plugin/search2.inc.php

   1 <?php
   2 // PukiWiki - Yet another WikiWikiWeb clone.
   3 // search2.inc.php
   4 // Copyright 2017-2022 PukiWiki Development Team
   5 // License: GPL v2 or (at your option) any later version
   6 //
   7 // Search2 plugin - Show detail result using JavaScript
   8
   9 define('PLUGIN_SEARCH2_MAX_BASE', 16); // #search(1,2,3,...,15,16)
  10
  11 define('PLUGIN_SEARCH2_RESULT_RECORD_LIMIT', 1000);
  12 define('PLUGIN_SEARCH2_RESULT_RECORD_LIMIT_START', 100);
  13 define('PLUGIN_SEARCH2_SEARCH_WAIT_MILLISECONDS', 1000);
  14 define('PLUGIN_SEARCH2_SEARCH_MAX_RESULTS', 500);
  15
  16 // Show a search box on a page
  17 function plugin_search2_convert()
  18 {
  19         return 'Usage: Please use #search()';
  20 }
  21
  22 function plugin_search2_action()
  23 {
  24         global $vars, $_title_search, $_title_result, $_msg_searching;
  25
  26         $action = isset($vars['action']) ? $vars['action'] : '';
  27         $base = isset($vars['base']) ? $vars['base'] : '';
  28         $start_s = isset($vars['start']) ? $vars['start'] : '';
  29         $start_index = pkwk_ctype_digit($start_s) ? intval($start_s) : 0;
  30         $bases = array();
  31         if ($base !== '') {
  32                 $bases[] = $base;
  33         }
  34         if ($action === '') {
  35                 $q = trim(isset($vars['q']) ? $vars['q'] : '');
  36                 $offset_s = isset($vars['offset']) ? $vars['offset'] : '';
  37                 $offset = pkwk_ctype_digit($offset_s) ? intval($offset_s) : 0;
  38                 $prev_offset_s = isset($vars['prev_offset']) ? $vars['prev_offset'] : '';
  39                 if ($q === '') {
  40                         return array('msg' => $_title_search,
  41                                 'body' => "<br>" . $_msg_searching . "\n" .
  42                                 plugin_search2_search_form($q, $bases, $offset));
  43                 } else {
  44                         $q2 = $q;
  45                         if (defined('PKWK_UTF8_ENABLE')) {
  46                                 $zen_space = "\xE3\x80\x80"; // IDEOGRAPHIC SPACE in UTF-8 - &#x3000;
  47                                 $q2 = str_replace($zen_space, ' ', $q);
  48                         }
  49                         $msg  = str_replace('$1', htmlsc($q2), $_title_result);
  50                         return array('msg' => $msg,
  51                                 'body' => plugin_search2_search_form($q2, $bases, $offset, $prev_offset_s));
  52                 }
  53         } else if ($action === 'query') {
  54                 $q = isset($vars['q']) ? $vars['q'] : '';
  55                 $search_start_time = isset($vars['search_start_time']) ?
  56                         $vars['search_start_time'] : null;
  57                 $modified_since = (int)(isset($vars['modified_since']) ?
  58                         $vars['modified_since'] : '0');
  59                 header('Content-Type: application/json; charset=UTF-8');
  60                 plugin_search2_do_search($q, $base, $start_index,
  61                         $search_start_time, $modified_since);
  62                 exit;
  63         }
  64 }
  65
  66 function plugin_search2_get_base_url($search_text)
  67 {
  68         global $vars;
  69         $params = array();
  70         if (!defined('PKWK_UTF8_ENABLE')) {
  71                 $params[] = 'encode_hint=' . rawurlencode($vars['encode_hint']);
  72         }
  73         $params[] = 'cmd=search2';
  74         if (isset($vars['encode_hint']) && $vars['encode_hint']) {
  75                 $params[] = 'encode_hint=' . rawurlencode($vars['encode_hint']);
  76         }
  77         if ($search_text) {
  78                 $params[] = 'q=' . plugin_search2_urlencode_searchtext($search_text);
  79         }
  80         if (isset($vars['base']) && $vars['base']) {
  81                 $params[] = 'base=' . rawurlencode($vars['base']);
  82         }
  83         $url = get_base_uri() . '?' . join('&', $params);
  84         return $url;
  85 }
  86
  87 function plugin_search2_urlencode_searchtext($search_text)
  88 {
  89         $s2 = preg_replace('#^\s+|\s+$#', '', $search_text);
  90         if (!$s2) return '';
  91         $sp = preg_split('#\s+#', $s2);
  92         $list = array();
  93         for ($i = 0; $i < count($sp); $i++) {
  94                 $list[] = rawurlencode($sp[$i]);
  95         }
  96         return join('+', $list);
  97 }
  98
  99 function plugin_search2_do_search($query_text, $base, $start_index,
 100         $search_start_time, $modified_since)
 101 {
 102         global $whatsnew, $non_list, $search_non_list;
 103         global $_msg_andresult, $_msg_orresult;
 104         global $search_auth, $auth_user;
 105
 106         $result_record_limit = $start_index === 0 ?
