diff --git a/perl/exit_strategy.pl b/perl/exit_strategy.pl
index f1b080f4c52c508ee6249d86ed806444d0462404..8e5aa819f039ae5f1240d0de4bee1ab836ffc638 100644
--- a/perl/exit_strategy.pl
+++ b/perl/exit_strategy.pl
@@ -30,6 +30,9 @@
#
# perl ./exit_strategy.pl --recover rosetta_exit_strategy/tmp.db /permanent/
#
+#
+# call help with:
+# perl ./exit_strategy.pl --help
###############################################################################
@@ -47,6 +50,7 @@ use XML::LibXML::XPathContext;
use DBD::SQLite;
use DBI;
use Getopt::Long;
+use Pod::Usage;
use constant DEBUG => 0; # no debug
use IO::Handle;
@@ -390,51 +394,6 @@ sub check_if_db_conform ($string, $filename) {
}
}
-###############################################################################
-# because ExLibris Rosetta produces filenames of following format:
-# V\d+-IE\d+\.xml
-# e.G.:
-# V1-IE23891.xml
-# V1-IE94621.xml
-# V2-IE23891.xml
-# …
-# we must find the relevant file with highest V-value, in example the file
-# "V2-IE23891.xml"
-#
-# this function gets an array reference with all possible files of given regEx
-# and returns an array reference with reduced files using only highest V-value
-# HINT, it only operates on sorted file list with fake versions (with zero-filled prefixes)
-################################################################################
-sub find_newest_iefile_version ($files_sorted, $files_truncated, $cnt_files) {
- my $cnt_truncated = 0;
- my $fh = $files_sorted->filehandle("<");
- my $last_entry;
- # FIXME, how many lines?
- my $p = Time::Progress->new(min=>0, max=> $cnt_files);
- my $i=0;
- while(<$fh>) {
- my $entry = $_;
- $entry =~ m/^(.+?V)(\d+)(-IE\d+\.xml)$/;
- if (!defined $last_entry) {
- $last_entry = $entry;
- }
- my ($prefix, $version, $suffix) = ($1, $2, $3);
- $last_entry =~ m/^(.+?V)(\d+)(-IE\d+\.xml)$/;
- my ($last_prefix, $last_version, $last_suffix) = ($1, $2, $3);
- if (($last_prefix eq $prefix ) && ($last_suffix eq $suffix) && ($last_version < $version)) {
- } else {
- $files_truncated->append($last_entry);
- $cnt_truncated++;
- }
- print $p->report("find newest IE files: %40b ETA: %E \r", $i++);
- $last_entry = $entry;
- }
- $files_truncated->append($last_entry);
- $cnt_truncated++;
- say "";
- return $cnt_truncated;
-}
-
sub searching_ie_files ($dir, $tmp_ies_unsorted_file) {
my $cnt_unsorted_files = 0;
my $first_two_levels_of_dirs = 0;
@@ -479,65 +438,85 @@ sub searching_ie_files ($dir, $tmp_ies_unsorted_file) {
return $cnt_unsorted_files;
}
-
-
###############################################################################
###############################################################################
############# main ############################################################
###############################################################################
###############################################################################
-my $tmp_ies_dir = Path::Tiny->tempdir( TEMPLATE => "exitstrategy_XXXXXXXXXXX", CLEANUP => 1);
+my $tmp_ies_dir = Path::Tiny->tempdir(TEMPLATE => "exitstrategy_XXXXXXXXXXX", CLEANUP => 1);
my $tmp_ies_unsorted_file = $tmp_ies_dir->child("unsorted_ies");
- my $flag_recovery = undef;
- my $flag_sqldump = undef;
- my $db_filename = $db_name.".db";
- my @ARGV_tail;
- GetOptions(
- "recovery" => \$flag_recovery,
- "sqlitedb-file=s" => \$db_filename,
- "enable_sqldump" => \$flag_sqldump,
- '<>' => sub {push @ARGV_tail, @_;}
- );
- if ($#ARGV_tail < 0) {
- die "you need a directory as argument\n";
- }
- if (defined $flag_recovery) { warn "recovery enabled for XML processing\n"; }
