function extract_gramtab($graminfoFile, $outDir, $asText)
{
$factory = new phpMorphy_Storage_Factory();
$graminfo = phpMorphy_GramInfo::create($factory->open(PHPMORPHY_STORAGE_FILE, $graminfoFile, false), false);
$poses = $graminfo->readAllPartOfSpeech();
$grammems = $graminfo->readAllGrammems();
$ancodes = $graminfo->readAllAncodes();
if ($asText) {
foreach ($ancodes as &$ancode) {
$pos_id = $ancode['pos_id'];
if (!isset($poses[$pos_id])) {
throw new Exception("Unknown pos_id '{$pos_id}' found");
}
$ancode['pos_id'] = $poses[$pos_id]['name'];
foreach ($ancode['grammem_ids'] as &$grammem_id) {
if (!isset($grammems[$grammem_id])) {
throw new Exception("Unknown grammem_id '{$grammem_id}' found");
}
$grammem_id = $grammems[$grammem_id]['name'];
}
}
//$poses = replace_keys_with_name($poses);
//$grammems = replace_keys_with_name($grammems);
}
$result = array('poses' => $poses, 'grammems' => $grammems, 'ancodes' => $ancodes);
$type = $asText ? '_txt' : '';
$out_file = 'gramtab' . $type . '.' . strtolower($graminfo->getLocale()) . '.bin';
$out_file = $outDir . '/' . $out_file;
if (false === file_put_contents($out_file, serialize($result))) {
throw new Exception("Can`t write '{$out_file}'");
}
}
#!/usr/bin/php
<?php
if (2 == (ini_get('mbstring.func_overload') & 2)) {
die("don`t overload string functions in mbstring extension, see mbstring.func_overload option");
}
if ($argc < 3) {
echo "Usage " . $argv[0] . " MORPH_DATA_FILE OUT_DIR";
exit;
}
require_once dirname(__FILE__) . '/../src/common.php';
$file = $argv[1];
$out_dir = $argv[2];
$use_references = true;
try {
$factory = new phpMorphy_Storage_Factory();
$graminfo = phpMorphy_GramInfo::create($factory->open(PHPMORPHY_STORAGE_FILE, $file, false), false);
$ancodes_map = new Map('ancodes');
$flexias_map = new Map('affixes');
$i = 0;
foreach ($graminfo->readAllFlexia() as $id => $flexia) {
$offset = $flexia['header']['offset'];
// + $graminfo->getGramInfoHeaderSize();
$ancodes_map->update($flexia, $offset);
//$flexias_map->update($flexia, $offset);
$i++;
}
echo "Total flexias = {$i}, unique ancodes = " . count($ancodes_map->getMap()) . ', unique flexias = ' . count($flexias_map->getMap()) . PHP_EOL;
$out_file_format = $out_dir . '/%s.' . strtolower($graminfo->getLocale()) . '.bin';
file_put_contents(sprintf($out_file_format, 'morph_data_ancodes_cache'), serialize($ancodes_map->compose($use_references)));
//file_put_contents(sprintf($out_file_format, 'morph_data_flexias_cache'), serialize($flexias_map->compose($use_references)));
} catch (Exception $e) {
function __get($propName)
{
if ($propName == 'info') {
$this->info = phpMorphy_GramInfo::create($this->storage, false);
unset($this->storage);
return $this->info;
}
throw new phpMorphy_Exception("Unknown prop name '{$propName}'");
}
