gpst-file

   1 #!/usr/bin/perl
   2
   3 #=======================================================================
   4 # gpst-file
   5 # File ID: 86736de4-00a1-11de-acc1-000475e441b9
   6 # Generate TAB streams of files with modification time for use with COPY
   7 # in Postgres
   8 #
   9 # Character set: UTF-8
  10 # ©opyleft 2009– Øyvind A. Holm <sunny@sunbase.org>
  11 # License: GNU General Public License version 3 or later, see end of
  12 # file for legal stuff.
  13 #=======================================================================
  14
  15 use strict;
  16 use warnings;
  17 use Getopt::Long;
  18
  19 BEGIN {
  20     push(@INC, "$ENV{'HOME'}/bin/src/gpstools");
  21     our @version_array;
  22 }
  23
  24 use GPST;
  25 use GPSTxml;
  26
  27 $| = 1;
  28
  29 our $Debug = 0;
  30 our $NA = '\N';
  31 our %Std = (
  32
  33     'output-format' => 'pgtab',
  34     'timezone' => '',
  35
  36 );
  37 our %Opt = (
  38
  39     'author' => '',
  40     'debug' => 0,
  41     'description' => '',
  42     'help' => 0,
  43     'output-format' => $Std{'output-format'},
  44     'strip-whitespace' => 0,
  45     'timezone' => $Std{'timezone'},
  46     'verbose' => 0,
  47     'version' => 0,
  48
  49 );
  50
  51 our $progname = $0;
  52 $progname =~ s/^.*\/(.*?)$/$1/;
  53 our $VERSION = "0.00";
  54
  55 Getopt::Long::Configure("bundling");
  56 GetOptions(
  57
  58     "author|a=s" => \$Opt{'author'},
  59     "debug" => \$Opt{'debug'},
  60     "description|d=s" => \$Opt{'description'},
  61     "help|h" => \$Opt{'help'},
  62     "output-format|o=s" => \$Opt{'output-format'},
  63     "strip-whitespace|w" => \$Opt{'strip-whitespace'},
  64     "timezone|T=s" => \$Opt{'timezone'},
  65     "verbose|v+" => \$Opt{'verbose'},
  66     "version" => \$Opt{'version'},
  67
  68 ) || die("$progname: Option error. Use -h for help.\n");
  69
  70 $Opt{'debug'} && ($Debug = 1);
  71 $Opt{'help'} && usage(0);
  72 if ($Opt{'version'}) {
  73     print_version();
  74     exit(0);
  75 }
  76
  77 $GPST::Spc = $Opt{'strip-whitespace'} ? "" : " ";
  78 my $Spc = $GPST::Spc; # FIXME
  79 my $tz_str = "";
  80 if (length($Opt{'timezone'})) {
  81     if ($Opt{'timezone'} =~ /^[\+\-][0-2][0-9]{3}$/) {
  82         $tz_str = $Opt{'timezone'};
  83     } elsif ($Opt{'timezone'} =~ /^z$/i) {
  84         $tz_str = $Opt{'timezone'};
  85     } elsif ($Opt{'timezone'} =~ /^[a-z]+$/i) {
  86         $tz_str = " $Opt{'timezone'}";
  87     } else {
  88         die("$progname: $Opt{'timezone'}: Invalid time zone\n");
  89     }
  90     $tz_str = uc($tz_str);
  91 }
  92
  93 if ($Opt{'output-format'} eq "xml") {
  94     print("<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<gpstfile>\n");
  95 }
  96
  97 if ($#ARGV < 0) {
  98     while (<>) {
  99         chomp();
 100         print_entry($_);
 101     }
 102 } else {
 103     for my $fname (@ARGV) {
 104         print_entry($fname);
 105     }
 106 }
 107
 108 if ($Opt{'output-format'} eq "xml") {
 109     print("</gpstfile>\n");
 110 }
 111
 112 sub print_entry {
 113     # {{{
 114     my $filename = shift;
 115     my $Retval = 0;
 116     my ($date, $coor) =
 117        ('', '');
 118     D("filename = '$filename'");
 119     if (my @stat_array = stat($filename)) {
 120         my @TA = localtime($stat_array[9]);
 121         $date = sprintf("%04u-%02u-%02uT%02u:%02u:%02u%s",
 122             $TA[5]+1900, $TA[4]+1, $TA[3], $TA[2], $TA[1], $TA[0], $tz_str);
 123         D("tz_str = '$tz_str'");
 124         $filename =~ s/^.*\/(.*?)$/$1/;
 125         my $Output = "";
 126         if ($Opt{'output-format'} eq "xml") {
 127             if (length("$filename$date")) {
 128                 $Output = join("",
