| 1 |
<?php
|
| 2 |
/* $Id: z3950.module,v 1.4 2007/09/01 19:49:13 douggreen Exp $ */
|
| 3 |
|
| 4 |
/**
|
| 5 |
* @file
|
| 6 |
* Lets users search metadata in z39.50 compliant servers.
|
| 7 |
*/
|
| 8 |
|
| 9 |
/**
|
| 10 |
* YAZ configuration defaults
|
| 11 |
* See http://www.loc.gov/marc/bibliographic/ecbdhome.html
|
| 12 |
* See http://www.loc.gov/z3950/lcserver.html
|
| 13 |
*/
|
| 14 |
define('YAZ_DEFAULT_SYNTAX', '1.2.840.10003.5.10');
|
| 15 |
define('YAZ_DEFAULT_SERVER', 'z3950.loc.gov:7090/Voyager'); // jenson.stanford.edu:2210/unicorn
|
| 16 |
|
| 17 |
/**
|
| 18 |
* Implementation of hook_menu()
|
| 19 |
*/
|
| 20 |
function z3950_menu($may_cache = true) {
|
| 21 |
$items = array();
|
| 22 |
$items['z3950'] = array(
|
| 23 |
'access' => 'search in z3950 servers',
|
| 24 |
'page callback' => 'z3950_get'
|
| 25 |
);
|
| 26 |
$items['admin/settings/z3950'] = array(
|
| 27 |
'title' => 'z3950',
|
| 28 |
'description' => 'Set z3950 configuration options',
|
| 29 |
'page callback' => 'drupal_get_form',
|
| 30 |
'page arguments' => 'z3950_admin_settings',
|
| 31 |
'access arguments' => array('administer site configuration'),
|
| 32 |
);
|
| 33 |
return $items;
|
| 34 |
}
|
| 35 |
|
| 36 |
/**
|
| 37 |
* Implementation of hook_perm().
|
| 38 |
*/
|
| 39 |
function z3950_perm() {
|
| 40 |
return array('search in z3950 servers');
|
| 41 |
}
|
| 42 |
|
| 43 |
/**
|
| 44 |
* Implementation of hook_settings().
|
| 45 |
*/
|
| 46 |
function z3950_admin_settings() {
|
| 47 |
if (!_z3950_install_check()) {
|
| 48 |
return;
|
| 49 |
}
|
| 50 |
|
| 51 |
$form = array();
|
| 52 |
$form['z3950_zurls'] = array(
|
| 53 |
'#type' => 'textarea',
|
| 54 |
'#title' => t('z39.50 zURLS'),
|
| 55 |
'#default_value' => variable_get('z3950_zurls', YAZ_DEFAULT_SERVER),
|
| 56 |
'#description' => t('Valid zURLs to perform searches of metadata. Add one zURL per row. zURL is a string that takes the form host[:port][/database]. If port is omitted, port 210 is used. If database is omitted Default is used.'),
|
| 57 |
);
|
| 58 |
$form['z3950_syntax'] = array(
|
| 59 |
'#type' => 'select',
|
| 60 |
'#title' => t('Record Syntax'),
|
| 61 |
'#default_value' => variable_get('z3950_syntax', YAZ_DEFAULT_SYNTAX),
|
| 62 |
'#options' => _z3950_syntax_options(),
|
| 63 |
'#description' => t('Specifies the preferred record syntax for retrieval (<a href="@url">see also</a>)', array('@url' => 'http://www.loc.gov/z3950/agency/defns/oids.html#5')),
|
| 64 |
);
|
| 65 |
return system_settings_form($form);
|
| 66 |
}
|
| 67 |
|
| 68 |
/**
|
| 69 |
* Check to make sure that the yaz functions are installed in php,
|
| 70 |
* and if not, display an error
|
| 71 |
*/
|
| 72 |
function _z3950_install_check() {
|
| 73 |
if (!function_exists('yaz_connect')) {
|
| 74 |
drupal_set_message(t('You must compile Php with <a href="@yazurl">YAZ</a> and enable the YAZ extension in <a href="@phpurl">php.ini</a> file.', array('@yazurl' => 'http://www.php.net/ref.yaz', '@phpurl' => 'http://www.php.net/configuration#configuration.file')), 'error');
|
| 75 |
return FALSE;
|
| 76 |
}
|
| 77 |
return TRUE;
