annotate archiver.pl @ 8:65895eec9e30

Added handling of modification date for HTTP feeds.
author casties
date Tue, 28 Sep 2004 14:20:26 +0200
parents 1a377102b1ce
children b19df18aa19a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
1 #!/usr/local/bin/perl -w
30497c6a3eca Initial revision
casties
parents:
diff changeset
2
30497c6a3eca Initial revision
casties
parents:
diff changeset
3 use strict;
30497c6a3eca Initial revision
casties
parents:
diff changeset
4
30497c6a3eca Initial revision
casties
parents:
diff changeset
5 use XML::LibXML;
30497c6a3eca Initial revision
casties
parents:
diff changeset
6
30497c6a3eca Initial revision
casties
parents:
diff changeset
7 # MPIWG libraries
30497c6a3eca Initial revision
casties
parents:
diff changeset
8 use lib '/usr/local/mpiwg/archive';
30497c6a3eca Initial revision
casties
parents:
diff changeset
9 use MPIWGStor;
30497c6a3eca Initial revision
casties
parents:
diff changeset
10
30497c6a3eca Initial revision
casties
parents:
diff changeset
11 # make output unbuffered
30497c6a3eca Initial revision
casties
parents:
diff changeset
12 $|=1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
13
30497c6a3eca Initial revision
casties
parents:
diff changeset
14 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
15 # internal parameters
30497c6a3eca Initial revision
casties
parents:
diff changeset
16 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
17
30497c6a3eca Initial revision
casties
parents:
diff changeset
18 # program version
5
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
19 my $version = "0.5.2 (12.7.2004)";
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
20
30497c6a3eca Initial revision
casties
parents:
diff changeset
21 # rewrite XML file (necessary for archive date!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
22 my $fix_xml = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
23 my $xml_changed = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
24 # XML namespace (not really implemented!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
25 my $namespace = "";
30497c6a3eca Initial revision
casties
parents:
diff changeset
26
30497c6a3eca Initial revision
casties
parents:
diff changeset
27 # archive name (archive-path element, usually == $docdir)
30497c6a3eca Initial revision
casties
parents:
diff changeset
28 my $archname;
30497c6a3eca Initial revision
casties
parents:
diff changeset
29 # archive storage date (now)
30497c6a3eca Initial revision
casties
parents:
diff changeset
30 my $archdate = stime(time);
30497c6a3eca Initial revision
casties
parents:
diff changeset
31
30497c6a3eca Initial revision
casties
parents:
diff changeset
32 # delete "junk" files before archiving
30497c6a3eca Initial revision
casties
parents:
diff changeset
33 my $delete_junk_files = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
34
30497c6a3eca Initial revision
casties
parents:
diff changeset
35 # delete data files after archiving
30497c6a3eca Initial revision
casties
parents:
diff changeset
36 my $delete_data_files = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
37
30497c6a3eca Initial revision
casties
parents:
diff changeset
38
30497c6a3eca Initial revision
casties
parents:
diff changeset
39 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
40 # external programs
30497c6a3eca Initial revision
casties
parents:
diff changeset
41 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
42 my $archprog = "/opt/tivoli/tsm/client/ba/bin/dsmc";
30497c6a3eca Initial revision
casties
parents:
diff changeset
43 if (! -x $archprog) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
44 print "ABORT: TSM client program '$archprog' missing!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
45 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
46 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
47 my $checkprog = "/usr/local/mpiwg/archive/metacheck";
30497c6a3eca Initial revision
casties
parents:
diff changeset
48 if (! -x $checkprog) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
49 print "ABORT: meta data checking program '$checkprog' missing!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
50 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
51 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
52 # log file for archiver
30497c6a3eca Initial revision
casties
parents:
diff changeset
53 my $log_file = "/var/log/mpiwg-archiver.log";
30497c6a3eca Initial revision
casties
parents:
diff changeset
54 if (! open LOG, ">>$log_file") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
55 print "ABORT: unable to write log file '$log_file'!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
56 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
57 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
58
30497c6a3eca Initial revision
casties
parents:
diff changeset
59 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
60 # check parameters that were passed to the program
30497c6a3eca Initial revision
casties
parents:
diff changeset
61 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
62 if ($#ARGV < 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
63 print "ABORT: no document directory given!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
64 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
65 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
66 my $docdir = $ARGV[0];
1
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
67 # strip double slashes
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
68 $docdir =~ s/\/\//\//;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
69 # strip trailing slashes
1
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
70 $docdir =~ s/\/+$//;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
71 if (! -d $docdir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
72 print "ABORT: document directory \'$docdir\' doesn't exist!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
73 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
74 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
75 if (($#ARGV > 0)&&($ARGV[1] eq "-premigrate")) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
76 $delete_data_files = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
77 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
78
30497c6a3eca Initial revision
casties
parents:
diff changeset
79 my $metafile = "$docdir/index.meta";
30497c6a3eca Initial revision
casties
parents:
diff changeset
