annotate archiver.pl @ 11:b19df18aa19a

updated command line option parsing and logging
author casties
date Wed, 23 Mar 2005 13:12:08 +0100
parents 1a377102b1ce
children fdf4ceb36db1
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
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
19 my $version = "0.6 (ROC 23.3.2005)";
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
20
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
21 # short help
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
22 my $shorthelp = "MPIWG archiver $version\nuse:\n archiver [options] docpath\noptions:\n -premigrate don't delete archived files\n";
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
23
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
24 # read command line parameters
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
25 my $args = MPIWGStor::parseargs;
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
26
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
27 # debug level
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
28 my $debug = (exists $$args{'debug'}) ? $$args{'debug'} : 0;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
29
30497c6a3eca Initial revision
casties
parents:
diff changeset
30 # rewrite XML file (necessary for archive date!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
31 my $fix_xml = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
32 my $xml_changed = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
33 # XML namespace (not really implemented!)
30497c6a3eca Initial revision
casties
parents:
diff changeset
34 my $namespace = "";
30497c6a3eca Initial revision
casties
parents:
diff changeset
35
30497c6a3eca Initial revision
casties
parents:
diff changeset
36 # archive name (archive-path element, usually == $docdir)
30497c6a3eca Initial revision
casties
parents:
diff changeset
37 my $archname;
30497c6a3eca Initial revision
casties
parents:
diff changeset
38 # archive storage date (now)
30497c6a3eca Initial revision
casties
parents:
diff changeset
39 my $archdate = stime(time);
30497c6a3eca Initial revision
casties
parents:
diff changeset
40
30497c6a3eca Initial revision
casties
parents:
diff changeset
41 # delete "junk" files before archiving
30497c6a3eca Initial revision
casties
parents:
diff changeset
42 my $delete_junk_files = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
43
30497c6a3eca Initial revision
casties
parents:
diff changeset
44 # delete data files after archiving
30497c6a3eca Initial revision
casties
parents:
diff changeset
45 my $delete_data_files = 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
46
30497c6a3eca Initial revision
casties
parents:
diff changeset
47
30497c6a3eca Initial revision
casties
parents:
diff changeset
48 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
49 # external programs
30497c6a3eca Initial revision
casties
parents:
diff changeset
50 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
51 my $archprog = "/opt/tivoli/tsm/client/ba/bin/dsmc";
30497c6a3eca Initial revision
casties
parents:
diff changeset
52 if (! -x $archprog) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
53 logger('ABORT', "TSM client program '$archprog' missing!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
54 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
55 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
56 my $checkprog = "/usr/local/mpiwg/archive/metacheck";
30497c6a3eca Initial revision
casties
parents:
diff changeset
57 if (! -x $checkprog) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
58 logger('ABORT', "meta data checking program '$checkprog' missing!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
59 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
60 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
61 # log file for archiver
30497c6a3eca Initial revision
casties
parents:
diff changeset
62 my $log_file = "/var/log/mpiwg-archiver.log";
30497c6a3eca Initial revision
casties
parents:
diff changeset
63 if (! open LOG, ">>$log_file") {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
64 logger('ABORT', "unable to write log file '$log_file'!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
65 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
66 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
67
30497c6a3eca Initial revision
casties
parents:
diff changeset
68 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
69 # check parameters that were passed to the program
30497c6a3eca Initial revision
casties
parents:
diff changeset
70 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
71 if ($#ARGV < 0) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
72 print $shorthelp;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
73 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
74 }
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
75 my $docdir = $$args{'path'};
1
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
76 # strip double slashes
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
77 $docdir =~ s/\/\//\//;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
78 # strip trailing slashes
1
9143a0c4c678 small fix for // in pathnames
casties
parents: 0
diff changeset
79 $docdir =~ s/\/+$//;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
80 if (! -d $docdir) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
