| 1 |
yamaji |
1.1 |
<?php |
| 2 |
|
|
// $Revision: 1.8 $ // |
| 3 |
|
|
// ------------------------------------------------------------------------ // |
| 4 |
|
|
// // |
| 5 |
|
|
// OAI-PMH based Institutional Repository Introduction Module // |
| 6 |
|
|
// Copyright (C) 2005 Kazutsuna YAMAJI. All rights reserved. // |
| 7 |
|
|
// // |
| 8 |
|
|
// ------------------------------------------------------------------------ // |
| 9 |
|
|
// This program is free software; you can redistribute it and/or modify // |
| 10 |
|
|
// it under the terms of the GNU General Public License as published by // |
| 11 |
|
|
// the Free Software Foundation; either version 2 of the License, or // |
| 12 |
|
|
// (at your option) any later version. // |
| 13 |
|
|
// // |
| 14 |
|
|
// You may not change or alter any portion of this comment or credits // |
| 15 |
|
|
// of supporting developers from this source code or any supporting // |
| 16 |
|
|
// source code which is considered copyrighted (c) material of the // |
| 17 |
|
|
// original comment or credit authors. // |
| 18 |
|
|
// // |
| 19 |
|
|
// This program is distributed in the hope that it will be useful, // |
| 20 |
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of // |
| 21 |
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the // |
| 22 |
|
|
// GNU General Public License for more details. // |
| 23 |
|
|
// // |
| 24 |
|
|
// You should have received a copy of the GNU General Public License // |
| 25 |
|
|
// along with this program; if not, write to the Free Software // |
| 26 |
|
|
// Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA // |
| 27 |
|
|
// ------------------------------------------------------------------------ // |
| 28 |
|
|
|
| 29 |
|
|
class OAIPMHHarvester { |
| 30 |
|
|
var $dateformat; |
| 31 |
|
|
var $earliestDatestamp; |
| 32 |
|
|
var $baseURL; |
| 33 |
|
|
var $metadataPrefix; |
| 34 |
|
|
var $metadataFormats; |
| 35 |
|
|
var $lastError; |
| 36 |
|
|
var $lastStatus; |
| 37 |
|
|
var $repositoryName; |
| 38 |
|
|
|
| 39 |
|
|
//public |
| 40 |
|
|
function OAIPMHHarvester( $_baseURL ) { |
| 41 |
|
|
$this->baseURL = $_baseURL; |
| 42 |
|
|
$this->lastError = null; |
| 43 |
|
|
$this->lastStatus = null; |
| 44 |
|
|
$this->metadataPrefix = null; |
| 45 |
|
|
$this->metadataFormats = array( ); |
| 46 |
|
|
$this->dateformat = null; |
| 47 |
|
|
$this->earliestDatestamp = null; |
| 48 |
|
|
$this->repositoryName = null; |
| 49 |
|
|
} function __construct( $_baseURL ) { |
| 50 |
|
|
/* constructer for PHP5 */ |
| 51 |
|
|
$this->OAIPMHHarvester( $_baseURL ); |
| 52 |
|
|
} |
| 53 |
|
|
function harvest( ) { |
| 54 |
|
|
global $xoopsDB; |
| 55 |
|
|
|
| 56 |
|
|
//update repositories table |
| 57 |
|
|
$tbl_repositories = $xoopsDB->prefix( 'irview_repositories' ); |
| 58 |
|
|
$sql = "UPDATE ${tbl_repositories} SET last_access_date=".gmdate( "U", time( ) ).", last_access_result=NULL WHERE URL='".addslashes( $this->baseURL )."'"; |
| 59 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 60 |
|
|
if( $result ) { |
| 61 |
|
|
////////////////// |
| 62 |
|
|
// 1st Identify // |
| 63 |
|
|
////////////////// |
| 64 |
|
|
if( $this->Identify( ) ) { |
| 65 |
|
|
// update (or set) repository name |
| 66 |
|
|
$sql = "UPDATE ${tbl_repositories} SET repository_name='".addslashes( $this->repositoryName)."' WHERE URL='".addslashes( $this->baseURL )."'"; |
| 67 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 68 |
|
|
///////////////////////////// |
| 69 |
|
|
// 2nd ListMetadataFormats // |
| 70 |
|
|
///////////////////////////// |
| 71 |
|
|
if( $result && $this->ListMetadataFormats( ) ) { |
| 72 |
|
|
$args = array( 'metadataPrefix' => $this->metadataPrefix ); |
| 73 |
|
|
$sql = "SELECT last_success_date FROM ${tbl_repositories} WHERE URL='".addslashes( $this->baseURL )."'"; |
| 74 |
|
|
$result = $xoopsDB->query( $sql ); |
| 75 |
|
|
if( $result && $xoopsDB->getRowsNum( $result ) > 0 ) { |
| 76 |
|
|
list( $from_timestamp ) = $xoopsDB->fetchRow( $result ); |
| 77 |
|
|
if( $from_timestamp == NULL && !is_null( $this->earliestDatestamp ) ){ |
