--- loncom/metadata_database/parse_activity_log.pl	2005/09/19 20:30:37	1.17
+++ loncom/metadata_database/parse_activity_log.pl	2005/09/20 18:01:01	1.20
@@ -2,7 +2,7 @@
 #
 # The LearningOnline Network
 #
-# $Id: parse_activity_log.pl,v 1.17 2005/09/19 20:30:37 albertel Exp $
+# $Id: parse_activity_log.pl,v 1.20 2005/09/20 18:01:01 matthew Exp $
 #
 # Copyright Michigan State University Board of Trustees
 #
@@ -144,6 +144,7 @@ if ($log) {
 my $sourcefilename;   # activity log data
 my $newfilename;      # $sourcefilename will be renamed to this
 my $error_filename;   # Errors in parsing the activity log will be written here
+my $chunk_filename;   # where we save data we are not going to write to db
 if ($srcfile) {
     $sourcefilename = $srcfile;
 } else {
@@ -153,6 +154,8 @@ my $sql_filename = $sourcefilename;
 $sql_filename =~ s|[^/]*$|activity.log.sql|;
 my $gz_sql_filename = $sql_filename.'.gz';
 #
+$chunk_filename = $sourcefilename.".unprocessed_chunks";
+#
 my $xml_filename = $sourcefilename;
 my $gz_xml_filename = $xml_filename.'.gz';
 if (defined($xmlfile)) {
@@ -486,8 +489,18 @@ sub process_courselog {
         if (! defined($host)) { $host = 'unknown'; }
         my $prevchunk = 'none';
         foreach my $chunk (split(/\&\&\&/,$log)) {
+            if (length($chunk) > 20000) {
+                # avoid putting too much data into the database
+                # (usually an uploaded file or something similar)
+                if (! &savechunk(\$chunk,$timestamp,$host)) {
+                    close(IN);
+                    return undef;
+                }
+                next;
+            }
             my $warningflag = '';
 	    my ($time,$res,$uname,$udom,$action,@values)= split(/:/,$chunk);
+            #
             if (! defined($res) || $res =~ /^\s*$/) {
                 $res = '/adm/roles';
                 $action = 'LOGIN';
@@ -501,8 +514,12 @@ sub process_courselog {
             if ($action !~ /^(LOGIN|VIEW|POST|CSTORE|STORE)$/) {
                 $warningflag .= 'action';
                 print $error_fh 'full log entry:'.$log.$/;
-                print $error_fh 'error on chunk:'.$chunk.$/;
-                $logthis->('(action) Unable to parse '.$/.$chunk.$/.
+                print $error_fh 'error on chunk (saving)'.$/;
+                if (! &savechunk(\$chunk,$timestamp,$host)) {
+                    close(IN);
+                    return undef;
+                }
+                $logthis->('(action) Unable to parse chunk'.$/.
                          'got '.
                          'time = '.$time.$/.
                          'res  = '.$res.$/.
@@ -539,6 +556,20 @@ sub process_courselog {
     }
     close IN;
     return $linecount;
+    ##
+    ##
+    sub savechunk {
+        my ($chunkref,$timestamp,$host) = @_;
+        my $chunk = &escape(${$chunkref});
+        if (! open(CHUNKFILE,">>$chunk_filename") ||
+            ! print CHUNKFILE $timestamp.':'.$host.':'.$chunk.$/) {
+            # abort
+            close(CHUNKFILE);
+            return 0;
+        }
+        close(CHUNKFILE);
+        return 1;
+    }
 }