Use PCRE syntax in host patterns when configured

with --enable-pcre-host-patterns.

diff --git a/acconfig.h b/acconfig.h
index 2274f84..81eb72a 100644 (file)
--- a/acconfig.h
+++ b/acconfig.h
@@ -37,6 +37,9 @@
  *
  * Revisions   :
  *    $Log: acconfig.h,v $
  *
  * Revisions   :
  *    $Log: acconfig.h,v $

+ *    Revision 1.33  2006/12/17 19:15:26  fabiankeil
+ *    Added ./configure switch for FEATURE_GRACEFUL_TERMINATION.
+ *

  *    Revision 1.32  2006/07/18 14:48:45  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 1.32  2006/07/18 14:48:45  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)


@@ -349,6 +352,11 @@
  */
 #undef FEATURE_GRACEFUL_TERMINATION
 
  */
 #undef FEATURE_GRACEFUL_TERMINATION
 

+/*
+ * Allow PCRE syntax in host patterns.
+ */
+#undef FEATURE_PCRE_HOST_PATTERNS
+

 /*
  * Use POSIX threads instead of native threads.
  */
 /*
  * Use POSIX threads instead of native threads.
  */

diff --git a/configure.in b/configure.in
index dbd2ec2..8a0e526 100644 (file)
--- a/configure.in
+++ b/configure.in
@@ -1,6 +1,6 @@
 dnl Process this file with autoconf to produce a configure script.
 dnl 
 dnl Process this file with autoconf to produce a configure script.
 dnl 

-dnl $Id: configure.in,v 1.111 2008/02/03 14:40:47 fabiankeil Exp $
+dnl $Id: configure.in,v 1.112 2008/03/27 18:27:22 fabiankeil Exp $

 dnl 
 dnl Written by and Copyright (C) 2001 - 2007 the SourceForge
 dnl Privoxy team. http://www.privoxy.org/
 dnl 
 dnl Written by and Copyright (C) 2001 - 2007 the SourceForge
 dnl Privoxy team. http://www.privoxy.org/


@@ -28,6 +28,9 @@ dnl or write to the Free Software Foundation, Inc., 59
 dnl Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 dnl 
 dnl $Log: configure.in,v $
 dnl Temple Place - Suite 330, Boston, MA  02111-1307, USA.
 dnl 
 dnl $Log: configure.in,v $

+dnl Revision 1.112  2008/03/27 18:27:22  fabiankeil
+dnl Remove kill-popups action.
+dnl

 dnl Revision 1.111  2008/02/03 14:40:47  fabiankeil
 dnl Remove unused OSX_DARWIN macro. Reported by Mark Miller in #1852529.
 dnl
 dnl Revision 1.111  2008/02/03 14:40:47  fabiankeil
 dnl Remove unused OSX_DARWIN macro. Reported by Mark Miller in #1852529.
 dnl


@@ -514,7 +517,7 @@ dnl =================================================================
 dnl AutoConf Initialization
 dnl =================================================================
 
 dnl AutoConf Initialization
 dnl =================================================================
 

-AC_REVISION($Revision: 1.111 $)
+AC_REVISION($Revision: 1.112 $)

 AC_INIT(jcc.c)
 
 if test ! -f config.h.in; then
 AC_INIT(jcc.c)
 
 if test ! -f config.h.in; then


@@ -1313,6 +1316,12 @@ AC_ARG_ENABLE(graceful-termination,
   AC_DEFINE(FEATURE_GRACEFUL_TERMINATION)
 fi])
 
   AC_DEFINE(FEATURE_GRACEFUL_TERMINATION)
 fi])
 

+AC_ARG_ENABLE(pcre-host-patterns,
+[  --enable-pcre-host-patterns     Allow PCRE syntax in host patterns.],
+[if test $enableval = yes; then
+  AC_DEFINE(FEATURE_PCRE_HOST_PATTERNS)
+fi])
+

 dnl pcre/pcrs is needed for CGI anyway, so
 dnl the choice is only between static and
 dnl dynamic:
 dnl pcre/pcrs is needed for CGI anyway, so
 dnl the choice is only between static and
 dnl dynamic:

diff --git a/project.h b/project.h
index 21c71d3..ae3836b 100644 (file)
--- a/project.h
+++ b/project.h
@@ -1,7 +1,7 @@
 #ifndef PROJECT_H_INCLUDED
 #define PROJECT_H_INCLUDED
 /** Version string. */
 #ifndef PROJECT_H_INCLUDED
 #define PROJECT_H_INCLUDED
 /** Version string. */

