X-Git-Url: http://v3vee.org/palacios/gitweb/gitweb.cgi?p=palacios.git;a=blobdiff_plain;f=misc%2Fdecoder_test%2FXED2%2Fdoc%2Fhtml%2Fsearch.php;fp=misc%2Fdecoder_test%2FXED2%2Fdoc%2Fhtml%2Fsearch.php;h=b8185c2f98af018854cae1d45e3352d0d466548d;hp=0000000000000000000000000000000000000000;hb=ddc16b0737cf58f7aa90a69c6652cdf4090aec51;hpb=626595465a2c6987606a6bc697df65130ad8c2d3 diff --git a/misc/decoder_test/XED2/doc/html/search.php b/misc/decoder_test/XED2/doc/html/search.php new file mode 100644 index 0000000..b8185c2 --- /dev/null +++ b/misc/decoder_test/XED2/doc/html/search.php @@ -0,0 +1,382 @@ + + +XED2: XED2 User Guide - Thu May 15 03:15:09 2008 + + + + +
+ \n
\n"; +} + +function readInt($file) +{ + $b1 = ord(fgetc($file)); $b2 = ord(fgetc($file)); + $b3 = ord(fgetc($file)); $b4 = ord(fgetc($file)); + return ($b1<<24)|($b2<<16)|($b3<<8)|$b4; +} + +function readString($file) +{ + $result=""; + while (ord($c=fgetc($file))) $result.=$c; + return $result; +} + +function readHeader($file) +{ + $header =fgetc($file); $header.=fgetc($file); + $header.=fgetc($file); $header.=fgetc($file); + return $header; +} + +function computeIndex($word) +{ + // Fast string hashing + //$lword = strtolower($word); + //$l = strlen($lword); + //for ($i=0;$i<$l;$i++) + //{ + // $c = ord($lword{$i}); + // $v = (($v & 0xfc00) ^ ($v << 6) ^ $c) & 0xffff; + //} + //return $v; + + // Simple hashing that allows for substring search + if (strlen($word)<2) return -1; + // high char of the index + $hi = ord($word{0}); + if ($hi==0) return -1; + // low char of the index + $lo = ord($word{1}); + if ($lo==0) return -1; + // return index + return $hi*256+$lo; +} + +function search($file,$word,&$statsList) +{ + $index = computeIndex($word); + if ($index!=-1) // found a valid index + { + fseek($file,$index*4+4); // 4 bytes per entry, skip header + $index = readInt($file); + if ($index) // found words matching the hash key + { + $start=sizeof($statsList); + $count=$start; + fseek($file,$index); + $w = readString($file); + while ($w) + { + $statIdx = readInt($file); + if ($word==substr($w,0,strlen($word))) + { // found word that matches (as substring) + $statsList[$count++]=array( + "word"=>$word, + "match"=>$w, + "index"=>$statIdx, + "full"=>strlen($w)==strlen($word), + "docs"=>array() + ); + } + $w = readString($file); + } + $totalHi=0; + $totalFreqHi=0; + $totalFreqLo=0; + for ($count=$start;$count $idx, + "freq" => $freq>>1, + "rank" => 0.0, + "hi" => $freq&1 + ); + if ($freq&1) // word occurs in high priority doc + { + $totalHi++; + $totalFreqHi+=$freq*$multiplier; + } + else // word occurs in low priority doc + { + $totalFreqLo+=$freq*$multiplier; + } + } + // read name and url info for the doc + for ($i=0;$i<$numDocs;$i++) + { + fseek($file,$docInfo[$i]["idx"]); + $docInfo[$i]["name"]=readString($file); + $docInfo[$i]["url"]=readString($file); + } + $statInfo["docs"]=$docInfo; + } + $totalFreq=($totalHi+1)*$totalFreqLo + $totalFreqHi; + for ($count=$start;$count$key, + "name"=>$di["name"], + "rank"=>$rank + ); + } + $docs[$key]["words"][] = array( + "word"=>$wordInfo["word"], + "match"=>$wordInfo["match"], + "freq"=>$di["freq"] + ); + } + } + return $docs; +} + +function filter_results($docs,&$requiredWords,&$forbiddenWords) +{ + $filteredDocs=array(); + while (list ($key, $val) = each ($docs)) + { + $words = &$docs[$key]["words"]; + $copy=1; // copy entry by default + if (sizeof($requiredWords)>0) + { + foreach ($requiredWords as $reqWord) + { + $found=0; + foreach ($words as $wordInfo) + { + $found = $wordInfo["word"]==$reqWord; + if ($found) break; + } + if (!