80 if (! -f $metafile) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
81 print "ABORT: metadata index file \'$metafile\' doesn't exist!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
82 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
83 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
84
30497c6a3eca Initial revision
casties
parents:
diff changeset
85 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
86 # internal variables
30497c6a3eca Initial revision
casties
parents:
diff changeset
87 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
88
30497c6a3eca Initial revision
casties
parents:
diff changeset
89 # number of errors
30497c6a3eca Initial revision
casties
parents:
diff changeset
90 my $errcnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
91 # number of warnings
30497c6a3eca Initial revision
casties
parents:
diff changeset
92 my $warncnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
93
30497c6a3eca Initial revision
casties
parents:
diff changeset
94 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
95 # subroutines
30497c6a3eca Initial revision
casties
parents:
diff changeset
96 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
97
30497c6a3eca Initial revision
casties
parents:
diff changeset
98 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
99 # $files = read_resource_meta($rootnode)
30497c6a3eca Initial revision
casties
parents:
diff changeset
100 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
101 # checks general resource meta information and reads the list of files
30497c6a3eca Initial revision
casties
parents:
diff changeset
102 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
103 sub read_resource_meta {
30497c6a3eca Initial revision
casties
parents:
diff changeset
104 my ($rootnode) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
105 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
106 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
107 # archive path
30497c6a3eca Initial revision
casties
parents:
diff changeset
108 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
109 # get archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
110 $archname = MPIWGStor::sstrip($rootnode->findvalue('child::archive-path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
111 if (! $archname) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
112 print "ABORT: archive-name element missing!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
113 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
114 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
115
30497c6a3eca Initial revision
casties
parents:
diff changeset
116 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
117 # files
30497c6a3eca Initial revision
casties
parents:
diff changeset
118 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
119 my @filenodes = $rootnode->findnodes('child::file');
30497c6a3eca Initial revision
casties
parents:
diff changeset
120 foreach my $fn (@filenodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
121 my $name = MPIWGStor::sstrip($fn->findvalue('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
122 my $path = MPIWGStor::sstrip($fn->findvalue('child::path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
123 print "FILE: ($path)$name\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
124 my $f = ($path) ? "$path/$name" : "$name";
30497c6a3eca Initial revision
casties
parents:
diff changeset
125 $files{$f} = $name;
30497c6a3eca Initial revision
casties
parents:
diff changeset
126 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
127
30497c6a3eca Initial revision
casties
parents:
diff changeset
128 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
129 # archive-storage-date
30497c6a3eca Initial revision
casties
parents:
diff changeset
130 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
131 my $stordatenode = ($rootnode->find('child::archive-storage-date'))->get_node(1);
30497c6a3eca Initial revision
casties
parents:
diff changeset
132 if ($stordatenode) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
133 print "WARNING: archive storage date exists! Resource already archived?\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
134 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
135 # delete old date
30497c6a3eca Initial revision
casties
parents:
diff changeset
136 $stordatenode->removeChildNodes;
30497c6a3eca Initial revision
casties
parents:
diff changeset
137 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
138 # create new storage date node
30497c6a3eca Initial revision
casties
parents:
diff changeset
139 $stordatenode = $rootnode->addNewChild($namespace, "archive-storage-date");
30497c6a3eca Initial revision
casties
parents:
diff changeset
140 # move after archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
141 $rootnode->insertAfter($stordatenode, ($rootnode->find('child::archive-path'))->get_node(1));
30497c6a3eca Initial revision
casties
parents:
diff changeset
142 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
143 $stordatenode->appendTextNode($archdate);
30497c6a3eca Initial revision
casties
parents:
diff changeset
144 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
145 return \%files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
146 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
147
30497c6a3eca Initial revision
casties
parents:
diff changeset
148
30497c6a3eca Initial revision
casties
parents:
diff changeset
149 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
150 # $%files = run_archive
30497c6a3eca Initial revision
casties
parents:
diff changeset
151 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
152 # runs the archiver program on $docdir and returns a list of archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
153 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
154 sub run_archive {
30497c6a3eca Initial revision
casties
parents:
diff changeset
155 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
156 print LOG "START archiver $version $archdate\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
157 my $archcmd = $archprog;
30497c6a3eca Initial revision
casties
parents:
diff changeset
158 $archcmd .= " archive -archsymlinkasfile=no -subdir=yes";
30497c6a3eca Initial revision
casties
parents:
diff changeset
159 $archcmd .= " -description='$archname'";
30497c6a3eca Initial revision
casties
parents:
diff changeset
160 $archcmd .= " '$docdir/'";
30497c6a3eca Initial revision
casties
