| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 1 | #!/usr/bin/env perl |
| Akron | 3587f36 | 2020-05-20 17:50:38 +0200 | [diff] [blame] | 2 | use strict; |
| 3 | use warnings; | ||||
| Akron | 85f5d9e | 2024-12-17 14:45:20 +0100 | [diff] [blame^] | 4 | use lib 'lib'; |
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 5 | use KorAP::Def; |
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 6 | use KorAP::DefList; |
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 7 | use Getopt::Long; |
| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 8 | |
| Akron | 99d2d08 | 2024-07-18 16:17:08 +0200 | [diff] [blame] | 9 | binmode(STDERR, ':encoding(UTF-8)'); |
| 10 | |||||
| Akron | 340a9cb | 2020-05-20 12:55:22 +0200 | [diff] [blame] | 11 | # 2020-05-20 |
| 12 | # Preliminary support for C2 def-files. | ||||
| Akron | 34a4f58 | 2020-05-27 12:18:57 +0200 | [diff] [blame] | 13 | # 2020-05-29 |
| 14 | # Introduce optimizable object system. | ||||
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 15 | # 2024-07-17 |
| 16 | # Add KorAP::Def. | ||||
| Akron | 0989cb3 | 2024-12-16 13:52:47 +0100 | [diff] [blame] | 17 | # 2024-12-16 |
| 18 | # Fix doc trimming | ||||
| Akron | adcc8de | 2024-12-16 15:10:50 +0100 | [diff] [blame] | 19 | # Fix country codes. |
| Akron | 340a9cb | 2020-05-20 12:55:22 +0200 | [diff] [blame] | 20 | |
| Akron | 0989cb3 | 2024-12-16 13:52:47 +0100 | [diff] [blame] | 21 | our $VERSION = 0.4; |
| Akron | 26b5970 | 2020-05-19 12:14:41 +0200 | [diff] [blame] | 22 | our @ARGV; |
| 23 | |||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 24 | my $cmd = shift @ARGV; |
| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 25 | |
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 26 | |
| Akron | b31321e | 2024-07-18 11:17:46 +0200 | [diff] [blame] | 27 | my ($input, $output, $copysrc); |
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 28 | |
| 29 | GetOptions ( | ||||
| Akron | b31321e | 2024-07-18 11:17:46 +0200 | [diff] [blame] | 30 | "input|i=s" => \$input, |
| 31 | "output|o=s" => \$output, | ||||
| 32 | "copy-src|c=s" => \$copysrc, | ||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 33 | ) |
| 34 | or die("Error in command line arguments\n"); | ||||
| 35 | |||||
| Akron | 4b489ed | 2024-12-16 15:19:40 +0100 | [diff] [blame] | 36 | if (!$cmd || ($cmd ne 'def' && $cmd ne 'list')) { |
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 37 | print <<'HELP'; |
| 38 | Convert a list of C2 VC definitions or a single definition into | ||||
| 39 | KoralQuery VCs. | ||||
| 40 | |||||
| 41 | $ perl cosmasvc2koralquery def my_vc.txt | gzip -vc > my_vc.jsonld.gz | ||||
| 42 | $ cat my_vc.txt | perl cosmasvc2koralquery def - | gzip -vc > my_vc.jsonld.gz | ||||
| 43 | |||||
| Akron | b31321e | 2024-07-18 11:17:46 +0200 | [diff] [blame] | 44 | Command: def |
| 45 | |||||
| 46 | Convert a def file or a list of sigles to a KoralQuery VC. | ||||
| 47 | |||||
| 48 | Takes the list or def from STDIN and exports to STDOUT. | ||||
| 49 | |||||
| 50 | Command: list | ||||
| 51 | |||||
| 52 | Convert a list with copy or regex instructions to KoralQuery VCs. | ||||
| 53 | |||||
| 54 | --output: The output directory | ||||
| 55 | --copy-src: The directory for def files to copy | ||||
| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 56 | |
| 57 | HELP | ||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 58 | exit 1; |
| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 59 | }; |
| 60 | |||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 61 | |
| 62 | # Process a list | ||||
| 63 | if ($cmd eq 'list') { | ||||
| Akron | b31321e | 2024-07-18 11:17:46 +0200 | [diff] [blame] | 64 | KorAP::DefList->new( |
| 65 | file => ($input || $ARGV[0]), | ||||
| 66 | copy => ($copysrc || '.'), | ||||
| 67 | output => ($output || '.') | ||||
| 68 | )->parse; | ||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 69 | exit(0); |
| 70 | }; | ||||
| 71 | |||||
| 72 | # Parse a single def | ||||
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 73 | my $def_parser; |
| Akron | 26b5970 | 2020-05-19 12:14:41 +0200 | [diff] [blame] | 74 | if ($ARGV[0] eq '-') { |
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 75 | $def_parser = KorAP::Def->new(\*STDIN); |
| 76 | } | ||||
| Akron | a3f74f5 | 2024-07-17 17:18:17 +0200 | [diff] [blame] | 77 | elsif ($input) { |
| 78 | $def_parser = KorAP::Def->new($input); | ||||
| 79 | } | ||||
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 80 | else { |
| 81 | $def_parser = KorAP::Def->new($ARGV[0]); | ||||
| Akron | 3f875be | 2020-05-11 14:57:19 +0200 | [diff] [blame] | 82 | }; |
| 83 | |||||
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 84 | $def_parser->parse; |
| Akron | 26b5970 | 2020-05-19 12:14:41 +0200 | [diff] [blame] | 85 | |
| Akron | 1c07045 | 2020-05-25 11:28:30 +0200 | [diff] [blame] | 86 | # Stringify current (extended?) virtual corpus |
| Akron | 71cf2f5 | 2024-07-17 08:54:15 +0200 | [diff] [blame] | 87 | print $def_parser->to_string; |