iapt-platform
/
mint
zrkadlo https://github.com/iapt-platform/mint.git


			
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337
							<!DOCTYPE html>
<html>
<head>
</head>
<body>
<h2>Insert to Index</h2>
<p><a href="index.php">Home</a></p>
<?php
include "./_pdo.php";
include "../path.php";
if(isset($_GET["from"])==false){
?>
<form action="db_insert_index.php" method="get">
From: <input type="text" name="from" value="0"><br>
To: <input type="text" name="to" value="216"><br>
<input type="submit">
</form>
<?php
return;
}

$from=$_GET["from"];
$to=$_GET["to"];


$dirLog=_DIR_LOG_."/";
$dirDb="db/";
$dirXmlBase=_DIR_PALI_CSV_."/";

$filelist=array();
$fileNums=0;
$log="";
echo "<h2>$from</h2>";

//已经存在的词
$g_wordCounter=0;
$g_wordIndexCounter=0;
$iAllWordIndex=array();
$sAllWord=array();
//新加入的词
$wordindex_max_index = 0;
$aNewWordIndex = array(); //词内容
$sNewWord = array(); //词头索引

global $dbh_word_index;
$dns = "sqlite:"._FILE_DB_WORD_INDEX_;
$dbh_word_index = new PDO($dns, "", "",array(PDO::ATTR_PERSISTENT=>true));
$dbh_word_index->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING);

$query = "select id from wordindex where 1 order by id DESC ";
	$stmt = $dbh_word_index->prepare($query);
	$stmt->execute(array());
	$id  = $stmt->fetch(PDO::FETCH_ASSOC);
	if($id === FALSE){
		$wordindex_max_index = 0;
	}
	else{
		$wordindex_max_index =$id["id"];
	}
	$db_file = _FILE_DB_PALI_INDEX_;
	PDO_Connect("sqlite:$db_file");
	$query = "select id from word where 1 order by id DESC ";
	$stmt = $PDO->prepare($query);
	$stmt->execute(array());
	$id  = $stmt->fetch(PDO::FETCH_ASSOC);	
	if($id === FALSE){
		$g_wordCounter = 0;
	}
	else{
		$g_wordCounter = $id["id"];
	}
function dict_lookup($word){
	global $dbh_word_index;
	$query = "select * from wordindex where \"word\" = ? ";
	$stmt = $dbh_word_index->prepare($query);
	$stmt->execute(array($word));
    return $stmt->fetch(PDO::FETCH_ASSOC);	
}

function getWordEn($strIn){
	$out=$strIn;
	$out=str_replace("ā","a",$out);
	$out=str_replace("ī","i",$out);
	$out=str_replace("ū","u",$out);
	$out=str_replace("ṅ","n",$out);
	$out=str_replace("ñ","n",$out);
	$out=str_replace("ṭ","t",$out);
	$out=str_replace("ḍ","d",$out);
	$out=str_replace("ṇ","n",$out);
	$out=str_replace("ḷ","l",$out);
	$out=str_replace("ṃ","m",$out);
	return($out);
}

if(($handle=fopen("filelist.csv",'r'))!==FALSE){
	while(($filelist[$fileNums]=fgetcsv($handle,0,','))!==FALSE){
		$fileNums++;
	}
}
if($to==0 || $to>=$fileNums) $to=$fileNums-1;


//for($iFile=$from;$iFile<=$to;$iFile++)
$iFile=$from;
{

	$FileName=$filelist[$iFile][1].".htm";
	$fileId=$filelist[$iFile][0];

	$inputFileName=$FileName;
	$outputFileNameHead=$filelist[$iFile][1];
	$bookId=$filelist[$iFile][2];

	$dirXml=$outputFileNameHead."/";

	$xmlfile = $inputFileName;
	echo "doing:".$xmlfile."<br>";
	$log=$log."$iFile,$FileName,open\r\n";

	$arrInserString=array();

	// 打开文件并读取数据
	$irow=0;
	if(($fp=fopen($dirXmlBase.$dirXml.$outputFileNameHead.".csv", "r"))!==FALSE){
		while(($data=fgetcsv($fp,0,','))!==FALSE){
			$irow++;
			if($irow>1){
				$params=$data;
				$arrInserString[]=$params;
			}
		}
		fclose($fp);
		echo "单词表load：".$dirXmlBase.$dirXml.$outputFileNameHead.".csv<br>";
	}
	else{
		echo "can not open csv file. filename=".$dirXmlBase.$dirXml.$outputFileNameHead.".csv";
	}
	

	// 开始一个事务，关闭自动提交
	$PDO->beginTransaction();
	$query="INSERT INTO word ('id','book','paragraph','wordindex','bold') VALUES (?,?,?,?,?)";
	$stmt = $PDO->prepare($query);
	$count=0;
	$count1=0;
	$sen="";
	$sen1="";
	$sen_en="";
	$sen_count=0;
	$book="";
	$paragraph="";
	foreach($arrInserString as $oneParam){
		if($oneParam[5]!=""){
			$g_wordCounter++;
			$book=substr($oneParam[2],1);
			$paragraph=$oneParam[3];
			$word=$oneParam[5];
			if($oneParam[15]=="bld" ){
				$bold=1;
			}
			else{
				$bold=0;
			}			
			
			if(isset($sAllWord[$word])){
				//已经存在的词
				$wordindex=$sAllWord[$word];
				