|
| 78 |
}
|
| 79 |
|
| 80 |
/**
|
| 81 |
* Return list of the z39.50 syntax options
|
| 82 |
* See http://www.loc.gov/z3950/agency/defns/oids.html
|
| 83 |
*/
|
| 84 |
function _z3950_syntax_options() {
|
| 85 |
$options = array('' => t('-- Select One --'));
|
| 86 |
foreach (_z3950_load_syntax_files() as $info) {
|
| 87 |
$options += $info['options'];
|
| 88 |
}
|
| 89 |
return $options;
|
| 90 |
}
|
| 91 |
|
| 92 |
function _z3950_get_syntax($id, $pos) {
|
| 93 |
$syntax = yaz_record($id, $pos, 'syntax');
|
| 94 |
$modules = _z3950_load_syntax_files();
|
| 95 |
if (isset($modules[$syntax])) {
|
| 96 |
return $modules[$syntax];
|
| 97 |
}
|
| 98 |
else {
|
| 99 |
$msg = t('z39.50 search invalid %syntax.', array('%syntax' => $syntax));
|
| 100 |
watchdog('search', $msg, WATCHDOG_WARNING);
|
| 101 |
drupal_set_message($msg);
|
| 102 |
}
|
| 103 |
}
|
| 104 |
|
| 105 |
function _z3950_load_syntax_files() {
|
| 106 |
static $syntax_info;
|
| 107 |
if (!isset($syntax_info)) {
|
| 108 |
$syntax_info = array();
|
| 109 |
$path = drupal_get_path('module', 'z3950') .'/syntax';
|
| 110 |
$files = drupal_system_listing('z3950_.*\.inc$', $path, 'name', 0);
|
| 111 |
foreach ($files as $file) {
|
| 112 |
require_once('./'. $file->filename);
|
| 113 |
$function = $file->name;
|
| 114 |
if (function_exists($function)) {
|
| 115 |
$info = call_user_func($function, 'info');
|
| 116 |
$info['function'] = $function;
|
| 117 |
$syntax_info[$info['value']] = $info;
|
| 118 |
}
|
| 119 |
}
|
| 120 |
}
|
| 121 |
return $syntax_info;
|
| 122 |
}
|
| 123 |
|
| 124 |
/**
|
| 125 |
* Implementation of hook_search().
|
| 126 |
*/
|
| 127 |
function z3950_search($op = 'search', $keys = null) {
|
| 128 |
if (function_exists('yaz_connect') && user_access('search in z3950 servers')) {
|
| 129 |
switch ($op) {
|
| 130 |
case 'name':
|
| 131 |
return t('Metadata');
|
| 132 |
case 'search':
|
| 133 |
return _z3950_search($keys);
|
| 134 |
}
|
| 135 |
}
|
| 136 |
}
|
| 137 |
|
| 138 |
/**
|
| 139 |
* internal implementation of hook_search op=$search
|
| 140 |
*/
|
| 141 |
function _z3950_search($keys) {
|
| 142 |
// get list of available servers
|
| 143 |
if ($servers = _z3950_get_servers()) {
|
| 144 |
// parse the search keys into it's form elements,
|
| 145 |
if (preg_match('/(.*)server:(.*)/', $keys, $match)) {
|
| 146 |
// replace the keys (which was keys and server) with just the keys
|
| 147 |
$keys = $match[1];
|
| 148 |
|
| 149 |
// remove unused servers from the list of available servers
|
| 150 |
$valid_servers = drupal_map_assoc(explode(',', $match[2]));
|
| 151 |
foreach ($servers as $server_index => $server) {
|
| 152 |
if (!isset($valid_servers[$server_index])) {
|
| 153 |
unset($servers[$server_index]);
|
| 154 |
}
|
| 155 |
}
|
| 156 |
}
|
| 157 |
|
| 158 |
// define the CCL fields the is used by yaz_ccl_conf
|
| 159 |
$fields = array(
|
| 160 |
'ti' => '1=4', // title
|
| 161 |
'au' => '1=1', // author
|
| 162 |
);
|
| 163 |
|
| 164 |
// setup the YAZ search on each server
|
| 165 |
$ids = array();
|
| 166 |
foreach ($servers as $server_index => $server) {
|
| 167 |
