added unit tests, and inserted the verbalizer class to the main context

Tests/mind3rd/API/languages/pt-BR/IgnoreFormsTest.php

@@ -0,0 +1,49 @@
+<?php
+
+require_once dirname(__FILE__) . '/../../../../../mind3rd/API/languages/pt-BR/IgnoreForms.php';
+
+/**
+ * Test class for IgnoreForms.
+ * Generated by PHPUnit on 2010-12-19 at 11:23:40.
+ */
+class IgnoreFormsTest extends PHPUnit_Framework_TestCase {
+
+	/**
+	 * @var IgnoreForms
+	 */
+	protected $object;
+
+	/**
+	 * Sets up the fixture, for example, opens a network connection.
+	 * This method is called before a test is executed.
+	 */
+	protected function setUp() {
+		$this->object = new IgnoreForms;
+	}
+
+	/**
+	 * Tears down the fixture, for example, closes a network connection.
+	 * This method is called after a test is executed.
+	 */
+	protected function tearDown() {
+
+	}
+
+	public function testShouldBeIgnored() {
+		$this->assertTrue(IgnoreForms::shouldBeIgnored('felizmente'));
+	}
+	public function testShouldBeIgnored1() {
+		$this->assertTrue(IgnoreForms::shouldBeIgnored('provavelmente'));
+	}
+
+
+	public function testShouldBeUsed() {
+		$this->assertTrue(IgnoreForms::shouldBeUsed('trabalhará'));
+	}
+	public function testShouldBeUsed1() {
+		$this->assertTrue(IgnoreForms::shouldBeUsed('mente'));
+	}
+
+}
+
+?>

Tests/mind3rd/API/languages/pt-BR/InflectTest.php

@@ -53,6 +53,12 @@ class InflectTest extends PHPUnit_Framework_TestCase {
 	public function testisSingular7() {
 		$this->assertTrue(Inflect::isSingular('atlas'));
 	}
+	public function testisSingular8() {
+		$this->assertTrue(Inflect::isSingular('professor'));
+	}
+	public function testisSingular9() {
+		$this->assertFalse(Inflect::isSingular('professores'));
+	}
 
 	public function testToPlural() {
 		$this->assertEquals(Inflect::toPlural('homem'), 'homens');
@@ -103,6 +109,9 @@ class InflectTest extends PHPUnit_Framework_TestCase {
 	public function testToSingular7() {
 		$this->assertEquals(Inflect::toSingular('alunas'), 'aluna');
 	}
+	public function testToSingular8() {
+		$this->assertEquals(Inflect::toSingular('professores'), 'professor');
+	}
 
 	public function testToFemale() {
 		$this->assertEquals(Inflect::toFemale('aluno'), 'aluna');
@@ -150,6 +159,9 @@ class InflectTest extends PHPUnit_Framework_TestCase {
 	public function testToMale8() {
 		$this->assertEquals(Inflect::toMale('doutora'), 'doutor');
 	}
+	public function testToMale9() {
+		$this->assertEquals(Inflect::toMale('professora'), 'professor');
+	}
 
 	public function testIsFemale() {
 		$this->assertTrue(Inflect::isFemale('mulher'));
@@ -181,5 +193,8 @@ class InflectTest extends PHPUnit_Framework_TestCase {
 	public function testIsFemale9() {
 		$this->assertTrue(Inflect::isFemale('pequenininha'));
 	}
+	public function testIsFemale10() {
+		$this->assertFalse(Inflect::isFemale('professor'));
+	}
 }
 ?>

Tests/mind3rd/API/languages/pt-BR/VerbalizerTest.php

@@ -1,5 +1,6 @@
 <?php
 
+require_once dirname(__FILE__) . '/../../../../../mind3rd/API/classes/Mind.php';
 require_once dirname(__FILE__) . '/../../../../../mind3rd/API/languages/pt-BR/Verbalizer.php';
 
 /**
@@ -92,7 +93,6 @@ class VerbalizerTest extends PHPUnit_Framework_TestCase {
 	public function testIsVerb20() {
 		$this->assertFalse(Verbalizer::isVerb('violão'));
 	}
-
 	public function testIsVerb21() {
 		$this->assertTrue(Verbalizer::isVerb('falhar'));
 	}
@@ -126,9 +126,12 @@ class VerbalizerTest extends PHPUnit_Framework_TestCase {
 	public function testIsVerb31() {
 		$this->assertTrue(Verbalizer::isVerb('comeu'));
 	}
+	public function testIsVerb32() {
+		$this->assertFalse(Verbalizer::isVerb('professor'));
+	}
 
