plugin/search2.inc.php

   1 <?php
   2 // PukiWiki - Yet another WikiWikiWeb clone.
   3 // search2.inc.php
   4 // Copyright 2017-2021 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                         $msg  = str_replace('$1', htmlsc($q), $_title_result);
  45                         return array('msg' => $msg,
  46                                 'body' => plugin_search2_search_form($q, $bases, $offset, $prev_offset_s));
  47                 }
  48         } else if ($action === 'query') {
  49                 $q = isset($vars['q']) ? $vars['q'] : '';
  50                 $search_start_time = isset($vars['search_start_time']) ?
  51                         $vars['search_start_time'] : null;
  52                 $modified_since = (int)(isset($vars['modified_since']) ?
  53                         $vars['modified_since'] : '0');
  54                 header('Content-Type: application/json; charset=UTF-8');
  55                 plugin_search2_do_search($q, $base, $start_index,
  56                         $search_start_time, $modified_since);
  57                 exit;
  58         }
  59 }
  60
  61 function plugin_search2_get_base_url($search_text)
  62 {
  63         global $vars;
  64         $params = array();
  65         if (!defined('PKWK_UTF8_ENABLE')) {
  66                 $params[] = 'encode_hint=' . rawurlencode($vars['encode_hint']);
  67         }
  68         $params[] = 'cmd=search2';
  69         if (isset($vars['encode_hint']) && $vars['encode_hint']) {
  70                 $params[] = 'encode_hint=' . rawurlencode($vars['encode_hint']);
  71         }
  72         if ($search_text) {
  73                 $params[] = 'q=' . plugin_search2_urlencode_searchtext($search_text);
  74         }
  75         if (isset($vars['base']) && $vars['base']) {
  76                 $params[] = 'base=' . rawurlencode($vars['base']);
  77         }
  78         $url = get_base_uri() . '?' . join('&', $params);
  79         return $url;
  80 }
  81
  82 function plugin_search2_urlencode_searchtext($search_text)
  83 {
  84         $s2 = preg_replace('#^\s+|\s+$#', '', $search_text);
  85         if (!$s2) return '';
  86         $sp = preg_split('#\s+#', $s2);
  87         $list = array();
  88         for ($i = 0; $i < count($sp); $i++) {
  89                 $list[] = rawurlencode($sp[$i]);
  90         }
  91         return join('+', $list);
  92 }
  93
  94 function plugin_search2_do_search($query_text, $base, $start_index,
  95         $search_start_time, $modified_since)
  96 {
  97         global $whatsnew, $non_list, $search_non_list;
  98         global $_msg_andresult, $_msg_orresult;
  99         global $search_auth, $auth_user;
 100
 101         $result_record_limit = $start_index === 0 ?
 102                 PLUGIN_SEARCH2_RESULT_RECORD_LIMIT_START : PLUGIN_SEARCH2_RESULT_RECORD_LIMIT;
 103         $retval = array();
 104
 105         $b_type_and = true; // AND:TRUE OR:FALSE
 106         $key_candidates = preg_split('/\s+/', $query_text, -1, PREG_SPLIT_NO_EMPTY);
 107         for ($i = count($key_candidates) - 2; $i >= 1; $i--) {
 108                 if ($key_candidates[$i] === 'OR') {
 109                         $b_type_and = false;
 110                         unset($key_candidates[$i]);
 111                 }
 112         }
 113         $key_candidates = array_merge($key_candidates);
 114         $keys = get_search_words($key_candidates);
 115         foreach ($keys as $key=>$value)
 116                 $keys[$key] = '/' . $value . '/S';
 117
 118         if ($modified_since > 0) {
 119                 // Recent search
 120                 $recent_files = get_recent_files();
 121                 $modified_loc = $modified_since - LOCALZONE;
 122                 $pages = array();
 123                 foreach ($recent_files as $p => $time) {
 124                         if ($time >= $modified_loc) {
 125                                 $pages[] = $p;
 126                         }
 127                 }
 128                 if ($base != '') {
 129                         $pages = preg_grep('/^' . preg_quote($base, '/') . '/S', $pages);
 130                 }
 131                 $page_names = $pages;
 132         } else {
 133                 // Normal search
 134                 $pages = get_existpages();
 135
 136                 // Avoid