 107                 PLUGIN_SEARCH2_RESULT_RECORD_LIMIT_START : PLUGIN_SEARCH2_RESULT_RECORD_LIMIT;
 108         $retval = array();
 109
 110         $b_type_and = true; // AND:TRUE OR:FALSE
 111         $key_candidates = preg_split('/\s+/', $query_text, -1, PREG_SPLIT_NO_EMPTY);
 112         for ($i = count($key_candidates) - 2; $i >= 1; $i--) {
 113                 if ($key_candidates[$i] === 'OR') {
 114                         $b_type_and = false;
 115                         unset($key_candidates[$i]);
 116                 }
 117         }
 118         $key_candidates = array_merge($key_candidates);
 119         $keys = get_search_words($key_candidates);
 120         foreach ($keys as $key=>$value)
 121                 $keys[$key] = '/' . $value . '/S';
 122
 123         if ($modified_since > 0) {
 124                 // Recent search
 125                 $recent_files = get_recent_files();
 126                 $modified_loc = $modified_since - LOCALZONE;
 127                 $pages = array();
 128                 foreach ($recent_files as $p => $time) {
 129                         if ($time >= $modified_loc) {
 130                                 $pages[] = $p;
 131                         }
 132                 }
 133                 if ($base != '') {
 134                         $pages = preg_grep('/^' . preg_quote($base, '/') . '/S', $pages);
 135                 }
 136                 $page_names = $pages;
 137         } else {
 138                 // Normal search
 139                 $pages = get_existpages();
 140
 141                 // Avoid
 142                 if ($base != '') {
 143                         $pages = preg_grep('/^' . preg_quote($base, '/') . '/S', $pages);
 144                 }
 145                 if (! $search_non_list) {
 146                         $pages = array_diff($pages, preg_grep('/' . $non_list . '/S', $pages));
 147                 }
 148                 $pages = array_flip($pages);
 149                 unset($pages[$whatsnew]);
 150                 $page_names = array_keys($pages);
 151         }
 152         natsort($page_names);
 153         // Cache collabolate
 154         if (is_null($search_start_time)) {
 155                 // Don't use client cache
 156                 $search_start_time = UTIME + LOCALZONE;
 157         }
 158         $found_pages = array();
 159         $readable_page_index = -1;
 160         $scan_page_index = -1;
 161         $saved_scan_start_index = -1;
 162         $last_read_page_name = null;
 163         foreach ($page_names as $page) {
 164                 $b_match = FALSE;
 165                 $pagename_only = false;
 166                 $scan_page_index++;
 167                 if (! is_page_readable($page)) {
 168                         if ($search_auth) {
 169                                 // $search_auth - 1: User can know page names that contain search text if the page is readable
 170                                 continue;
 171                         }
 172                         // $search_auth - 0: All users can know page names that conntain search text
 173                         $pagename_only = true;
 174                 }
 175                 $readable_page_index++;
 176                 if ($readable_page_index < $start_index) {
 177                         // Skip: It's not time to read
 178                         continue;
 179                 }
 180                 if ($saved_scan_start_index === -1) {
 181                         $saved_scan_start_index = $scan_page_index;
 182                 }
 183                 if (count($keys) > 0) {
 184                         // Search for page name and contents
 185                         $body = get_source($page, TRUE, TRUE, TRUE);
 186                         $target = $page . "\n" . remove_author_header($body);
 187                         foreach ($keys as $key) {
 188                                 $b_match = preg_match($key, $target);
 189                                 if ($b_type_and xor $b_match) break; // OR
 190                         }
 191                 } else {
 192                         // No search target. get_source($page) is meaningless.
 193                         // $b_match is always false.
 194                 }
 195                 if ($b_match) {
 196                         // Found!