 	// testing the toInfinitive method
-	// NOTICE that its goas is about the present/future words...
+	// NOTICE that its goals is about the present/future words...
 	// past is not supported
 	public function testToInfinitive1() {
 		$this->assertEquals('saber', Verbalizer::toInfinitive('sei'));

Tests/mind3rd/API/languages/pt-BR/ignore.list

@@ -0,0 +1,4 @@
+de
+que
+da
+do

mind3rd/API/classes/Mind.php

@@ -24,6 +24,7 @@
 		public static $tokenizer;
 		public static $langPath= "";
 		public static $content= "";
+		public static $curLang= 'en';
 
 		/**
 		 * This method returns or outputs messages using the L10N library
@@ -94,8 +95,12 @@
 			$this->defaults= parse_ini_file($path.'/mind3rd/env/defaults.ini');
 			$this->conf= parse_ini_file($path.'/mind3rd/env/mind.ini');
 			include($path.'/mind3rd/API/L10N/'.$this->defaults['default_human_language'].'.php');
+			Mind::$curLang= $this->defaults['default_human_languageName'];
 			Mind::$l10n= new $this->defaults['default_human_language']();
 			Mind::$langPath= $path.'/mind3rd/API/languages/';
+			Mind::$curLang= $this->defaults['default_human_language'];
+			$langPath= $path.'/mind3rd/API/languages/'.$this->defaults['default_human_languageName'].'/';
+			set_include_path(get_include_path() . PATH_SEPARATOR . $langPath);
 		}
 		/**
 		* function taken from: http://www.dasprids.de/blog/2008/08/22/getting-a-password-hidden-from-stdin-with-php-cli

mind3rd/API/cortex/Lexer/Lexer.php

@@ -1,12 +1,8 @@
 <?php
-/* 
- * To change this template, choose Tools | Templates
- * and open the template in the editor.
- */
-
 /**
- * Description of Lexer
- *
+ * This class sweeps the passed words and validates if
+ * its content matches with the valid chars
+ * @package cortex.analyst
  * @author felipe
  */
 class Lexer
@@ -16,6 +12,7 @@ class Lexer
 	public  $lang= 'en';
 	private $content= "";
 	private $tokens= Array();
+	private $glue= '';
 
 	/**
 	 * Thanks to saeedco (no more information found about him!)
@@ -58,6 +55,11 @@ class Lexer
 		return $array;
 	}
 
+	/**
+	 * Validates the sent char
+	 * @param char $letter
+	 * @return boolean
+	 */
 	private function isValidChar($letter)
 	{
 		$letter= ($letter);
@@ -74,6 +76,12 @@ class Lexer
 		return false;
 	}
 
+	/**
+	 * Performs a sweep over the sent content and replace any
+	 * special char, also removing any invalid char.
+	 * @param string $content
+	 * @return string The cleaned content
+	 */
     public function sweep($content)
 	{
 		$this->content= trim(str_replace("\n", ' ', $content));
@@ -84,13 +92,15 @@ class Lexer
 		// the fixed content;
 		$fixed= "";
 		// for each charactere on the content
+		// let's remove th invalid ones
 		for($i=0, $j=sizeof($this->content); $i<$j; $i++)
 		{
 			$letter= $this->content[$i];
 			if($this->isValidChar($letter))
 				$fixed.= $letter;
 		}
-		
+
+		// preparing the tokens
 		foreach($this->tokens as $char=>$token)
 		{
 			$fixed= str_replace($char, $token, $fixed);
@@ -101,16 +111,22 @@ class Lexer
 		return sizeof(Mind::$content)>0? Mind::$content: false;
 	}
 
+	/**
+	 * Returns the fixed char to the sent char
+	 * @param string $char
+	 * @return string
+	 */
 	public function translateChars($str)
 	{
 		$from = $this->replacements[0];
 		$to   = $this->replacements[1];
 		return strtr($str, $from, $to);
 	}
-
+	
 	public function __construct()
 	{
 		GLOBAL $_MIND;
+		$this->glue= chr('176');
 		$this->lang= Mind::$l10n->name;
 		$xml= simplexml_load_file(Mind::$langPath.$this->lang.'/lexics.xml');
 		include(Mind::$langPath.$this->lang.'/Inflect.php');
@@ -125,8 +141,9 @@ class Lexer
 		$this->replacements[0]= (string)$xml->replacements->from;
 		$this->replacements[1]= (string)$xml->replacements->to;
 		