| 78 |
|
|
$from_timestamp = $this->earliestDatestamp; |
| 79 |
|
|
} |
| 80 |
|
|
$args['from'] = gmdate( $this->dateformat, $from_timestamp ); |
| 81 |
|
|
//$to_timestamp = ISO8601toUTC( ( string )( gmdate( "Y", $from_timestamp )+1 )."-01-01" ); |
| 82 |
|
|
//$args['until']= date( $this->dateformat, $to_timestamp ); |
| 83 |
|
|
} |
| 84 |
|
|
$sql = "UPDATE ${tbl_repositories} SET metadata_formats='".addslashes(implode($this->metadataFormats, ', '))."' WHERE URL='".addslashes( $this->baseURL )."'"; |
| 85 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 86 |
|
|
///////////////////// |
| 87 |
|
|
// 3rd ListRecords // |
| 88 |
|
|
///////////////////// |
| 89 |
|
|
if( $result && $this->ListRecords( $args ) ) { |
| 90 |
|
|
//update repositories table |
| 91 |
|
|
$tbl_repositories = $xoopsDB->prefix( 'irview_repositories' ); |
| 92 |
|
|
//$success_date = min( (int)ISO8601toUTC($args['until']), (int)gmdate( "U", time( ) ) ); |
| 93 |
|
|
$success_date = (int)gmdate( "U", time( ) ); |
| 94 |
|
|
$sql = "UPDATE ${tbl_repositories} SET last_success_date=".$success_date.", last_access_result='".addslashes( $this->lastStatus )."' WHERE URL='".addslashes( $this->baseURL )."'"; |
| 95 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 96 |
|
|
if( $result ) |
| 97 |
|
|
return true; |
| 98 |
|
|
else |
| 99 |
|
|
$this->lastError = $xoopsDB->error( ); |
| 100 |
|
|
} |
| 101 |
|
|
} |
| 102 |
|
|
} |
| 103 |
|
|
} else |
| 104 |
|
|
$this->lastError = $xoopsDB->error( ); |
| 105 |
|
|
|
| 106 |
|
|
//update repositories table(last_access_result) |
| 107 |
|
|
$tbl_repositories = $xoopsDB->prefix( 'irview_repositories' ); |
| 108 |
|
|
$sql = "UPDATE ${tbl_repositories} SET last_access_result='".addslashes( $this->lastError )."' WHERE URL='".addslashes( $this->baseURL )."'"; |
| 109 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 110 |
|
|
return false; |
| 111 |
|
|
} |
| 112 |
|
|
function last_error( ) { |
| 113 |
|
|
return $this->lastError; |
| 114 |
|
|
} |
| 115 |
|
|
|
| 116 |
|
|
//private |
| 117 |
|
|
function Identify( ) { |
| 118 |
|
|
$data = parse_url($this->baseURL."?verb=Identify"); |
| 119 |
|
|
$host = $data["host"]; |
| 120 |
|
|
$path = $data["path"]; |
| 121 |
|
|
$query= $data["query"]; |
| 122 |
|
|
$sock = fsockopen($host, 80, $errno, $errmsg, 100); |
| 123 |
|
|
$request = "GET {$path}?{$query} HTTP/1.0\r\n" |
| 124 |
|
|
. "Accept-Charset: UTF-8\r\n" |
| 125 |
|
|
. "Connetction: Keep-Alive\r\n" |
| 126 |
|
|
. "Host: {$host}\r\n" |
| 127 |
|
|
. "\r\n"; |
| 128 |
|
|
fwrite($sock, $request); |
| 129 |
|
|
|
| 130 |
|
|
if( !$sock ) { |
| 131 |
|
|
$this->lastError = "can't retrieve ".$this->baseURL."verb=Identify"; |
| 132 |
|
|
return false; |
| 133 |
|
|
} |
| 134 |
|
|
$metadata = ""; |
| 135 |
|
|
$this->lastError = $http_status = fgets($sock); |
| 136 |
|
|
do{ |
| 137 |
|
|
$data = fgets($sock); |
| 138 |
|
|
$metadata .= $data; |
| 139 |
|
|
}while($data != "\r\n"); |
| 140 |
|
|
|
| 141 |
|
|
$this->parser = xml_parser_create( 'UTF-8' ); |
| 142 |
|
|
if( !$this->parser ) { |
| 143 |
|
|
fclose( $sock ); |
| 144 |
|
|
$this->lastError = "can't create XML parser"; |
| 145 |
|
|
return false; |
| 146 |
|
|
} |
| 147 |
|
|
$handler = new IdentifyHandler( $this->parser ); |
| 148 |
|
|
|
| 149 |
|
|
$result = $this->parse( $sock ); |
| 150 |
|
|
xml_parser_free( $this->parser ); |
| 151 |
|
|
fclose( $sock ); |
| 152 |
|
|
if( !$result ) |
| 153 |
|
|
return false; //some error has occured in parse( $sock ); |
| 154 |
|
|
|
| 155 |
|
|
$this->dateformat = $handler->getDateFormat( ); |
| 156 |
|
|
if( !$this->dateformat ) { |
| 157 |
|
|
$this->lastError = "value of <granularity> is wrong"; |
| 158 |
|
|
return false; |
| 159 |
|
|
} |
| 160 |
|
|
|
| 161 |
|
|
$this->earliestDatestamp = $handler->getEarliestDatestamp( ); |
| 162 |
|
|
$this->repositoryName = $handler->getRepositoryName( ); |
| 163 |
|
|
|
| 164 |
|
|
return true; |
| 165 |
|
|
} |
| 166 |
|
|
|
| 167 |
|
|
function ListMetadataFormats( ) { |
| 168 |
|
|
$data = parse_url($this->baseURL."?verb=ListMetadataFormats"); |
| 169 |
|
|
