Initial commit

.gitignore

@@ -0,0 +1,4 @@
+*.oxt
+en_us
+dict-*
+wordforms-*

Makefile

@@ -0,0 +1,13 @@
+help:
+	@echo "help   - napoveda"
+	@echo "dump   - dump slovniku"
+	@echo "clean  - smaze stazene a generovane soubory"
+
+dump:
+	./download.sh
+	./unpack.sh
+	./dump.sh
+	./pack.sh
+
+clean:
+	rm -rf *.oxt en_us* dict-* wordforms-*

README.md

@@ -0,0 +1 @@
+CZ a SK wordforms slovníky pro Sphinx search.

download.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+
+for foo in \
+	http://downloads.sourceforge.net/project/aoo-extensions/1078/0/dict-cs-2.0.oxt \
+	http://downloads.sourceforge.net/project/aoo-extensions/1143/11/dict-sk.oxt \
+	http://downloads.sourceforge.net/project/aoo-extensions/1470/1/en_us.oxt \
+	http://downloads.sourceforge.net/project/aoo-extensions/1317/0/dict-en-us-names-2008-10-12.oxt
+do
+	wget --no-clobber $foo
+done

dump.sh

@@ -0,0 +1,16 @@
+#!/bin/bash
+
+spelldump en_us/en_US.dic en_us/en_US.aff wordforms-en.txt
+spelldump dict-en-us-names-2008-10-12/en_USNames.dic dict-en-us-names-2008-10-12/en_USNames.aff wordforms-en-names.txt
+spelldump dict-cs-2.0/cs_CZ.dic dict-cs-2.0/cs_CZ.aff wordforms-cz.txt
+spelldump dict-sk/sk_SK/sk_SK.dic dict-sk/sk_SK/sk_SK.aff wordforms-sk.txt
+
+iconv -f ISO-8859-2 -t ASCII//TRANSLIT wordforms-cz.txt | tr '[:upper:]' '[:lower:]' | sort | uniq > wordforms-cz.txt.tmp
+cat wordforms-sk.txt | tr '[:upper:]' '[:lower:]' | sort | uniq > wordforms-sk.txt.tmp
+iconv -f ISO-8859-1 -t ASCII//TRANSLIT wordforms-en.txt | tr '[:upper:]' '[:lower:]' | sort | uniq > wordforms-en.txt.tmp
+iconv -f ISO-8859-1 -t ASCII//TRANSLIT wordforms-en-names.txt | tr '[:upper:]' '[:lower:]' | sort | uniq > wordforms-en-names.txt.tmp
+
+for foo in wordforms-*.txt.tmp
+do
+	./uniq.php $foo > `echo $foo | sed "s/\.tmp$//"` && rm $foo
+done

pack.sh

@@ -0,0 +1,3 @@
+#!/bin/bash
+
+cd .. && tar zcf sphinxsearch-wordforms.tar.gz sphinxsearch-wordforms/wordforms-*.txt

uniq.php

@@ -0,0 +1,16 @@
+#!/usr/bin/php -d memory_limit=2G
+<?php
+
+$vstup = file($argv[1]);
+
+$vystup = array();
+foreach($vstup as $line){
+        $line = preg_split('/ > /',trim($line));
+        $vystup[$line[0]] = $line[1];
+}
+
+foreach($vystup as $foo=>$bar){
+        if($foo != $bar){
+                print "$foo > $bar\n";
+        }
+}

unpack.sh

@@ -0,0 +1,6 @@
+#!/bin/bash
+
+for foo in *.oxt
+do
+	dtrx --overwrite $foo
+done