src/search_functions_php.h

   1 "<script language=\"PHP\">\n"
   2 "require_once \"search-config.php\";\n"
   3 "\n"
   4 "function end_form($value)\n"
   5 "{\n"
   6 "  global $config;\n"
   7 "  global $translator;\n"
   8 "  if ($config['DISABLE_INDEX'] == false)\n"
   9 "  {\n"
  10 "  echo \"            <input type=\\\"text\\\" id=\\\"MSearchField\\\" name=\\\"query\\\" value=\\\"$value\\\" size=\\\"20\\\" accesskey=\\\"S\\\" onfocus=\\\"searchBox.OnSearchFieldFocus(true)\\\" onblur=\\\"searchBox.OnSearchFieldFocus(false)\\\"/>\\n            </form>\\n          </div><div class=\\\"right\\\"></div>\\n        </div>\\n      </li>\\n    </ul>\\n  </div>\\n</div>\\n\";\n"
  11 "  }\n"
  12 "  if ($config['GENERATE_TREEVIEW'])\n"
  13 "  {\n"
  14 "    echo $translator['split_bar'];\n"
  15 "  }\n"
  16 "}\n"
  17 "\n"
  18 "function end_page()\n"
  19 "{\n"
  20 "  //global $config;\n"
  21 "  //global $translator;\n"
  22 "  //if ($config['GENERATE_TREEVIEW'])\n"
  23 "  //{\n"
  24 "  //  echo \"</div>\\n<div id=\\\"nav-path\\\" class=\\\"navpath\\\">\\n  <ul>\\n    <li class=\\\"footer\\\">\";\n"
  25 "  //  echo $translator['logo'];\n"
  26 "  //  echo \"</li>\\n  </ul>\\n</div>\";\n"
  27 "  //}\n"
  28 "  echo \"</body></html>\";\n"
  29 "}\n"
  30 "\n"
  31 "function search_results()\n"
  32 "{\n"
  33 "  global $translator;\n"
  34 "  return $translator['search_results_title'];\n"
  35 "}\n"
  36 "\n"
  37 "function matches_text($num)\n"
  38 "{\n"
  39 "  global $translator;\n"
  40 "  $string = $translator['search_results'][($num>2)?2:$num];\n"
  41 "  // The eval is used so that translator strings can contain $num.\n"
  42 "  eval(\"\\$result = \\\"$string\\\";\");\n"
  43 "  return $result;\n"
  44 "}\n"
  45 "\n"
  46 "function report_matches()\n"
  47 "{\n"
  48 "  global $translator;\n"
  49 "  return $translator['search_matches'];\n"
  50 "}\n"
  51 "\n"
  52 "function readInt($file)\n"
  53 "{\n"
  54 "  $b1 = ord(fgetc($file)); $b2 = ord(fgetc($file));\n"
  55 "  $b3 = ord(fgetc($file)); $b4 = ord(fgetc($file));\n"
  56 "  return ($b1<<24)|($b2<<16)|($b3<<8)|$b4;\n"
  57 "}\n"
  58 "\n"
  59 "function readString($file)\n"
  60 "{\n"
  61 "  $result=\"\";\n"
  62 "  while (ord($c=fgetc($file))) $result.=$c;\n"
  63 "  return $result;\n"
  64 "}\n"
  65 "\n"
  66 "function readHeader($file)\n"
  67 "{\n"
  68 "  $header =fgetc($file); $header.=fgetc($file);\n"
  69 "  $header.=fgetc($file); $header.=fgetc($file);\n"
  70 "  return $header;\n"
  71 "}\n"
  72 "\n"
  73 "function computeIndex($word)\n"
  74 "{\n"
  75 "  // Simple hashing that allows for substring search\n"
  76 "  if (strlen($word)<2) return -1;\n"
  77 "  // high char of the index\n"
  78 "  $hi = ord($word{0});\n"
  79 "  if ($hi==0) return -1;\n"
  80 "  // low char of the index\n"
  81 "  $lo = ord($word{1});\n"
  82 "  if ($lo==0) return -1;\n"
  83 "  // return index\n"
  84 "  return $hi*256+$lo;\n"
  85 "}\n"
  86 "\n"
  87 "function search($file,$word,&$statsList)\n"
  88 "{\n"
  89 "  $index = computeIndex($word);\n"
  90 "  if ($index!=-1) // found a valid index\n"
  91 "  {\n"
  92 "    fseek($file,$index*4+4); // 4 bytes per entry, skip header\n"
  93 "    $index = readInt($file);\n"
  94 "    if ($index) // found words matching the hash key\n"
  95 "    {\n"
  96 "      $start=sizeof($statsList);\n"