 197                         $author_info = get_author_info($body);
 198                         if ($author_info) {
 199                                 $updated_at = get_update_datetime_from_author($author_info);
 200                                 $updated_time = strtotime($updated_at);
 201                         } else {
 202                                 $updated_time = filemtime(get_filename($page));
 203                                 $updated_at = get_date_atom($updated_time);
 204                         }
 205                         if ($pagename_only) {
 206                                 // The user cannot read this page body
 207                                 $found_pages[] = array('name' => (string)$page,
 208                                         'url' => get_page_uri($page), 'updated_at' => $updated_at,
 209                                         'updated_time' => $updated_time,
 210                                         'body' => '', 'pagename_only' => 1);
 211                         } else {
 212                                 $found_pages[] = array('name' => (string)$page,
 213                                         'url' => get_page_uri($page), 'updated_at' => $updated_at,
 214                                         'updated_time' => $updated_time,
 215                                         'body' => (string)$body);
 216                         }
 217                 }
 218                 $last_read_page_name = $page;
 219                 if ($start_index + $result_record_limit <= $readable_page_index + 1) {
 220                         // Read page limit
 221                         break;
 222                 }
 223         }
 224         $message = str_replace('$1', htmlsc($query_text), str_replace('$2', count($found_pages),
 225                 str_replace('$3', count($page_names), $b_type_and ? $_msg_andresult : $_msg_orresult)));
 226         $search_done = (boolean)($scan_page_index + 1 === count($page_names));
 227         $result_obj = array(
 228                 'message' => $message,
 229                 'q' => $query_text,
 230                 'start_index' => $start_index,
 231                 'limit' => $result_record_limit,
 232                 'read_page_count' => $readable_page_index - $start_index + 1,
 233                 'scan_page_count' => $scan_page_index - $saved_scan_start_index + 1,
 234                 'page_count' => count($page_names),
 235                 'last_read_page_name' => $last_read_page_name,
 236                 'next_start_index' => $readable_page_index + 1,
 237                 'search_done' => $search_done,
 238                 'search_start_time' => $search_start_time,
 239                 'auth_user' => $auth_user,
 240                 'results' => $found_pages);
 241         $obj = $result_obj;
 242         if (!defined('PKWK_UTF8_ENABLE')) {
 243                 if (SOURCE_ENCODING === 'EUC-JP') {
 244                         mb_convert_variables('UTF-8', 'CP51932', $obj);
 245                 } else {
 246                         mb_convert_variables('UTF-8', SOURCE_ENCODING, $obj);
 247                 }
 248         }
 249         print(json_encode($obj, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES));
 250 }
 251
 252 function plugin_search2_search_form($search_text = '', $bases = array(),
 253         $offset, $prev_offset_s = null)
 254 {
 255         global $_btn_search;
 256         global $_search_pages, $_search_all;
 257         global $_msg_andresult, $_msg_orresult, $_msg_notfoundresult;
 258         global $_search_detail, $_search_searching, $_search_showing_result;
 259         global $_msg_unsupported_webbrowser, $_msg_use_alternative_link;
 260         global $_msg_more_results, $_msg_prev_results, $_msg_general_error;
 261         global $auth_user;
 262
 263         static $search2_form_total_count = 0;
 264         $search2_form_total_count++;
 265         $script = get_base_uri();
 266         $h_search_text = htmlsc($search_text);
 267
 268         $base_option = '';
 269         if (!empty($bases)) {
 270                 $base_msg = '';
 271                 $_num = 0;
 272                 $check = ' checked';
 273                 foreach($bases as $base) {
 274                         ++$_num;
 275                         if (PLUGIN_SEARCH2_MAX_BASE < $_num) break;
 276                         $s_base   = htmlsc($base);
 277                         $base_str = '<strong>' . $s_base . '</strong>';
 278                         $base_label = str_replace('$1', $base_str, $_search_pages);
 279                         $base_msg  .=<<<EOD
 280  <div>
 281   <label>
 282    <input type="radio" name="base" value="$s_base" $check> $base_label
 283   </label>
 284  </div>
 285 EOD;
 286                         $check = '';
 287                 }
 288                 $base_msg .=<<<EOD
 289 <label><input type="radio" name="base" value=""> $_search_all</label>
 290 EOD;
 291                 $base_option = '<div class="small">' . $base_msg . '</div>';
 292         }
 293         $_search2_result_notfound = htmlsc($_msg_notfoundresult);
 294         $_search2_result_found = htmlsc($_msg_andresult);
 295         $_search2_search_wait_milliseconds = PLUGIN_SEARCH2_SEARCH_WAIT_MILLISECONDS;
 296         $result_page_panel =<<<EOD
 297 <input type="checkbox" id="_plugin_search2_detail" checked><label for="_plugin_search2_detail">$_search_detail</label>
 298 <ul id="_plugin_search2_result-list">
 299 </ul>
 300 EOD;
 301         if ($h_search_text == '' || $search2_form_total_count > 1) {
 302                 $result_page_panel = '';
 303         }
 304
 305         $plain_search_link = '<a href="' . $script . '?cmd=search' . '">' . htmlsc($_btn_search) . '</a>';
 306         $alt_msg = str_replace('$1', $plain_search_link, $_msg_use_alternative_link);
 307         $status_span_text = '<span class="_plugin_search2_search_status_text1"></span>' .
