From: eldy <>
Date: Mon, 21 Oct 2002 12:37:37 +0000 (+0000)
Subject: Added URLReferrerWithQuery parameter.
X-Git-Tag: AWSTATS_5_1_RELEASE~24
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=234dc3e24c9582ed02cbea8016eb84d32fc53567;p=thirdparty%2FAWStats.git

Added URLReferrerWithQuery parameter.
---

diff --git a/docs/awstats_changelog.txt b/docs/awstats_changelog.txt
index 3e37f6e2..5404eb64 100644
--- a/docs/awstats_changelog.txt
+++ b/docs/awstats_changelog.txt
@@ -17,8 +17,10 @@ AWStats Changelog
 - New parameter: Added URLNotCaseSensitive.
 - New parameter: Added URLWithQueryWithoutFollowingParameters to exclude some
   parameters from URL when URLWithQuery is on.
+- New parameter: Added URLReferrerWithquery.
 - Added tag %Dw-n for LogFile parameter (replaced with the day number in week
   but differs from %DW-n because start with 0).
+- Fixed: Log analyze is no more stopped if log file contains binary chars.
 - Fixed: -debug option is allowed in migrate.
 - Fixed: Wrong window was opened when clicking on flag link when
   UseFramesWhenCGI was on.
diff --git a/docs/awstats_config.html b/docs/awstats_config.html
index 945e3a1a..6d9e2af3 100644
--- a/docs/awstats_config.html
+++ b/docs/awstats_config.html
@@ -83,6 +83,7 @@ when reading it), follow the example:<br>
 <li><a href="#URLNotCaseSensitive">URLNotCaseSensitive</a>
 <li><a href="#URLWithQuery">URLWithQuery</a>
 <li><a href="#URLWithQueryWithoutFollowingParameters">URLWithQueryWithoutFollowingParameters</a>
+<li><a href="#URLReferrerWithQuery">URLReferrerWithQuery</a>
 <li><a href="#WarningMessages">WarningMessages</a>
 <li><a href="#NbOfLinesForCorruptedLog">NbOfLinesForCorruptedLog</a>
 <li><a href="#SplitSearchString">SplitSearchString</a>
@@ -762,7 +763,7 @@ when reading it), follow the example:<br>
 
 <a name="URLWithQuery"><b>URLWithQuery</b></a><br>
 <b>Version : </b>3.2 +<br>
-<br># Keep or attach the query string to the URL in the statistics for individual
+<br># Keep or remove the query string to the URL in the statistics for individual
 <br># pages. This is primarily used to differentiate between the URLs of dynamic
 <br># pages. If set to 1, mypage.html?id=x and mypage.html?id=y are counted as two
 <br># different pages.
@@ -798,6 +799,21 @@ when reading it), follow the example:<br>
 
 <br><br><hr>
 
+<a name="URLReferrerWithQuery"><b>URLReferrerWithQuery</b></a><br>
+<b>Version : </b>5.1 +<br>
+<br># Keep or remove the query string to the referrer URL in the statistics for
+<br># external referrer pages. This is used to differentiate between the URLs of
+<br># dynamic referrer pages. If set to 1, mypage.html?id=x and mypage.html?id=y
+<br># are counted as two different referrer pages.
+<br># Possible values:
+<br># 0 - Referrer URLs are cleaned from the query string (ie: "/mypage.html")
+<br># 1 - Full URL with query string is used      (ie: "/mypage.html?p=x&q=y")
+<br># Default: 0
+<br># 
+<br>URLReferrerWithQuery=0
+
+<br><br><hr>
+
 <a name="WarningMessages"><b>WarningMessages</b></a><br>
 <b>Version : </b>1.0 +<br>
 <br># AWStats can detect setup problems or show you important informations to have
diff --git a/wwwroot/cgi-bin/awstats.model.conf b/wwwroot/cgi-bin/awstats.model.conf
index f1d3c271..b1b17a7c 100644
--- a/wwwroot/cgi-bin/awstats.model.conf
+++ b/wwwroot/cgi-bin/awstats.model.conf
@@ -443,7 +443,7 @@ ValidHTTPCodes="200 304"
 URLNotCaseSensitive=0
 