  97 "      $count=$start;\n"
  98 "      fseek($file,$index);\n"
  99 "      $w = readString($file);\n"
 100 "      while ($w)\n"
 101 "      {\n"
 102 "        $statIdx = readInt($file);\n"
 103 "        if ($word==substr($w,0,strlen($word)))\n"
 104 "        { // found word that matches (as substring)\n"
 105 "          $statsList[$count++]=array(\n"
 106 "              \"word\"=>$word,\n"
 107 "              \"match\"=>$w,\n"
 108 "              \"index\"=>$statIdx,\n"
 109 "              \"full\"=>strlen($w)==strlen($word),\n"
 110 "              \"docs\"=>array()\n"
 111 "              );\n"
 112 "        }\n"
 113 "        $w = readString($file);\n"
 114 "      }\n"
 115 "      $totalHi=0;\n"
 116 "      $totalFreqHi=0;\n"
 117 "      $totalFreqLo=0;\n"
 118 "      for ($count=$start;$count<sizeof($statsList);$count++)\n"
 119 "      {\n"
 120 "        $statInfo = &$statsList[$count];\n"
 121 "        $multiplier = 1;\n"
 122 "        // whole word matches have a double weight\n"
 123 "        if ($statInfo[\"full\"]) $multiplier=2;\n"
 124 "        fseek($file,$statInfo[\"index\"]); \n"
 125 "        $numDocs = readInt($file);\n"
 126 "        $docInfo = array();\n"
 127 "        // read docs info + occurrence frequency of the word\n"
 128 "        for ($i=0;$i<$numDocs;$i++)\n"
 129 "        {\n"
 130 "          $idx=readInt($file); \n"
 131 "          $freq=readInt($file); \n"
 132 "          $docInfo[$i]=array(\"idx\"  => $idx,\n"
 133 "                             \"freq\" => $freq>>1,\n"
 134 "                             \"rank\" => 0.0,\n"
 135 "                             \"hi\"   => $freq&1\n"
 136 "                            );\n"
 137 "          if ($freq&1) // word occurs in high priority doc\n"
 138 "          {\n"
 139 "            $totalHi++;\n"
 140 "            $totalFreqHi+=$freq*$multiplier;\n"
 141 "          }\n"
 142 "          else // word occurs in low priority doc\n"
 143 "          {\n"
 144 "            $totalFreqLo+=$freq*$multiplier;\n"
 145 "          }\n"
 146 "        }\n"
 147 "        // read name and url info for the doc\n"
 148 "        for ($i=0;$i<$numDocs;$i++)\n"
 149 "        {\n"
 150 "          fseek($file,$docInfo[$i][\"idx\"]);\n"
 151 "          $docInfo[$i][\"name\"]=readString($file);\n"
 152 "          $docInfo[$i][\"url\"]=readString($file);\n"
 153 "        }\n"
 154 "        $statInfo[\"docs\"]=$docInfo;\n"
 155 "      }\n"
 156 "      $totalFreq=($totalHi+1)*$totalFreqLo + $totalFreqHi;\n"
 157 "      for ($count=$start;$count<sizeof($statsList);$count++)\n"
 158 "      {\n"
 159 "        $statInfo = &$statsList[$count];\n"
 160 "        $multiplier = 1;\n"
 161 "        // whole word matches have a double weight\n"
 162 "        if ($statInfo[\"full\"]) $multiplier=2;\n"
 163 "        for ($i=0;$i<sizeof($statInfo[\"docs\"]);$i++)\n"
 164 "        {\n"
 165 "          $docInfo = &$statInfo[\"docs\"];\n"
 166 "          // compute frequency rank of the word in each doc\n"
 167 "          $freq=$docInfo[$i][\"freq\"];\n"
 168 "          if ($docInfo[$i][\"hi\"])\n"
 169 "          {\n"
 170 "            $statInfo[\"docs\"][$i][\"rank\"]=\n"
 171 "              (float)($freq*$multiplier+$totalFreqLo)/$totalFreq;\n"
 172 "          }\n"
 173 "          else\n"
 174 "          {\n"
 175 "            $statInfo[\"docs\"][$i][\"rank\"]=\n"