// connect to the yaz server
|
| 168 |
$id = $ids[$server_index] = yaz_connect($server);
|
| 169 |
|
| 170 |
// tell the yaz server the preferred syntax
|
| 171 |
yaz_syntax($id, variable_get('z3950_syntax', YAZ_DEFAULT_SYNTAX));
|
| 172 |
|
| 173 |
// parse the query and setup the search
|
| 174 |
yaz_ccl_conf($id, $fields);
|
| 175 |
if (yaz_ccl_parse($id, $keys, $cclresult)) {
|
| 176 |
if (!yaz_search($id, 'rpn', $cclresult['rpn'])) {
|
| 177 |
unset($ids[$server_index]);
|
| 178 |
}
|
| 179 |
}
|
| 180 |
else {
|
| 181 |
// display the error message, with server name (if more than one server)
|
| 182 |
$error = $cclresult['errorstring'];
|
| 183 |
if (count($servers) > 1) {
|
| 184 |
$error .= '('. $servers[$server_index] .')';
|
| 185 |
}
|
| 186 |
drupal_set_message($error, 'error');
|
| 187 |
|
| 188 |
// on failure, remove this server from the waiting list
|
| 189 |
unset($ids[$server_index]);
|
| 190 |
}
|
| 191 |
}
|
| 192 |
|
| 193 |
// if there are any available servers that didn't fail, wait for, then process results
|
| 194 |
if (count($ids)) {
|
| 195 |
// wait for all YAZ servers to respond, then process results
|
| 196 |
$options = array('timeout' => 30);
|
| 197 |
if (yaz_wait($options)) {
|
| 198 |
// count the number of results
|
| 199 |
$server_hits = array();
|
| 200 |
$server_parts = array();
|
| 201 |
$server_pos = array();
|
| 202 |
$total_hits = 0;
|
| 203 |
foreach ($ids as $server_index => $id) {
|
| 204 |
// check for error and get number of 'hits' (results) from this server
|
| 205 |
$error = yaz_error($id);
|
| 206 |
if (empty($error)) {
|
| 207 |
$server_hits[$server_index] = yaz_hits($id);
|
| 208 |
|
| 209 |
// if this server has any results
|
| 210 |
if (!empty($server_hits[$server_index])) {
|
| 211 |
// keep track of total number of hits, for use in the pager
|
| 212 |
$total_hits += $server_hits[$server_index];
|
| 213 |
|
| 214 |
// set the starting position
|
| 215 |
$server_pos[$server_index] = 1;
|
| 216 |
|
| 217 |
// parse this server:port/database into it's three parts
|
| 218 |
preg_match('/^([^\/|^:]+):*([0-9]*)\/*(.*)/i', $servers[$server_index], $zurl_parts);
|
| 219 |
$server_part = $zurl_parts[1];
|
| 220 |
if ($server_part == 'localhost' or $server_part == '127.0.0.1') {
|
| 221 |
$server_part = $_SERVER['SERVER_NAME'];
|
| 222 |
}
|
| 223 |
$server_parts[$server_index] = $server_part;
|
| 224 |
}
|
| 225 |
}
|
| 226 |
else {
|
| 227 |
// display the error message, with server name (if more than one server)
|
| 228 |
if (count($servers) > 1) {
|
| 229 |
$error .= ' ('. $servers[$server_index] .')';
|
| 230 |
}
|
| 231 |
drupal_set_message($error, 'error');
|
| 232 |
}
|
| 233 |
}
|
| 234 |
|
| 235 |
if ($total_hits > 0) {
|
| 236 |
// initialize pager variables
|
| 237 |
$total_pos = pager_init(10, 0, $total_hits);
|
| 238 |
$total_per_server = $total_pos / count($ids);
|
| 239 |
|
| 240 |
// initialize the current position in each server
|
| 241 |
// @TODO: there's a faster way to do this, this is just the easy way!
|
| 242 |
// @TODO: test this around the edge cases, is it losing any results?