81 logger('ABORT', "document directory \'$docdir\' doesn't exist!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
82 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
83 }
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
84 # don't delete archived files with "-premigrate"
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
85 if (exists $$args{'premigrate'}) {
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
86 $delete_data_files = not $$args{'premigrate'};
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
87 }
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
88 if ($delete_data_files) {
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
89 logger('INFO', "going to remove successfully archived files from disk");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
90 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
91
30497c6a3eca Initial revision
casties
parents:
diff changeset
92 my $metafile = "$docdir/index.meta";
30497c6a3eca Initial revision
casties
parents:
diff changeset
93 if (! -f $metafile) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
94 logger('ABORT', "metadata index file \'$metafile\' doesn't exist!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
95 exit 1;
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 # internal variables
30497c6a3eca Initial revision
casties
parents:
diff changeset
100 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
101
30497c6a3eca Initial revision
casties
parents:
diff changeset
102 # number of errors
30497c6a3eca Initial revision
casties
parents:
diff changeset
103 my $errcnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
104 # number of warnings
30497c6a3eca Initial revision
casties
parents:
diff changeset
105 my $warncnt = 0;
30497c6a3eca Initial revision
casties
parents:
diff changeset
106
30497c6a3eca Initial revision
casties
parents:
diff changeset
107 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
108 # subroutines
30497c6a3eca Initial revision
casties
parents:
diff changeset
109 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
110
30497c6a3eca Initial revision
casties
parents:
diff changeset
111 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
112 # $files = read_resource_meta($rootnode)
30497c6a3eca Initial revision
casties
parents:
diff changeset
113 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
114 # checks general resource meta information and reads the list of files
30497c6a3eca Initial revision
casties
parents:
diff changeset
115 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
116 sub read_resource_meta {
30497c6a3eca Initial revision
casties
parents:
diff changeset
117 my ($rootnode) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
118 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
119 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
120 # archive path
30497c6a3eca Initial revision
casties
parents:
diff changeset
121 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
122 # get archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
123 $archname = MPIWGStor::sstrip($rootnode->findvalue('child::archive-path'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
124 if (! $archname) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
125 logger('ABORT', "archive-name element missing!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
126 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
127 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
128
30497c6a3eca Initial revision
casties
parents:
diff changeset
129 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
130 # files
30497c6a3eca Initial revision
casties
parents:
diff changeset
131 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
132 my @filenodes = $rootnode->findnodes('child::file');
30497c6a3eca Initial revision
casties
parents:
diff changeset
133 foreach my $fn (@filenodes) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
134 my $name = MPIWGStor::sstrip($fn->findvalue('child::name'));
30497c6a3eca Initial revision
casties
parents:
diff changeset
135 my $path = MPIWGStor::sstrip($fn->findvalue('child::path'));
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
136 logger('DEBUG', "FILE ($path)$name\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
137 my $f = ($path) ? "$path/$name" : "$name";
30497c6a3eca Initial revision
casties
parents:
diff changeset
138 $files{$f} = $name;
30497c6a3eca Initial revision
casties
parents:
diff changeset
139 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
140
30497c6a3eca Initial revision
casties
parents:
diff changeset
141 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
142 # archive-storage-date
30497c6a3eca Initial revision
casties
parents:
diff changeset
143 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
144 my $stordatenode = ($rootnode->find('child::archive-storage-date'))->get_node(1);
30497c6a3eca Initial revision
casties
parents:
diff changeset
145 if ($stordatenode) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
146 logger('WARNING', "archive storage date exists! Resource already archived?\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
147 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
148 # delete old date
30497c6a3eca Initial revision
casties
parents:
diff changeset
149 $stordatenode->removeChildNodes;
30497c6a3eca Initial revision
casties
parents:
diff changeset
150 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
151 # create new storage date node