 176 "              (float)($freq*$multiplier)/$totalFreq;\n"
 177 "          }\n"
 178 "        }\n"
 179 "      }\n"
 180 "    }\n"
 181 "  }\n"
 182 "  return $statsList;\n"
 183 "}\n"
 184 "\n"
 185 "function combine_results($results,&$docs)\n"
 186 "{\n"
 187 "  foreach ($results as $wordInfo)\n"
 188 "  {\n"
 189 "    $docsList = &$wordInfo[\"docs\"];\n"
 190 "    foreach ($docsList as $di)\n"
 191 "    {\n"
 192 "      $key=$di[\"url\"];\n"
 193 "      $rank=$di[\"rank\"];\n"
 194 "      if (isset($docs[$key]))\n"
 195 "      {\n"
 196 "        $docs[$key][\"rank\"]+=$rank;\n"
 197 "      }\n"
 198 "      else\n"
 199 "      {\n"
 200 "        $docs[$key] = array(\"url\"=>$key,\n"
 201 "            \"name\"=>$di[\"name\"],\n"
 202 "            \"rank\"=>$rank\n"
 203 "            );\n"
 204 "      }\n"
 205 "      $docs[$key][\"words\"][] = array(\n"
 206 "               \"word\"=>$wordInfo[\"word\"],\n"
 207 "               \"match\"=>$wordInfo[\"match\"],\n"
 208 "               \"freq\"=>$di[\"freq\"]\n"
 209 "               );\n"
 210 "    }\n"
 211 "  }\n"
 212 "  return $docs;\n"
 213 "}\n"
 214 "\n"
 215 "function filter_results($docs,&$requiredWords,&$forbiddenWords)\n"
 216 "{\n"
 217 "  $filteredDocs=array();\n"
 218 "  while (list ($key, $val) = each ($docs)) \n"
 219 "  {\n"
 220 "    $words = &$docs[$key][\"words\"];\n"
 221 "    $copy=1; // copy entry by default\n"
 222 "    if (sizeof($requiredWords)>0)\n"
 223 "    {\n"
 224 "      foreach ($requiredWords as $reqWord)\n"
 225 "      {\n"
 226 "        $found=0;\n"
 227 "        foreach ($words as $wordInfo)\n"
 228 "        { \n"
 229 "          $found = $wordInfo[\"word\"]==$reqWord;\n"
 230 "          if ($found) break;\n"
 231 "        }\n"
 232 "        if (!$found) \n"
 233 "        {\n"
 234 "          $copy=0; // document contains none of the required words\n"
 235 "          break;\n"
 236 "        }\n"
 237 "      }\n"
 238 "    }\n"
 239 "    if (sizeof($forbiddenWords)>0)\n"
 240 "    {\n"
 241 "      foreach ($words as $wordInfo)\n"
 242 "      {\n"
 243 "        if (in_array($wordInfo[\"word\"],$forbiddenWords))\n"
 244 "        {\n"
 245 "          $copy=0; // document contains a forbidden word\n"
 246 "          break;\n"
 247 "        }\n"
 248 "      }\n"
 249 "    }\n"
 250 "    if ($copy) $filteredDocs[$key]=$docs[$key];\n"
 251 "  }\n"
 252 "  return $filteredDocs;\n"
 253 "}\n"
 254 "\n"
 255 "function compare_rank($a,$b)\n"
 256 "{\n"
 257 "  if ($a[\"rank\"] == $b[\"rank\"]) \n"
 258 "  {\n"
 259 "    return 0;\n"
 260 "  }\n"
 261 "  return ($a[\"rank\"]>$b[\"rank\"]) ? -1 : 1; \n"
 262 "}\n"
 263 "\n"
 264 "function sort_results($docs,&$sorted)\n"
 265 "{\n"
 266 "  $sorted = $docs;\n"
 267 "  usort($sorted,\"compare_rank\");\n"
 268 "  return $sorted;\n"
 269 "}\n"
 270 "\n"
 271 "function report_results(&$docs)\n"
 272 "{\n"
 273 "  echo \"<div class=\\\"header\\\">\";\n"
 274 "  echo \"  <div class=\\\"headertitle\\\">\\n\";\n"
 275 "  echo \"    <h1>\".search_results().\"</h1>\\n\";\n"
 276 "  echo \"  </div>\\n\";\n"
 277 "  echo \"</div>\\n\";\n"
 278 "  echo \"<div class=\\\"searchresults\\\">\\n\";\n"
 279 "  echo \"<table cellspacing=\\\"2\\\">\\n\";\n"
 280 "  $numDocs = sizeof($docs);\n"
 281 "  if ($numDocs==0)\n"
 282 "  {\n"
 283 "    echo \"  <tr>\\n\";\n"
