<?php
require_once "../uti/ChineseCharacterStatistic.php";
require_once "../uti/EnglishWordsStatistic.php";
require_once "../uti/Ureaddir.php";
require_once "../uti/MysqlJiaguwen.php";
$BOOKDIR = $_GET["book"];
if (isset($_GET["fu"])) {
$out = ChineseCharacterStatistic::BookDir2outfile($BOOKDIR, true);
unlink($out);
}
syslog(0, $BOOKDIR);
$ud = new Ureaddir();
$arr = $ud->readdir2arr_RecursFiles($BOOKDIR);
$ccs = null;
if (ChineseCharacterStatistic::IsChaninese($BOOKDIR)) {
syslog(0, $BOOKDIR);
$ccs = new ChineseCharacterStatistic($arr);
} else {
$ccs = new EnglishWordsStatistic($arr);
}
$ccs->run();
//$ccs->display();
print $ccs->get_view_content($BOOKDIR);
//reload if already exist to save time.
print "<br>Uniq Chinese from book:<br>";
$chinesezid = array();
foreach ($ccs->WordFreqArr as $key => $arr) {
$chinesezid[] = substr($key, 2, 5);
}
sort($chinesezid);
public static function FindStatisticFromFile($bookdir, $keyword)
{
if (!isset($keyword)) {
return "";
}
$pathfilename = ChineseCharacterStatistic::BookDir2outfile($bookdir, true);
syslog(0, $bookdir . $keyword);
if (!file_exists($pathfilename)) {
return $keyword;
}
//return $pathfilename . $keyword;
$data = file_get_contents($pathfilename);
if (false == $data) {
syslog(0, "failed to read file:" . $pathfilename);
}
$bChinese = ChineseCharacterStatistic::IsChaninese($bookdir);
$map = new MapChineseGbEnglishWord();
$keyword = $map->getChineseGB($keyword, $bChinese);
//ChineseCharacterStatistic::LanguageTranslate($bookdir, $keyword);
$out = ChineseCharacterStatistic::FindStatisticFromData($data, $keyword);
return $out;
}