|
| 243 |
while ($total_pos > 1) {
|
| 244 |
foreach ($ids as $server_index => $id) {
|
| 245 |
if ($server_pos[$server_index] < $server_hits[$server_index]) {
|
| 246 |
$server_pos[$server_index] ++;
|
| 247 |
$total_pos --;
|
| 248 |
}
|
| 249 |
}
|
| 250 |
}
|
| 251 |
|
| 252 |
// process results from search server, evenly mixing results from each server
|
| 253 |
$limit = min($total_hits, 10);
|
| 254 |
$find = array();
|
| 255 |
while (count($find) <= $limit) {
|
| 256 |
foreach ($ids as $server_index => $id) {
|
| 257 |
// loop through the results from this server
|
| 258 |
if ($server_pos[$server_index] < $server_hits[$server_index]) {
|
| 259 |
// store the position in a scalar variable to make code easier to read
|
| 260 |
$pos = $server_pos[$server_index];
|
| 261 |
|
| 262 |
// store the database name for use in the source string below
|
| 263 |
$rec_database = yaz_record($id, $pos, 'database');
|
| 264 |
|
| 265 |
// process the result based on the z39.50 syntax
|
| 266 |
if ($syntax = _z3950_get_syntax($id, $pos)) {
|
| 267 |
// get the syntax summary for this record
|
| 268 |
$summary = call_user_func($syntax['function'], 'summary', $id, $pos);
|
| 269 |
|
| 270 |
// add the link to the summary
|
| 271 |
$summary['link'] = url('z3950/'. $server_index .'/'. $pos .'/'. $keys);
|
| 272 |
|
| 273 |
// add the syntax and source to the summary
|
| 274 |
if (!isset($summary['extra'])) {
|
| 275 |
$summary['extra'] = array();
|
| 276 |
}
|
| 277 |
$summary['extra']['syntax'] = t('syntax: %syntax', array('%syntax' => $syntax['value']));
|
| 278 |
$summary['extra']['source'] = t('source: <a href="@url">@url</a> (%database)', array('@url' => 'http://'. $server_parts[$server_index], '%database' => $rec_database));
|
| 279 |
|
| 280 |
// save this summary, and don't process it again
|
| 281 |
$find[] = $summary;
|
| 282 |
$server_pos[$server_index] ++;
|
| 283 |
}
|
| 284 |
}
|
| 285 |
}
|
| 286 |
}
|
| 287 |
return $find;
|
| 288 |
}
|
| 289 |
}
|
| 290 |
}
|
| 291 |
}
|
| 292 |
}
|
| 293 |
|
| 294 |
|
| 295 |
/**
|
| 296 |
* Implementation of hook_form_alter().
|
| 297 |
*/
|
| 298 |
function z3950_form_alter(&$form, $form_state, $form_id) {
|
| 299 |
// alter the z39.50 search form
|
| 300 |
if ($form_id == 'search_form' && arg(1) == 'z3950') {
|
| 301 |
// get list of available servers
|
| 302 |
$options = _z3950_get_servers();
|
| 303 |
$default_values = array_keys($options);
|
| 304 |
|
| 305 |
// if search keys have already been used, parse them into form elements
|
| 306 |
$keys = $form['basic']['inline']['keys']['#default_value'];
|
| 307 |
if (preg_match('/(.*)server:(.*)/', $keys, $match)) {
|
| 308 |
// replace the keys (which was keys and server) with just the keys
|
| 309 |
$form['basic']['inline']['keys']['#default_value'] = $match[1];
|
| 310 |
|
| 311 |
// remove unused servers from the list of available servers
|
| 312 |
$valid_servers = drupal_map_assoc(explode(',', $match[2]));
|
| 313 |
foreach ($default_values as $server_index => $server) {
|
| 314 |
if (!isset($valid_servers[$server_index])) {
|
| 315 |
unset($default_values[$server_index]);
|
| 316 |
}
|
| 317 |
}
|
| 318 |
}
|
| 319 |
|
| 320 |
// add checkboxes for available servers to the basic form
|
| 321 |
if (count($options) > 1) {
|
| 322 |
$form['basic']['server'] = array(
|
| 323 |
'#type' => 'checkboxes',
|
| 324 |
'#options' => $options,
|
| 325 |
'#default_value' => $default_values,
|
| 326 |
);
|
| 327 |
}
|
| 328 |
|
| 329 |
// add help text to the basic form
|
| 330 |
$form['basic']['#description'] = t('enter the z39.50 search query (<A HREF="@url">see also</A>) and select the servers to search', array('@url' => 'http://www.indexdata.dk/yaz/doc/tools.tkl#CCL'));
|
| 331 |
|
| 332 |
// add advanced form
|
| 333 |
// @TODO
|
| 334 |
|
| 335 |
// add validate hook to process these new form elements
|
| 336 |
$form['#validate'][] = 'z3950_search_validate';
|
| 337 |
}
|
| 338 |
}
|
| 339 |
|
| 340 |
/**
|
| 341 |
* Form API callback for the search form. Registered in node_form_alter().
|
| 342 |
*
|
| 343 |
* hook_form_validate() is used solely to set the 'processed_keys' form
|
| 344 |
* value for the basic search form.
