Mercurial > hg > foxridge-archiver

changeset 10:4417be0e2f07

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
adapted to dsmc's problem with mount points saves index.meta from overwriting by archived version
author casties
date Fri, 21 Jan 2005 15:33:26 +0100
parents 9c61f624d802
children b19df18aa19a
files MPIWGStor.pm unarchiver.pl
diffstat 2 files changed, 100 insertions(+), 25 deletions(-) [+]
line wrap: on
line diff
--- a/MPIWGStor.pm	Tue Nov 30 13:39:51 2004 +0100
+++ b/MPIWGStor.pm	Fri Jan 21 15:33:26 2005 +0100
@@ -13,14 +13,14 @@
 	    $file_perm);
 use Exporter;
 
-$VERSION = 0.3; #  ROC 24.9.2003
+$VERSION = 0.4; #  ROC 20.1.2005
 
 @ISA = qw(Exporter);
 
 @EXPORT = qw($debug %junk_files %index_files $index_file_perm $archived_index_file_perm 
 	     $file_perm
 	     &parseargs &logger &stime  &s2stime &ymd_date &split_file_path &sstrip 
-	     &valid_file_name &valid_dir_name &read_xml &write_xml);
+	     &valid_file_name &valid_dir_name &park_file &unpark_file &read_xml &write_xml);
 
 # debug level
 $debug = 0;
@@ -211,6 +211,44 @@
 
 
 #
+# $newfilename = park_file($filename)
+#
+# parks a file under a new name (*.bak)
+#
+sub park_file {
+    my ($filename) = @_;
+    my $newfn = "";
+    if (-f $filename) {
+	$newfn = "$filename.bak";
+	if (! rename $filename, $newfn) {
+	    logger("ABORT", "unable to rename file $filename!");
+	    exit 1;
+	}
+    }
+    return $newfn;
+}
+    
+#
+# $filename = unpark_file($filename)
+#
+# unparks a file
+#
+sub unpark_file {
+    my ($filename) = @_;
+    my $newfn = "$filename.bak";
+    if (-f $newfn) {
+	if (! rename $newfn, $filename) {
+	    logger("ABORT", "unable to rename file $newfn!");
+	    exit 1;
+	}
+	return $filename;
+    }
+    return "";
+}
+    
+
+
+#
 # ($document, $rootnode) = read_xml($file)
 #
 # reads xml meta file $file
--- a/unarchiver.pl	Tue Nov 30 13:39:51 2004 +0100
+++ b/unarchiver.pl	Fri Jan 21 15:33:26 2005 +0100
@@ -18,7 +18,7 @@
 #
 
 # program version
-my $version = "0.2 (23.9.2004)";
+my $version = "0.3 (20.1.2004)";
 
 # read command line parameters
 my $args = parseargs;
@@ -79,10 +79,6 @@
     exit 1;
 }
 
-# construct document's parent dir
-my $docparent = $docdir;
-$docparent =~ s!/[^/]+$!!;
-
 #######################################################
 # internal variables
 #
@@ -96,6 +92,14 @@
 # subroutines
 #
 
+# construct document's parent dir
+sub get_parent {
+    my ($dirname) = @_;
+    my $dirparent = $dirname;
+    $dirparent =~ s!/[^/]+$!!;
+    return $dirparent;
+}
+
 
