEL_RE_FILTER, " ...produced %d hits (new size %d).", current_hits, size);
+ log_error(LOG_LEVEL_RE_FILTER, "re_filtering %s%s (size %d) with filter %s produced %d hits (new size %d).",
+ csp->http->hostport, csp->http->path, prev_size, b->name, current_hits, size);
+
hits += current_hits;
}
}
}
+ }
/*
* If there were no hits, destroy our copy and let
@@ -1407,6 +1593,80 @@ char *gif_deanimate_response(struct client_state *csp)
}
+/*********************************************************************
+ *
+ * Function : jpeg_inspect_response
+ *
+ * Description :
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : a pointer to the (newly allocated) modified buffer
+ * or NULL in case something went wrong.
+ *
+ *********************************************************************/
+char *jpeg_inspect_response(struct client_state *csp)
+{
+ struct binbuffer *in = NULL, *out = NULL;
+ char *p = NULL;
+ size_t size = csp->iob->eod - csp->iob->cur;
+
+ /*
+ * If the body has a "chunked" transfer-encoding,
+ * get rid of it first, adjusting size and iob->eod
+ */
+ if (csp->flags & CSP_FLAG_CHUNKED)
+ {
+ log_error(LOG_LEVEL_DEANIMATE, "Need to de-chunk first");
+ if (0 == (size = remove_chunked_transfer_coding(csp->iob->cur, size)))
+ {
+ return(NULL);
+ }
+ csp->iob->eod = csp->iob->cur + size;
+ csp->flags |= CSP_FLAG_MODIFIED;
+ }
+
+ if (NULL == (in = (struct binbuffer *)zalloc(sizeof *in )))
+ {
+ log_error(LOG_LEVEL_DEANIMATE, "failed! (jpeg no mem 1)");
+ return NULL;
+ }
+
+ if (NULL == (out = (struct binbuffer *)zalloc(sizeof *out)))
+ {
+ log_error(LOG_LEVEL_DEANIMATE, "failed! (jpeg no mem 2)");
+ return NULL;
+ }
+
+ in->buffer = csp->iob->cur;
+ in->size = size;
+
+ /*
+ * Calling jpeg_inspect has the side-effect of creating and
+ * modifying the image buffer of "out" directly.
+ */
+ if (jpeg_inspect(in, out))
+ {
+ log_error(LOG_LEVEL_DEANIMATE, "failed! (jpeg parsing)");
+ free(in);
+ buf_free(out);
+ return(NULL);
+
+ }
+ else
+ {
+ csp->content_length = out->offset;
+ csp->flags |= CSP_FLAG_MODIFIED;
+ p = out->buffer;
+ free(in);
+ free(out);
+ return(p);
+ }
+
+}
+
+
/*********************************************************************
*
* Function : remove_chunked_transfer_coding
@@ -1445,7 +1705,12 @@ int remove_chunked_transfer_coding(char *buffer, const size_t size)
log_error(LOG_LEVEL_ERROR, "Parse error while stripping \"chunked\" transfer coding");
return(0);
}
- newsize += chunksize;
+
+ if ((newsize += chunksize) >= size)
+ {
+ log_error(LOG_LEVEL_ERROR, "Chunksize exceeds buffer in \"chunked\" transfer coding");
+ return(0);
+ }
from_p += 2;
memmove(to_p, from_p, (size_t) chunksize);
@@ -1484,16 +1749,21 @@ void url_actions(struct http_request *http,
{
struct file_list *fl;
struct url_actions *b;
+ int i;
init_current_action(csp->action);
- if (((fl = csp->actions_list) == NULL) || ((b = fl->f) == NULL))
+ for (i = 0; i < MAX_AF_FILES; i++)
{
- return;
- }
+ if (((fl = csp->actions_list[i]) == NULL) || ((b = fl->f) == NULL))
+ {
+ return;
+ }
- apply_url_actions(csp->action, http, b);
+ apply_url_actions(csp->action, http, b);
+ }
+ return;
}
@@ -1568,6 +1838,63 @@ const struct forward_spec * forward_url(struct http_request *http,
}
+/*********************************************************************
+ *
+ * Function : direct_response
+ *
+ * Description : Check if Max-Forwards == 0 for an OPTIONS or TRACE
+ * request and if so, return a HTTP 501 to the client.
+ *
+ * FIXME: I have a stupid name and I should handle the
+ * requests properly. Still, what we do here is rfc-
+ * compliant, whereas ignoring or forwarding are not.
+ *
+ * Parameters :
+ * 1 : csp = Current client state (buffers, headers, etc...)
+ *
+ * Returns : http_response if , NULL if nonmatch or handler fail
+ *
+ *********************************************************************/
+struct http_response *direct_response(struct client_state *csp)
+{
+ struct http_response *rsp;
+ struct list_entry *p;
+
+ if ((0 == strcmpic(csp->http->gpc, "trace"))
+ || (0 == strcmpic(csp->http->gpc, "options")))
+ {
+ for (p = csp->headers->first; (p != NULL) ; p = p->next)
+ {
+ if (!strncmp("Max-Forwards:", p->str, 13)
+ && (*(p->str+13) != '\0') && (atoi(p->str+13) == 0))
+ {
+ /* FIXME: We could handle at least TRACE here,
+ but that would require a verbatim copy of
+ the request which we don't have anymore */
+
+ log_error(LOG_LEVEL_HEADER, "Found Max-Forwards:0 in OPTIONS or TRACE request -- Returning 501");
+
+ /* Get mem for response or fail*/
+ if (NULL == (rsp = alloc_http_response()))
+ {
+ return cgi_error_memory();
+ }
+
+ if (NULL == (rsp->status = strdup("501 Not Implemented")))
+ {
+ free_http_response(rsp);
+ return cgi_error_memory();
+ }
+
+ rsp->is_static = 1;
+ return(finish_http_response(rsp));
+ }
+ }
+ }
+ return NULL;
+}
+
+
/*
Local Variables:
tab-width: 3