 137                 if ($base != '') {
 138                         $pages = preg_grep('/^' . preg_quote($base, '/') . '/S', $pages);
 139                 }
 140                 if (! $search_non_list) {
 141                         $pages = array_diff($pages, preg_grep('/' . $non_list . '/S', $pages));
 142                 }
 143                 $pages = array_flip($pages);
 144                 unset($pages[$whatsnew]);
 145                 $page_names = array_keys($pages);
 146         }
 147         natsort($page_names);
 148         // Cache collabolate
 149         if (is_null($search_start_time)) {
 150                 // Don't use client cache
 151                 $search_start_time = UTIME + LOCALZONE;
 152         }
 153         $found_pages = array();
 154         $readable_page_index = -1;
 155         $scan_page_index = -1;
 156         $saved_scan_start_index = -1;
 157         $last_read_page_name = null;
 158         foreach ($page_names as $page) {
 159                 $b_match = FALSE;
 160                 $pagename_only = false;
 161                 $scan_page_index++;
 162                 if (! is_page_readable($page)) {
 163                         if ($search_auth) {
 164                                 // $search_auth - 1: User can know page names that contain search text if the page is readable
 165                                 continue;
 166                         }
 167                         // $search_auth - 0: All users can know page names that conntain search text
 168                         $pagename_only = true;
 169                 }
 170                 $readable_page_index++;
 171                 if ($readable_page_index < $start_index) {
 172                         // Skip: It's not time to read
 173                         continue;
 174                 }
 175                 if ($saved_scan_start_index === -1) {
 176                         $saved_scan_start_index = $scan_page_index;
 177                 }
 178                 if (count($keys) > 0) {
 179                         // Search for page name and contents
 180                         $body = get_source($page, TRUE, TRUE, TRUE);
 181                         $target = $page . "\n" . remove_author_header($body);
 182                         foreach ($keys as $key) {
 183                                 $b_match = preg_match($key, $target);
 184                                 if ($b_type_and xor $b_match) break; // OR
 185                         }
 186                 } else {
 187                         // No search target. get_source($page) is meaningless.
 188                         // $b_match is always false.
 189                 }
 190                 if ($b_match) {
 191                         // Found!
 192                         $author_info = get_author_info($body);
 193                         if ($author_info) {
 194                                 $updated_at = get_update_datetime_from_author($author_info);
 195                                 $updated_time = strtotime($updated_at);
 196                         } else {
 197                                 $updated_time = filemtime(get_filename($page));
 198                                 $updated_at = get_date_atom($updated_time);
 199                         }
 200                         if ($pagename_only) {
 201                                 // The user cannot read this page body
 202                                 $found_pages[] = array('name' => (string)$page,
 203                                         'url' => get_page_uri($page), 'updated_at' => $updated_at,
 204                                         'updated_time' => $updated_time,
 205                                         'body' => '', 'pagename_only' => 1);
 206                         } else {
 207                                 $found_pages[] = array('name' => (string)$page,
 208                                         'url' => get_page_uri($page), 'updated_at' => $updated_at,
 209                                         'updated_time' => $updated_time,
 210                                         'body' => (string)$body);
 211                         }
 212                 }
 213                 $last_read_page_name = $page;
 214                 if ($start_index + $result_record_limit <= $readable_page_index + 1) {