$host = $data["host"]; |
| 170 |
|
|
$path = $data["path"]; |
| 171 |
|
|
$query= $data["query"]; |
| 172 |
|
|
$sock = fsockopen($host, 80, $errno, $errmsg, 100); |
| 173 |
|
|
$request = "GET {$path}?{$query} HTTP/1.0\r\n" |
| 174 |
|
|
. "Accept-Charset: UTF-8\r\n" |
| 175 |
|
|
. "Connetction: Keep-Alive\r\n" |
| 176 |
|
|
. "Host: {$host}\r\n" |
| 177 |
|
|
. "\r\n"; |
| 178 |
|
|
fwrite($sock, $request); |
| 179 |
|
|
|
| 180 |
|
|
if( !$sock ) { |
| 181 |
|
|
$this->lastError = "can't retrieve ".$this->baseURL."verb=ListMetadataFormats"; |
| 182 |
|
|
return false; |
| 183 |
|
|
} |
| 184 |
|
|
$metadata = ""; |
| 185 |
|
|
$this->lastError = $http_status = fgets($sock); |
| 186 |
|
|
do{ |
| 187 |
|
|
$data = fgets($sock); |
| 188 |
|
|
$metadata .= $data; |
| 189 |
|
|
}while($data != "\r\n"); |
| 190 |
|
|
|
| 191 |
|
|
$this->parser = xml_parser_create( 'UTF-8' ); |
| 192 |
|
|
if( !$this->parser ) { |
| 193 |
|
|
fclose( $sock ); |
| 194 |
|
|
$this->lastError = "can't create XML parser"; |
| 195 |
|
|
return false; |
| 196 |
|
|
} |
| 197 |
|
|
$handler = new ListMetadataFormatsHandler( $this->parser ); |
| 198 |
|
|
|
| 199 |
|
|
$result = $this->parse( $sock ); |
| 200 |
|
|
xml_parser_free( $this->parser ); |
| 201 |
|
|
fclose( $sock ); |
| 202 |
|
|
if( !$result ) |
| 203 |
|
|
return false; //some error has occured in parse( $sock ); |
| 204 |
|
|
|
| 205 |
|
|
$this->metadataPrefix = $handler->getMetadataPrefix( ); |
| 206 |
|
|
$this->metadataFormats = $handler->getMetadataFormats( ); |
| 207 |
|
|
if( !$this->metadataPrefix && !$this->metadataFormats) { |
| 208 |
|
|
$this->metadataPrefix = null; |
| 209 |
|
|
$this->lastError = "can't retrieve <metadataPrefix>"; |
| 210 |
|
|
return false; |
| 211 |
|
|
} |
| 212 |
|
|
return true; |
| 213 |
|
|
} |
| 214 |
|
|
|
| 215 |
|
|
function ListRecords( $args ) { |
| 216 |
|
|
global $xoopsDB; |
| 217 |
|
|
if( !isset( $args['metadataPrefix'] ) ) { |
| 218 |
|
|
$this->lastError = "'metadataPrefix' is not specified."; |
| 219 |
|
|
return false; |
| 220 |
|
|
} |
| 221 |
|
|
$resumptionToken = null; |
| 222 |
|
|
$resumptionToken_num = 0; |
| 223 |
|
|
do { |
| 224 |
|
|
$url = $this->baseURL."?verb=ListRecords"; |
| 225 |
|
|
if( $resumptionToken == null ){ |
| 226 |
|
|
foreach( array( 'metadataPrefix', 'from', 'until', 'set' ) as $k ) { |
| 227 |
|
|
if( isset( $args[$k] ) ) { |
| 228 |
|
|
$url .= "&".urlencode( $k )."=".urlencode( $args[$k] ); |
| 229 |
|
|
} |
| 230 |
|
|
} |
| 231 |
|
|
} else { |
| 232 |
|
|
$url .= "&resumptionToken=".htmlspecialchars( $resumptionToken ); |
| 233 |
|
|
} |
| 234 |
|
|
|
| 235 |
|
|
$data = parse_url($url); |
| 236 |
|
|
$host = $data["host"]; |
| 237 |
|
|
$path = $data["path"]; |
| 238 |
|
|
$query= $data["query"]; |
| 239 |
|
|
$sock = fsockopen( $host, 80, $errno, $errmsg, 100 ); |
| 240 |
|
|
$request = "GET {$path}?{$query} HTTP/1.0\r\n" |
| 241 |
|
|
. "Accept-Charset: UTF-8\r\n" |
| 242 |
|
|
. "Connetction: Keep-Alive\r\n" |
| 243 |
|
|
. "Host: {$host}\r\n" |
| 244 |
|
|
. "\r\n"; |
| 245 |
|
|
fwrite($sock, $request); |
| 246 |
|
|
|
| 247 |
|
|
if( !$sock ) { |
| 248 |
|
|
$this->lastError = "can't retrieve ${url}"; |
| 249 |
|
|
$xoopsDB -> setLogger( XoopsLogger::instance() ); |
| 250 |
|
|
return false; |
| 251 |
|
|
} |
| 252 |
|
|
$metadata = ""; |
| 253 |
|
|
$this->lastError = $this->lastStatus = fgets($sock); |
| 254 |
|
|
do{ |
| 255 |
|
|
$data = fgets($sock); |
| 256 |
|
|
$metadata .= $data; |
| 257 |
|
|
}while($data != "\r\n"); |
| 258 |
|
|
|
| 259 |
|
|
$this->parser = xml_parser_create( 'UTF-8' ); |
| 260 |
|
|
if( !$this->parser ) { |
| 261 |
|
|
fclose( $sock ); |
| 262 |
|
|
$this->lastError = "can't create XML parser"; |
| 263 |
|
|
$xoopsDB -> setLogger( XoopsLogger::instance() ); |
| 264 |
|
|
return false; |
| 265 |
|
|
} |
| 266 |
|
|
|
| 267 |
|
|
$handler = "${resumptionToken}_handler"; |
| 268 |
|
|
$${'handler'} = new ListRecordsHandler( $this->parser, $this->baseURL, $args['metadataPrefix'] ); |
| 269 |
|
|
|
| 270 |
|
|
$result = $this->parse( $sock ); |
| 271 |
|
|
xml_parser_free( $this->parser ); |