 129                     "$Spc$Spc<file>\n",
 130                     length($filename)
 131                         ? sprintf("$Spc$Spc$Spc$Spc<filename>%s</filename>\n",
 132                             txt_to_xml($filename))
 133                         : "",
 134                     length($date)
 135                         ? sprintf("$Spc$Spc$Spc$Spc<date>%s</date>\n",
 136                             txt_to_xml($date))
 137                         : "",
 138                     length($Opt{'description'})
 139                         ? sprintf("$Spc$Spc$Spc$Spc<desc>%s</desc>\n",
 140                             txt_to_xml($Opt{'description'}))
 141                         : "",
 142                     length($Opt{'author'})
 143                         ? sprintf("$Spc$Spc$Spc$Spc<author>%s</author>\n",
 144                             txt_to_xml($Opt{'author'}))
 145                         : "",
 146                     "$Spc$Spc</file>\n",
 147                 );
 148             }
 149         } elsif ($Opt{'output-format'} eq "pgtab") {
 150             # Version information {{{
 151             # Version 1:
 152             #   "1" \t
 153             #   date \t
 154             #   "(lat,lon)"-coordinates \t
 155             #   description \t
 156             #   filename \t
 157             #   author \n
 158             # }}}
 159             $Output = pgtab_entry(
 160                 1, # Version number
 161                 $date,
 162                 $coor,
 163                 $Opt{'description'},
 164                 $filename,
 165                 $Opt{'author'}
 166             );
 167         } else {
 168             die("$progname: $Opt{'output-format'}: Unknown output format\n");
 169         }
 170         print($Output);
 171         $Opt{'verbose'} && print(STDERR $Output);
 172     } else {
 173         warn("$progname: $filename: Cannot stat file: $!\n");
 174     }
 175     return($Retval);
 176     # }}}
 177 } # print_entry()
 178
 179 sub pgtab_entry {
 180     # {{{
 181     my ($Version, $Date, $Coor, $Descr, $Filename, $Author) = @_;
 182     defined($Date)     || ($Date = $NA);
 183     defined($Coor)     || ($Coor = $NA);
 184     defined($Descr)    || ($Descr = $NA);
 185     defined($Filename) || ($Filename = $NA);
 186     defined($Author)   || ($Author = $NA);
 187     my $Retval = "";
 188     if ($Version == 1) {
 189         $Retval = join("\t",
 190             1, # Version number
 191             postgresql_copy_safe($Date),
 192             length($Coor)
 193                 ? postgresql_copy_safe($Coor)
 194                 : $NA,
 195             length($Opt{'description'})
 196                 ? postgresql_copy_safe($Opt{'description'})
 197                 : $NA,
 198             length($Filename)
 199                 ? postgresql_copy_safe($Filename)
 200                 : $NA,
 201             length($Opt{'author'})
 202                 ? postgresql_copy_safe($Opt{'author'})
 203                 : $NA
 204         ) . "\n";
 205     }
 206     return($Retval);
 207     # }}}
 208 } # pgtab_entry()
 209
 210 sub print_version {
 211     # Print program version {{{
 212     print("$progname v$VERSION\n");
 213     # }}}
 214 } # print_version()
 215
 216 sub usage {
 217     # Send the help message to stdout {{{
 218     my $Retval = shift;
 219
 220     if ($Opt{'verbose'}) {
 221         print("\n");
 222         print_version();
 223     }
 224     print(<<END);
 225
 226 Usage: $progname [options] [file [files [...]]]
 227
 228 Generate TAB streams with filenames and file modification time for use
 229 with PostgreSQL’s COPY command. If no filenames are specified on the
 230 command line, file names are read from stdin.
