plugin/search2.inc.php

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