30497c6a3eca Initial revision
casties
parents:
diff changeset
152 $stordatenode = $rootnode->addNewChild($namespace, "archive-storage-date");
30497c6a3eca Initial revision
casties
parents:
diff changeset
153 # move after archive-path
30497c6a3eca Initial revision
casties
parents:
diff changeset
154 $rootnode->insertAfter($stordatenode, ($rootnode->find('child::archive-path'))->get_node(1));
30497c6a3eca Initial revision
casties
parents:
diff changeset
155 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
156 $stordatenode->appendTextNode($archdate);
30497c6a3eca Initial revision
casties
parents:
diff changeset
157 $xml_changed++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
158 return \%files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
159 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
160
30497c6a3eca Initial revision
casties
parents:
diff changeset
161
30497c6a3eca Initial revision
casties
parents:
diff changeset
162 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
163 # $%files = run_archive
30497c6a3eca Initial revision
casties
parents:
diff changeset
164 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
165 # runs the archiver program on $docdir and returns a list of archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
166 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
167 sub run_archive {
30497c6a3eca Initial revision
casties
parents:
diff changeset
168 my %files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
169 print LOG "START archiver $version $archdate\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
170 my $archcmd = $archprog;
30497c6a3eca Initial revision
casties
parents:
diff changeset
171 $archcmd .= " archive -archsymlinkasfile=no -subdir=yes";
30497c6a3eca Initial revision
casties
parents:
diff changeset
172 $archcmd .= " -description='$archname'";
30497c6a3eca Initial revision
casties
parents:
diff changeset
173 $archcmd .= " '$docdir/'";
30497c6a3eca Initial revision
casties
parents:
diff changeset
174
30497c6a3eca Initial revision
casties
parents:
diff changeset
175 print LOG "CMD: $archcmd\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
176 if (open ARCH, "$archcmd 2>&1 |") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
177 while (<ARCH>) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
178 chomp;
30497c6a3eca Initial revision
casties
parents:
diff changeset
179 print LOG "ARCH: $_\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
180 if (/Normal File-->\s+[\d,]+\s+(.*)\s+\[Sent\]/) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
181 print " ARCH: file '$1'\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
182 $files{$1} = "ok";
30497c6a3eca Initial revision
casties
parents:
diff changeset
183 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
184 if (/^Archive processing of .* finished without failure./) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
185 print " ARCH: OK\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
186 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
187 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
188 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
189 logger('ABORT', "unable to start archive command '$archcmd'!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
190 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
191 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
192
30497c6a3eca Initial revision
casties
parents:
diff changeset
193 return \%files;
30497c6a3eca Initial revision
casties
parents:
diff changeset
194 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
195
30497c6a3eca Initial revision
casties
parents:
diff changeset
196
30497c6a3eca Initial revision
casties
parents:
diff changeset
197 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
198 # check_files(\%files_to_archive, \%archived_files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
199 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
200 # compares the list of archived and to be archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
201 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
202 sub check_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
203 my ($to_archive, $archived) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
204
30497c6a3eca Initial revision
casties
parents:
diff changeset
205 my $nt = scalar keys %$to_archive;
30497c6a3eca Initial revision
casties
parents:
diff changeset
206 my $na = scalar keys %$archived;
30497c6a3eca Initial revision
casties
parents:
diff changeset
207
30497c6a3eca Initial revision
casties
parents:
diff changeset
208 foreach my $ft (sort keys %$to_archive) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
209 my $fp = "$docdir/$ft";
30497c6a3eca Initial revision
casties
parents:
diff changeset
210 #print " fp: $fp\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
211 if ($$archived{$fp}) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
212 logger('DEBUG', "$ft archived OK\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
213 $$archived{$fp} = "OK";
30497c6a3eca Initial revision
casties
parents:
diff changeset
214 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
215 logger('ERROR', "file '$ft' missing from archive!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
216 $errcnt++;
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 foreach my $fa (sort keys %$archived) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