 
-# Keep or attach the query string to the URL in the statistics for individual
+# Keep or remove the query string to the URL in the statistics for individual
 # pages. This is primarily used to differentiate between the URLs of dynamic
 # pages. If set to 1, mypage.html?id=x and mypage.html?id=y are counted as two
 # different pages.
@@ -475,6 +475,18 @@ URLWithQuery=0
 URLWithQueryWithoutFollowingParameters=""
 
 
+# Keep or remove the query string to the referrer URL in the statistics for
+# external referrer pages. This is used to differentiate between the URLs of
+# dynamic referrer pages. If set to 1, mypage.html?id=x and mypage.html?id=y
+# are counted as two different referrer pages.
+# Possible values:
+# 0 - Referrer URLs are cleaned from the query string (ie: "/mypage.html")
+# 1 - Full URL with query string is used      (ie: "/mypage.html?p=x&q=y")
+# Default: 0
+# 
+URLReferrerWithQuery=0
+
+
 # AWStats can detect setup problems or show you important informations to have
 # a better use. Keep this to 1, except if AWStats says you can change it.
 # Possible values: 0 or 1
diff --git a/wwwroot/cgi-bin/awstats.pl b/wwwroot/cgi-bin/awstats.pl
index c759f191..7db4c147 100644
--- a/wwwroot/cgi-bin/awstats.pl
+++ b/wwwroot/cgi-bin/awstats.pl
@@ -110,7 +110,8 @@ $NbOfLinesShowsteps $NewLinePhase $NbOfLinesForCorruptedLog $PurgeLogFile
 $ShowAuthenticatedUsers $ShowFileSizesStats
 $ShowDropped $ShowCorrupted $ShowUnknownOrigin $ShowLinksToWhoIs
 $ShowEMailSenders $ShowEMailReceivers
-$Expires $UpdateStats $MigrateStats $URLNotCaseSensitive $URLWithQuery $UseFramesWhenCGI $DecodeUA
+$Expires $UpdateStats $MigrateStats $URLNotCaseSensitive $URLWithQuery $URLReferrerWithQuery
+$UseFramesWhenCGI $DecodeUA
 /;
 ($EnableLockForUpdate, $DNSLookup, $AllowAccessFromWebToAuthenticatedUsersOnly,
 $BarHeight, $BarWidth, $CreateDirDataIfNotExists, $KeepBackupOfHistoricFiles, $MaxLengthOfURL,
@@ -123,8 +124,9 @@ $NbOfLinesShowsteps, $NewLinePhase, $NbOfLinesForCorruptedLog, $PurgeLogFile,
 $ShowAuthenticatedUsers, $ShowFileSizesStats,
 $ShowDropped, $ShowCorrupted, $ShowUnknownOrigin, $ShowLinksToWhoIs,
 $ShowEMailSenders, $ShowEMailReceivers,
-$Expires, $UpdateStats, $MigrateStats, $URLNotCaseSensitive, $URLWithQuery, $UseFramesWhenCGI, $DecodeUA)=
-(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
+$Expires, $UpdateStats, $MigrateStats, $URLNotCaseSensitive, $URLWithQuery, $URLReferrerWithQuery,
+$UseFramesWhenCGI, $DecodeUA)=
+(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0);
 use vars qw/
 $AllowToUpdateStatsFromBrowser $ArchiveLogRecords $DetailedReportsOnNewWindows
 $FirstDayOfWeek $KeyWordsNotSensitive $SaveDatabaseFilesWithPermissionsForEveryone
@@ -1111,16 +1113,17 @@ sub Parse_Config {
 			next;
 			}
 		if ($param =~ /^URLNotCaseSensitive$/)		{ $URLNotCaseSensitive=$value; next; }
-		if ($param =~ /^URLWithQuery$/)			{ $URLWithQuery=$value; next; }
+		if ($param =~ /^URLWithQuery$/)				{ $URLWithQuery=$value; next; }