parents:
diff changeset
161
30497c6a3eca Initial revision
casties
parents:
diff changeset
162 print LOG "CMD: $archcmd\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
163 if (open ARCH, "$archcmd 2>&1 |") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
164 while (<ARCH>) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
165 chomp;
30497c6a3eca Initial revision
casties
parents:
diff changeset
166 print LOG "ARCH: $_\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
167 if (/Normal File-->\s+[\d,]+\s+(.*)\s+\[Sent\]/) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
168 print " ARCH: file '$1'\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
169 $files{$1} = "ok";
30497c6a3eca Initial revision
casties
parents:
diff changeset
170 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
171 if (/^Archive processing of .* finished without failure./) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
172 print " ARCH: OK\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
173 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
174 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
175 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
176 print "ABORT: unable to start archive command '$archcmd'!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
177 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
178 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
179
30497c6a3eca Initial revision
casties
parents:
diff changeset
180 return \%files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
181 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
182
30497c6a3eca Initial revision
casties
parents:
diff changeset
183
30497c6a3eca Initial revision
casties
parents:
diff changeset
184 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
185 # check_files(\%files_to_archive, \%archived_files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
186 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
187 # compares the list of archived and to be archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
188 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
189 sub check_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
190 my ($to_archive, $archived) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
191
30497c6a3eca Initial revision
casties
parents:
diff changeset
192 my $nt = scalar keys %$to_archive;
30497c6a3eca Initial revision
casties
parents:
diff changeset
193 my $na = scalar keys %$archived;
30497c6a3eca Initial revision
casties
parents:
diff changeset
194
30497c6a3eca Initial revision
casties
parents:
diff changeset
195 foreach my $ft (sort keys %$to_archive) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
196 my $fp = "$docdir/$ft";
30497c6a3eca Initial revision
casties
parents:
diff changeset
197 #print " fp: $fp\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
198 if ($$archived{$fp}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
199 print "DEBUG: $ft archived OK\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
200 $$archived{$fp} = "OK";
30497c6a3eca Initial revision
casties
parents:
diff changeset
201 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
202 print "ERROR: file '$ft' missing from archive!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
203 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
204 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
205 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
206
30497c6a3eca Initial revision
casties
parents:
diff changeset
207 foreach my $fa (sort keys %$archived) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
208 if ($$archived{$fa} ne "OK") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
209 my ($fn, $fp) = MPIWGStor::split_file_path($fa);
30497c6a3eca Initial revision
casties
parents:
diff changeset
210 if ($MPIWGStor::index_files{$fn}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
211 print "DEBUG: $fa ignored\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
212 $na--;
30497c6a3eca Initial revision
casties
parents:
diff changeset
213 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
214 print "WARNING: $fa archived but not in list!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
215 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
216 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
217 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
218 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
219
30497c6a3eca Initial revision
casties
parents:
diff changeset
220 if ($nt > $na) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
221 print "WARNING: less files were archived ($na vs. $nt)!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
222 } elsif ($na > $nt) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
223 print "WARNING: more files were archived ($na vs. $nt)!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
224 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
225
30497c6a3eca Initial revision
casties
parents:
diff changeset
226 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
227
30497c6a3eca Initial revision
casties
parents:
diff changeset
228
30497c6a3eca Initial revision
casties
parents:
diff changeset
229 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
230 # delete_files(\%files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
231 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
232 # deletes the files from the list (of absolute files) and their directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
233 # if they are empty
30497c6a3eca Initial revision
casties
parents:
diff changeset
234 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
235 sub delete_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
236 my ($files) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
237 my %dirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
238
30497c6a3eca Initial revision
casties
parents:
diff changeset
239 foreach my $f (sort keys %$files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
240 my ($fn, $fp) = MPIWGStor::split_file_path($f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
241 # collect all unique directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
242 if ($fp && (! $dirs{$fp})) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
243 $dirs{$fp} = $fp;
30497c6a3eca Initial revision
casties
parents:
diff changeset
244 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
245 # don't delete index files