 215                         // Read page limit
 216                         break;
 217                 }
 218         }
 219         $message = str_replace('$1', htmlsc($query_text), str_replace('$2', count($found_pages),
 220                 str_replace('$3', count($page_names), $b_type_and ? $_msg_andresult : $_msg_orresult)));
 221         $search_done = (boolean)($scan_page_index + 1 === count($page_names));
 222         $result_obj = array(
 223                 'message' => $message,
 224                 'q' => $query_text,
 225                 'start_index' => $start_index,
 226                 'limit' => $result_record_limit,
 227                 'read_page_count' => $readable_page_index - $start_index + 1,
 228                 'scan_page_count' => $scan_page_index - $saved_scan_start_index + 1,
 229                 'page_count' => count($page_names),
 230                 'last_read_page_name' => $last_read_page_name,
 231                 'next_start_index' => $readable_page_index + 1,
 232                 'search_done' => $search_done,
 233                 'search_start_time' => $search_start_time,
 234                 'auth_user' => $auth_user,
 235                 'results' => $found_pages);
 236         $obj = $result_obj;
 237         if (!defined('PKWK_UTF8_ENABLE')) {
 238                 if (SOURCE_ENCODING === 'EUC-JP') {
 239                         mb_convert_variables('UTF-8', 'CP51932', $obj);
 240                 } else {
 241                         mb_convert_variables('UTF-8', SOURCE_ENCODING, $obj);
 242                 }
 243         }
 244         print(json_encode($obj, JSON_UNESCAPED_UNICODE | JSON_UNESCAPED_SLASHES));
 245 }
 246
 247 function plugin_search2_search_form($search_text = '', $bases = array(),
 248         $offset, $prev_offset_s = null)
 249 {
 250         global $_btn_search;
 251         global $_search_pages, $_search_all;
 252         global $_msg_andresult, $_msg_orresult, $_msg_notfoundresult;
 253         global $_search_detail, $_search_searching, $_search_showing_result;
 254         global $_msg_unsupported_webbrowser, $_msg_use_alternative_link;
 255         global $_msg_more_results, $_msg_prev_results, $_msg_general_error;
 256         global $auth_user;
 257
 258         static $search2_form_total_count = 0;
 259         $search2_form_total_count++;
 260         $script = get_base_uri();
 261         $h_search_text = htmlsc($search_text);
 262
 263         $base_option = '';
 264         if (!empty($bases)) {
 265                 $base_msg = '';
 266                 $_num = 0;
 267                 $check = ' checked';
 268                 foreach($bases as $base) {
 269                         ++$_num;
 270                         if (PLUGIN_SEARCH2_MAX_BASE < $_num) break;
 271                         $s_base   = htmlsc($base);
 272                         $base_str = '<strong>' . $s_base . '</strong>';
 273                         $base_label = str_replace('$1', $base_str, $_search_pages);
 274                         $base_msg  .=<<<EOD
 275  <div>
 276   <label>
 277    <input type="radio" name="base" value="$s_base" $check> $base_label
 278   </label>
 279  </div>
 280 EOD;
 281                         $check = '';
 282                 }
 283                 $base_msg .=<<<EOD
 284 <label><input type="radio" name="base" value=""> $_search_all</label>
 285 EOD;
 286                 $base_option = '<div class="small">' . $base_msg . '</div>';
 287         }
 288         $_search2_result_notfound = htmlsc($_msg_notfoundresult);
 289         $_search2_result_found = htmlsc($_msg_andresult);
 290         $_search2_search_wait_milliseconds = PLUGIN_SEARCH2_SEARCH_WAIT_MILLISECONDS;
 291         $result_page_panel =<<<EOD
 292 <input type="checkbox" id="_plugin_search2_detail" checked><label for="_plugin_search2_detail">$_search_detail</label>
 293 <ul id="_plugin_search2_result-list">
 294 </ul>
 295 EOD;
 296         if ($h_search_text == '' || $search2_form_total_count > 1) {
 297                 $result_page_panel = '';
 298         }
 299
 300         $plain_search_link = '<a href="' . $script . '?cmd=search' . '">' . htmlsc($_btn_search) . '</a>';
 301         $alt_msg = str_replace('$1', $plain_search_link, $_msg_use_alternative_link);
 302         $status_span_text = '<span class="_plugin_search2_search_status_text1"></span>' .