-		$this->tokens[' ']= chr('176');
-		$this->tokens['.']= chr('176').'.'.chr('176');
+		$this->tokens[' ']= $this->glue;
+		$this->tokens['.']= $this->glue.'.'.$this->glue;
+		$this->tokens[',']= $this->glue.','.$this->glue;
 
 		// prepare it to the next step
 		Mind::$canonic= new Canonic();

mind3rd/API/cortex/canonic/Canonic.php

@@ -1,33 +1,56 @@
 <?php
-/* 
- * To change this template, choose Tools | Templates
- * and open the template in the editor.
- */
-
 /**
- * Description of Canonic
- *
+ * This class will take the content to be used, and take
+ * it to its canonical form
+ * @package cortex.analyst
  * @author felipe
  */
 class Canonic extends Inflect{
 
+	/**
+	 * Takes a word to its canonic form
+	 * @param string$word
+	 * @return string
+	 */
 	public static function canonize($word)
 	{
 		if(!self::isSingular($word))
 			$word= self::toSingular($word);
 		if(self::isFemale($word))
 			$word= self::toMale($word);
-		echo $word."<br>";
+		return $word;
 	}
 
+	/**
+	 * Brings all the words in the Array to their canonical form
+	 * @param array $content
+	 * @return Array
+	 */
 	public function sweep(Array $content)
 	{
-		array_map(function ($word)
+		$v= new Verbalizer;
+		$newContent= Array();
+		foreach($content as $word)
+		{
+			if(!Verbalizer::isVerb($word) && strlen($word) > 1)
+			{
+				//echo 'the canonic form of '.$word.' is '.Canonic::canonize($word);
+				$word= Canonic::canonize($word);
+			}
+			$newContent[]= $word;
+		}
+		/*array_map(function ($word)
 				  {
+					print_r($v);
+					//echo Verbalizer::toInfinitive($word);
+					//if(Verbalizer::isVerb($word))
+					//return Verbalizer::toInfinitive($word);
 					Canonic::canonize($word);
 				  },
 				  $content);
-		print_r($content);
+		*/
+		print_r($newContent);
+		return $newContent;
 	}
 	
     public function __construct()

mind3rd/API/languages/pt-BR/IgnoreForms.php

@@ -0,0 +1,74 @@
+<?php
+	/**
+	 * This class provides a list of instructtions
+	 * which define when a word should be ignored,
+	 * plus a list of key words to ignore
+	 * @package cortex.analyst
+	 * @author felipe
+	 */
+	class IgnoreForms {
+		/**
+		 * The list of words to be ignored
+		 * @var Array
+		 * @static
+		 */
+		public static $ignoreList= false;
+
+		/**
+		 * A list of rules, to identify words to be ignored
+		 * @var Array
+		 * @static
+		 */
+		public static $ignoreRules= Array(
+			'/(.)mente$/'
+		);
+
+		/**
+		 * This method reads the ignore.list file and
+		 * parses it to an indexed array
+		 * @static
+		 * @static
+		 * @name loadVerbs
+		 */
+		public static function loadIgnoreList()
+		{
+			$fR= fopen('ignore.list', 'rb');
+			self::$ignoreList= Array();
+			while (!feof($fR)){
+				$word= preg_replace('/\s/', '', fgets($fR, 4096));
+				self::$ignoreList[$word]= true;
+			}
+		}
+
+		/**
+		 * Returns wheter the word should or not be ignored
+		 * @name shouldBeIgnored
+		 * @param string $word
+		 * @return boolean
+		 */
+		public static function shouldBeIgnored($word)
+		{
+			if(!self::$ignoreList)
+				self::loadIgnoreList();
+			if(isset(self::$ignoreList[$word]))
+				return true;
+			foreach(self::$ignoreRules as $rule)
+			{
+				if(preg_match($rule, $word))
+					return true;
+			}
+			return false;
+		}
+
+		/**
+		 * Returns if the word should be used
+		 * @name shouldBeUsed
+		 * @static
+		 * @param string $word
+		 * @return boolean
+		 */
+		public static function shouldBeUsed($word)
+		{
+			return !self::shouldBeIgnored($word);
+		}
+	}