 284 "    echo \"    <td colspan=\\\"2\\\">\".matches_text(0).\"</td>\\n\";\n"
 285 "    echo \"  </tr>\\n\";\n"
 286 "  }\n"
 287 "  else\n"
 288 "  {\n"
 289 "    echo \"  <tr>\\n\";\n"
 290 "    echo \"    <td colspan=\\\"2\\\">\".matches_text($numDocs);\n"
 291 "    echo \"\\n\";\n"
 292 "    echo \"    </td>\\n\";\n"
 293 "    echo \"  </tr>\\n\";\n"
 294 "    $num=1;\n"
 295 "    foreach ($docs as $doc)\n"
 296 "    {\n"
 297 "      echo \"  <tr>\\n\";\n"
 298 "      echo \"    <td align=\\\"right\\\">$num.</td>\";\n"
 299 "      echo     \"<td><a class=\\\"el\\\" href=\\\"\".$doc[\"url\"].\"\\\">\".$doc[\"name\"].\"</a></td>\\n\";\n"
 300 "      echo \"  <tr>\\n\";\n"
 301 "      echo \"    <td></td><td class=\\\"tiny\\\">\".report_matches().\" \";\n"
 302 "      foreach ($doc[\"words\"] as $wordInfo)\n"
 303 "      {\n"
 304 "        $word = $wordInfo[\"word\"];\n"
 305 "        $matchRight = substr($wordInfo[\"match\"],strlen($word));\n"
 306 "        echo \"<b>$word</b>$matchRight(\".$wordInfo[\"freq\"].\") \";\n"
 307 "      }\n"
 308 "      echo \"    </td>\\n\";\n"
 309 "      echo \"  </tr>\\n\";\n"
 310 "      $num++;\n"
 311 "    }\n"
 312 "  }\n"
 313 "  echo \"</table>\\n\";\n"
 314 "  echo \"</div>\\n\";\n"
 315 "}\n"
 316 "\n"
 317 "function run_query($query)\n"
 318 "{\n"
 319 "  if(strcmp('4.1.0', phpversion()) > 0) \n"
 320 "  {\n"
 321 "    die(\"Error: PHP version 4.1.0 or above required!\");\n"
 322 "  }\n"
 323 "  if (!($file=fopen(\"search/search.idx\",\"rb\"))) \n"
 324 "  {\n"
 325 "    die(\"Error: Search index file could NOT be opened!\");\n"
 326 "  }\n"
 327 "  if (readHeader($file)!=\"DOXS\")\n"
 328 "  {\n"
 329 "    die(\"Error: Header of index file is invalid!\");\n"
 330 "  }\n"
 331 "  $results = array();\n"
 332 "  $requiredWords = array();\n"
 333 "  $forbiddenWords = array();\n"
 334 "  $foundWords = array();\n"
 335 "  $word=strtok($query,\" \");\n"
 336 "  while ($word) // for each word in the search query\n"
 337 "  {\n"
 338 "    if (($word{0}=='+')) { $word=substr($word,1); $requiredWords[]=$word; }\n"
 339 "    if (($word{0}=='-')) { $word=substr($word,1); $forbiddenWords[]=$word; }\n"
 340 "    if (!in_array($word,$foundWords))\n"
 341 "    {\n"
 342 "      $foundWords[]=$word;\n"
 343 "      search($file,strtolower($word),$results);\n"
 344 "    }\n"
 345 "    $word=strtok(\" \");\n"
 346 "  }\n"
 347 "  fclose($file);\n"
 348 "  $docs = array();\n"
 349 "  combine_results($results,$docs);\n"
 350 "  // filter out documents with forbidden word or that do not contain\n"
 351 "  // required words\n"
 352 "  $filteredDocs = filter_results($docs,$requiredWords,$forbiddenWords);\n"
 353 "  // sort the results based on rank\n"
 354 "  $sorted = array();\n"
 355 "  sort_results($filteredDocs,$sorted);\n"
 356 "  return $sorted;\n"
 357 "}\n"
 358 "\n"
 359 "function main()\n"
 360 "{\n"
 361 "  $query = \"\";\n"
 362 "  if (array_key_exists(\"query\", $_GET))\n"
 363 "  {\n"
 364 "    $query=$_GET[\"query\"];\n"
 365 "  }\n"
 366 "  $sorted = run_query($query);\n"
 367 "  // Now output the HTML stuff...\n"
 368 "  // End the HTML form\n"
 369 "  end_form(preg_replace(\"/[^a-zA-Z0-9\\-\\_\\.]/i\", \" \", $query ));\n"
 370 "  // report results to the user\n"
 371 "  report_results($sorted);\n"
 372 "  end_page();\n"
 373 "}\n"
 374 "</script>\n"