| 272 |
|
|
fclose( $sock ); |
| 273 |
|
|
if( !$result ) { |
| 274 |
|
|
//some erorr has occured |
| 275 |
|
|
if( $${'handler'}->getIdentifier( ) != null ) |
| 276 |
|
|
$this->lastError .= "[identifier]".$${'handler'}->getIdentifier( ); |
| 277 |
|
|
$xoopsDB -> setLogger( XoopsLogger::instance() ); |
| 278 |
|
|
return false; //some error has occured in parse( $sock ); |
| 279 |
|
|
} |
| 280 |
|
|
$resumptionToken = $${'handler'}->getResumptionToken( ); |
| 281 |
|
|
$resumptionToken_num ++; |
| 282 |
|
|
} while( $resumptionToken != null ); |
| 283 |
|
|
if( $resumptionToken_num-1 < 2 ){ |
| 284 |
|
|
echo "resumptionToken\t".($resumptionToken_num-1)." time issued<br>\n"; |
| 285 |
|
|
}else{ |
| 286 |
|
|
echo "resumptionToken\t".($resumptionToken_num-1)." times issued<br>\n"; |
| 287 |
|
|
} |
| 288 |
|
|
$xoopsDB -> setLogger( XoopsLogger::instance() ); |
| 289 |
|
|
return true; |
| 290 |
|
|
} |
| 291 |
|
|
|
| 292 |
|
|
function parse( $sock ) { |
| 293 |
|
|
while( !feof( $sock ) ) { |
| 294 |
|
|
$data = fread( $sock, 8192 ); |
| 295 |
|
|
|
| 296 |
|
|
if( !xml_parse( $this->parser, $data ) ) { |
| 297 |
|
|
$this->lastError = "[XMLParser]".xml_error_string( xml_get_error_code( $this->parser ) )." at line ".xml_get_current_line_number( $this->parser ).", column ".xml_get_current_column_number( $this->parser ); |
| 298 |
|
|
|
| 299 |
|
|
fclose( $sock ); |
| 300 |
|
|
xml_parser_free( $this->parser ); |
| 301 |
|
|
return false; |
| 302 |
|
|
} |
| 303 |
|
|
} |
| 304 |
|
|
return true; |
| 305 |
|
|
} |
| 306 |
|
|
} |
| 307 |
|
|
|
| 308 |
|
|
class HarvesterHandler { |
| 309 |
|
|
var $parser; |
| 310 |
|
|
var $lastError; |
| 311 |
|
|
|
| 312 |
|
|
function HarvesterHandler( $_parser ) { |
| 313 |
|
|
$this->lastError = null; |
| 314 |
|
|
$this->parser = $_parser; |
| 315 |
|
|
xml_set_object( $this->parser, &$this ); |
| 316 |
|
|
xml_set_element_handler( $this->parser, 'startElementHandler', 'endElementHandler' ); |
| 317 |
|
|
xml_set_character_data_handler( $this->parser, 'characterDataHandler' ); |
| 318 |
|
|
} function __construct( $_parser ) { |
| 319 |
|
|
$this->HarvesterHandler( $_parser ); |
| 320 |
|
|
} |
| 321 |
|
|
function startElementHandler( $parser, $name, $attribs ) { |
| 322 |
|
|
} |
| 323 |
|
|
function endElementHandler( $parser, $name ) { |
| 324 |
|
|
} |
| 325 |
|
|
function characterDataHandler( $parser, $data ) { |
| 326 |
|
|
} |
| 327 |
|
|
|
| 328 |
|
|
function last_error( ) { |
| 329 |
|
|
return $this->lastError; |
| 330 |
|
|
} |
| 331 |
|
|
} |
| 332 |
|
|
|
| 333 |
|
|
class IdentifyHandler extends HarvesterHandler { |
| 334 |
|
|
var $dateformat; |
| 335 |
|
|
var $earliestDatestamp; |
| 336 |
|
|
var $tagstack; |
| 337 |
|
|
var $repositoryName; |
| 338 |
|
|
function IdentifyHandler( $_parser ) { |
| 339 |
|
|
parent::HarvesterHandler( $_parser ); |
| 340 |
|
|
$this->earliestDatestamp = null; |
| 341 |
|
|
$this->dateformat = null; |
| 342 |
|
|
$this->tagstack = array( ); |
| 343 |
|
|
$this->repositoryName = null; |
| 344 |
|
|
} function __construct( $_parser ) { |
| 345 |
|
|
$this->IdentifyHandler( $_parser ); |
| 346 |
|
|
} |
| 347 |
|
|
|
| 348 |
|
|
function startElementHandler( $parser, $name, $attribs ) { |
| 349 |
|
|
array_push( $this->tagstack, $name ); |
| 350 |
|
|
} |
| 351 |
|
|
|
| 352 |
|
|
function endElementHandler( $parser, $name ) { |
| 353 |
|
|
array_pop( $this->tagstack ); |
| 354 |
|
|
} |
| 355 |
|
|
|
| 356 |
|
|
function characterDataHandler( $parser, $data ) { |
| 357 |
|
|
if( end( $this->tagstack ) == 'GRANULARITY' ) { |
| 358 |
|
|
if( $data == "YYYY-MM-DDThh:mm:ssZ" ) |
| 359 |
|
|
$this->dateformat = "Y-m-d\TH:i:s\Z"; |
| 360 |
|
|
else if( $data == "YYYY-MM-DD" ) |
| 361 |
|
|
$this->dateformat = "Y-m-d"; |
| 362 |
|
|
else |
| 363 |
|
|
$this->dateformat = false; |
| 364 |
|
|
} else if( end( $this->tagstack ) == 'EARLIESTDATESTAMP' ) { |
| 365 |
|
|
$this->earliestDatestamp = ISO8601toUTC($data); |
| 366 |
|
|
} else if( end( $this->tagstack ) == 'REPOSITORYNAME' ) { |
| 367 |
|
|
// Set repository name to reponame |
| 368 |