 308                 '<span class="_plugin_search2_search_status_text2"></span>';
 309         $form =<<<EOD
 310 <form action="$script" method="GET" class="_plugin_search2_form">
 311  <div>
 312   <input type="hidden" name="cmd" value="search2">
 313   <input type="search" name="q" value="$h_search_text" data-original-q="$h_search_text" size="40">
 314   <input type="submit" value="$_btn_search">
 315  </div>
 316 $base_option
 317 </form>
 318 EOD;
 319         $second_form =<<<EOD
 320 <div class="_plugin_search2_second_form" style="display:none;">
 321 <div class="_plugin_search2_search_status">$status_span_text</span></div>
 322 <div class="_plugin_search2_message"></div>
 323 $form
 324 </div>
 325 EOD;
 326
 327         $h_auth_user = htmlsc($auth_user);
 328         $h_base_url = htmlsc(plugin_search2_get_base_url($search_text));
 329         $h_msg_more_results = htmlsc($_msg_more_results);
 330         $h_msg_prev_results = htmlsc($_msg_prev_results);
 331         $max_results = PLUGIN_SEARCH2_SEARCH_MAX_RESULTS;
 332         $prev_offset = pkwk_ctype_digit($prev_offset_s) ? $prev_offset_s : '';
 333         $search_props =<<<EOD
 334 <div style="display:none;">
 335   <input type="hidden" id="_plugin_search2_auth_user" value="$h_auth_user">
 336   <input type="hidden" id="_plugin_search2_base_url" value="$h_base_url">
 337   <input type="hidden" id="_plugin_search2_msg_searching" value="$_search_searching">
 338   <input type="hidden" id="_plugin_search2_msg_showing_result" value="$_search_showing_result">
 339   <input type="hidden" id="_plugin_search2_msg_result_notfound" value="$_search2_result_notfound">
 340   <input type="hidden" id="_plugin_search2_msg_result_found" value="$_search2_result_found">
 341   <input type="hidden" id="_plugin_search2_msg_more_results" value="$h_msg_more_results">
 342   <input type="hidden" id="_plugin_search2_msg_prev_results" value="$h_msg_prev_results">
 343   <input type="hidden" id="_plugin_search2_search_wait_milliseconds" value="$_search2_search_wait_milliseconds">
 344   <input type="hidden" id="_plugin_search2_max_results" value="$max_results">
 345   <input type="hidden" id="_plugin_search2_offset" value="$offset">
 346   <input type="hidden" id="_plugin_search2_prev_offset" value="$prev_offset">
 347   <input type="hidden" id="_plugin_search2_msg_error" value="$_msg_general_error">
 348 </div>
 349 EOD;
 350         if ($search2_form_total_count > 1) {
 351                 $search_props = '';
 352         }
 353
 354         return <<<EOD
 355 <noscript>
 356  <p>$_msg_unsupported_webbrowser $alt_msg</p>
 357 </noscript>
 358 <style>
 359 input#_plugin_search2_detail:checked ~ ul > li > div.search-result-detail {
 360   display:block;
 361 }
 362 input#_plugin_search2_detail ~ ul > li > div.search-result-detail {
 363   display:none;
 364 }
 365 ._plugin_search2_search_status {
 366   min-height:1.5em;
 367 }
 368 @keyframes plugin-search2-searching {
 369   10% { opacity: 1; }
 370   40% { opacity: 0; }
 371   70% { opacity: 0; }
 372   90% { opacity: 1; }
 373 }
 374 span.plugin-search2-progress {
 375   animation: plugin-search2-searching 1.5s infinite ease-out;
 376 }
 377 span.plugin-search2-progress1 {
 378   animation-delay: -1s;
 379 }
 380 span.plugin-search2-progress2 {
 381   animation-delay: -0.8s;
 382 }
 383 span.plugin-search2-progress3 {
 384   animation-delay: -0.6s;
 385 }
 386 </style>
 387 <p class="_plugin_search2_nosupport_message" style="display:none;">
 388   $_msg_unsupported_webbrowser $alt_msg
 389 </p>
 390 $search_props
 391 $form
 392 <div class="_plugin_search2_search_status">$status_span_text</div>
 393 <div class="_plugin_search2_message"></div>
 394 $result_page_panel
 395 $second_form
 396 EOD;
 397 }