Changeset 37194

Timestamp:

2023-01-28T19:01:43+13:00 (15 months ago)

Author:

davidb

Message:

Tested version of file-level document-version history (fldv-history) using the newer mv archives to archives_keepold and then hardlink-copy back to archiaves. Uses the newer technique of storing the timestamp in archives-timestamp.out, rather than rely on the timestamp of the file (which now changes due to the copy step

Location:

main/trunk/greenstone2/perllib

Files:

• DocHistoryFileUtils.pm (modified) (4 diffs)
• FileUtils.pm (modified) (3 diffs)
• arcinfo.pm (modified) (2 diffs)
• inexport.pm (modified) (4 diffs)
• plugins/DirectoryPlugin.pm (modified) (1 diff)

main/trunk/greenstone2/perllib/DocHistoryFileUtils.pm

@@ -379 +379 @@
 
 
+sub doc_copy_archivedir_keepold_to_archivedir
+{
+
+    my ($keepold_doc_dirname,$doc_dirname) = @_;
+    
+    my $status_ok = &FileUtils::hardlinkFilesRefRecursive([$keepold_doc_dirname],$doc_dirname, { 'strict' => 1 } );
+    
+    if (!$status_ok) {
+    print STDERR "Error: Failed to recursively hardlink copy the top-level 'keepold_archives' doc to 'archives'\n";
+    }   
+
+    if (!$status_ok) {
+    print STDERR "\n";
+    print STDERR "**** A critical error occurred in creating/updating file-level document-version history\n";
+    print STDERR "**** After determining and correcting the cause of the error, to reset, delete\n";
+    print STDERR "****     your 'archives' folder, and move 'archives_keep' back to 'archives'\n";
+    print STDERR "\n";
+    exit 1;
+    }        
+}
+
 
 sub archivedir_keepold_to_archivedir
@@ -385 +406 @@
 
     # Action Step 5 (from inexport.pm)
-    
-    # 5.1  a keepold doc's '_fldv_history' goes first
-    # 5.2  then the keepold doc's top-level content for new 'nminus 1'
+
+    # If needing to turn content in 'archives_keep' into a doc's 'archive' fldv_history/ directory, then
+    #   5.1  a keepold doc's '_fldv_history' goes first
+    #   5.2  then the keepold doc's top-level content for new 'nminus 1'
     
     my $perform_firsttime_init = 1;
@@ -446 +468 @@
         # **** Need additional work if -groupsize option > 1 is to be supported!
         # **** otherwise need to have earlier test to prevent -groupsize > 1 from been used with fldv-history
-        
+
+        my $keepold_doc_filename = &FileUtils::filenameConcatenate($archivedir_keepold, $keepold_doc_file);
+        my $keepold_doc_dirname  = &util::get_parent_folder($keepold_doc_filename);
+
+        my $doc_filename = &FileUtils::filenameConcatenate($archivedir, $doc_file);
+        my $doc_dirname  = &util::get_parent_folder($doc_filename);
+
         if ($index_status eq "D") {
         #
         # (1) if archive_info entry marked for deletion (D)
-        #           
+        #
+
+        # Even though delete is the ultimate outcome for this doc/dir, due to moving 'archives' to 'archives_keep' at the very start of import.pl
+        # we need to recursively hardlink copy it back, so the rest of the incremental import.pl process can work as expected
+        doc_copy_archivedir_keepold_to_archivedir($keepold_doc_dirname,$doc_dirname);
+        
         print STDERR "  The latest version of $keepold_OID is marked for deletion\n";
         print STDERR "  => nothing to do right now (will be deleted when buildcol.pl run)\n";           
         }
         else {
-        my $keepold_doc_filename = &FileUtils::filenameConcatenate($archivedir_keepold, $keepold_doc_file);
-        my $keepold_doc_dirname  = &util::get_parent_folder($keepold_doc_filename);
-
-        my $doc_filename = &FileUtils::filenameConcatenate($archivedir, $doc_file);
-        my $doc_dirname = &util::get_parent_folder($doc_filename);
-        
         if ($keepold) {
             #
@@ -482 +509 @@
             }
             elsif ($index_status eq "B") {
+            # This doc's archive doc folder hasn't change, but due to moving 'archives' to 'archives_keep' at the very start of import.pl
+            # we need to recursively hardlink copy it back
+            
+            doc_copy_archivedir_keepold_to_archivedir($keepold_doc_dirname,$doc_dirname);
+            
             if ($incremental_mode ne "all") {
                 print STDERR "  Unchanged version of document in 'archives/' (Index-Status=B)\n";

main/trunk/greenstone2/perllib/FileUtils.pm

@@ -33 +33 @@
 
 use FileHandle;
+use File::stat;
 
