<?php
/**
* php-mecab/examples
* dump all nodes (OO-API)
* 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' => '-');
<?php
/**
* php-mecab/examples
* dump all nodes (OO-API 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' => $rpath->cost, 'rnode' => '-', 'lnode' => '-');
if ($key == 0) {
array_push($time_line_texts, $resolved);
}
}
}
}
}
foreach ($jset3['statuses'] as $result) {
$timeee = date('H', strtotime((string) $result['user']['created_at']));
//時間を取得
$lists = array("time" => $timeee, "id" => $result['id_str'], "userID" => $result['user']['screen_name'], "name" => $result['user']['name'], "link" => $result['user']['profile_image_url'], "text" => $result['text']);
// ツイートのリストをliで出力
array_push($list, $lists);
ini_set('mecab.default_userdic', 'onomasticon2.dic');
$mecab = new MeCab_Tagger();
for ($node = $mecab->parseToNode($result['text']); $node; $node = $node->getNext()) {
if ($node->getStat() != 2 && $node->getStat() != 3) {
$resolved = array("Surface" => $node->getSurface(), "Feature" => $node->getFeature(), "Time" => $timeee);
$feature = '/固有名詞(.*)/u';
// フィルターに掛ける品詞 形容詞(.*)|感動詞(.*)|
if (preg_match($feature, $resolved["Feature"])) {
// フィルター
unset($resolved["Feature"]);
$search = '/http|RT|\\w+|ｗ+|笑|bot|あ+|あー|い+|え+|お+|ない|ー/u';
// フィルターに掛ける文字
$key = preg_match($search, $resolved["Surface"]);
// フィルター
if ($key == 0) {
array_push($time_line_texts, $resolved);
}
}
<?php
/**
* php-mecab/examples
* parse string, wakati output format (OO-API 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 $node;
}
border();
function funcGetKana($str)
{
$mecab = new MeCab_Tagger();
$kna = "";
for ($node = $mecab->parseToNode($str); $node; $node = $node->getNext()) {
if ($node->getStat() != 2 && $node->getStat() != 3) {
$word = $node->getSurface();
switch (true) {
case isZenKana($word, "UTF-8"):
$kna .= $word;
//echo "全角カタカナ".$word."<br />";
break;
case isHanKana($word, "UTF-8"):
$kna .= mb_convert_kana($word, "K");
//echo "半角カタカナ".$word."<br />";
break;
case isHiragana($word, "UTF-8"):
$kna .= mb_convert_kana($word, "C");
//echo "ひらがな".$word."<br />";
break;
case isZenNum($word, "UTF-8"):
$kna .= $word;
//echo "全角数値".$word."<br />";
break;
case isHanNum($word, "UTF-8"):
$kna .= mb_convert_kana($word, "N");
//echo "半角数値".$word."<br />";
break;
case isZenEiji($word, "UTF-8"):
$kna .= $word;
//echo "全角英字".$word."<br />";
break;
case isHanEiji($word, "UTF-8"):
$kna .= mb_convert_kana($word, "R");
//echo "半角英字".$word."<br />";
break;
case isKigou($word, "UTF-8"):
$kna .= $word;
//echo "記号".$word."<br />";
break;
default:
$fet = $node->getFeature();
$fettbl = explode(",", $fet);
$kna .= $fettbl[7];
}
}
}
//for
return $kna;
}
/**
* php-mecab/examples
* test like official bindings examples (OO-API)
* 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 (OO-API)
* 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 = $node->getNext();
}
border();
function create($input)
{
$lists = array();
$completion_data = array("name" => "tagcloud", "children" => array());
// タグクラウドで使うデータ（配列）
$time_line_texts = array();
$test_m = array();
for ($ii = 0; $ii <= count($input) - 1; $ii++) {
