Changeset 20390

Timestamp:

2009-08-25T14:43:41+12:00 (15 years ago)

Author:

mdewsnip

Message:

Improved cfgread::read_cfg_line() so it doesn't take forever to parse collect.cfg files containing large multi-line values (e.g. format statements, collectionextra values etc.). It was very slow previously because it was doing a fairly complex regular expression on the full value each time it read a line for the multi-line value. Since the value was getting longer and longer with each line read, the regular expression was taking longer and longer.

By Michael Dewsnip, DL Consulting Ltd.

File:

• gsdl/trunk/perllib/cfgread.pm (modified) (1 diff)

gsdl/trunk/perllib/cfgread.pm

@@ -44 +44 @@
     if ($line =~ m/^([\"\'])/ || $line =~ m/[^\\]([\"\'])/) {
         my $quote=$1;
-        while ($line !~ m/$quote(.*?[^\\])?(\\\\)*$quote/) {
-        my $nextline = <$handle>;
-        if (defined($nextline)) {
+
+        # Improve speed substantially by not doing the regular expression on $line in the while loop
+        #   (since $line gets longer each iteration, the regular expression gets slower and slower)
+        # Instead we just check each new line to see if it finishes the quoted multi-line value
+        if ($line !~ m/$quote(.*?[^\\])?(\\\\)*$quote/)
+        {
+        my $nextline;
+        while (defined($nextline = <$handle>))
+        {
             $nextline =~ s/\r?\n//; # remove end-of-line
             $line .= " " . $nextline;
-        } else {
-            return undef; # parse error?
+
+            # Break out of the while loop if we've found the end of the multi-line value
+            last if ($nextline =~ m/^(.*?[^\\])?(\\\\)*$quote/);
         }
         }