mind3rd/API/languages/pt-BR/Inflect.php

@@ -17,9 +17,9 @@
  * This class should inflect words for different idioms
  * changing its genre and number
  * IN THIS CASE: pt-BR
- *
  * @name Inflect
  * @author Felipe Nascimento de Moura <felipenmoura@gmail.com>
+ * @package cortex.analyst
  */
 class Inflect implements inflection
 {

mind3rd/API/languages/pt-BR/Verbalizer.php

@@ -1,9 +1,9 @@
 <?php
 /**
  * This class should identify verbs
- * NOTICE that its goas is about the present/future words...
+ * NOTICE that its goals is about the present/future words...
  * past is not supported
- *
+ * @package cortex.analyst
  * @author felipe
  */
 class Verbalizer {
@@ -14,7 +14,6 @@ class Verbalizer {
 	public static $verbs= false;
 
 	/**
-	 *
 	 * @var AssocArray An array of verbs that may indicate a possibility
 	 */
 	static $possibilities= Array(
@@ -27,7 +26,7 @@ class Verbalizer {
 	);
 
 	/**
-	 *
+	 * @static
 	 * @var AssocArray the possible flections a verb may suffer
 	 */
     static $flections = array(
@@ -55,7 +54,7 @@ class Verbalizer {
 	);
 
 	/**
-	 *
+	 * @static
 	 * @var AssocArray A list of fixed exceptions
 	 */
 	static $exceptions= Array(
@@ -77,6 +76,7 @@ class Verbalizer {
 	/**
 	 * Returns if the passed word means the verb is a possibility
 	 * @param string $word
+	 * @static
 	 * @return boolean
 	 */
 	public static function isAPosibility($string)
@@ -87,13 +87,15 @@ class Verbalizer {
 	/**
 	 * Returns if the passed word is in the infinitive form
 	 * @param string $word
+	 * @static
 	 * @return boolean
 	 */
 	public static function isInfinitive($string)
 	{
+		$isInVerBlist= self::isInVerbList($string);
 		foreach(self::$flections as $pattern=>$result)
 		{
-			if(preg_match('/'.$result.'$/i', $string) && self::isInVerbList($string))
+			if(preg_match('/'.$result.'$/i', $string) && $isInVerBlist)
 				return true;
 		}
 		return false;
@@ -104,12 +106,14 @@ class Verbalizer {
 	 * Returns false if it is not in the verbs.list
 	 * @param string $word
 	 * @return string
+	 * @static
 	 */
 	public static function toInfinitive($string)
     {
 		$string= strtolower($string);
 		if(self::isInfinitive($string))
 			return $string;
+		
         if(isset(self::$exceptions[$string]))
 		{
 			return self::$exceptions[$string];
@@ -136,6 +140,7 @@ class Verbalizer {
 	 * Verifies if the passed word is in the verbs.list
 	 * @param string $word
 	 * @return boolean
+	 * @static
 	 */
 	public static function isInVerbList($string)
 	{
@@ -147,6 +152,7 @@ class Verbalizer {
 	/**
 	 * Returns if the received word is a verb or not
 	 * @param string $word
+	 * @static
 	 */
 	public static function isVerb($word)
 	{
@@ -157,11 +163,17 @@ class Verbalizer {
 	}
 
 	/**
+	 * This method reads the verbs.list file and
+	 * parses it to an indexed array
 	 * @name loadVerbs
+	 * @static
 	 */
 	public static function loadVerbs()
 	{
-		$fR= fopen('verbs.list', 'rb');
+		if(!file_exists('verbs.list'))
+			$fR= fopen(Mind::$langPath.Mind::$l10n->name.'/verbs.list', 'rb');
+		else
+			$fR= fopen('verbs.list', 'rb');
 		self::$verbs= Array();
 		while (!feof($fR)){
 			$verb= preg_replace('/\s/', '', fgets($fR, 4096));