 231
 232 Options:
 233
 234   -a, --author x
 235     Specify author of file.
 236   -d, --description x
 237     Specify description for file.
 238   -h, --help
 239     Show this help.
 240   -v, --verbose
 241     Increase level of verbosity. Can be repeated.
 242   -o x, --output-format x
 243     Create output of type x:
 244       xml
 245       pgtab
 246     Default: "$Std{'output-format'}".
 247   -T X, --timezone X
 248     Prepend X as timezone to the date. Valid formats:
 249       UTC offset
 250         A '+' or '-' followed by a four-digit number (HHMM) which
 251         indicates the offset relative to UTC. Examples:
 252           +0000
 253           -1600
 254           +0630
 255       Time zone abbreviation. Examples:
 256         UTC
 257         Z
 258         CET
 259   -w, --strip-whitespace
 260     Strip all unnecessary whitespace.
 261   --version
 262     Print version information.
 263   --debug
 264     Print debugging messages.
 265
 266 END
 267     exit($Retval);
 268     # }}}
 269 } # usage()
 270
 271 sub msg {
 272     # Print a status message to stderr based on verbosity level {{{
 273     my ($verbose_level, $Txt) = @_;
 274
 275     if ($Opt{'verbose'} >= $verbose_level) {
 276         print(STDERR "$progname: $Txt\n");
 277     }
 278     # }}}
 279 } # msg()
 280
 281 sub D {
 282     # Print a debugging message {{{
 283     $Debug || return;
 284     my @call_info = caller;
 285     chomp(my $Txt = shift);
 286     my $File = $call_info[1];
 287     $File =~ s#\\#/#g;
 288     $File =~ s#^.*/(.*?)$#$1#;
 289     print(STDERR "$File:$call_info[2] $$ $Txt\n");
 290     return("");
 291     # }}}
 292 } # D()
 293
 294 __END__
 295
 296 # Plain Old Documentation (POD) {{{
 297
 298 =pod
 299
 300 =head1 NAME
 301
 302
 303
 304 =head1 SYNOPSIS
 305
 306  [options] [file [files [...]]]
 307
 308 =head1 DESCRIPTION
 309
 310
 311
 312 =head1 OPTIONS
 313
 314 =over 4
 315
 316 =item B<-h>, B<--help>
 317
 318 Print a brief help summary.
 319
 320 =item B<-v>, B<--verbose>
 321
 322 Increase level of verbosity. Can be repeated.
 323
 324 =item B<--version>
 325
 326 Print version information.
 327
 328 =item B<--debug>
 329
 330 Print debugging messages.
 331
 332 =back
 333
 334 =head1 BUGS
 335
 336
 337
 338 =head1 AUTHOR
 339
 340 Made by Øyvind A. Holm S<E<lt>sunny@sunbase.orgE<gt>>.
 341
 342 =head1 COPYRIGHT
 343
 344 Copyleft © Øyvind A. Holm E<lt>sunny@sunbase.orgE<gt>
 345 This is free software; see the file F<COPYING> for legalese stuff.
 346
 347 =head1 LICENCE
 348
 349 This program is free software: you can redistribute it and/or modify it
 350 under the terms of the GNU General Public License as published by the
 351 Free Software Foundation, either version 3 of the License, or (at your
 352 option) any later version.
 353
 354 This program is distributed in the hope that it will be useful, but
 355 WITHOUT ANY WARRANTY; without even the implied warranty of
 356 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 357 See the GNU General Public License for more details.
 358
 359 You should have received a copy of the GNU General Public License along
 360 with this program.
 361 If not, see L<http://www.gnu.org/licenses/>.
 362
 363 =head1 SEE ALSO
 364
 365 =cut
 366
 367 # }}}
 368
 369 # vim: set fenc=UTF-8 ft=perl fdm=marker ts=4 sw=4 sts=4 et fo+=w :