extraction-interface: search/search

annotate search/search_function.php @ 0:b12c99b7c3f0

commit for previous development

author	Zoe Hong <zhong@mpiwg-berlin.mpg.de>
date	Mon, 19 Jan 2015 17:13:49 +0100
parents
children

rev	line source
0 b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	1 <?php
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	2
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	3 include_once('../interface/Lib_mb_utf8.php');
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	4 include_once('../interface/config.php');
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	5
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	6 set_time_limit(0);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	7 ini_set('memory_limit', '-1');
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	8
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	9 $link_mysql = mysql_connect($mysql_server, $mysql_user, $mysql_password);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	10 mysql_query("SET NAMES utf8");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	11
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	12 if (!$link_mysql) {
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	13 die('Could not connect: ' . mysql_error());
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	14 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	15
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	16 $db_selected = mysql_select_db($mysql_database, $link_mysql);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	17 if (!$db_selected) {
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	18 die ('Can\'t use foo : ' . mysql_error());
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	19 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	20 function trimKeyword($keywords){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	21 $token=strtok($keywords,",");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	22 while($token!=false){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	23 $token=trim($token);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	24 $keywordArray[]=$token;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	25 $token=strtok(",");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	26 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	27 return $keywordArray;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	28 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	29
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	30 function search($keywordArray){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	31 $condition="";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	32 foreach($keywordArray as $i=>$keyword){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	33 if($i!=0){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	34 $condition.=" OR ";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	35 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	36 $condition.=" contents.content LIKE '%".$keyword."%' ";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	37 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	38 //books.name AS book_name, books.level1, books.level2, books.period, contents.books_id, contents.line, contents.content
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	39 $query="SELECT books.level1 AS LEVEL1, books.level2 AS LEVEL2,
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	40 books.name AS Name, books.period AS PERIOD,
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	41 books.start_year AS 'TimeSpan:begin', books.end_year AS 'TimeSpan:end',
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	42 books.id AS BOOK_ID, contents.line AS PAGE, contents.content AS CONTENT,
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	43 info.volume AS VOLUME, info.author AS AUTHOR, info.edition AS EDITION
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	44 FROM contents
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	45 JOIN books ON contents.books_id = books.id
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	46 JOIN books_info info ON contents.books_id=info.books_id
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	47 WHERE ".$condition."
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	48 ORDER BY contents.books_id, contents.line";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	49 //WHERE contents.content LIKE '%".$keyword."%'
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	50 $result = mysql_query($query);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	51 //echo "result length: ".mysql_num_rows($result)."<br>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	52 while ($row = mysql_fetch_assoc($result)) { //Find the section(s) where the page belongs to
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	53 //$resultArray[$i]=$row;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	54 $subQuery="SELECT id, name, start_page, end_page
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	55 FROM sections
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	56 WHERE books_id=".$row['BOOK_ID']." AND start_page<=".$row['PAGE']." AND end_page>=".$row['PAGE'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	57 $subResult=mysql_query($subQuery);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	58 $sectionArray=array();
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	59 while($subRow=mysql_fetch_assoc($subResult)){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	60 $sectionArray[]=$subRow;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	61 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	62 $row['SECTION']=$sectionArray;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	63 $resultArray[]=$row;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	64 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	65 //echo "array length: ".sizeof($resultArray)."<br>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	66 return $resultArray;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	67 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	68 $NO_TAG=0;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	69 $LOCUST_TEMPLE=1;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	70
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	71 function printTable($array,$keywordArray,$filename,$tag){ //print HTML
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	72 global $NO_TAG, $LOCUST_TEMPLE;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	73 $header='<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	74 <html>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	75 <head>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	76 <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	77 <link href="../search.css" type="text/css" rel="stylesheet"/>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	78 <script src="http://ajax.googleapis.com/ajax/libs/jquery/1.6.2/jquery.min.js" type="text/javascript"></script>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	79 <script src="../search.js" charset="utf-8"></script>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	80 </head>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	81 <body>';
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	82 $keywordStr="";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	83 foreach($keywordArray as $i=>$keyword){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	84 if($i!=0){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	85 $keywordStr.=", ";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	86 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	87 $keywordStr.=$keyword;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	88 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	89
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	90