|
|
$this->repositoryName = $data; |
| 369 |
|
|
} |
| 370 |
|
|
} |
| 371 |
|
|
|
| 372 |
|
|
function getDateFormat( ) { |
| 373 |
|
|
return $this->dateformat; |
| 374 |
|
|
} |
| 375 |
|
|
function getEarliestDatestamp( ) { |
| 376 |
|
|
return $this->earliestDatestamp; |
| 377 |
|
|
} |
| 378 |
|
|
function getRepositoryName( ) { |
| 379 |
|
|
return $this->repositoryName; |
| 380 |
|
|
} |
| 381 |
|
|
} |
| 382 |
|
|
|
| 383 |
|
|
class ListMetadataFormatsHandler extends HarvesterHandler { |
| 384 |
|
|
var $metadataPrefix; |
| 385 |
|
|
var $tagstack; |
| 386 |
|
|
var $metadataFormats; |
| 387 |
|
|
|
| 388 |
|
|
function ListMetadataFormatsHandler( $_parser ) { |
| 389 |
|
|
parent::HarvesterHandler( $_parser ); |
| 390 |
|
|
|
| 391 |
|
|
$this->metadataPrefix = "oai_dc"; |
| 392 |
|
|
$this->tagstack = array( ); |
| 393 |
|
|
$this->metadataFormats = array( ); |
| 394 |
|
|
} function __construct( $_parser ) { |
| 395 |
|
|
$this->ListMetadataFormatsHandler( $_parser ); |
| 396 |
|
|
} |
| 397 |
|
|
|
| 398 |
|
|
function startElementHandler( $parser, $name, $attribs ) { |
| 399 |
|
|
array_push( $this->tagstack, $name ); |
| 400 |
|
|
} |
| 401 |
|
|
|
| 402 |
|
|
function endElementHandler( $parser, $name ) { |
| 403 |
|
|
array_pop( $this->tagstack ); |
| 404 |
|
|
} |
| 405 |
|
|
|
| 406 |
|
|
function characterDataHandler( $parser, $data ) { |
| 407 |
|
|
if( end( $this->tagstack ) == 'METADATAPREFIX' ) { |
| 408 |
|
|
$this->metadataFormats[] = $data; |
| 409 |
|
|
if( $data == "junii" ) { |
| 410 |
|
|
$this->metadataPrefix = $data; |
| 411 |
|
|
} |
| 412 |
|
|
} |
| 413 |
|
|
} |
| 414 |
|
|
|
| 415 |
|
|
function getMetadataPrefix( ) { |
| 416 |
|
|
return $this->metadataPrefix; |
| 417 |
|
|
} |
| 418 |
|
|
function getMetadataFormats( ) { |
| 419 |
|
|
return $this->metadataFormats; |
| 420 |
|
|
} |
| 421 |
|
|
} |
| 422 |
|
|
|
| 423 |
|
|
class ListRecordsHandler extends HarvesterHandler { |
| 424 |
|
|
var $resumptionToken; |
| 425 |
|
|
var $identifier; |
| 426 |
|
|
var $title; |
| 427 |
|
|
var $metadata; |
| 428 |
|
|
var $metadataPrefix; |
| 429 |
|
|
var $search_text; |
| 430 |
|
|
var $datastamp; |
| 431 |
|
|
var $baseURL; |
| 432 |
|
|
var $delete_flag; |
| 433 |
|
|
|
| 434 |
|
|
function ListRecordsHandler( $_parser, $_baseURL, $_metadataPrefix ) { |
| 435 |
|
|
parent::HarvesterHandler( $_parser ); |
| 436 |
|
|
|
| 437 |
|
|
$this->resumptionToken = null; |
| 438 |
|
|
$this->identifier = null; |
| 439 |
|
|
$this->title = null; |
| 440 |
|
|
$this->metadata = array( ); |
| 441 |
|
|
$this->metadataPrefix = $_metadataPrefix; |
| 442 |
|
|
$this->search_text = array( ); |
| 443 |
|
|
$this->datestamp = array( ); |
| 444 |
|
|
$this->tagstack = array( ); |
| 445 |
|
|
$this->baseURL = $_baseURL; |
| 446 |
|
|
$this->delete_flag = false; |
| 447 |
|
|
} function __construct( $_parser, $_baseURL, $_metadataPrefix ) { |
| 448 |
|
|
$this->ListRecordsHandler( $_parser, $_baseURL, $_metadataPrefix ); |
| 449 |
|
|
} |
| 450 |
|
|
|
| 451 |
|
|
function startElementHandler( $parser, $name, $attribs ) { |
| 452 |
|
|
array_push( $this->tagstack, $name ); |
| 453 |
|
|
if( $name == 'RECORD' ){ |
| 454 |
|
|
global $xoopsDB; |
| 455 |
|
|
$xoopsDB -> setLogger( new XoopsLogger() ); |
| 456 |
|
|
|
| 457 |
|
|
// initialize following value for each records |
| 458 |
|
|
$this->title = null; |
| 459 |
|
|
$this->metadata = array( ); |
| 460 |
|
|
$this->search_text = array( ); |
| 461 |
|
|
$this->datestamp = null; |
| 462 |
|
|
} |
| 463 |
|
|
if( $name == 'HEADER' ) { |
| 464 |
|
|
if( isset( $attribs['STATUS'] ) |
| 465 |
|
|
&& $attribs['STATUS'] == 'deleted' ) { |
| 466 |
|
|
$this->delete_flag = true; |
| 467 |
|
|
} else { |
| 468 |
|
|
$this->delete_flag = false; |
| 469 |
|
|
} |
| 470 |
|
|
} |
| 471 |
|
|
if( !isset( $this->tagstack[2] ) || $this->tagstack[2] != 'RECORD' ) |
| 472 |
|
|
return; |
| 473 |
|
|
|
| 474 |
|
|
array_push( $this->metadata, "<".strtolower( $name )."" ); |
| 475 |
|
|
foreach( $attribs as $k => $v ) { |
| 476 |
|
|
array_push( $this->metadata, " ".strtolower($k)."='${v}'" ); |
