initial implementation

cebe · Apr 24, 2014 · 90226a3 · 90226a3
1 parent 698ffab
commit 90226a3
Show file tree

Hide file tree

Showing 12 changed files with 653 additions and 0 deletions.
diff --git a/.gitignore b/.gitignore
@@ -0,0 +1,3 @@
+jssearch.index.js
+
+/vendor/
diff --git a/bin/jsindex b/bin/jsindex
@@ -0,0 +1,133 @@
+#!/usr/bin/env php
+<?php
+
+// Send all errors to stderr
+ini_set('display_errors', 'stderr');
+
+// setup composer autoloading
+$composerAutoload = [
+	__DIR__ . '/../vendor/autoload.php', // standalone with "composer install" run
+	__DIR__ . '/../../../autoload.php', // script is installed as a composer binary
+];
+foreach ($composerAutoload as $autoload) {
+	if (file_exists($autoload)) {
+		require($autoload);
+		break;
+	}
+}
+
+if (!class_exists('cebe\jssearch\Indexer')) {
+	error('Autoloading does not seem to work. Looks like you should run `composer install` first.');
+}
+
+// check arguments
+$src = [];
+foreach($argv as $k => $arg) {
+	if ($k == 0) {
+		continue;
+	}
+	if ($arg[0] == '-') {
+		$arg = explode('=', $arg);
+		switch($arg[0]) {
+			// TODO allow baseUrl to be set via arg
+			case '-h':
+			case '--help':
+				echo "jssearch index builder\n";
+				echo "----------------------\n\n";
+				echo "by Carsten Brandt <mail@cebe.cc>\n\n";
+				usage();
+				break;
+			default:
+				error("Unknown argument " . $arg[0], "usage");
+		}
+	} else {
+		$src[] = $arg;
+	}
+}
+
+if (empty($src)) {
+	error("You have to give an input directory.", "usage");
+}
+
+$indexer = new \cebe\jssearch\Indexer();
+
+$files = [];
+foreach($src as $dir) {
+	$files = array_merge($files, findFiles($dir));
+
+	if (empty($files)) {
+		error("No files where found in $dir.");
+	}
+
+	$indexer->indexFiles($files, $dir);
+}
+
+$js = $indexer->exportJs();
+file_put_contents('jssearch.index.js', $js);
+
+
+// functions
+
+/**
+ * Display usage information
+ */
+function usage() {
+	global $argv;
+	$cmd = $argv[0];
+	echo <<<EOF
+Usage:
+    $cmd [src-directory]
+
+    --help    shows this usage information.
+
+    creates and jssearch.index.js file in the current directory.
+
+EOF;
+	exit(1);
+}
+
+/**
+ * Send custom error message to stderr
+ * @param $message string
+ * @param $callback mixed called before script exit
+ * @return void
+ */
+function error($message, $callback = null) {
+	$fe = fopen("php://stderr", "w");
+	fwrite($fe, "Error: " . $message . "\n");
+
+	if (is_callable($callback)) {
+		call_user_func($callback);
+	}
+
+	exit(1);
+}
+
+function findFiles($dir, $ext = '.html')
+{
+	if (!is_dir($dir)) {
+		error("$dir is not a directory.");
+	}
+	$dir = rtrim($dir, DIRECTORY_SEPARATOR);
+	$list = [];
+	$handle = opendir($dir);
+	if ($handle === false) {
+		error('Unable to open directory: ' . $dir);
+	}
+	while (($file = readdir($handle)) !== false) {
+		if ($file === '.' || $file === '..') {
+			continue;
+		}
+		$path = $dir . DIRECTORY_SEPARATOR . $file;
+		if (substr($file, -($l = strlen($ext)), $l) === $ext) {
+			if (is_file($path)) {
+				$list[] = $path;
+			} else {
+				$list = array_merge($list, findFiles($path, $ext));
+			}
+		}
+	}
+	closedir($handle);
+
+	return $list;
+}
diff --git a/composer.json b/composer.json
@@ -0,0 +1,19 @@
+{
+    "name": "cebe/js-search",
+    "description": "A client side search engine for use on static pages.",
+    "license": "MIT",
+    "authors": [
+        {
+            "name": "Carsten Brandt",
+            "email": "mail@cebe.cc"
+        }
+    ],
+    "require": {
+		"php": ">=5.4.0"
+    },
+	"autoload": {
+		"psr-4": {
+			"cebe\\jssearch\\": "lib/"
+		}
+	}
+}
diff --git a/example.html b/example.html
@@ -0,0 +1,50 @@
+<!doctype html>
+<html>
+<head>
