diff options
| author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-04 18:07:14 +0000 |
|---|---|---|
| committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-05-04 18:07:14 +0000 |
| commit | a175314c3e5827eb193872241446f2f8f5c9d33c (patch) | |
| tree | cd3d60ca99ae00829c52a6ca79150a5b6e62528b /storage/mroonga/vendor/groonga/examples/dictionary/gene95 | |
| parent | Initial commit. (diff) | |
| download | mariadb-10.5-a175314c3e5827eb193872241446f2f8f5c9d33c.tar.xz mariadb-10.5-a175314c3e5827eb193872241446f2f8f5c9d33c.zip | |
Adding upstream version 1:10.5.12.upstream/1%10.5.12upstream
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to 'storage/mroonga/vendor/groonga/examples/dictionary/gene95')
3 files changed, 63 insertions, 0 deletions
diff --git a/storage/mroonga/vendor/groonga/examples/dictionary/gene95/Makefile.am b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/Makefile.am new file mode 100644 index 00000000..e89f13f5 --- /dev/null +++ b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/Makefile.am @@ -0,0 +1,4 @@ +gene95dir = $(examples_dictionarydir)/gene95 +dist_gene95_SCRIPTS = \ + gene2grn.rb \ + gene-import.sh diff --git a/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene-import.sh b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene-import.sh new file mode 100755 index 00000000..488d6c83 --- /dev/null +++ b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene-import.sh @@ -0,0 +1,26 @@ +#!/bin/sh + +base_dir=$(dirname $0) + +if [ 1 != $# -a 2 != $# ]; then + echo "usage: $0 db_path [gene.txt_path]" + exit 1 +fi + +if [ -z $2 ]; then + dictionary_dir=gene95-dictionary + gene_txt=${dictionary_dir}/gene.txt + if [ ! -f $gene_txt ]; then + gene95_tar_gz=gene95.tar.gz + wget -O $gene95_tar_gz \ + http://www.namazu.org/~tsuchiya/sdic/data/gene95.tar.gz + mkdir -p ${dictionary_dir} + tar xvzf ${gene95_tar_gz} -C ${dictionary_dir} + fi +else + gene_txt=$2 +fi + +if cat $gene_txt | ${base_dir}/gene2grn.rb | groonga $1 > /dev/null; then + echo "gene95 data loaded." +fi diff --git a/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene2grn.rb b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene2grn.rb new file mode 100755 index 00000000..c9d9a593 --- /dev/null +++ b/storage/mroonga/vendor/groonga/examples/dictionary/gene95/gene2grn.rb @@ -0,0 +1,33 @@ +#!/usr/bin/env ruby + +require "json" + +print(<<HEADER.chomp) +column_create item_dictionary gene95_desc COLUMN_SCALAR ShortText +column_create bigram item_dictionary_gene95_desc COLUMN_INDEX|WITH_POSITION item_dictionary gene95_desc +load --table item_dictionary +[ +["_key","gene95_desc"] +HEADER + +loop do + raw_key = gets + break if raw_key.nil? + raw_body = gets + + key = nil + body = nil + begin + key = raw_key.encode("UTF-8", "Windows-31J").strip + body = raw_body.encode("UTF-8", "Windows-31J").strip + rescue EncodingError + $stderr.puts("Ignore:") + $stderr.puts(" key: <#{raw_key}>") + $stderr.puts(" body: <#{raw_body}>") + next + end + puts(",") + print([key, body].to_json) +end +puts +puts("]") |