| 477 |
|
|
} |
| 478 |
|
|
array_push( $this->metadata, ">" ); |
| 479 |
|
|
} |
| 480 |
|
|
|
| 481 |
|
|
function endElementHandler( $parser, $name ) { |
| 482 |
|
|
array_push( $this->search_text, ' ' ); |
| 483 |
|
|
if( isset( $this->tagstack[2] ) && $this->tagstack[2] == 'RECORD' ) |
| 484 |
|
|
array_push( $this->metadata, "</".strtolower( $name ).">" ); |
| 485 |
|
|
if( $name == 'RECORD' ) { |
| 486 |
|
|
global $xoopsDB; |
| 487 |
|
|
|
| 488 |
|
|
$tbl_repositories = $xoopsDB->prefix( 'irview_repositories' ); |
| 489 |
|
|
$tbl_metadata = $xoopsDB->prefix( 'irview_metadata' ); |
| 490 |
|
|
|
| 491 |
|
|
if( $this->delete_flag ) { |
| 492 |
|
|
//delete a metadata |
| 493 |
|
|
$sql = "DELETE FROM ${tbl_metadata} WHERE identifier='".addslashes( $this->identifier )."'"; |
| 494 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 495 |
|
|
} else { |
| 496 |
|
|
//retrieve a repository id |
| 497 |
|
|
$sql = "SELECT repository_id FROM ${tbl_repositories} WHERE URL='".addslashes( $this->baseURL )."'"; |
| 498 |
|
|
$result = $xoopsDB->query( $sql ); |
| 499 |
|
|
list( $id ) = $xoopsDB->fetchRow( $result ); |
| 500 |
|
|
if( $id ) { |
| 501 |
|
|
$this->search_text = array_unique( $this->search_text ); |
| 502 |
|
|
array_walk( $this -> search_text, 'array_walk_bin2hex' ); |
| 503 |
|
|
//insert data into metadata table |
| 504 |
|
|
$values = array( addslashes( $this->identifier ), |
| 505 |
|
|
$id, |
| 506 |
|
|
addslashes( $this->metadataPrefix ), |
| 507 |
|
|
addslashes( $this->title ), |
| 508 |
|
|
addslashes( implode( '', $this->metadata ) ), |
| 509 |
|
|
addslashes( implode( ' ', $this->search_text ) ), |
| 510 |
|
|
addslashes( $this->datestamp ) ); |
| 511 |
|
|
$sql = "INSERT DELAYED INTO ${tbl_metadata} (identifier, repository_id, format, title, metadata, search_text, datestamp) VALUES ('".implode( "', '", $values )."')"; |
| 512 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 513 |
|
|
if( !$result && $xoopsDB->errno( ) == 1062 ) { //1062: Can't write; duplicate key in table '%s' |
| 514 |
|
|
$update = array( 'title' => addslashes( $this->title ), |
| 515 |
|
|
'format' => addslashes( $this->metadataPrefix ), |
| 516 |
|
|
'metadata' => addslashes( implode( '', $this->metadata ) ), |
| 517 |
|
|
'search_text' => addslashes( implode( ' ', $this->search_text ) ), |
| 518 |
|
|
'datestamp' => addslashes( $this->datestamp ) ); |
| 519 |
|
|
$sql = "UPDATE ${tbl_metadata} SET "; |
| 520 |
|
|
$update2 = array( ); |
| 521 |
|
|
foreach( $update as $k => $v ) $update2[] = "${k}='${v}'"; |
| 522 |
|
|
$sql .= implode( ', ', $update2 ); |
| 523 |
|
|
$sql .= " WHERE identifier='".addslashes( $this->identifier )."'"; |
| 524 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 525 |
|
|
} |
| 526 |
|
|
} |
| 527 |
|
|
} |
| 528 |
|
|
$this->metadata = array( ); |
| 529 |
|
|
$this->search_text = array( ); |
| 530 |
|
|
} |
| 531 |
|
|
array_pop( $this->tagstack ); |
| 532 |
|
|
} |
| 533 |
|
|
|
| 534 |
|
|
function characterDataHandler( $parser, $data ) { |
| 535 |
|
|
if( end( $this->tagstack ) == 'RESUMPTIONTOKEN' ) { |
| 536 |
|
|
$this->resumptionToken = $data; |
| 537 |
|
|
return; |
| 538 |
|
|
} |
| 539 |
|
|
if( end( $this->tagstack ) == 'IDENTIFIER' && prev( $this->tagstack ) == 'HEADER' ) { |
| 540 |
|
|
$this->identifier = $data; |
| 541 |
|
|
echo "${data}<br>\n"; |
| 542 |
|
|
} |
| 543 |
|
|
|
| 544 |
|
|
//<xxx:title> -> <title> |
| 545 |
|
|
$tmp = split( ':', end( $this->tagstack ) ); |
| 546 |
|
|
if( end( $tmp ) == 'TITLE' ) { |
| 547 |
|
|
$this->title .= $data; |
| 548 |
|
|
} |
| 549 |
|
|
|
| 550 |
|
|
if( !isset( $this->tagstack[2] ) || $this->tagstack[2] != 'RECORD' ) |
| 551 |
|
|
return; |
| 552 |
|
|
array_push( $this->metadata, $data ); |
| 553 |
|
|
array_push( $this->search_text, $data ); |
| 554 |
|
|
$this->search_text = array_merge( $this->search_text, $this -> word_separation( $data ) ); |
| 555 |
|
|
|
| 556 |
|
|
// Extracting the datestamp data from the XML |
| 557 |
|
|
if( end( $this->tagstack ) == 'DATESTAMP' && prev( $this->tagstack ) == 'HEADER' ) { |
| 558 |
|
|