-#define PROJECT_H_VERSION "$Id: project.h,v 1.109 2008/03/28 15:13:41 fabiankeil Exp $"
+#define PROJECT_H_VERSION "$Id: project.h,v 1.110 2008/03/29 12:13:46 fabiankeil Exp $"

 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/project.h,v $
 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/project.h,v $


@@ -37,6 +37,9 @@
  *
  * Revisions   :
  *    $Log: project.h,v $
  *
  * Revisions   :
  *    $Log: project.h,v $

+ *    Revision 1.110  2008/03/29 12:13:46  fabiankeil
+ *    Remove send-wafer and send-vanilla-wafer actions.
+ *

  *    Revision 1.109  2008/03/28 15:13:41  fabiankeil
  *    Remove inspect-jpegs action.
  *
  *    Revision 1.109  2008/03/28 15:13:41  fabiankeil
  *    Remove inspect-jpegs action.
  *


@@ -930,10 +933,14 @@ struct url_spec
        Used for debugging or display only.  */
    char  *spec;
 
        Used for debugging or display only.  */
    char  *spec;
 

+#ifdef FEATURE_PCRE_HOST_PATTERNS
+   regex_t *host_regex;/**< Regex for host matching                          */
+#else

    char  *dbuffer;     /**< Buffer with '\0'-delimited domain name, or NULL to match all hosts. */
    char **dvec;        /**< List of pointers to the strings in dbuffer.       */
    int    dcount;      /**< How many parts to this domain? (length of dvec)   */
    int    unanchored;  /**< Bitmap - flags are ANCHOR_LEFT and ANCHOR_RIGHT.  */
    char  *dbuffer;     /**< Buffer with '\0'-delimited domain name, or NULL to match all hosts. */
    char **dvec;        /**< List of pointers to the strings in dbuffer.       */
    int    dcount;      /**< How many parts to this domain? (length of dvec)   */
    int    unanchored;  /**< Bitmap - flags are ANCHOR_LEFT and ANCHOR_RIGHT.  */

+#endif /* defined FEATURE_PCRE_HOST_PATTERNS */

 
    char  *port_list;   /**< List of acceptable ports, or NULL to match all ports */
 
 
    char  *port_list;   /**< List of acceptable ports, or NULL to match all ports */
 


@@ -945,7 +952,11 @@ struct url_spec
 /**
  * If you declare a static url_spec, this is the value to initialize it to zero.
  */
 /**
  * If you declare a static url_spec, this is the value to initialize it to zero.
  */

+#ifndef FEATURE_PCRE_HOST_PATTERNS

 #define URL_SPEC_INITIALIZER { NULL, NULL, NULL, 0, 0, NULL, NULL, NULL, NULL }
 #define URL_SPEC_INITIALIZER { NULL, NULL, NULL, 0, 0, NULL, NULL, NULL, NULL }

+#else
+#define URL_SPEC_INITIALIZER { NULL, NULL, NULL, NULL, NULL, NULL }
+#endif /* def FEATURE_PCRE_HOST_PATTERNS */

 
 /**
  * Constant for host part matching in URLs.  If set, indicates that the start of
 
 /**
  * Constant for host part matching in URLs.  If set, indicates that the start of

diff --git a/urlmatch.c b/urlmatch.c
index 735e8f8..54089a7 100644 (file)
--- a/urlmatch.c
+++ b/urlmatch.c
@@ -1,4 +1,4 @@
-const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.22 2008/03/30 15:02:32 fabiankeil Exp $";
+const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.23 2008/04/05 12:19:20 fabiankeil Exp $";

 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/urlmatch.c,v $
 /*********************************************************************
  *
  * File        :  $Source: /cvsroot/ijbswa/current/urlmatch.c,v $


@@ -6,7 +6,7 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.22 2008/03/30 15:02:32 fabianke
  * Purpose     :  Declares functions to match URLs against URL
  *                patterns.
  *
  * Purpose     :  Declares functions to match URLs against URL
  *                patterns.
  *

- * Copyright   :  Written by and Copyright (C) 2001-2003, 2006-2007 the SourceForge
+ * Copyright   :  Written by and Copyright (C) 2001-2003, 2006-2008 the SourceForge

  *                Privoxy team. http://www.privoxy.org/
  *
  *                Based on the Internet Junkbuster originally written
  *                Privoxy team. http://www.privoxy.org/
  *
  *                Based on the Internet Junkbuster originally written


@@ -33,6 +33,9 @@ const char urlmatch_rcs[] = "$Id: urlmatch.c,v 1.22 2008/03/30 15:02:32 fabianke
  *
  * Revisions   :
  *    $Log: urlmatch.c,v $
  *
  * Revisions   :
  *    $Log: urlmatch.c,v $