mind3rd/API/languages/pt-BR/ignore.list

@@ -0,0 +1,4 @@
+de
+que
+da
+do

mind3rd/API/utils.php

@@ -8,48 +8,8 @@
 	*/
 	session_start();
 	define('_CONSOLE_LINE_LENGTH_', 80);
-
-	function __autoload($what)
-	{
-		GLOBAL $_MIND;
-		$what= preg_replace("/[ '\"\.\/]/", '', $what);
-		if(file_exists(_MINDSRC_.'/mind3rd/API/programs/'.$what.'.php'))
-		{
-			include(_MINDSRC_.'/mind3rd/API/programs/'.$what.'.php');
-			return true;
-		}
-		if(strpos($what, '\\')>=0)
-		{
-			$what= explode('\\', $what);
-			$what= array_pop($what);
-		}
-		$dirs= Array(
-			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Input/',
-			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Output/',
-			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Command/',
-			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Helper/',
-			_MINDSRC_.'/mind3rd/API/cortex/Lexer/',
-			_MINDSRC_.'/mind3rd/API/interfaces/',
-			_MINDSRC_.'/mind3rd/API/programs/',
-			_MINDSRC_.'/mind3rd/API/L10N/',
-			_MINDSRC_.'/mind3rd/API/classes/',
-			_MINDSRC_.'/mind3rd/API/cortex/tokenizer/',
-			_MINDSRC_.'/mind3rd/API/cortex/canonic/',
-			_MINDSRC_.'/mind3rd/API/cortex/sintaxer/'
-		);
-		for($i=0; $i<sizeof($dirs); $i++)
-		{
-			if(file_exists($dirs[$i].$what.'.php'))
-			{
-				require_once($dirs[$i].$what.'.php');
-				return true;
-			}
-		}
-		echo " [ERROR] Class not found: ".$what."\n";
-		exit;
-		return false;
-	}
-
+	require(_MINDSRC_.'/mind3rd/API/classes/Mind.php');
+	
 	/*require(_MINDSRC_.'/mind3rd/API/interfaces/program.php');
 	require(_MINDSRC_.'/mind3rd/API/classes/Mind.php');
 	require(_MINDSRC_.'/mind3rd/API/classes/MindDB.php');
@@ -106,6 +66,56 @@
 
 	Mind::$lexer= new Lexer();
 	
+	function __autoload($what)
+	{
+		GLOBAL $_MIND;
+		$what= preg_replace("/[ '\"\.\/]/", '', $what);
+
+		if(file_exists(_MINDSRC_.'/mind3rd/API/programs/'.$what.'.php'))
+		{
+			include(_MINDSRC_.'/mind3rd/API/programs/'.$what.'.php');
+			return true;
+		}
+		if(strpos($what, '\\')>=0)
+		{
+			$what= explode('\\', $what);
+			$what= array_pop($what);
+		}
+		$dirs= Array(
+			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Input/',
+			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Output/',
+			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Command/',
+			_MINDSRC_.'/mind3rd/API/external/Symfony/Component/Console/Helper/',
+			_MINDSRC_.'/mind3rd/API/cortex/Lexer/',
+			_MINDSRC_.'/mind3rd/API/interfaces/',
+			_MINDSRC_.'/mind3rd/API/programs/',
+			_MINDSRC_.'/mind3rd/API/L10N/',
+			_MINDSRC_.'/mind3rd/API/classes/',
+			_MINDSRC_.'/mind3rd/API/cortex/tokenizer/',
+			_MINDSRC_.'/mind3rd/API/cortex/canonic/',
+			_MINDSRC_.'/mind3rd/API/cortex/sintaxer/'
+		);
+		for($i=0; $i<sizeof($dirs); $i++)
+		{
+			if(file_exists($dirs[$i].$what.'.php'))
+			{
+				require_once($dirs[$i].$what.'.php');
+				return true;
+			}
+		}
+		// let's check if it is a language
+		$langPath= _MINDSRC_.'/mind3rd/API/languages/'.$_MIND->defaults['default_human_languageName'].'/'.$what.'.php';
+		if(file_exists($langPath))
+		{
+			require_once($langPath);
+			return true;
+		}
+
+		echo " [ERROR] Class not found: ".$what."\n";
+		exit;
+		return false;
+	}
+
 	if($_REQ['env']=='shell')
             include('shell.php');
 	else

mind3rd/env/defaults.ini

@@ -3,6 +3,7 @@
 ;
 ; the default idiom
 default_human_language=ptBR
+default_human_languageName=pt-BR
 ; default timezone
 timezone=America/Sao_paulo
 ; the default programming language

mind3rd/projects/x/sources/main.mnd

@@ -1,2 +1,2 @@
 bem, sabemos que todo professor pode ter muitos alunos & que cada aluno pode ter vários professores.
-Tamabém, que cada professora tem um chefe.
+Também, que cada professora tem um chefe.