$this->datestamp = ISO8601toUTC($data); |
| 559 |
|
|
} |
| 560 |
|
|
} |
| 561 |
|
|
|
| 562 |
|
|
function getResumptionToken( ) { |
| 563 |
|
|
return $this->resumptionToken; |
| 564 |
|
|
} |
| 565 |
|
|
function getIdentifier( ) { |
| 566 |
|
|
return $this->identifier; |
| 567 |
|
|
} |
| 568 |
|
|
|
| 569 |
|
|
//private |
| 570 |
|
|
function utf8bytes( $c1 ){ |
| 571 |
|
|
if( ( $c1 & 0x80 ) == 0 ) return 1; |
| 572 |
|
|
if( ( $c1 & 0xe0 ) == 0xc0 ) return 2; |
| 573 |
|
|
if( ( $c1 & 0xe0 ) == 0xe0 ) return 3; |
| 574 |
|
|
if( ( $c1 & 0xf8 ) == 0xf0 ) return 4; |
| 575 |
|
|
return 0; |
| 576 |
|
|
} |
| 577 |
|
|
|
| 578 |
|
|
function window( $bytes ){ |
| 579 |
|
|
$length = 8; |
| 580 |
|
|
$ret = array(); |
| 581 |
|
|
$word = array(); |
| 582 |
|
|
$len = 0; |
| 583 |
|
|
|
| 584 |
|
|
for( $len = 0; $len < $length && count( $bytes ) > 0 ; $len++ ){ |
| 585 |
|
|
$c1 = reset( $bytes ); |
| 586 |
|
|
for( $i = 0; $i < $this -> utf8bytes( $c1 ); $i++ ){ |
| 587 |
|
|
array_push( $word, pack( "C", array_shift( $bytes ) ) ); |
| 588 |
|
|
} |
| 589 |
|
|
} |
| 590 |
|
|
while( true ){ |
| 591 |
|
|
array_push( $ret, implode( '', $word ) ); |
| 592 |
|
|
if( count( $bytes ) > 0 && $len >= $length || $len >= 3 ){ |
| 593 |
|
|
$tmp = unpack( "C", reset( $word ) ); |
| 594 |
|
|
$c1 = reset( $tmp ); |
| 595 |
|
|
for( $i = 0; $i < $this -> utf8bytes( $c1 ); $i++ ){ |
| 596 |
|
|
array_shift( $word ); |
| 597 |
|
|
} |
| 598 |
|
|
$len--; |
| 599 |
|
|
}else break; |
| 600 |
|
|
if( count( $bytes ) > 0 ){ |
| 601 |
|
|
$c1 = reset( $bytes ); |
| 602 |
|
|
for( $i = 0; $i < $this -> utf8bytes( $c1 ); $i++ ){ |
| 603 |
|
|
array_push( $word, pack( "C", array_shift( $bytes ) ) ); |
| 604 |
|
|
} |
| 605 |
|
|
$len++; |
| 606 |
|
|
} |
| 607 |
|
|
} |
| 608 |
|
|
return $ret; |
| 609 |
|
|
} |
| 610 |
|
|
|
| 611 |
|
|
/** |
| 612 |
|
|
* |
| 613 |
|
|
* �������������������������� |
| 614 |
|
|
* ��������������������������������8�������������������������������� |
| 615 |
|
|
* 1�������������������������� |
| 616 |
|
|
* |
| 617 |
|
|
*/ |
| 618 |
|
|
function word_separation( $text ){ |
| 619 |
|
|
$multi = array(); |
| 620 |
|
|
$bytes = unpack( "C*", $text ); |
| 621 |
|
|
$c1 = 0; |
| 622 |
|
|
$words = array(); |
| 623 |
|
|
while( count( $bytes ) >= 0 ){ |
| 624 |
|
|
$tmp = array(); |
| 625 |
|
|
if( count( $multi ) > 0 && ( $c1 & 0x80 ) != 0 && ( reset( $bytes ) & 0x80 ) == 0 ){ |
| 626 |
|
|
$words = array_merge( $words, $this -> window( $multi ) ); |
| 627 |
|
|
$multi = array(); |
| 628 |
|
|
}else if( count( $multi ) > 0 && ( $c1 & 0x80 ) == 0 && ( reset( $bytes ) & 0x80 ) != 0 ){ |
| 629 |
|
|
$tmp = array(); |
| 630 |
|
|
while( count( $multi ) > 0 ) array_push( $tmp, pack( "C", array_shift( $multi ) ) ); |
| 631 |
|
|
$words = array_merge( $words, explode( ' ', implode( '', $tmp ) ) ); |
| 632 |
|
|
$multi = array(); |
| 633 |
|
|
} |
| 634 |
|
|
if( count( $bytes ) == 0 ) break; |
| 635 |
|
|
$c1 = reset( $bytes ); |
| 636 |
|
|
if( $this -> utf8bytes( $c1 ) == 1 ){ |
| 637 |
|
|
array_push( $multi, array_shift( $bytes ) ); |
| 638 |
|
|
}else{ |
| 639 |
|
|
for( $i = 0; $i < $this -> utf8bytes( $c1 ); $i++ ){ |
| 640 |
|
|
array_push( $multi, array_shift( $bytes ) ); |
| 641 |
|
|
} |
| 642 |
|
|
} |
| 643 |
|
|
} |
| 644 |
|
|
if( count( $multi ) > 0 ){ |
| 645 |
|
|
$tmp = array(); |
| 646 |
|
|
while( count( $multi ) > 0 ) array_push( $tmp, pack( "C", array_shift( $multi ) ) ); |
| 647 |
|
|
$words = array_merge( $words, explode( ' ', implode( '', $tmp ) ) ); |
| 648 |
|
|
} |
| 649 |
|
|
|
| 650 |
|
|
return $words; |
| 651 |
|
|
} |
| 652 |
|
|
|
| 653 |
|
|
} |
| 654 |
|
|
|
| 655 |
|
|
function array_walk_bin2hex( &$item1, $key ){ |
| 656 |
|
|
if ($item1 == '') return; |
| 657 |
|
|
$tmp = unpack( "C", $item1 ); |
| 658 |
|
|
if( ( reset( $tmp ) & 0x80 ) == 0 || $item1 == ' ' || $item1 == "\n" ) return; |
| 659 |
|
|
$item1 = bin2hex( $item1 ); |
| 660 |
|
|
} |
| 661 |
|
|
|
| 662 |
|
|
|
| 663 |
|
|
/** |
| 664 |
|
|
* |
| 665 |
|
|
* resumptionToken������������������������ |
| 666 |
|
|
* |
| 667 |
|
|