221 if ($$archived{$fa} ne "OK") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
222 my ($fn, $fp) = MPIWGStor::split_file_path($fa);
30497c6a3eca Initial revision
casties
parents:
diff changeset
223 if ($MPIWGStor::index_files{$fn}) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
224 logger('DEBUG', "$fa ignored\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
225 $na--;
30497c6a3eca Initial revision
casties
parents:
diff changeset
226 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
227 logger('WARNING', "$fa archived but not in list!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
228 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
229 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
230 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
231 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
232
30497c6a3eca Initial revision
casties
parents:
diff changeset
233 if ($nt > $na) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
234 logger('WARNING', "less files were archived ($na vs. $nt)!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
235 } elsif ($na > $nt) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
236 logger('WARNING', "more files were archived ($na vs. $nt)!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
237 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
238
30497c6a3eca Initial revision
casties
parents:
diff changeset
239 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
240
30497c6a3eca Initial revision
casties
parents:
diff changeset
241
30497c6a3eca Initial revision
casties
parents:
diff changeset
242 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
243 # delete_files(\%files)
30497c6a3eca Initial revision
casties
parents:
diff changeset
244 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
245 # deletes the files from the list (of absolute files) and their directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
246 # if they are empty
30497c6a3eca Initial revision
casties
parents:
diff changeset
247 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
248 sub delete_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
249 my ($files) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
250 my %dirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
251
30497c6a3eca Initial revision
casties
parents:
diff changeset
252 foreach my $f (sort keys %$files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
253 my ($fn, $fp) = MPIWGStor::split_file_path($f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
254 # collect all unique directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
255 if ($fp && (! $dirs{$fp})) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
256 $dirs{$fp} = $fp;
30497c6a3eca Initial revision
casties
parents:
diff changeset
257 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
258 # don't delete index files
30497c6a3eca Initial revision
casties
parents:
diff changeset
259 next if ($MPIWGStor::index_files{$fn});
30497c6a3eca Initial revision
casties
parents:
diff changeset
260 # no file no delete
30497c6a3eca Initial revision
casties
parents:
diff changeset
261 next unless (-f $f);
30497c6a3eca Initial revision
casties
parents:
diff changeset
262 # delete files
30497c6a3eca Initial revision
casties
parents:
diff changeset
263 if (unlink $f) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
264 logger('INFO', "remove $f ($fn)\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
265 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
266 logger('ERROR', "unable to delete $f!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
267 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
268 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
269 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
270 # try to delete all empty directories
30497c6a3eca Initial revision
casties
parents:
diff changeset
271 my @dirkeys = sort keys %dirs;
30497c6a3eca Initial revision
casties
parents:
diff changeset
272 # starting at the end to get to the subdirectories first
30497c6a3eca Initial revision
casties
parents:
diff changeset
273 for (my $i = $#dirkeys; $i >= 0; $i--) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
274 my $d = $dirkeys[$i];
30497c6a3eca Initial revision
casties
parents:
diff changeset
275 # dont't remove document dir (shouldn't be empty anyway)
30497c6a3eca Initial revision
casties
parents:
diff changeset
276 next if ($d eq $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
277 if (-d $d) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
278 logger('INFO', "remove dir $d\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
279 rmdir $d;
30497c6a3eca Initial revision
casties
parents:
diff changeset
280 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
281 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
282 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
283
30497c6a3eca Initial revision
casties
parents:
diff changeset
284
30497c6a3eca Initial revision
casties
parents:
diff changeset
285 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
286 # delete_all_files(\%files, $dir)
30497c6a3eca Initial revision
casties
parents:
diff changeset
287 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
288 # deletes all files with names from the list %files
30497c6a3eca Initial revision
casties
parents:
diff changeset
289 # in the directory $dir and its subdirectories
30497c6a3eca Initial revision