30497c6a3eca Initial revision
casties
parents:
diff changeset
246 next if ($MPIWGStor::index_files{$fn});
30497c6a3eca Initial revision
casties
parents:
diff changeset
247 # no file no delete
30497c6a3eca Initial revision
casties
parents:
diff changeset
248 next unless (-f $f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
249 # delete files
30497c6a3eca Initial revision
casties
parents:
diff changeset
250 if (unlink $f) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
251 print "INFO: remove $f ($fn)\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
252 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
253 print "ERROR: unable to delete $f!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
254 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
255 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
256 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
257 # try to delete all empty directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
258 my @dirkeys = sort keys %dirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
259 # starting at the end to get to the subdirectories first
30497c6a3eca Initial revision
casties
parents:
diff changeset
260 for (my $i = $#dirkeys; $i >= 0; $i--) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
261 my $d = $dirkeys[$i];
30497c6a3eca Initial revision
casties
parents:
diff changeset
262 # dont't remove document dir (shouldn't be empty anyway)
30497c6a3eca Initial revision
casties
parents:
diff changeset
263 next if ($d eq $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
264 if (-d $d) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
265 print "INFO: remove dir $d\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
266 rmdir $d;
30497c6a3eca Initial revision
casties
parents:
diff changeset
267 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
268 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
269 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
270
30497c6a3eca Initial revision
casties
parents:
diff changeset
271
30497c6a3eca Initial revision
casties
parents:
diff changeset
272 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
273 # delete_all_files(\%files, $dir)
30497c6a3eca Initial revision
casties
parents:
diff changeset
274 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
275 # deletes all files with names from the list %files
30497c6a3eca Initial revision
casties
parents:
diff changeset
276 # in the directory $dir and its subdirectories
30497c6a3eca Initial revision
casties
parents:
diff changeset
277 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
278 sub delete_all_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
279 my ($files, $dir) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
280
30497c6a3eca Initial revision
casties
parents:
diff changeset
281 if (! opendir DIR, $dir) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
282 print "ERROR: unable to read directory $dir!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
283 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
284 return;
30497c6a3eca Initial revision
casties
parents:
diff changeset
285 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
286 my @fl = readdir DIR;
30497c6a3eca Initial revision
casties
parents:
diff changeset
287 closedir DIR;
30497c6a3eca Initial revision
casties
parents:
diff changeset
288
30497c6a3eca Initial revision
casties
parents:
diff changeset
289 foreach my $f (@fl) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
290 next if ($f =~ /^\.{1,2}$/);
30497c6a3eca Initial revision
casties
parents:
diff changeset
291 if ($$files{$f}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
292 # $f is in the file list
30497c6a3eca Initial revision
casties
parents:
diff changeset
293 if (-f "$dir/$f") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
294 # $f is a file
30497c6a3eca Initial revision
casties
parents:
diff changeset
295 if (unlink "$dir/$f") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
296 print "INFO: removed $f\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
297 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
298 print "ERROR: unable to delete $f!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
299 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
300 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
301 } elsif (-d _) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
302 # $f is a directory (unlink won't work)
30497c6a3eca Initial revision
casties
parents:
diff changeset
303 if ((system 'rm', '-r', "$dir/$f") == 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
304 print "INFO: removed directory $f\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
305 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
306 print "ERROR: unable to delete directory $f!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
307 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
308 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
309 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
310 print "ERROR: funny object $dir/$f!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
311 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
312 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
313 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
314 # $f is not in the list
30497c6a3eca Initial revision
casties
parents:
diff changeset
315 if (-d "$dir/$f") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
316 # recurse into directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
317 print "DEBUG: enter $dir/$f\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
318 delete_all_files($files, "$dir/$f");
30497c6a3eca Initial revision
casties
parents:
diff changeset
319 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
320 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
321 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
322 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
323
30497c6a3eca Initial revision
casties
parents:
diff changeset
324
30497c6a3eca Initial revision
casties
parents:
diff changeset
325 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
326 # main
30497c6a3eca Initial revision
casties
parents:
diff changeset
327 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
328
30497c6a3eca Initial revision
casties
parents:
diff changeset
329 print "START: archiver $version at $archdate\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
330