* @param resumption_token |
| 668 |
|
|
* @param metadata_prefix |
| 669 |
|
|
* @param verb |
| 670 |
|
|
* @param args verb�������������������������� |
| 671 |
|
|
* @param last_item_id ����������������������������ID�������� |
| 672 |
|
|
* @param limit_row ����resumptionToken�������������������������������� |
| 673 |
|
|
* @param expire_date ����resumptionToken���������� |
| 674 |
|
|
* @param publish_date ����resumptionToken����������(������) |
| 675 |
|
|
* @return ���� |
| 676 |
|
|
* |
| 677 |
|
|
*/ |
| 678 |
|
|
function |
| 679 |
|
|
setResumptionToken( $resumption_token, $metadata_prefix, $verb, $args, $last_item_id, $limit_row, $expire_date, $publish_date = null ) |
| 680 |
|
|
{ |
| 681 |
|
|
global $xoopsDB; |
| 682 |
|
|
|
| 683 |
|
|
if( $publish_date == null ) |
| 684 |
|
|
$publish_date = time( ); |
| 685 |
|
|
$table = $xoopsDB->prefix( "irview_resumption_token" ); |
| 686 |
|
|
$sql = "INSERT INTO ${table} VALUES ('".addslashes( $resumption_token ) |
| 687 |
|
|
."', '${metadata_prefix}', '${verb}', '".addslashes( serialize( $args ) ) |
| 688 |
|
|
."', ${last_item_id}, ${limit_row}, ${publish_date}, ${expire_date} )"; |
| 689 |
|
|
return $xoopsDB->queryF( $sql ); |
| 690 |
|
|
} |
| 691 |
|
|
|
| 692 |
|
|
/** |
| 693 |
|
|
* |
| 694 |
|
|
* resumptionToken������������������ |
| 695 |
|
|
* |
| 696 |
|
|
* @param resumption_token |
| 697 |
|
|
* @return array( |
| 698 |
|
|
* 'metadata_prefix ' => metadataPrefix, |
| 699 |
|
|
* 'verb ' => verb, |
| 700 |
|
|
* 'args' => verb��������������������������, |
| 701 |
|
|
* 'last_item_id' => ����������������������������ID��������, |
| 702 |
|
|
* 'limit_row' => ����resumptionToken�������������������������������� , |
| 703 |
|
|
* 'expire_date' => ����resumptionToken���������� , |
| 704 |
|
|
* 'publish_date' => ����resumptionToken���������� ) |
| 705 |
|
|
* @return false resumptionToken�������������� |
| 706 |
|
|
* |
| 707 |
|
|
*/ |
| 708 |
|
|
function |
| 709 |
|
|
getResumptionToken( $resumption_token ) |
| 710 |
|
|
{ |
| 711 |
|
|
global $xoopsDB; |
| 712 |
|
|
|
| 713 |
|
|
$table = $xoopsDB->prefix( "irview_resumption_token" ); |
| 714 |
|
|
$sql = "SELECT resumption_token, metadata_prefix, args, last_item_id, publish_date, expire_date, verb FROM ${table} WHERE resumption_token=\"".stripslashes( $resumption_token ) |
| 715 |
|
|
.'"'; |
| 716 |
|
|
$result = $xoopsDB->query( $sql ); |
| 717 |
|
|
$ret = $xoopsDB->fetchArray( $result ); |
| 718 |
|
|
$ret['args'] = unserialize( $ret['args'] ); |
| 719 |
|
|
return $ret; |
| 720 |
|
|
} |
| 721 |
|
|
|
| 722 |
|
|
/** |
| 723 |
|
|
* |
| 724 |
|
|
* resumptionToken������������ |
| 725 |
|
|
* �������������������������������������������� |
| 726 |
|
|
* |
| 727 |
|
|
* @param resumption_token ������������������ |
| 728 |
|
|
* |
| 729 |
|
|
*/ |
| 730 |
|
|
function |
| 731 |
|
|
expireResumptionToken( $resumptionToken ) |
| 732 |
|
|
{ |
| 733 |
|
|
global $xoopsDB; |
| 734 |
|
|
|
| 735 |
|
|
$table = $xoopsDB->prefix( "irview_resumption_token" ); |
| 736 |
|
|
$sql = "DELETE FROM ${table} WHERE resumption_token=\"".stripslashes( $resumptionToken ) |
| 737 |
|
|
.'" OR expire_date < UNIX_TIMESTAMP( NOW() )'; |
| 738 |
|
|
$result = $xoopsDB->queryF( $sql ); |
| 739 |
|
|
} |
| 740 |
|
|
|
| 741 |
|
|
/** |
| 742 |
|
|
* |
| 743 |
|
|
* ISO8601������UTC����������. ������������������false������ |
| 744 |
|
|
* Usage: ISO8601toUTC( "2005-08-01T12:00:00Z" ); |
| 745 |
|
|
* Usage: ISO8601toUTC( "2005-08-01" ); |
| 746 |
|
|
* |
| 747 |
|
|
*/ |
| 748 |
|
|
function |
| 749 |
|
|
ISO8601toUTC( $str ) |
| 750 |
|
|
{ |
| 751 |
|
|
if( preg_match( "/([0-9]{4})-([0-9]{2})-([0-9]{2})(T([0-9]{2}):([0-9]{2}):([0-9]{2})Z)?/", $str, $match ) == 0 ) |
| 752 |
|
|
return 0; |
| 753 |
|
|
if( !isset( $match[5] ) ) |
| 754 |
|
|
$match[5] = 0; |
| 755 |
|
|
if( !isset( $match[6] ) ) |
| 756 |
|
|
$match[6] = 0; |
| 757 |
|
|
if( !isset( $match[7] ) ) |
| 758 |
|
|
$match[7] = 0; |
| 759 |
|
|
return gmmktime( $match[5], $match[6], $match[7], $match[2], $match[3], $match[1] ); |
| 760 |
|
|
} |
| 761 |
|
|
|
| 762 |
|
|
|
| 763 |
|
|
?> |