/**
* mecab classifier
* @param $document
* @return array|void
*/
public static function toWords($document)
{
if (!extension_loaded('mecab')) {
Log::error('mecab extension not loaded');
return;
}
$mecab = new \MeCab\Tagger();
$nodes = $mecab->parseToNode($document);
$words = [];
foreach ($nodes as $n) {
$word = $n->getSurface();
if (static::isInvalidWord($word)) {
continue;
}
if (strpos($n->getFeature(), '名詞') !== false) {
$words[] = $word;
}
if (strpos($n->getFeature(), '動詞') !== false) {
$words[] = $word;
}
if (strpos($n->getFeature(), '助動詞') !== false) {
$words[] = $word;
}
if (strpos($n->getFeature(), '形容詞') !== false) {
$words[] = $word;
}
}
return $words;
}
<?php
/**
* php-mecab/examples
* parse string, wakati output format (with SPL and autocast)
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg_format);
border();
foreach ($mecab->parseToNode($str_long) as $node) {
echo $mecab->formatNode($node);
}
border();
foreach ($mecab->parseToNode($str_long) as $node) {
echo $node;
}
border();
<?php /** * php-mecab/examples * parse string * charset=utf-8 */ require_once dirname(__FILE__) . '/common.inc.php'; $mecab = new MeCab\Tagger(); border(); echo $mecab->parse($str_long); border();
/**
* php-mecab/examples
* test like official bindings examples
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$sentence = '太郎はこの本を二郎を見た女性に渡した。';
if (isset($_SERVER['argv'])) {
$options = $_SERVER['argv'];
array_shift($options);
} else {
$options = array();
}
writeln(MeCab\VERSION);
$t = new MeCab\Tagger($options);
writeln($t->parse($sentence));
foreach ($t->parseToNode($sentence) as $m) {
writeln($m->surface . "\t" . $m->feature);
}
writeln('EOS');
$di = $t->dictionaryInfo();
foreach ($di as $d) {
writefln('filename: %s', $d['filename']);
writefln('charset: %s', $d['charset']);
writefln('size: %d', $d['size']);
writefln('type: %d', $d['type']);
writefln('lsize: %d', $d['lsize']);
writefln('rsize: %d', $d['rsize']);
writefln('version: %d', $d['version']);
}
<?php /** * php-mecab/examples * parse string, wakati output format * charset=utf-8 */ require_once dirname(__FILE__) . '/common.inc.php'; $mecab = new MeCab\Tagger($arg_all_morphs); border(); echo $mecab->parse($str); border();
<?php
/**
* php-mecab/examples
* parse string, wakati output format
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg_format);
border();
$node = $mecab->parseToNode($str_long);
while ($node) {
echo $mecab->formatNode($node);
$node = $node->getNext();
}
border();
$node = $mecab->parseToNode($str_long);
while ($node) {
echo $node->toString();
$node = $node->getNext();
}
border();
<?php
/**
* php-mecab/examples
* parse N-Best
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg);
if ($mecab->parseNBestInit($str_long)) {
border();
for ($i = 0; $i < NBEST_MAX_RESULT && ($next = $mecab->next()); $i++) {
echo $next;
border();
}
}
<?php
/**
* php-mecab/examples
* show dictionary information
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg);
border();
writefln('MeCab Version: %s', MeCab\VERSION);
border();
writeln('Dictionary Information:');
print_r($mecab->dictionaryInfo());
border();
<?php
/**
* php-mecab/examples
* dump all nodes (with SPL and Overloading)
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg);
function call_format($node)
{
return format($node->surface, $node->feature, $node->id, $node->stat);
}
if ($iter = $mecab->parseToNode($str)) {
border();
foreach ($iter as $node) {
$ar = array('node' => $node->toArray(), 'prev' => '-', 'next' => '-', 'enext' => '-', 'bnext' => '-', 'rpath' => '-', 'lpath' => '-');
if ($node->prev) {
$ar['prev'] = call_format($node->prev);
}
if ($node->next) {
$ar['next'] = call_format($node->next);
}
if ($node->enext) {
$ar['enext'] = call_format($node->enext);
}
if ($node->bnext) {
$ar['bnext'] = call_format($node->bnext);
}
if ($rpath = $node->rpath) {
$ar['rpath'] = array('prob' => $rpath->prob, 'cost' => $rpath->cost, 'rnode' => '-', 'lnode' => '-');
<?php
/**
* php-mecab/examples
* dump all nodes
* charset=utf-8
*/
require_once dirname(__FILE__) . '/common.inc.php';
$mecab = new MeCab\Tagger($arg);
function call_format($node)
{
return format($node->getSurface(), $node->getFeature(), $node->getId(), $node->getStat());
}
if ($node = $mecab->parseToNode($str)) {
border();
while ($node) {
$ar = array('node' => $node->toArray(), 'prev' => '-', 'next' => '-', 'enext' => '-', 'bnext' => '-', 'rpath' => '-', 'lpath' => '-');
if ($prev = $node->getPrev()) {
$ar['prev'] = call_format($prev);
}
if ($next = $node->getNext()) {
$ar['next'] = call_format($next);
}
if ($enext = $node->getENext()) {
$ar['enext'] = call_format($enext);
}
if ($bnext = $node->getBNext()) {
$ar['bnext'] = call_format($bnext);
}
if ($rpath = $node->getRPath()) {
$ar['rpath'] = array('prob' => $rpath->getProb(), 'cost' => $rpath->getCost(), 'rnode' => '-', 'lnode' => '-');