|
| 345 |
*/
|
| 346 |
function z3950_search_validate($form, &$form_state) {
|
| 347 |
if (isset($form_state['values']['server'])) {
|
| 348 |
$keys = $form_state['values']['processed_keys'];
|
| 349 |
$keys .= ' server:'. implode(',', $form_state['values']['server']);
|
| 350 |
form_set_value($form['basic']['inline']['processed_keys'], $keys, $form_state);
|
| 351 |
}
|
| 352 |
}
|
| 353 |
|
| 354 |
/**
|
| 355 |
* Return the Page results for one search item
|
| 356 |
* This page is linked from each record in the search results
|
| 357 |
*/
|
| 358 |
function z3950_get() {
|
| 359 |
// The URL from the search results is made from the server-index/position/keys
|
| 360 |
$server_index = arg(1);
|
| 361 |
$pos = arg(2);
|
| 362 |
$keys = arg(3);
|
| 363 |
|
| 364 |
$servers = _z3950_get_servers();
|
| 365 |
if ($servers && is_numeric($server_index) && is_numeric($pos)) {
|
| 366 |
// connect to the yaz server
|
| 367 |
$id = yaz_connect($servers[$server_index]);
|
| 368 |
|
| 369 |
// tell the yaz server the preferred syntax
|
| 370 |
yaz_syntax($id, variable_get('z3950_syntax', YAZ_DEFAULT_SYNTAX));
|
| 371 |
|
| 372 |
// setup the search
|
| 373 |
if (yaz_search($id, 'rpn', $keys)) {
|
| 374 |
// wait for YAZ server to respond, then display page results
|
| 375 |
$options = array('timeout' => 30);
|
| 376 |
if (yaz_wait($options)) {
|
| 377 |
// check for errors, then display page results
|
| 378 |
$error = yaz_error($id);
|
| 379 |
if (empty($error)) {
|
| 380 |
// display the page results based on the z39.50 syntax
|
| 381 |
if ($syntax = _z3950_get_syntax($id, $pos)) {
|
| 382 |
if ($record = call_user_func($syntax['function'], 'detail', $id, $pos)) {
|
| 383 |
return theme('yaz_record', $record);
|
| 384 |
}
|
| 385 |
}
|
| 386 |
}
|
| 387 |
else {
|
| 388 |
drupal_set_message(t('z39.50 search error: %error in server %server', array('%error' => $error, '%server' => $servers[$server_index])));
|
| 389 |
}
|
| 390 |
}
|
| 391 |
}
|
| 392 |
}
|
| 393 |
|
| 394 |
// if we get here, there was some error, so go back to the search page
|
| 395 |
drupal_goto('search/z3950/'. $keys);
|
| 396 |
}
|
| 397 |
|
| 398 |
function _z3950_get_servers() {
|
| 399 |
if ($servers = variable_get('z3950_zurls', YAZ_DEFAULT_SERVER)) {
|
| 400 |
// split the servers into an array, and trim leading and trailing white space
|
| 401 |
return preg_replace('/^\s+|\s+$/', '', split("/\r\n|\n|\r/", $servers));
|
| 402 |
}
|
| 403 |
drupal_set_message(t('No servers defined in the <a href="@url">settings</a>', array('@url' => url('admin/settings/z3950'))), 'error');
|
| 404 |
}
|
| 405 |
|
| 406 |
/**
|
| 407 |
* Implementation of hook_theme().
|
| 408 |
*/
|
| 409 |
function z3950_theme() {
|
| 410 |
return array('yaz_record' => array());
|
| 411 |
}
|
| 412 |
|
| 413 |
/**
|
| 414 |
* Theme function to display one record in the search results
|
| 415 |
*/
|
| 416 |
function theme_yaz_record($result) {
|
| 417 |
$output .= '<div><pre>';
|
| 418 |
$output .= empty($result) ? t('No records found.') : $result;
|
| 419 |
$output .= '</pre></div>';
|
| 420 |
return $output;
|
| 421 |
}
|
| 422 |
|
| 423 |
/**
|
| 424 |
* Initialize the global pager variables without using the pager_query
|
| 425 |
* This should be in pager.inc, but isn't
|
| 426 |
*/
|
| 427 |
function pager_init($limit = 10, $element = 0, $count) {
|
| 428 |
global $pager_page_array, $pager_total, $pager_total_items;
|
| 429 |
$page = isset($_GET['page']) ? $_GET['page'] : '';
|
| 430 |
|
| 431 |
// Convert comma-separated $page to an array, used by other functions.
|
| 432 |
$pager_page_array = explode(',', $page);
|
| 433 |
|
| 434 |
// calculate the total of pages as ceil(items / limit).
|
| 435 |
$pager_total_items[$element] = $count;
|
| 436 |
$pager_total[$element] = ceil($pager_total_items[$element] / $limit);
|
| 437 |
$pager_page_array[$element] = max(0, min((int)$pager_page_array[$element], ((int)$pager_total[$element]) - 1));
|
| 438 |
|
| 439 |
// return the current position
|
| 440 |
return $pager_page_array[$element] * $limit;
|
| 441 |
}
|
Parent Directory
| 
Revision Log
| 
Revision Graph