casties
parents:
diff changeset
290 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
291 sub delete_all_files {
30497c6a3eca Initial revision
casties
parents:
diff changeset
292 my ($files, $dir) = @_;
30497c6a3eca Initial revision
casties
parents:
diff changeset
293
30497c6a3eca Initial revision
casties
parents:
diff changeset
294 if (! opendir DIR, $dir) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
295 logger('ERROR', "unable to read directory $dir!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
296 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
297 return;
30497c6a3eca Initial revision
casties
parents:
diff changeset
298 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
299 my @fl = readdir DIR;
30497c6a3eca Initial revision
casties
parents:
diff changeset
300 closedir DIR;
30497c6a3eca Initial revision
casties
parents:
diff changeset
301
30497c6a3eca Initial revision
casties
parents:
diff changeset
302 foreach my $f (@fl) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
303 next if ($f =~ /^\.{1,2}$/);
30497c6a3eca Initial revision
casties
parents:
diff changeset
304 if ($$files{$f}) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
305 # $f is in the file list
30497c6a3eca Initial revision
casties
parents:
diff changeset
306 if (-f "$dir/$f") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
307 # $f is a file
30497c6a3eca Initial revision
casties
parents:
diff changeset
308 if (unlink "$dir/$f") {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
309 logger('INFO', "removed $f\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
310 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
311 logger('ERROR', "unable to delete $f!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
312 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
313 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
314 } elsif (-d _) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
315 # $f is a directory (unlink won't work)
30497c6a3eca Initial revision
casties
parents:
diff changeset
316 if ((system 'rm', '-r', "$dir/$f") == 0) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
317 logger('INFO', "removed directory $f\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
318 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
319 logger('ERROR', "unable to delete directory $f!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
320 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
321 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
322 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
323 logger('ERROR', "funny object $dir/$f!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
324 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
325 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
326 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
327 # $f is not in the list
30497c6a3eca Initial revision
casties
parents:
diff changeset
328 if (-d "$dir/$f") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
329 # recurse into directories
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
330 logger('DEBUG', "enter $dir/$f\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
331 delete_all_files($files, "$dir/$f");
30497c6a3eca Initial revision
casties
parents:
diff changeset
332 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
333 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
334 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
335 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
336
30497c6a3eca Initial revision
casties
parents:
diff changeset
337
30497c6a3eca Initial revision
casties
parents:
diff changeset
338 #######################################################
30497c6a3eca Initial revision
casties
parents:
diff changeset
339 # main
30497c6a3eca Initial revision
casties
parents:
diff changeset
340 #
30497c6a3eca Initial revision
casties
parents:
diff changeset
341
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
342 logger('START', "archiver $version at $archdate\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
343
30497c6a3eca Initial revision
casties
parents:
diff changeset
344 # make shure the right user is running this program
30497c6a3eca Initial revision
casties
parents:
diff changeset
345 my $user = getlogin;
30497c6a3eca Initial revision
casties
parents:
diff changeset
346 #if (($user ne "archive")||($user ne "root")) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
347 # logger("ABORT", "you ($user) must be archive or root user to run this program!");
30497c6a3eca Initial revision
casties
parents:
diff changeset
348 # exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
349 #}
30497c6a3eca Initial revision
casties
parents:
diff changeset
350
30497c6a3eca Initial revision
casties
parents:
diff changeset
351 # use metacheck first
30497c6a3eca Initial revision
casties
parents:
diff changeset
352 if (system("$checkprog $docdir >/dev/null") == 0) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
353 logger('INFO', "resource '$docdir' check OK\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
354 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
355 logger('ABORT', "resource '$docdir' check failed!!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
356 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
357 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
358
30497c6a3eca Initial revision
casties
parents:
diff changeset
359 # read index.meta file
30497c6a3eca Initial revision
casties
parents:
diff changeset
360 my ($document, $rootnode) = MPIWGStor::read_xml($metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
361
30497c6a3eca Initial revision
casties
parents:
diff changeset
362 # check file and add archive date
30497c6a3eca Initial revision
casties
parents:
diff changeset
363 my $files_to_archive = read_resource_meta($rootnode);
30497c6a3eca Initial revision
casties
parents:
diff changeset
364
30497c6a3eca Initial revision
casties
parents:
diff changeset
365 print "INFO: ", scalar keys %$files_to_archive, " files to archive\n";
30497c6a3eca Initial revision
casties
parents:
diff changeset
366
30497c6a3eca Initial revision
casties
parents:
diff changeset
367 # check for .archived file
30497c6a3eca Initial revision
casties
parents:
diff changeset
368 if (-f "$docdir/.archived") {
30497c6a3eca Initial revision
casties
parents:
diff changeset
369 if (unlink "$docdir/.archived") {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
370 logger('WARNING', "existing .archived file has been removed! Resource already archived?\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
371 $warncnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
372 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
373 logger('ERROR', "unable to remove existing .archived file!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
374 $errcnt++;
30497c6a3eca Initial revision
casties
parents:
diff changeset
375 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
376 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
377
30497c6a3eca Initial revision
casties
parents:
diff changeset
378 # remove junk files
30497c6a3eca Initial revision
casties
parents:
diff changeset
379 if ($delete_junk_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
380 delete_all_files(\%MPIWGStor::junk_files, $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
381 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
382
30497c6a3eca Initial revision
casties
parents:
diff changeset
383 # write new index.meta
30497c6a3eca Initial revision
casties
parents:
diff changeset
384 if ($errcnt > 0) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
385 logger('ABORT', "there were errors!\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
386 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
387 } else {
30497c6a3eca Initial revision
casties
parents:
diff changeset
388 if ($fix_xml) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
389 MPIWGStor::write_xml($document, $metafile);
30497c6a3eca Initial revision
casties
parents:
diff changeset
390 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
391 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
392
30497c6a3eca Initial revision
casties
parents:
diff changeset
393 # start archiving
5
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
394 my $archived_files = run_archive();
1a377102b1ce small fix for number of hash keys
casties
parents: 1
diff changeset
395 my $num_archfiles = scalar keys %$archived_files;
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
396
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
397 logger('INFO', "$num_archfiles files archived\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
398
30497c6a3eca Initial revision
casties
parents:
diff changeset
399 # check list of archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
400 check_files($files_to_archive, $archived_files);
30497c6a3eca Initial revision
casties
parents:
diff changeset
401
30497c6a3eca Initial revision
casties
parents:
diff changeset
402 # delete files if all went OK
30497c6a3eca Initial revision
casties
parents:
diff changeset
403 if ($errcnt == 0) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
404 system("touch", "$docdir/.archived");
30497c6a3eca Initial revision
casties
parents:
diff changeset
405 # remove junk files (again)
30497c6a3eca Initial revision
casties
parents:
diff changeset
406 if ($delete_junk_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
407 delete_all_files(\%MPIWGStor::junk_files, $docdir);
30497c6a3eca Initial revision
casties
parents:
diff changeset
408 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
409 # remove archived files
30497c6a3eca Initial revision
casties
parents:
diff changeset
410 if ($delete_data_files) {
30497c6a3eca Initial revision
casties
parents:
diff changeset
411 delete_files($archived_files);
30497c6a3eca Initial revision
casties
parents:
diff changeset
412 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
413 }
30497c6a3eca Initial revision
casties
parents:
diff changeset
414
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
415 logger('INFO', "$warncnt warnings\n");
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
416 logger('INFO', "$errcnt errors\n");
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
417 if ($errcnt > 0) {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
418 logger('ABORT', "there were errors! ($num_archfiles files archived) at " . stime(time));
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
419 exit 1;
30497c6a3eca Initial revision
casties
parents:
diff changeset
420 } else {
11
b19df18aa19a updated command line option parsing and logging
casties
parents: 5
diff changeset
421 logger('DONE', "$num_archfiles files archived at " . stime(time));
0
30497c6a3eca Initial revision
casties
parents:
diff changeset
422 }