Changeset 28067 for other-projects/nightly-tasks/diffcol

Timestamp:

2013-08-15T16:46:52+12:00 (11 years ago)

Author:

ak19

Message:

Fixing the numerous overly greedy regular expressions concerning timestamped tmp folders, after testing that the fixes still work with affected previous collections (PDFBox) and that they also behave correctly correctly with a new affected collection (Multimedia)

File:

• other-projects/nightly-tasks/diffcol/trunk/diffcol/gdbdiff.pm (modified) (4 diffs)

other-projects/nightly-tasks/diffcol/trunk/diffcol/gdbdiff.pm

@@ -186 +186 @@
     (my $gsdlhome_re = $ENV{'GSDLHOME'}) =~ s@\\@\/@g;      
     $gsdlhome_re = ".*" unless $$ENV{'GSDLHOME'};
-    my $tmpfile_regex = "<URL>http://$gsdlhome_re/tmp/([^\.]*)(\..{3,4})"; # $gsdlhome/tmp/randomfilename.html, file ext can be 3 or 4 chars long           
+    my $tmpfile_regex = "<URL>http://$gsdlhome_re/tmp/([^\.]*?)(\..{3,4})"; # $gsdlhome/tmp/randomfilename.html, file ext can be 3 or 4 chars long          
     if($test_text =~ m@$tmpfile_regex@g) {          
         # found a match, replace the tmp file name with "random", keeping the original file extension 
@@ -198 +198 @@
     
         
-        $tmpfile_regex = "(<(URL|UTF8URL|gsdlconvertedfilename|OrigSource)>(http://)?)($gsdlhome_re)?(/tmp/)?.*(\..{3,4})";
+        $tmpfile_regex = "(<(URL|UTF8URL|gsdlconvertedfilename|OrigSource)>(http://)?)($gsdlhome_re)?(/tmp/)?.*?(\..{3,4})";
         if($5) { 
             $test_text =~ s@$tmpfile_regex@$1$5$new_tmp_filename$6@mg;
@@ -206 +206 @@
         
         # modelcol used a different gsdlhome, but also a tmp dir, so make the same changes to its random filename           
-        $tmpfile_regex = "(<(URL|UTF8URL|gsdlconvertedfilename|OrigSource)>(http://)?)(.*)?(/tmp/)?.*(\..{3,4})";
+        $tmpfile_regex = "(<(URL|UTF8URL|gsdlconvertedfilename|OrigSource)>(http://)?)(.*)?(/tmp/)?.*?(\..{3,4})";
         if($5) { 
             $model_text =~ s@$tmpfile_regex@$1$5$new_tmp_filename$6@mg;
@@ -219 +219 @@
         #$test_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]@tmp/random$1@mg; # HASH OIDs can appear in different order
         #$model_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]@tmp/random$1@mg;
-        $test_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]\n<section>[^\n]*\n@tmp/random$1\n<section>RandomHash\n@sg;
-        $model_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]\n<section>[^\n]*\n@tmp/random$1\n<section>RandomHash\n@sg;
+        
+        $test_text  =~ s@\[http://[^\n]*?/tmp/.*?(\..{3,4})\]\n<section>([^\n]*?)\n@tmp/random$1\n<section>$2\n@sg;
+        $model_text =~ s@\[http://[^\n]*?/tmp/.*?(\..{3,4})\]\n<section>([^\n]*?)\n@tmp/random$1\n<section>$2\n@sg;
     }