+ *    Revision 1.23  2008/04/05 12:19:20  fabiankeil
+ *    Factor compile_host_pattern() out of create_url_spec().
+ *

  *    Revision 1.22  2008/03/30 15:02:32  fabiankeil
  *    SZitify unknown_method().
  *
  *    Revision 1.22  2008/03/30 15:02:32  fabiankeil
  *    SZitify unknown_method().
  *


@@ -632,6 +635,62 @@ jb_err parse_http_request(const char *req,
 }
 
 
 }
 
 

+#ifdef FEATURE_PCRE_HOST_PATTERNS
+/*********************************************************************
+ *
+ * Function    :  compile_host_pattern
+ *
+ * Description :  Parses and compiles a PCRE host pattern..
+ *
+ * Parameters  :
+ *          1  :  url = Target url_spec to be filled in.
+ *          2  :  host_pattern = Host pattern to compile.
+ *
+ * Returns     :  JB_ERR_OK - Success
+ *                JB_ERR_MEMORY - Out of memory
+ *                JB_ERR_PARSE - Cannot parse regex
+ *
+ *********************************************************************/
+static jb_err compile_host_pattern(struct url_spec *url, const char *host_pattern)
+{
+   int errcode;
+   char rebuf[BUFFER_SIZE];
+
+   assert(host_pattern);
+   assert(strlen(host_pattern) < sizeof(rebuf) - 2);
+
+   url->host_regex = zalloc(sizeof(*url->host_regex));
+   if (NULL == url->host_regex)
+   {
+      free_url_spec(url);
+      return JB_ERR_MEMORY;
+   }
+
+   snprintf(rebuf, sizeof(rebuf), "%s$", host_pattern);
+
+   errcode = regcomp(url->host_regex, rebuf,
+      (REG_EXTENDED|REG_NOSUB|REG_ICASE));
+
+   if (errcode)
+   {
+      size_t errlen = regerror(errcode, url->host_regex, rebuf, sizeof(rebuf));
+      if (errlen > (sizeof(rebuf) - (size_t)1))
+      {
+         errlen = sizeof(rebuf) - (size_t)1;
+      }
+      rebuf[errlen] = '\0';
+      log_error(LOG_LEVEL_ERROR, "error compiling %s: %s", url->spec, rebuf);
+      free_url_spec(url);
+
+      return JB_ERR_PARSE;
+   }
+
+   return JB_ERR_OK;
+
+}
+
+#else
+

 /*********************************************************************
  *
  * Function    :  compile_host_pattern
 /*********************************************************************
  *
  * Function    :  compile_host_pattern


@@ -848,6 +907,7 @@ static int domain_match(const struct url_spec *pattern, const struct http_reques
    }
 
 }
    }
 
 }

+#endif /* def FEATURE_PCRE_HOST_PATTERNS */

 
 
 /*********************************************************************
 
 
 /*********************************************************************


@@ -1021,8 +1081,16 @@ void free_url_spec(struct url_spec *url)
    if (url == NULL) return;
 
    freez(url->spec);
    if (url == NULL) return;
 
    freez(url->spec);

+#ifdef FEATURE_PCRE_HOST_PATTERNS
+   if (url->host_regex)
+   {
+      regfree(url->host_regex);
+      freez(url->host_regex);
+   }
+#else

    freez(url->dbuffer);
    freez(url->dvec);
    freez(url->dbuffer);
    freez(url->dvec);

+#endif /* ndef FEATURE_PCRE_HOST_PATTERNS */

    freez(url->path);
    freez(url->port_list);
    if (url->preg)
    freez(url->path);
    freez(url->port_list);
    if (url->preg)


@@ -1056,7 +1124,11 @@ int url_match(const struct url_spec *pattern,
 {
    /* XXX: these should probably be functions. */
 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, url->port))
 {
    /* XXX: these should probably be functions. */
 #define PORT_MATCHES ((NULL == pattern->port_list) || match_portlist(pattern->port_list, url->port))

+#ifdef FEATURE_PCRE_HOST_PATTERNS
+#define DOMAIN_MATCHES ((NULL == pattern->host_regex) || (0 == regexec(pattern->host_regex, url->host, 0, NULL, 0)))
+#else

 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, url)))
 #define DOMAIN_MATCHES ((NULL == pattern->dbuffer) || (0 == domain_match(pattern, url)))

+#endif

 #define PATH_MATCHES ((NULL == pattern->path) || (0 == regexec(pattern->preg, url->path, 0, NULL, 0)))
 
    if (pattern->tag_regex != NULL)
 #define PATH_MATCHES ((NULL == pattern->path) || (0 == regexec(pattern->preg, url->path, 0, NULL, 0)))
 
    if (pattern->tag_regex != NULL)