$found) + { + $copy=0; // document contains none of the required words + break; + } + } + } + if (sizeof($forbiddenWords)>0) + { + foreach ($words as $wordInfo) + { + if (in_array($wordInfo["word"],$forbiddenWords)) + { + $copy=0; // document contains a forbidden word + break; + } + } + } + if ($copy) $filteredDocs[$key]=$docs[$key]; + } + return $filteredDocs; +} + +function compare_rank($a,$b) +{ + if ($a["rank"] == $b["rank"]) + { + return 0; + } + return ($a["rank"]>$b["rank"]) ? -1 : 1; +} + +function sort_results($docs,&$sorted) +{ + $sorted = $docs; + usort($sorted,"compare_rank"); + return $sorted; +} + +function report_results(&$docs) +{ + echo "\n"; + echo " \n"; + echo " \n"; + echo " \n"; + $numDocs = sizeof($docs); + if ($numDocs==0) + { + echo " \n"; + echo " \n"; + echo " \n"; + } + else + { + echo " \n"; + echo " \n"; + echo " \n"; + $num=1; + foreach ($docs as $doc) + { + echo " \n"; + echo " "; + echo "\n"; + echo " \n"; + echo " \n"; + echo " \n"; + $num++; + } + } + echo "

".search_results()."

".matches_text(0)."
".matches_text($numDocs); + echo "\n"; + echo "
$num.".$doc["name"]."
".report_matches()." "; + foreach ($doc["words"] as $wordInfo) + { + $word = $wordInfo["word"]; + $matchRight = substr($wordInfo["match"],strlen($word)); + echo "$word$matchRight(".$wordInfo["freq"].") "; + } + echo "
\n"; +} + +function main() +{ + if(strcmp('4.1.0', phpversion()) > 0) + { + die("Error: PHP version 4.1.0 or above required!"); + } + if (!($file=fopen("search.idx","rb"))) + { + die("Error: Search index file could NOT be opened!"); + } + if (readHeader($file)!="DOXS") + { + die("Error: Header of index file is invalid!"); + } + $query=""; + if (array_key_exists("query", $_GET)) + { + $query=$_GET["query"]; + } + end_form($query); + echo " \n
\n"; + $results = array(); + $requiredWords = array(); + $forbiddenWords = array(); + $foundWords = array(); + $word=strtok($query," "); + while ($word) // for each word in the search query + { + if (($word{0}=='+')) { $word=substr($word,1); $requiredWords[]=$word; } + if (($word{0}=='-')) { $word=substr($word,1); $forbiddenWords[]=$word; } + if (!in_array($word,$foundWords)) + { + $foundWords[]=$word; + search($file,strtolower($word),$results); + } + $word=strtok(" "); + } + $docs = array(); + combine_results($results,$docs); + // filter out documents with forbidden word or that do not contain + // required words + $filteredDocs = filter_results($docs,$requiredWords,$forbiddenWords); + // sort the results based on rank + $sorted = array(); + sort_results($filteredDocs,$sorted); + // report results to the user + report_results($sorted); + echo "
\n"; + fclose($file); +} + +main(); + + +?> +
Generated on Thu May 15 03:15:09 2008 for XED2 by  + +doxygen 1.4.6
+ +