				$iAllWordIndex[$wordindex][1]++;
				if($bold==1){
					$iAllWordIndex[$wordindex][3]++;
				}
				else{
					$iAllWordIndex[$wordindex][2]++;
				}
				
			}
			else if(isset($sNewWord[$word])){
				//是新家入的词
				$wordindex=$sNewWord[$word];
				
				$aNewWordIndex[$wordindex][1]++;
				if($bold==1){
					$aNewWordIndex[$wordindex][3]++;
				}
				else{
					$aNewWordIndex[$wordindex][2]++;
				}
			}
			else if(($lookup=dict_lookup($word)) !== FALSE){
				//在数据库中找到
				$wordindex=$lookup["id"];
				$sAllWord[$word]=$wordindex;
				$iAllWordIndex[$wordindex][0]=$word;
				
				$iAllWordIndex[$wordindex][1]=$lookup["count"] + 1;//all word count
				if($bold==1){
					$iAllWordIndex[$wordindex][2] = $lookup["normal"] ;
					$iAllWordIndex[$wordindex][3] = $lookup["bold"] + 1;
				}
				else{
					$iAllWordIndex[$wordindex][2] = $lookup["normal"] + 1;
					$iAllWordIndex[$wordindex][3] = $lookup["bold"] ;
				}
			}
			else{
				//数据库里也没找到 怎么办呢？我想呀想 想呀想
				$wordindex=$wordindex_max_index + 1;
				$sNewWord[$word]=$wordindex;
				
				$aNewWordIndex[$wordindex][0]=$word;
				
				$aNewWordIndex[$wordindex][1]=1;//all word count
				if($bold==1){
					$aNewWordIndex[$wordindex][2]=0;
					$aNewWordIndex[$wordindex][3]=1;
				}
				else{
					$aNewWordIndex[$wordindex][2]=1;
					$aNewWordIndex[$wordindex][3]=0;
				}
				
				$wordindex_max_index++;
			}
		
	
			$newWord=array($g_wordCounter,$book,$paragraph,$wordindex,$bold);
			$stmt->execute($newWord);
			$count++;
		}

	}
	
	// 提交更改 
	$PDO->commit();
	if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
		$error = PDO_ErrorInfo();
		echo "error - $error[2] <br>";
		$log.="$from, $FileName, error, $error[2] \r\n";
	}
	else{
		echo "updata $count recorders.<br />";
		$log.="updata $count recorders.\r\n";
	}

}
//更新单词索引表

//首先插入新的词
	// 开始一个事务，关闭自动提交
	$dbh_word_index->beginTransaction();
	$query="INSERT INTO wordindex ('id','word','word_en','count','normal','bold','is_base','len') VALUES ( ? , ? , ? , ? , ? , ? , ? , ? )";
	$stmt = $dbh_word_index->prepare($query);
	
	echo "INSERT:".count($aNewWordIndex)."words<br>";
	foreach($aNewWordIndex as $wIndex => $info){
		$wordindex=$iword;
		$newWord=array(
								$wIndex,
								$info[0],
								getWordEn($info[0]),
								$info[1],
								$info[2],
								$info[3],
								0,
								mb_strlen($info[0],"UTF-8")
							);
		$stmt->execute($newWord);
	}
	// 提交更改 
	$dbh_word_index->commit();
	if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
		$error = $dbh_word_index->errorInfo();
		echo "error - $error[2] <br>";
		$log.="$from, $FileName, error, $error[2] \r\n";
	}
	else{
		echo "updata iword recorders.<br />";
		$log.="updata iword recorders.\r\n";
	}
	

//然后修改已经有的词
	// 开始一个事务，关闭自动提交
	$dbh_word_index->beginTransaction();
	$query="UPDATE wordindex SET count = ? , normal = ? , bold = ?   where  id = ?  ";
	$stmt = $dbh_word_index->prepare($query);
	
	echo "UPDATE:".count($iAllWordIndex)."words<br>";
	foreach($iAllWordIndex as $wIndex => $info){
		$wordindex=$iword;
		$newWord=array(
								$info[1],
								$info[2],
								$info[3],
								$wIndex
							);
		$stmt->execute($newWord);
	}
	// 提交更改 
	$dbh_word_index->commit();
	if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
		$error = $dbh_word_index->errorInfo();
		echo "error - $error[2] <br>";
		$log.="$from, $FileName, error, $error[2] \r\n";
	}
	else{
		echo "updata iword recorders.<br />";
		$log.="updata iword recorders.\r\n";
	}


	$myLogFile = fopen($dirLog."insert_index.log", "a");
	fwrite($myLogFile, $log);
	fclose($myLogFile);
	

?>


<?php 

if($from>=$to){
	echo "<h2>齐活！功德无量！all done!</h2>";
}
else{
	echo "<script>";
	echo "window.location.assign(\"db_insert_index.php?from=".($from+1)."&to=".$to."\")";
	echo "</script>";
	echo "正在载入:".($from+1)."——".$filelist[$from+1][0];
}
?>
</body>
</html>