+	<meta charset="UTF-8">
+
+	<script src="http://code.jquery.com/jquery-1.11.0.min.js"></script>
+	<script src="./jssearch.js"></script>
+	<script src="./jssearch.index.js"></script>
+
+	<script type="text/javascript">
+
+		$( document ).ready(function() {
+			$('#searchbox').on("keyup", function() {
+				var result = $.jssearch.search($(this).val());
+
+				$('#query').html($.jssearch.queryWords.join(' '));
+
+				$('#results').html('');
+				var i = 0;
+				result.forEach(function(item) {
+					if (i++ > 20) {
+						return;
+					}
+					var div = $('#results');
+					div.html(div.html() + '<li>"' + item.file.title + '" ' + item.file.url + ' w:' + item.weight + '</li>');
+				});
+			});
+		});
+	</script>
+
+	<title>Example</title>
+</head>
+<body>
+
+	<h1>Example</h1>
+
+	<label for="searchbox" style="display: inline-block; width: 160px;">Search: </label>
+	<input id="searchbox" type="text" value="">
+
+	<br/>
+
+	<label for="query" style="display: inline-block; width: 160px;">Actual query: </label>
+	<span id="query"></span>
+
+	<ul id="results">
+		<li>No results</li>
+	</ul>
+
+</body>
+</html>
diff --git a/jssearch.js b/jssearch.js
@@ -0,0 +1,80 @@
+
+$.jssearch = {
+
+	/**
+	 * the actual words finally used to query (set by last search call)
+	 */
+	queryWords: [],
+
+	search: function(query) {
+		var words = $.jssearch.tokenizeString(query);
+		var result = {};
+
+		$.jssearch.queryWords = words.map(function(i) { return i.t; });
+
+		// do not search when no words given
+		if (words.length == 0) {
+			return result;
+		}
+
+//		result = $.jssearch.searchForWords(words);
+//		if ($.isEmptyObject(result)) {
+			words = $.jssearch.completeWords(words);
+			$.jssearch.queryWords = words.map(function(i) { return i.t; });
+			result = $.jssearch.searchForWords(words);
+//		}
+
+		var res = [];
+		for (var i in result) {
+			res.push(result[i]);
+		}
+		res.sort(function(a,b) { return b.weight - a.weight; });
+		return res;
+	},
+
+	searchForWords: function(words) {
+		var result = {};
+		words.forEach(function(word) {
+			if ($.jssearch.index[word.t]) {
+				$.jssearch.index[word.t].forEach(function(file) {
+					if (result[file.f]) {
+						result[file.f].weight *= file.w * word.w;
+					} else {
+						result[file.f] = {
+							file: $.jssearch.files[file.f],
+							weight: file.w * word.w
+						};
+					}
+				});
+			}
+		});
+		return result;
+	},
+
+	completeWords: function(words) {
+		var result = [];
+
+		words.forEach(function(word) {
+			if (!$.jssearch.index[word.t] && word.t.length > 2) {
+				// complete words that are not in the index
+				for(var w in $.jssearch.index) {
+					if (w.substr(0, word.t.length) == word.t) {
+						result.push({t: w, w: 1});
+					}
+				}
+			} else {
+				// keep existing words
+				result.push(word);
+			}
+		});
+		return result;
+	},
+
+	tokenizeString: function(string)
+	{
+		if (console) {
+			console.log('Error: tokenizeString should have been overwritten by index JS file.')
+		}
+		return [{t: string, w: 1}];
+	}
+};
diff --git a/lib/AnalyzerInterface.php b/lib/AnalyzerInterface.php
@@ -0,0 +1,28 @@
+<?php
+/**
+ * @copyright Copyright (c) 2014 Carsten Brandt
+ * @license https://github.com/cebe/js-search/blob/master/LICENSE
+ * @link https://github.com/cebe/js-search#readme
+ */
+
+namespace cebe\jssearch;
+
+/**
+ * Interface for all Tokenizers.
+ *
+ * @author Carsten Brandt <mail@cebe.cc>
+ */
+interface AnalyzerInterface
+{
+	/**
+	 * Analyzes a string and returns an array of the following format:
+	 *
+	 * TODO
+	 * ```
+	 * ```
+	 *
+	 * @param string $string the string to analyze
+	 * @return array
+	 */
+	public function analyze($string, TokenizerInterface $tokenizer);
+}