 #
 # $files = read_resource_meta($rootnode)
@@ -172,23 +176,36 @@
 
 
 #
-# $%files = run_retrieve
+# $num_files = run_retrieve($docdir, $docmount, \%files)
 #
-# runs the retriever program on $docdir and returns a list of archived files
+# Runs the retriever program on $docdir and returns the number of unarchived files.
+# All filenames are put in %files. 
+# $docmount is the mount point of the doc partition in cases when the new mount point
+# is different.
 #
 # Sample output:
 # (old!) Retrieving          17,234 /mpiwg/archive/data/test/auto_titit_123/pageimg/essen-wind1.jpg [Done]
 # Retrieving      42,406,326 /mpiwg/archive/data/library/B980G582/raw/00015.tif --> /mpiwg/archive/data/library/B980G582/raw/00015.tif [Done]
 sub run_retrieve {
-    my %files;
+    my ($archdir, $archmount, $files) = @_;
+    my $archparent;
+    if ($archmount eq $archdir) {
+	# no explicit mount point
+	$archparent = get_parent($archdir);
+    } else {
+	# destination dir is mount point
+	$archparent = $archmount;
+    }
+
     print LOG "START unarchive $version on ", scalar localtime, "\n";
     my $archcmd = $archprog;
     $archcmd .= " retrieve -subdir=yes -replace=all";
-    $archcmd .= " -description='$archname'";
-    $archcmd .= " '$docdir/'"; # archive name
-    $archcmd .= " '$docparent/'"; # destination dir name
+    $archcmd .= " -description='$archname'"; # archive name
+    $archcmd .= " '$archmount/'"; # archive mount point
+    $archcmd .= " '$archparent/'"; # destination dir name
 
     my $archcnt = 0;
+    my $numfiles = 0;
     print LOG "CMD: $archcmd\n";
     if (open ARCH, "$archcmd 2>&1 |") {
 	while (<ARCH>) {
@@ -207,19 +224,20 @@
 		$size =~ s/,//g;
 		logger("DEBUG", "  RETRIEVE: file '$file'");
 		$archcnt++;
-		if ($files{$file}) {
+		if ($$files{$file}) {
 		    logger("WARNING", "file $file seems to be archived multiple times.");
 		    $warncnt++;
 		} 
-		$files{$file} = [$size];
+		$$files{$file} = [$size];
 	    }
 	}
-	logger("INFO", "$archcnt archives of " . (scalar keys %files) . " files.");
+	$numfiles =  (scalar keys %$files);
+	logger("INFO", "$archcnt archives of $numfiles files (in $archmount).");
     } else {
 	logger("ABORT", "unable to start archive command '$archcmd'!!");
 	exit 1;
     }
-    return \%files;
+    return $numfiles;
 }
 
 
@@ -286,9 +304,9 @@
 
 # use checkarchive first
 if (system("$checkprog $docdir >/dev/null") == 0) {
-    logger("INFO", "archive '$docdir' check OK");
+    logger("INFO", "archive \"$docdir\" check OK");
 } else {
-    logger("ABORT", "archive '$docdir' check failed!!");
+    logger("ABORT", "archive \"$docdir\" check failed!!");
     exit 1;
 }
 
@@ -309,18 +327,37 @@
 
 logger("INFO", "$num_archived_files files to retrieve.");
 
+# save current index.meta
+park_file($metafile); 
+
 # retrieve
-my $retrieved_files = run_retrieve;
+my %retrieved_files = ();
+my $archcnt = 0;
 
-my $num_arch_files = (scalar keys %$retrieved_files);
-if ($num_arch_files == 0) {
+if ($docdir =~ /\/mpiwg\/archive\/data\/(.*)/) {
+    # TSM needs two different paths because of historical mount points :-(
+    # try the new one first
+    $archcnt = run_retrieve($docdir, "/mpiwg/archive", \%retrieved_files);
+    if ($archcnt == 0) {
+	# and then the old one
+	$archcnt = run_retrieve($docdir, "/mpiwg/archive/data", \%retrieved_files);
+    }
+} else {
+    # otherwise we assume that it works
+    $archcnt += run_retrieve($docdir, $docdir, \%retrieved_files);
+}
+
+# restore current index.meta
+unpark_file($metafile);
+
+if ($archcnt == 0) {
     logger("ABORT", "no files retrieved!!");
     exit 1;
 }
-logger("INFO", "$num_arch_files files retrieved");
+logger("INFO", "$archcnt files retrieved");
 
 # check list of archived files
-check_files($archived_files, $retrieved_files);
+check_files($archived_files, \%retrieved_files);
 
 # rewrite index.meta file
 write_xml($document, $metafile);
@@ -328,7 +365,7 @@
 logger("INFO", "$warncnt warnings");
 logger("INFO", "$errcnt errors");
 if ($errcnt == 0) {
-    logger("DONE", "" . (scalar keys %$retrieved_files) . " archived files retrieved");
+    logger("DONE", "$archcnt archived files retrieved");
 } else {
     logger("ABORT", "there were $errcnt errors!!");
     exit 1;