 303                 '<span class="_plugin_search2_search_status_text2"></span>';
 304         $form =<<<EOD
 305 <form action="$script" method="GET" class="_plugin_search2_form">
 306  <div>
 307   <input type="hidden" name="cmd" value="search2">
 308   <input type="search" name="q" value="$h_search_text" data-original-q="$h_search_text" size="40">
 309   <input type="submit" value="$_btn_search">
 310  </div>
 311 $base_option
 312 </form>
 313 EOD;
 314         $second_form =<<<EOD
 315 <div class="_plugin_search2_second_form" style="display:none;">
 316 <div class="_plugin_search2_search_status">$status_span_text</span></div>
 317 <div class="_plugin_search2_message"></div>
 318 $form
 319 </div>
 320 EOD;
 321
 322         $h_auth_user = htmlsc($auth_user);
 323         $h_base_url = htmlsc(plugin_search2_get_base_url($search_text));
 324         $h_msg_more_results = htmlsc($_msg_more_results);
 325         $h_msg_prev_results = htmlsc($_msg_prev_results);
 326         $max_results = PLUGIN_SEARCH2_SEARCH_MAX_RESULTS;
 327         $prev_offset = pkwk_ctype_digit($prev_offset_s) ? $prev_offset_s : '';
 328         $search_props =<<<EOD
 329 <div style="display:none;">
 330   <input type="hidden" id="_plugin_search2_auth_user" value="$h_auth_user">
 331   <input type="hidden" id="_plugin_search2_base_url" value="$h_base_url">
 332   <input type="hidden" id="_plugin_search2_msg_searching" value="$_search_searching">
 333   <input type="hidden" id="_plugin_search2_msg_showing_result" value="$_search_showing_result">
 334   <input type="hidden" id="_plugin_search2_msg_result_notfound" value="$_search2_result_notfound">
 335   <input type="hidden" id="_plugin_search2_msg_result_found" value="$_search2_result_found">
 336   <input type="hidden" id="_plugin_search2_msg_more_results" value="$h_msg_more_results">
 337   <input type="hidden" id="_plugin_search2_msg_prev_results" value="$h_msg_prev_results">
 338   <input type="hidden" id="_plugin_search2_search_wait_milliseconds" value="$_search2_search_wait_milliseconds">
 339   <input type="hidden" id="_plugin_search2_max_results" value="$max_results">
 340   <input type="hidden" id="_plugin_search2_offset" value="$offset">
 341   <input type="hidden" id="_plugin_search2_prev_offset" value="$prev_offset">
 342   <input type="hidden" id="_plugin_search2_msg_error" value="$_msg_general_error">
 343 </div>
 344 EOD;
 345         if ($search2_form_total_count > 1) {
 346                 $search_props = '';
 347         }
 348
 349         return <<<EOD
 350 <noscript>
 351  <p>$_msg_unsupported_webbrowser $alt_msg</p>
 352 </noscript>
 353 <style>
 354 input#_plugin_search2_detail:checked ~ ul > li > div.search-result-detail {
 355   display:block;
 356 }
 357 input#_plugin_search2_detail ~ ul > li > div.search-result-detail {
 358   display:none;
 359 }
 360 ._plugin_search2_search_status {
 361   min-height:1.5em;
 362 }
 363 @keyframes plugin-search2-searching {
 364   10% { opacity: 1; }
 365   40% { opacity: 0; }
 366   70% { opacity: 0; }
 367   90% { opacity: 1; }
 368 }
 369 span.plugin-search2-progress {
 370   animation: plugin-search2-searching 1.5s infinite ease-out;
 371 }
 372 span.plugin-search2-progress1 {
 373   animation-delay: -1s;
 374 }
 375 span.plugin-search2-progress2 {
 376   animation-delay: -0.8s;
 377 }
 378 span.plugin-search2-progress3 {
 379   animation-delay: -0.6s;
 380 }
 381 </style>
 382 <p class="_plugin_search2_nosupport_message" style="display:none;">
 383   $_msg_unsupported_webbrowser $alt_msg
 384 </p>
 385 $search_props
 386 $form
 387 <div class="_plugin_search2_search_status">$status_span_text</div>
 388 <div class="_plugin_search2_message"></div>
 389 $result_page_panel
 390 $second_form
 391 EOD;
 392 }