Jan Tattermusch | f209c58 | 2015-08-24 17:47:35 -0700 | [diff] [blame] | 1 | <? |
| 2 | // Contributed to the Sandcastle Help File Builder project by Thomas Levesque |
| 3 | |
| 4 | class Ranking |
| 5 | { |
| 6 | public $filename; |
| 7 | public $pageTitle; |
| 8 | public $rank; |
| 9 | |
| 10 | function __construct($file, $title, $rank) |
| 11 | { |
| 12 | $this->filename = $file; |
| 13 | $this->pageTitle = $title; |
| 14 | $this->rank = $rank; |
| 15 | } |
| 16 | } |
| 17 | |
| 18 | |
| 19 | /// <summary> |
| 20 | /// Split the search text up into keywords |
| 21 | /// </summary> |
| 22 | /// <param name="keywords">The keywords to parse</param> |
| 23 | /// <returns>A list containing the words for which to search</returns> |
| 24 | function ParseKeywords($keywords) |
| 25 | { |
| 26 | $keywordList = array(); |
| 27 | $words = preg_split("/[^\w]+/", $keywords); |
| 28 | |
| 29 | foreach($words as $word) |
| 30 | { |
| 31 | $checkWord = strtolower($word); |
| 32 | $first = substr($checkWord, 0, 1); |
| 33 | if(strlen($checkWord) > 2 && !ctype_digit($first) && !in_array($checkWord, $keywordList)) |
| 34 | { |
| 35 | array_push($keywordList, $checkWord); |
| 36 | } |
| 37 | } |
| 38 | |
| 39 | return $keywordList; |
| 40 | } |
| 41 | |
| 42 | |
| 43 | /// <summary> |
| 44 | /// Search for the specified keywords and return the results as a block of |
| 45 | /// HTML. |
| 46 | /// </summary> |
| 47 | /// <param name="keywords">The keywords for which to search</param> |
| 48 | /// <param name="fileInfo">The file list</param> |
| 49 | /// <param name="wordDictionary">The dictionary used to find the words</param> |
| 50 | /// <param name="sortByTitle">True to sort by title, false to sort by |
| 51 | /// ranking</param> |
| 52 | /// <returns>A block of HTML representing the search results.</returns> |
| 53 | function Search($keywords, $fileInfo, $wordDictionary, $sortByTitle) |
| 54 | { |
| 55 | $sb = "<ol>"; |
| 56 | $matches = array(); |
| 57 | $matchingFileIndices = array(); |
| 58 | $rankings = array(); |
| 59 | |
| 60 | $isFirst = true; |
| 61 | |
| 62 | foreach($keywords as $word) |
| 63 | { |
| 64 | if (!array_key_exists($word, $wordDictionary)) |
| 65 | { |
| 66 | return "<strong>Nothing found</strong>"; |
| 67 | } |
| 68 | $occurrences = $wordDictionary[$word]; |
| 69 | |
| 70 | $matches[$word] = $occurrences; |
| 71 | $occurrenceIndices = array(); |
| 72 | |
| 73 | // Get a list of the file indices for this match |
| 74 | foreach($occurrences as $entry) |
| 75 | array_push($occurrenceIndices, ($entry >> 16)); |
| 76 | |
| 77 | if($isFirst) |
| 78 | { |
| 79 | $isFirst = false; |
| 80 | foreach($occurrenceIndices as $i) |
| 81 | { |
| 82 | array_push($matchingFileIndices, $i); |
| 83 | } |
| 84 | } |
| 85 | else |
| 86 | { |
| 87 | // After the first match, remove files that do not appear for |
| 88 | // all found keywords. |
| 89 | for($idx = 0; $idx < count($matchingFileIndices); $idx++) |
| 90 | { |
| 91 | if (!in_array($matchingFileIndices[$idx], $occurrenceIndices)) |
| 92 | { |
| 93 | array_splice($matchingFileIndices, $idx, 1); |
| 94 | $idx--; |
| 95 | } |
| 96 | } |
| 97 | } |
| 98 | } |
| 99 | |
| 100 | if(count($matchingFileIndices) == 0) |
| 101 | { |
| 102 | return "<strong>Nothing found</strong>"; |
| 103 | } |
| 104 | |
| 105 | // Rank the files based on the number of times the words occurs |
| 106 | foreach($matchingFileIndices as $index) |
| 107 | { |
| 108 | // Split out the title, filename, and word count |
| 109 | $fileIndex = explode("\x00", $fileInfo[$index]); |
| 110 | |
| 111 | $title = $fileIndex[0]; |
| 112 | $filename = $fileIndex[1]; |
| 113 | $wordCount = intval($fileIndex[2]); |
| 114 | $matchCount = 0; |
| 115 | |
| 116 | foreach($keywords as $words) |
| 117 | { |
| 118 | $occurrences = $matches[$word]; |
| 119 | |
| 120 | foreach($occurrences as $entry) |
| 121 | { |
| 122 | if(($entry >> 16) == $index) |
| 123 | $matchCount += $entry & 0xFFFF; |
| 124 | } |
| 125 | } |
| 126 | |
| 127 | $r = new Ranking($filename, $title, $matchCount * 1000 / $wordCount); |
| 128 | array_push($rankings, $r); |
| 129 | |
| 130 | if(count($rankings) > 99) |
| 131 | break; |
| 132 | } |
| 133 | |
| 134 | // Sort by rank in descending order or by page title in ascending order |
| 135 | if($sortByTitle) |
| 136 | { |
| 137 | usort($rankings, "cmprankbytitle"); |
| 138 | } |
| 139 | else |
| 140 | { |
| 141 | usort($rankings, "cmprank"); |
| 142 | } |
| 143 | |
| 144 | // Format the file list and return the results |
| 145 | foreach($rankings as $r) |
| 146 | { |
| 147 | $f = $r->filename; |
| 148 | $t = $r->pageTitle; |
| 149 | $sb .= "<li><a href=\"$f\" target=\"_blank\">$t</a></li>"; |
| 150 | } |
| 151 | |
| 152 | $sb .= "</ol"; |
| 153 | |
| 154 | if(count($rankings) < count($matchingFileIndices)) |
| 155 | { |
| 156 | $c = count(matchingFileIndices) - count(rankings); |
| 157 | $sb .= "<p>Omitted $c more results</p>"; |
| 158 | } |
| 159 | |
| 160 | return $sb; |
| 161 | } |
| 162 | |
| 163 | function cmprank($x, $y) |
| 164 | { |
| 165 | return $y->rank - $x->rank; |
| 166 | } |
| 167 | |
| 168 | function cmprankbytitle($x, $y) |
| 169 | { |
| 170 | return strcmp($x->pageTitle, $y->pageTitle); |
| 171 | } |
| 172 | |
| 173 | ?> |