30497c6a3eca Initial revision
casties
parents:
diff changeset
331 # make shure the right user is running this program
30497c6a3eca Initial revision
casties
parents:
diff changeset
332 my $user = getlogin;
30497c6a3eca Initial revision
casties
parents:
diff changeset
333 #if (($user ne "archive")||($user ne "root")) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
334 # logger("ABORT", "you ($user) must be archive or root user to run this program!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
335 # exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
336 #}
30497c6a3eca Initial revision
casties
parents:
diff changeset
337
30497c6a3eca Initial revision
casties
parents:
diff changeset
338 # use metacheck first
30497c6a3eca Initial revision
casties
parents:
diff changeset
339 if (system("$checkprog $docdir >/dev/null") == 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
340 print "INFO: resource '$docdir' check OK\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
341 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
342 print "ABORT: resource '$docdir' check failed!!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
343 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
344 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
345
30497c6a3eca Initial revision
casties
parents:
diff changeset
346 # read index.meta file
30497c6a3eca Initial revision
casties
parents:
diff changeset
347 my ($document, $rootnode) = MPIWGStor::read_xml($metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
348
30497c6a3eca Initial revision
casties
parents:
diff changeset
349 # check file and add archive date
30497c6a3eca Initial revision
casties
parents:
diff changeset
350 my $files_to_archive = read_resource_meta($rootnode);
30497c6a3eca Initial revision
casties
parents:
diff changeset
351
30497c6a3eca Initial revision
casties
parents:
diff changeset
352 print "INFO: ", scalar keys %$files_to_archive, " files to archive\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
353
30497c6a3eca Initial revision
casties
parents:
diff changeset
354 # check for .archived file
30497c6a3eca Initial revision
casties
parents:
diff changeset
355 if (-f "$docdir/.archived") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
356 if (unlink "$docdir/.archived") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
357 print "WARNING: existing .archived file has been removed! Resource already archived?\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
358 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
359 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
360 print "ERROR: unable to remove existing .archived file!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
361 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
362 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
363 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
364
30497c6a3eca Initial revision
casties
parents:
diff changeset
365 # remove junk files
30497c6a3eca Initial revision
casties
parents:
diff changeset
366 if ($delete_junk_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
367 delete_all_files(\%MPIWGStor::junk_files, $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
368 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
369
30497c6a3eca Initial revision
casties
parents:
diff changeset
370 # write new index.meta
30497c6a3eca Initial revision
casties
parents:
diff changeset
371 if ($errcnt > 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
372 print "ABORT: there were errors!\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
373 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
374 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
375 if ($fix_xml) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
376 MPIWGStor::write_xml($document, $metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
377 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
378 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
379
30497c6a3eca Initial revision
casties
parents:
diff changeset
380 # start archiving
5
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
381 my $archived_files = run_archive();
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
382 my $num_archfiles = scalar keys %$archived_files;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
383
5
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
384 print "INFO: $num_archfiles files archived\n";
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
385
30497c6a3eca Initial revision
casties
parents:
diff changeset
386 # check list of archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
387 check_files($files_to_archive, $archived_files);
30497c6a3eca Initial revision
casties
parents:
diff changeset
388
30497c6a3eca Initial revision
casties
parents:
diff changeset
389 # delete files if all went OK
30497c6a3eca Initial revision
casties
parents:
diff changeset
390 if ($errcnt == 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
391 system("touch", "$docdir/.archived");
30497c6a3eca Initial revision
casties
parents:
diff changeset
392 # remove junk files (again)
30497c6a3eca Initial revision
casties
parents:
diff changeset
393 if ($delete_junk_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
394 delete_all_files(\%MPIWGStor::junk_files, $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
395 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
396 # remove archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
397 if ($delete_data_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
398 delete_files($archived_files);
30497c6a3eca Initial revision
casties
parents:
diff changeset
399 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
400 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
401
30497c6a3eca Initial revision
casties
parents:
diff changeset
402 print "INFO: $warncnt warnings\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
403 print "INFO: $errcnt errors\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
404 if ($errcnt > 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
405 print "ABORT: there were errors! ($num_archfiles files archived) at ", stime(time), "\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
406 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
407 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
408 print "DONE: $num_archfiles files archived at ", stime(time), "\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
409 }