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