Changeset 28019 for other-projects

Timestamp:

2013-08-09T22:46:32+12:00 (11 years ago)

Author:

ak19

Message:

Better detection of whether a gdb file is a windows-generated one or not. The weakness of the earlier test was only discovered with the Multimedia tutorial collection.

Location:

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

Files:

• diffcol.pl (modified) (1 diff)
• gdbdiff.pm (modified) (3 diffs)

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

@@ -676 +676 @@
                 # Doing so is okay, since we're not modifying the doc.xml in the model or test collections, just normalising them in-memory for comparison
                 $$lin_contents =~ s@([^\\])\\([^\\])@$1\/$2@g;
+                
+                # Advanced Beatles collection, 
+                # linux version contains: IMG SRC=_httpextlink_&rl=1&href=http:///\\"http://www.boskowan.com/ (extra / slash)
+                # while windows contains: IMG SRC=_httpextlink_&rl=1&href=http://\\"http://www.boskowan.com/
+                # Normalising to windows version for doing a diff
+                $$lin_contents =~ s@href=http:///@href=http://@g;
             }
             

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

@@ -143 +143 @@
                 # E.g. On windows, the Word-PDF collection(s) contains double backslashes in the ex.File.Directory field
                 # the MARC-Exploded collection contains double backslashes in the null_file entry field of the .gdb file                
-                if($line =~ m@^<(ex.File.Directory|null_file)>(.*)@s) {
+                if($line =~ m@^<(ex.File.Directory|null_file)>(.*)@s) {                 
                     my ($fieldname, $escaped_path) = ($1, $2);
-                    $escaped_path =~ s@\\\\@\\@g; #(my $escaped_path = $2) =~ s@\\\\@\\@g;
+                    $escaped_path =~ s@\\\\@/@g; #(my $escaped_path = $2) =~ s@\\\\@\\@g;
                     $line = "<$fieldname>$escaped_path";
                 } 
@@ -160 +160 @@
                 $tmp .= $line;
             }
-            $$win_text = $tmp;          
-        }
-        
-        
-        # slashes in windows metadata text need to be turned into linux style slashes
-        $$win_text =~ s@\\@/@g; #$$win_text =~ s@\\([^n|r|\|"])@/$1@g; # filepath something\rtf remains something\rtf
+            $$win_text = $tmp;
+        
+            # slashes in windows metadata text need to be turned into linux style slashes. 
+            # index\col.gdb uses double backslashes, and single for \n,\t
+            #$$win_text =~ s@\\\\@/@g;
+        }
+        else { # archiveinf gdb file
+        
+            # slashes in windows metadata text need to be turned into linux style slashes. 
+            # In the two archivesinf gdb files, filepaths may use single backslashes
+            $$win_text =~ s@\\@/@g; #$$win_text =~ s@\\([^n|r|\|"])@/$1@g; # filepath something\rtf remains something\rtf
+        }
         
         # cut down absolute paths to files to just collect/colname/.../file, same as before
@@ -244 +250 @@
     my ($dbtailname, $db_contents) = @_; # db filename without suffix
     
-    #if($dbtailname !~ m/archiveinf/) { # only archiveinf-doc and archive-inf source need special Windows processing, not col.gdb
-    #   return 0;
-    #}
-    return ($db_contents =~ m/\\/) ? 1 : 0; # windows slashes detected. Better test would be: [Something\something] OR <tag>something\something
-    # for doc.xml:
-    #     <Metadata name="gsdlsourcefilename">import/html_files/cleves.html</Metadata>
+#   return ($db_contents =~ m/\\/) ? 1 : 0; # windows slashes detected. Better test would be: [Something\something] OR <tag>something\something 
+    
+    if($dbtailname =~ m/^archiveinf-doc/) {
+        return ($db_contents =~ m@<src-file>[a-zA-Z]:\\@) ? 1 : 0; # <src-file>C:\path
+    } 
+    elsif($dbtailname =~ m/^archiveinf-src/) { # <src-file>C:\path
+        return ($db_contents =~ m@\[[a-zA-Z]:\\@) ? 1 : 0; # [C:\path]
+    } 
+    else { # index/col.gdb file
+        if ($db_contents =~ m@<URL>http://[a-zA-Z]:/@) { # <URL>http://C:/path
+            return 1;
+        }       
+        elsif ($db_contents =~ m@^(<URL>http://[a-zA-Z]:/)|(<null_file>[^\\]*\\)@m) { # <URL>http://C:/path OR <null_file>CMSwp-all.00000001\\00000035.nul          
+            return 1;
+        }
+        return 0;       
+    }   
 }