<?php
require "strus.php";
$config = "path=storage; metadata=doclen UINT16";
$ctx = new StrusContext();
try {
$ctx->destroyStorage($config);
# ... delete the storage files if they already exists
} catch (Exception $e) {
}
# Create a new storage:
$ctx->createStorage($config);
# Get a client for the new created storage:
$storage = $ctx->createStorageClient($config);
# Define the document analyzer to use:
$analyzer = $ctx->createDocumentAnalyzer();
# Define the features and attributes to store:
$analyzer->defineAttribute("title", "/doc/title()", "content", "orig");
$analyzer->addSearchIndexFeature("word", "/doc/text()", "word", array(array("stem", "en"), "lc", array("convdia", "en")), "");
$analyzer->addForwardIndexFeature("orig", "/doc/text()", "split", "orig", "");
# Read input files, analyze and insert them:
$datadir = "./data/";
$datafiles = scandir($datadir);
$transaction = $storage->createTransaction();
foreach ($datafiles as &$datafile) {
if ($datafile[0] != '.') {
print "process document " . $datafile . "\n";
$docid = substr($datafile, 0, strlen($datafile) - 4);
$fullname = $datadir . $datafile;
$doc = $analyzer->analyze(file_get_contents($fullname));
$transaction->insertDocument($docid, $doc);
<!DOCTYPE html>
<html>
<body>
<?php
require "strus.php";
try {
$context = new StrusContext("localhost:7181");
$storage = $context->createStorageClient("");
echo '<p>';
echo "Number of documents inserted: " . $storage->nofDocumentsInserted() . "!";
echo '</p>';
} catch (Exception $e) {
echo '<p><font color="red">Caught exception: ' . "{$e->getMessage}()</font></p>";
}
?>
</body>
</html>
$restrictdoc = NULL;
parse_str(getenv('QUERY_STRING'), $_GET);
$queryString = $_GET['q'];
if (array_key_exists('n', $_GET)) {
$nofRanks = intval($_GET['n']);
}
if (array_key_exists('i', $_GET)) {
$minRank = intval($_GET['i']);
}
if (array_key_exists('s', $_GET)) {
$scheme = $_GET['s'];
}
if (array_key_exists('d', $_GET)) {
$restrictdoc = intval($_GET['d']);
}
$context = new StrusContext("localhost:7181");
$storage = $context->createStorageClient("");
$BM25_checked = "";
$BM25pff_checked = "";
$NBLNK_checked = "";
if (!$scheme || $scheme == 'BM25') {
$BM25_checked = "checked";
} else {
if ($scheme == 'BM25pff') {
$BM25pff_checked = "checked";
} else {
if ($scheme == 'NBLNK') {
$NBLNK_checked = "checked";
}
}
}
<?php
require "strus.php";
$queryphrase = "city";
$config = "path=storage";
$ctx = new StrusContext();
try {
# Get a client for the storage:
$storage = $ctx->createStorageClient($config);
# Define the query analyzer to use:
$analyzer = $ctx->createQueryAnalyzer();
$analyzer->addSearchIndexElement("word", "word", "word", array(array("stem", "en"), "lc", array("convdia", "en")));
# Define the query evaluation scheme:
$queryeval = $ctx->createQueryEval();
# Here we define what query features decide, what is ranked for the result:
$queryeval->addSelectionFeature("select");
# Here we define how we rank a document selected. We use the 'BM25' weighting scheme:
$queryeval->addWeightingFunction("BM25", array("k1" => 0.75, "b" => 2.1, "avgdoclen" => 1000, ".match" => "seek"));
# Now we define what attributes of the documents are returned and how they are build.
# The functions that extract stuff from documents for presentation are called summarizers.
# First we add a summarizer that extracts us the title of the document:
$queryeval->addSummarizer("attribute", array("name" => "title"));
# Then we add a summarizer that collects the sections that enclose the best matches
# in a ranked document:
$queryeval->addSummarizer("matchphrase", array("type" => "orig", "sentencesize" => 40, "windowsize" => 60, ".match" => "seek"));
# Now we build the query to issue:
$query = $queryeval->createQuery($storage);
# First we analyze the query phrase to get the terms to find in the form as they are stored in the storage:
$terms = $analyzer->analyzeField("word", $queryphrase);
if (count($terms) == 0) {
throw new Exception("query is empty");
<?php
require "strus.php";
$ctx = new StrusContext();
$config = "path=storage; metadata=doclen UINT16";
$storage = $ctx->createStorageClient($config);
# Query the number of documents inserted:
$nofDocuments = $storage->nofDocumentsInserted();
# Output:
print "Number of documents inserted: {$nofDocuments}\n";
?>