-
- my $dir = shift @ARGV_tail;
+my $flag_recovery = undef;
+my $flag_sqldump = undef;
+my $db_filename = $db_name . ".db";
+my @ARGV_tail;
+GetOptions(
+ "help|?" => sub {
+ say <<"HELP";
+call $0 with following options
+ --help | this help
+ --recovery | set special recovery flag
+ --sqlitedb-file=FILE | set database to file FILE
+ --enable-sqldump | dumps a given database as SQL to STDOUT
+
+ create an exit-DB:
+ $0 [--recovery] [--sqlitedb-file=FILE] DIR
+
+ dump an exit-DB as SQL
+ $0 [--sqlitedb-file=FILE] --enable-sqldump
+HELP
+ exit;
+ },
+ "recovery" => \$flag_recovery,
+ "sqlitedb-file=s" => \$db_filename,
+ "enable_sqldump" => \$flag_sqldump,
+ '<>' => sub {push @ARGV_tail, @_;}
+) or pod2usage(2);
+
+if (defined $flag_recovery) {warn "recovery enabled for XML processing\n";}
+if (defined $flag_sqldump) {
+ open my $cmd, '|-', '/usr/bin/sqlite3', $db_filename, '.dump' or die "Failed to dump DB: $!\n";
+ close $cmd or die "Failed to dumb DB: $!\n";
+ exit;
+}
+if ($#ARGV_tail < 0) {
+ die "you need a directory as argument\n";
+}
- if (defined $dir && -d "$dir") {
- say "preparing SQL";
- $tmp_ies_unsorted_file->touch();
- say "searching IE files";
- my $cnt_unsorted_files = searching_ie_files($dir, $tmp_ies_unsorted_file);
- # /permanent_storage/2020/04/02/IE201080/V1-FL201091.xml
- # /permanent_storage/2020/04/02/IE201080/V2-FL201091.xml
- my $fh_unsorted_IEs = $tmp_ies_unsorted_file->openr();
- my $count=0;
- my $progressbar =Time::Progress->new(min => 0, max => $cnt_unsorted_files, smoothing => 1);
- my $dbh = DBI->connect("dbi:SQLite:dbname=$db_filename", "", "", {
- RaiseError => 1,
- sqlite_unicode => 1,
- }) or die "could not connect to database (file '$db_filename')", $DBI::errstr;
- write_database_creation($dbh);
- write_tables_creation($dbh);
- write_prepare_insert($dbh);
- write_index_creation($dbh);
- while( <$fh_unsorted_IEs> ) {
- chomp;
- print $progressbar->report("parse IE files: %40b ETA: %E \r", $count++);
- s/V0*(\d+-IE)/V$1/; # revert fake version
- my $ret = parse_iexml($_, $flag_recovery);
- write_addsql($dbh, $ret);
- }
- say "";
- $dbh->disconnect or warn("disconnecting problems, ", $dbh->errstr);
- say "processed $count uniq IEs";
- } else {
- die "no directory given on commandline"
+my $dir = shift @ARGV_tail;
+
+if (defined $dir && -d "$dir") {
+ say "preparing SQL";
+ $tmp_ies_unsorted_file->touch();
+ say "searching IE files";
+ my $cnt_unsorted_files = searching_ie_files($dir, $tmp_ies_unsorted_file);
+ # /permanent_storage/2020/04/02/IE201080/V1-FL201091.xml
+ # /permanent_storage/2020/04/02/IE201080/V2-FL201091.xml
+ my $fh_unsorted_IEs = $tmp_ies_unsorted_file->openr();
+ my $count = 0;
+ my $progressbar = Time::Progress->new(min => 0, max => $cnt_unsorted_files, smoothing => 1);
+ my $dbh = DBI->connect("dbi:SQLite:dbname=$db_filename", "", "", {
+ RaiseError => 1,
+ sqlite_unicode => 1,
+ }) or die "could not connect to database (file '$db_filename')", $DBI::errstr;
+ write_database_creation($dbh);
+ write_tables_creation($dbh);
+ write_prepare_insert($dbh);
+ write_index_creation($dbh);
+ while (<$fh_unsorted_IEs>) {
+ chomp;
+ print $progressbar->report("parse IE files: %40b ETA: %E \r", $count++);
+ s/V0*(\d+-IE)/V$1/; # revert fake version
+ my $ret = parse_iexml($_, $flag_recovery);
+ write_addsql($dbh, $ret);
}
- print "\n";
-
+ say;
+ $dbh->disconnect or warn("disconnecting problems, ", $dbh->errstr);
+ say "processed $count uniq IEs";
+}
+else {
+ die "no directory given on commandline"
+}
+say;
1;