$twObj = new TwitterOAuth(CONSUMER_KEY, CONSUMER_SECRET, $_SESSION['oauthToken'], $_SESSION['oauthTokenSecret']);
// REST_API指定(今回はタイムラインのREST_API)
$tw_rest_api = 'https://api.twitter.com/1.1/statuses/user_timeline.json';
// メソッド指定
$request_method = 'GET';
// クエリーオプション
$options = array('screen_name' => $input[$ii], 'count' => '100');
// ユーザータイムライン取得
$tw_obj_request = $twObj->OAuthRequest($tw_rest_api, $request_method, $options);
// json形式で取得
$tw_obj_request_json = json_decode($tw_obj_request, true);
// 変数生成
$list = array();
$lists_test = array();
// 取得したデータを回して入れていく
foreach ($tw_obj_request_json as $key => $value) {
$time = date('H', strtotime((string) $value['created_at']));
//時間を取得
$str = $value["text"];
$list = array("time" => $time, "text" => $str, "id" => $value['id_str'], "userID" => $value['user']['screen_name'], "name" => $value['user']['name'], "link" => $value['user']['profile_image_url']);
array_push($lists_test, $list);
// mecab
ini_set('mecab.default_userdic', 'onomasticon2.dic');
$mecab = new MeCab_Tagger();
for ($node = $mecab->parseToNode($str); $node; $node = $node->getNext()) {
if ($node->getStat() != 2 && $node->getStat() != 3) {
$resolved = array("Surface" => $node->getSurface(), "Feature" => $node->getFeature(), "Time" => $time, "Id" => $value['user']['screen_name']);
$feature = '/形容詞(.*)|感動詞(.*)|固有名詞(.*)/u';
// フィルターに掛ける品詞 形容詞(.*)|感動詞(.*)|
if (preg_match($feature, $resolved["Feature"])) {
// フィルター
unset($resolved["Feature"]);
$search = '/http|RT|\\w+|ｗ+|笑|bot|あ+|あー|い+|え+|お+|ない|ー/u';
// フィルターに掛ける文字
$key = preg_match($search, $resolved["Surface"]);
// フィルター
if ($key == 0) {
array_push($time_line_texts, $resolved);
}
}
}
}
}
array_push($lists, $lists_test);
}
// 重複を消したリストを作成($search_list)
$tmp = array();
$search_list = array();
foreach ($time_line_texts as $key => $value) {
if (!in_array($value['Surface'], $tmp)) {
$tmp[] = $value['Surface'];
$search_list[] = $value;
}
}
// タグクラウドで使うデータを作成（$completion_data）
for ($i = 0, $size = count($search_list) - 1; $i <= $size; $i++) {
$mix_data = array();
// リセット
$search_text = $search_list[$i]["Surface"];
$plane_data = array();
foreach ($time_line_texts as $Surface => $c) {
if (in_array($search_text, $c)) {
// マッチしたら
/* thisText 追加 */
array_push($plane_data, $c);
}
}
$count_data = count($plane_data);
// $plane_dataの数
$mix_data = array("Surface" => "", "Time" => array(), "Count" => $count_data, "Id" => array());
// タグクラウドで使うデータ（配列）の１つ１つのデータ
$mix_data["Surface"] = $plane_data[0]["Surface"];
for ($t = 0; $t <= $count_data - 1; $t++) {
$mix_data["Time"][$t] = array();
$mix_data["Id"][$t] = array();
array_push($mix_data["Time"][$t], $plane_data[$t]["Time"]);
array_push($mix_data["Id"][$t], $plane_data[$t]["Id"]);
}
/* var_dump($mix_data); */
// フィルター
if ($mix_data["Count"] > 1) {
$test_key = array();
for ($t = 0; $t <= count($mix_data["Id"]) - 1; $t++) {
if ($mix_data["Id"][$t] == $mix_data["Id"][0]) {
array_push($test_key, "1");
} else {
array_push($test_key, "0");
}
}
//array_push($test_m , $test_key);
if (in_array("0", $test_key)) {
array_push($completion_data["children"], $mix_data);
}
}
}
// return $test_m;
// 配列をJSONにエンコード
return array($completion_data, $lists);
// 表示
/* var_dump($completion_data); */
}