Changeset 27695 for other-projects/nightly-tasks/diffcol/trunk/diffcol/diffcol.pl

Timestamp:

2013-06-21T22:54:56+12:00 (11 years ago)

Author:

ak19

Message:

Better diffing on Windows. If either the test or model collection was built on windows AND the other one was built on linux, there is now special handling for doc.xml and archiveinf-doc/src database files in order to normalise them to the linux situation for better results when diffing.

File:

• other-projects/nightly-tasks/diffcol/trunk/diffcol/diffcol.pl (modified) (5 diffs)

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

@@ -537 +537 @@
 }
 
+
+# this function is only called on DocXMLFiles. 
+# so far, only doc.xml files need special Windows processing (db files' OS-sensitivity are handled in gdbdiff.pm)
+# Returns true if the doc.xml contains windows style slashes in the gsdlsourcefilename meta field
+sub isDocXMLFileWindows
+{
+    my ($file_contents) = @_;
+    
+    #return ($file_contents =~ m/\\/) ? 1 : 0; # windows slashes detected. 
+    
+    # Is this a better test? look for gsdlsourcefilename, see if it contains windows slashes.
+    # what if $gsdlsourcefilename is not guaranteed to exist in all doc.xml files?
+    
+    # for doc.xml:
+    #     <Metadata name="gsdlsourcefilename">import/html_files/cleves.html</Metadata>
+    if($file_contents =~ m@<Metadata name="gsdlsourcefilename">([^>]*)</Metadata>@m) {
+        $gsdlsourcefilename = $1;
+        if($gsdlsourcefilename =~ m/\\/) { # windows slashes detected.
+            return 1;
+        }
+    }
+    
+    return 0;   
+}
+
 sub TestEach
 {
@@ -547 +572 @@
         my @aryInModel = &diffutil::files_in_dir($strModel);
         my @aryInTest = &diffutil::files_in_dir($strTest);
+        
+        # Files to be skipped because they get generated on one OS but not the other
+        # On windows, files of the form col.invf.state.\d\d\d\d get generated (e.g. Small-HTML.invf.state.1228) that aren't there on linux
+        my $skipfiles_re = qr/\.invf\.state\.\d+$/; # Create a regex of all files to be skipped, see http://perldoc.perl.org/perlop.html
+        @aryInModel = grep { $_ !~ m/$skipfiles_re/ } @aryInModel; # http://stackoverflow.com/questions/174292/what-is-the-best-way-to-delete-a-value-from-an-array-in-perl
+        @aryInTest = grep { $_ !~ m/$skipfiles_re/ } @aryInTest;
+        
+        # Now check all remaining files in the folder exist in both model and test collections
         my @aryTwoPointers = FolderTesting(\@aryInModel,\@aryInTest,$strModel,$strTest,$intLevel);
         my @aryCorrectFiles = @{$aryTwoPointers[1]};
@@ -557 +590 @@
                 my $strNewModel = &FileUtils::filenameConcatenate($strModel,$strEachFile);
                 my $strNewTest = &FileUtils::filenameConcatenate($strTest,$strEachFile);
-                # now additionally ignoring the earliestDatestamp file and the index/idx/*.idh binary file when diffing file
-                if(!($strEachFile eq "log" || $strEachFile eq "earliestDatestamp" || $strEachFile =~ m/\.cfg$/g || $strEachFile =~ m/\.((g|j|l|b)db|idh)$/g))
+                # now additionally ignoring the earliestDatestamp file and the index/idx/*.idh binary file when diffing file 
+                if(!($strEachFile eq "log" || $strEachFile eq "earliestDatestamp" || $strEachFile =~ m/\.cfg$/g || $strEachFile =~ m/\.((g|j|l|b)db|idh|i.*|wa|td|tsd|ti|t|tl|w)$/g)) #$strEachFile =~ m/\.invf\.state\.\d+/
                 {
                     push(@Errors,TestEach($strNewModel,$strNewTest,$intLevel));
@@ -578 +611 @@
                 my $strNewModel = &FileUtils::filenameConcatenate($strModel,$strEachFile);
                 my $strNewTest = &FileUtils::filenameConcatenate($strTest,$strEachFile);
-                if(!($strEachFile eq "log" || $strEachFile eq "earliestDatestamp" || $strEachFile =~ m/\.cfg$/g || $strEachFile =~ m/\.((g|j|l|b)db|idh)$/g))
+                if(!($strEachFile eq "log" || $strEachFile eq "earliestDatestamp" || $strEachFile =~ m/\.cfg$/g || $strEachFile =~ m/\.((g|j|l|b)db|idh|i.*|wa|td|tsd|ti|t|tl|w)$/g))
                 {
                     push(@Errors,TestEach($strNewModel,$strNewTest,$intLevel));
@@ -611 +644 @@
         $test_contents =~ s/$ignore_line_re//g;
         
+            # equalise/normalise the two doc.xml files for OS differences, if there are any
+            my $testIsWin = &isDocXMLFileWindows($test_contents);
+            my $modelIsWin = &isDocXMLFileWindows($model_contents);
+            
+            if($testIsWin != $modelIsWin) { # one of the 2 collections is built on windows, the other on linux, so need to make newlines constant
+            
+                my $win_contents = $testIsWin ? \$test_contents : \$model_contents;
+            
+                # make all windows slashes into unix slashes
+                $$win_contents =~ s@\\@\/@g;
+                # make windows \r newlines into constant \n newlines        
+                $$win_contents =~ s@\r\n@\n@mg; # #http://stackoverflow.com/questions/650743/in-perl-how-to-do-you-remove-m-from-a-file
+                
+                #FOR MAC: old macs use CR carriage return (see http://www.perlmonks.org/?node_id=745018), so replace with \n ?
+                # $$win_contents =~ s@\r@\n@mg;
+            }
+        
         $strResult = diff \$model_contents, \$test_contents, { STYLE => "OldStyle" };