Changeset 27767 for other-projects/nightly-tasks/diffcol

Timestamp:

2013-07-05T22:55:09+12:00 (11 years ago)

Author:

ak19

Message:

Fixes to previous commit: the random file names (created by PDFBox for its intermediate html files) are generated with random numbers. As a result, the HASH OIDs of these intermediate source files, although constant between model and test collections, can therefore appear in a different order in the model and the test collection when both are ordered alphabetically in the index/col.gdb file.

Location:

other-projects/nightly-tasks/diffcol/trunk/diffcol

Files:

• diffcol.pl (modified) (1 diff)
• gdbdiff.pm (modified) (2 diffs)

other-projects/nightly-tasks/diffcol/trunk/diffcol/diffcol.pl

@@ -679 +679 @@
         # The following block of code is necessary to deal with tmp (html) source files generated when using PDFBox
         # These tmpdirs are located inside the toplevel *greenstone* directory
-        (my $gsdlhome_re = $ENV{'GSDLHOME'}) =~ s@\\@\/@g;      
+        (my $gsdlhome_re = $ENV{'GSDLHOME'}) =~ s@\\@\/@g;
+        $gsdlhome_re = ".*" unless $$ENV{'GSDLHOME'};
         my $tmpfile_regex = "<Metadata name=\"URL\">http://$gsdlhome_re/tmp/([^\.]*)(\..{3,4})</Metadata>"; # $gsdlhome/tmp/randomfilename.html, file ext can be 3 or 4 chars long  
         if($test_contents =~ m@$tmpfile_regex@) {           

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

@@ -172 +172 @@
     # These tmpdirs are located inside the toplevel *greenstone* directory
     (my $gsdlhome_re = $ENV{'GSDLHOME'}) =~ s@\\@\/@g;      
+    $gsdlhome_re = ".*" unless $$ENV{'GSDLHOME'};
     my $tmpfile_regex = "<URL>http://$gsdlhome_re/tmp/([^\.]*)(\..{3,4})"; # $gsdlhome/tmp/randomfilename.html, file ext can be 3 or 4 chars long           
     if($test_text =~ m@$tmpfile_regex@g) {          
@@ -200 +201 @@
         
         # index/col.gdb also has entries for the random tmp file names in the form: [http://research/ak19/GS2bin_5July2013/tmp/F639.html]
-        # need to equalise these also
-        $test_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]@tmp/random$1@mg;
-        $model_text =~ s@\[http://.*/tmp/.*(\..{3,4})\]@tmp/random$1@mg;
+        # need to equalise these also. Sadly, when there are multiple intermediate files, their random tmp filenames are not 
+        # guaranteed to be generated in the same (alphabetical/numerical) order between model and test collection, so the 
+        # HASH OIDs, although all of them accounted for, appear in a different order. So we have to remove the Hash OIDs.
+        #$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;
     }