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	91 $fp=fopen("search_results/".$filename.".html","w");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	92 fwrite($fp,$header);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	93 //$str="<a href='http://".$_SERVER['HTTP_HOST']."/map/map.php?mode=1&file=".$filename.".csv&name=".$keywordStr."' target='_blank'>view the distribution on the map</a><br><br>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	94 $str="<a href='http://".$_SERVER['HTTP_HOST']."/map/get_coordinates_for_listed_books.php?file=".$filename.".csv&name=".$keywordStr."' target='_blank'>view the distribution on the map</a><br><br>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	95 fwrite($fp,$str);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	96 $str.="<a href='http://".$_SERVER['HTTP_HOST']."/search/search_results/".$filename.".html' target='_blank'>html version</a><br><br>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	97 echo $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	98 echo sizeof($array)." result(s) of \"".$keywordStr."\" ";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	99 fwrite($fp,sizeof($array)." result(s) of \"".$keywordStr."\"<br>");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	100 $str="<table>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	101 $str.="<tr>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	102 $str.="<td class='sequence'>#<td class='bookId'>book id<td class='bookName'>book name<td class='level1'>level1<td class='level2'>level2<td class='period'>period<td class='sectionName'>section info<td class='page'>page<td class='content'>content";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	103 fwrite($fp,$str);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	104 echo $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	105 $i=1;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	106 foreach($array as $row){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	107 $str="<tr>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	108 $str.="<td>".$i;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	109 $str.="<td>".$row['BOOK_ID'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	110 $str.="<td>".$row['Name'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	111 $str.="<td>".$row['LEVEL1'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	112 $str.="<td>".$row['LEVEL2'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	113 $str.="<td>".$row['PERIOD'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	114 $str.="<td>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	115 fwrite($fp,$str);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	116 echo $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	117 foreach($row['SECTION'] as $section){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	118 $str="<div class='section'><a href='/interface/tagging_text.php?id=".$section['id']."' target='_blank'>".$section['name']."</a>p".$section['start_page']."-".$section['end_page']."</div>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	119 echo $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	120 $str="<div class='section'><a href='/interface/tagging_text.php?id=".$section['id']."' target='_blank'>".$section['name']."</a>p".$section['start_page']."-".$section['end_page']."</div>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	121 fwrite($fp,$str);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	122 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	123 $str="<td>".$row['PAGE'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	124 if($tag==$NO_TAG){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	125 $str.="<td>".$row['CONTENT'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	126 }else if($tag==$LOCUST_TEMPLE){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	127 $str.="<td>".findLocustTempleDescription($row['CONTENT'],$keywordArray,0);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	128 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	129 fwrite($fp,$str);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	130 echo $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	131 $i++;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	132 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	133
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	134 echo "</table>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	135 fwrite($fp,"</table></body></html>");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	136 fclose($fp);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	137 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	138
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	139 function findLocustTempleDescription($str,$keywordArray,$findUnknownTemple){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	140 $result="";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	141 $pattern[0]="/(.*)(";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	142 foreach($keywordArray as $i=>$keyword){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	143 $pattern[0].=$keyword;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	144 if($i!=sizeof($keywordArray)-1){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	145 $pattern[0].="\|";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	146 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	147 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	148 $pattern[0].=")(.*)/u";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	149 //0=whole,1=text,2=locust temple,3=description
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	150 $pattern[1]="/(.\s+)(\S{1,5}[廟寺祠])(.)/u";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	151 //0=whole,1=descripption,2=unknown temple,3=text
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	152 $success=preg_match($pattern[$findUnknownTemple],$str,$match); //note that preg_match only matches the last match!
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	153 if($findUnknownTemple==0){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	154 if($success==1){//find string "locust temple"
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	155 $result=findLocustTempleDescription($match[1],$keywordArray,0);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	156 $result.="<keyword>".$match[2]."</keyword>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	157 $result.=findLocustTempleDescription($match[3],$keywordArray,1);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	158 return $result;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	159 }else{//no string "locust temple"
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	160 return "<irrelevant>".$str."</irrelevant>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	161 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	162 }else{
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	163 if($success==1){//find the pattern "locust temple......unknown temple"
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	164 $result=findLocustTempleDescription($match[1],$keywordArray,1);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	165 $result.="<irrelevant>".$match[2].$match[3]."</irrelevant>";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	166 //$result.="<keyword>".$match[2]."</keyword>".$match[3];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	167 return $result;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	168 }else{//can't find the pattern
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	169 return $str;
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	170 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	171
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	172 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	173 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	174
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	175 function writeCsvFile($array,$fileName){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	176 $columnNameArray=['BOOK_ID','LEVEL1','LEVEL2',
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	177 'Name','PERIOD','TimeSpan:begin','TimeSpan:end','PAGE','SECTION','CONTENT',
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	178 'Description'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	179 $fp=fopen("./csv_files/".$fileName.".csv","w");
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	180 fputcsv($fp,$columnNameArray);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	181 foreach($array as $row){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	182 $book=array();
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	183 $row['Description']='';
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	184 foreach($columnNameArray as $column){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	185 $book[$column]=$row[$column];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	186 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	187 $book['Name']="(".$row['PERIOD'].") ".$row['Name'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	188 $row['AUTHOR']=str_replace("（","(",$row['AUTHOR']);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	189 $row['AUTHOR']=str_replace("）",") ",$row['AUTHOR']);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	190 $book['Description']=$row['VOLUME']." ╱ ".$row['AUTHOR']." ╱ ".$row['EDITION'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	191 $book['SECTION']='';
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	192 foreach($row['SECTION'] as $idx=>$section){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	193 $book['SECTION'].=$section['name']." ".$section['start_page']."-".$section['end_page'];
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	194 if($idx!=sizeof($row['SECTION'])){
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	195 $book['SECTION'].=" ╱ ";
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	196 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	197 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	198 fputcsv($fp,$book);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	199 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	200 fclose($fp);
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	201 }
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	202
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	203 ?>
b12c99b7c3f0 commit for previous development Zoe Hong <zhong@mpiwg-berlin.mpg.de> parents: diff changeset	204

Mercurial > hg > extraction-interface

annotate search/search_function.php @ 0:b12c99b7c3f0