 		if ($param =~ /^URLWithQueryWithoutFollowingParameters$/)	{
 			foreach my $elem (split(/\s+/,$value))	{ push @URLWithQueryWithoutFollowingParameters,$elem; }
 			next;
 			}
-		if ($param =~ /^WarningMessages/)       { $WarningMessages=$value; next; }
-		if ($param =~ /^NbOfLinesForCorruptedLog/) { $NbOfLinesForCorruptedLog=$value; next; }
-		if ($param =~ /^Expires/)               { $Expires=$value; next; }
-		if ($param =~ /^WrapperScript/)         { $WrapperScript=$value; next; }
-		if ($param =~ /^DecodeUA/)         		{ $DecodeUA=$value; next; }
+		if ($param =~ /^URLReferrerWithQuery$/)		{ $URLReferrerWithQuery=$value; next; }
+		if ($param =~ /^WarningMessages/)       	{ $WarningMessages=$value; next; }
+		if ($param =~ /^NbOfLinesForCorruptedLog/) 	{ $NbOfLinesForCorruptedLog=$value; next; }
+		if ($param =~ /^Expires/)               	{ $Expires=$value; next; }
+		if ($param =~ /^WrapperScript/)         	{ $WrapperScript=$value; next; }
+		if ($param =~ /^DecodeUA/)         			{ $DecodeUA=$value; next; }
 		# Read optional accuracy setup section
 		if ($param =~ /^LevelForRobotsDetection/)			{ $LevelForRobotsDetection=$value; next; }
 		if ($param =~ /^LevelForBrowsersDetection/)			{ $LevelForBrowsersDetection=$value; next; }
@@ -1480,6 +1483,7 @@ sub Check_Config {
 	if (! $DefaultFile[0])                          { $DefaultFile[0]="index.html"; }
 	if ($URLNotCaseSensitive !~ /[0-1]/)           	{ $URLNotCaseSensitive=0; }
 	if ($URLWithQuery !~ /[0-1]/)                 	{ $URLWithQuery=0; }
+	if ($URLReferrerWithQuery !~ /[0-1]/)          	{ $URLReferrerWithQuery=0; }
 	if ($WarningMessages !~ /[0-1]/)              	{ $WarningMessages=1; }
 	if ($NbOfLinesForCorruptedLog !~ /^\d+/ || $NbOfLinesForCorruptedLog<1)	{ $NbOfLinesForCorruptedLog=50; }
 	if ($Expires !~ /^\d+/)                 		{ $Expires=0; }
@@ -5330,7 +5334,7 @@ if ($UpdateStats && $FrameName ne "index" && $FrameName ne "mainleft") {	# Updat
 							$found=1;
 						}
 						else {
-							# This hit came from the search engine
+							# This hit came from a search engine
 							if ($PageBool) { $_from_p[2]++; }
 							$_from_h[2]++;
 							$_se_referrals_h{$TmpRefererServer{$refererserver}}++;
@@ -5376,8 +5380,14 @@ if ($UpdateStats && $FrameName ne "index" && $FrameName ne "mainleft") {	# Updat
 						if ($PageBool) { $_from_p[3]++; }
 						$_from_h[3]++;
 						# http://www.mysite.com/ must be same referer than http://www.mysite.com but .../mypage/ differs of .../mypage
-						#if ($refurl[0] =~ /^[^\/]+\/$/) { $field[$pos_referer] =~ s/\/$//; }	# Code moved in save
-						$_pagesrefs_h{$field[$pos_referer]}++;
+						#if ($refurl[0] =~ /^[^\/]+\/$/) { $field[$pos_referer] =~ s/\/$//; }	# Code moved in SaveHistory
+						if ($URLReferrerWithQuery) {
+							$_pagesrefs_h{$field[$pos_referer]}++;
+						}
+						else {
+							if ($field[$pos_referer]=~/^(.*)\?/) { $_pagesrefs_h{"$1"}++; }
+							else { $_pagesrefs_h{$field[$pos_referer]}++; }
+						}
 						$found=1;
 					}
 				}