 # Greenstone modules
@@ -72 +73 @@
 
 #canRead
+#getTimestamp
 #isSymbolicLink
 #modificationTime
@@ -106 +108 @@
 
 
+## @function getTimestamp()
+#
+sub getTimestamp
+{
+    my ($filename) = @_;
+    
+    my $file_stat = stat($filename);
+    my $mtime = $file_stat->mtime;
+
+    return $mtime;
+}
+## getTimestamp()
+    
 ## @function closeFileHandle
 #

main/trunk/greenstone2/perllib/arcinfo.pm

@@ -123 +123 @@
 
     my ($cpackage,$cfilename,$cline,$csubr,$chas_args,$cwantarray) = caller(1);
-    print STDERR "\n\n!!!!!!!!!!!! load_info() -- Calling method: $cfilename:$cline $cpackage->$csubr\n\n";
     
     $self->{'info'} = {};
@@ -361 +360 @@
     my $self = shift (@_);
     my ($arcinfo_doc_filename) = @_;
-    print STDERR "\n\n!!!!!!!!!! SAVE save_arcinfo_doc_timestamp()\n\n";
     
     my $arcinfo_timestamp_filename = $self->get_timestamp_filename($arcinfo_doc_filename);

main/trunk/greenstone2/perllib/inexport.pm

@@ -715 +715 @@
     }
 
-    print STDERR "\n\n\n";
-    print STDERR "*****!!!!! No easy way in perl to perform file copy and perserve timestamps\n";
-    print STDERR "*****!!!!! SO => need to implement file with timestamp within (and DBinfo type for good measure)\n";
-    print STDERR "*****!!!!!       and change plugin/incremental building that depends on/uses -M \n";
-    # ArchiveInfoPlugin, DirectoryPlugin inexport.pm, arcinfo.pma
-    # DirectoryPlugin inexport.pm, arcinfo.pm (convertutil.pm OK, as working on the two files passed to it) 
-    print STDERR "\n\n\n";
-    }
-
-
+    if ($self->{'groupsize'} > 1) {
+        print STDERR "\n";
+        print STDERR "******\n";
+        print STDERR "Warning: Minus option '-groupsize' has not been tested with file-level document version history!\n";
+        print STDERR "   If the groups formed between subsequent invocations of import.pl stay the same, then\n";
+        print STDERR "   the formation of file-level document-version history 'nminus-<n> bundles' in _fldv_history directories\n";
+        print STDERR "   should remain correct\n";
+        print STDERR "******\n";
+        print STDERR "\n";
+    }
+    }
+
+    
     # Read the archive information file
     # coincidentally fldv-history: Action Step 4
@@ -1338 +1341 @@
     my $arcinfo_doc_filename = &dbutil::get_infodb_file_path($infodbtype, "archiveinf-doc", $archivedir);
 
-    #my $archiveinf_timestamp = -M $arcinfo_doc_filename;
     my ($unused_infodbtype,$archiveinf_timestamp) = $archive_info->load_timestamp($arcinfo_doc_filename);
-
+    
     # First convert all files to absolute form
     # This is to support the situation where the import folder is not
@@ -1387 +1389 @@
         if ($block_hash->{'metadata_files'}->{$full_curr_file}) {
         # is it modified??
-        if (-M $full_curr_file < $archiveinf_timestamp) {
+        my $full_curr_file_timestamp = &FileUtils::getTimestamp($full_curr_file);
+
+        if ($full_curr_file_timestamp > $archiveinf_timestamp) {
             print STDERR "*** Detected a *modified metadata* file: $full_curr_file\n" if $verbosity >= 2;
             # its newer than last build
@@ -1478 +1482 @@
 
     foreach my $existing_filename (@existing_files) {
-    if (-M $existing_filename < $archiveinf_timestamp) {
+    my $existing_filename_timestamp = &FileUtils::getTimestamp($existing_filename);
+    if ($existing_filename_timestamp > $archiveinf_timestamp) {
         # file is newer than last build
         

main/trunk/greenstone2/perllib/plugins/DirectoryPlugin.pm

@@ -141 +141 @@
         my $archive_info = new arcinfo($infodbtype); 
         my ($unused_infodbtype,$archiveinf_timestamp) = $archive_info->load_timestamp($arcinfo_doc_filename);
-        
-        # $self->{'inf_timestamp'} = -M $archives_inf;
         $self->{'inf_timestamp'} = $archiveinf_timestamp;
     }