[0.3.1] [UKR] Распознавание фамилии имя отчества

Library/NCL.NameCase.ua.php

@@ -696,7 +696,11 @@ class NCLNameCaseUa extends NCLNameCaseCore implements NCLNameCaseInterface
                 }
                 if (in_array($FLastTwo, array('он', 'ов', 'ав', 'ам', 'ол', 'ан', 'рд', 'мп', 'ко', 'ло')))
                 {
-                    $man+=0.3;
+                    $man+=0.5;
+                }
+                if (in_array($FLastThree, array('бов', 'нка', 'яра', 'ила')))
+                {
+                    $woman+=0.5;
                 }
                 if ($this->in($FLastSymbol, $this->consonant))
                 {
@@ -718,10 +722,12 @@ class NCLNameCaseUa extends NCLNameCaseCore implements NCLNameCaseInterface
                 }
             }
 
+//            $man*=1.2;
+//            $woman*=1.2;
 
             if (isset($this->secondName) and $this->secondName)
             {
-                if (in_array($SLastTwo, array('ов', 'ин', 'ев','ич', 'єв', 'ін', 'їн', 'ий', 'їв', 'ів', 'ій', 'ой','ей')))
+                if (in_array($SLastTwo, array('ов', 'ин', 'ев', 'єв', 'ін', 'їн', 'ий', 'їв', 'ів', 'ой', 'ей')))
                 {
                     $man+=0.4;
                 }
@@ -757,8 +763,65 @@ class NCLNameCaseUa extends NCLNameCaseCore implements NCLNameCaseInterface
 
     protected function detectNamePart($namepart)
     {
+        $LastSymbol = mb_substr($namepart, -1, 1, 'utf-8');
+        $LastTwo = mb_substr($namepart, -2, 2, 'utf-8');
+        $LastThree = mb_substr($namepart, -3, 3, 'utf-8');
+        $LastFour = mb_substr($namepart, -4, 4, 'utf-8');
 
-        return 'N';
+        //Считаем вероятность
+        $first = 0;
+        $second = 0;
+        $father = 0;
+
+        //если смахивает на отчество
+        if (in_array($LastThree, array('вна', 'чна', 'ліч')) or in_array($LastFour, array('ьмич', 'ович')))
+        {
+            $father+=3;
+        }
+
+        //Похоже на имя
+        if (in_array($LastThree, array('тин')) or in_array($LastFour, array('ьмич', 'юбов')))
+        {
+            $first+=0.5;
+        }
+
+        //Исключения
+        if (in_array($namepart, array('Лев', 'Гаїна', 'Афіна', 'Антоніна', 'Ангеліна', 'Альвіна', 'Альбіна', 'Аліна', 'Павло', 'Олесь')))
+        {
+            $first+=10;
+        }
+
+        //похоже на фамилию
+        if (in_array($LastTwo, array('ов', 'ін', 'ев', 'єв', 'ий', 'ин', 'ой', 'ко', 'ук', 'як', 'ца', 'их', 'ик', 'ун', 'ок', 'ша', 'ая', 'га', 'єк', 'аш', 'ив', 'юк', 'ус', 'це', 'ак', 'бр', 'яр', 'іл', 'ів', 'ич', 'сь', 'ей', 'нс', 'яс', 'ер', 'ай', 'ян', 'ах', 'ць', 'ющ', 'іс', 'ач', 'уб', 'ох', 'юх','ут','ча','ул','вк','зь'  /*{endings2}*/)))
+        {
+            $second+=0.4;
+        }
+
+        if (in_array($LastThree, array('ова', 'ева', 'єва', 'тих', 'рик', 'вач', 'аха', 'шен', 'мей', 'арь', 'вка', 'шир', 'бан', 'чий', 'іна', 'їна', 'ька', 'ань', 'ива', 'аль','ура','ран','ало','ола','кур','оба','оль','нта','зій','ґан','іло','шта', 'юпа', 'рна', 'бла', 'еїн', 'има', 'мар', 'кар', 'оха', 'чур', 'ниш', 'ета', 'тна', 'зур', 'нір', 'йма', 'орж', 'рба'  /*{endings3}*/)))
+        {
+            $second+=0.4;
+        }
+
+        if (in_array($LastFour, array('ьник', 'нчук', 'тник', 'кирь', 'ский', 'шена', 'шина', 'вина', 'нина' /*{endings4}*/)))
+        {
+            $second+=0.4;
+        }
+
+
+        $max = max(array($first, $second, $father));
+
+        if ($first == $max)
+        {
+            return 'N';
+        }
+        elseif ($second == $max)
+        {
+            return 'S';
+        }
+        else
+        {
+            return 'F';
+        }
     }
 
 }

Tests/AutoTest/Endings/.gitignore

@@ -0,0 +1,2 @@
+*.txt
+compiled/

Tests/AutoTest/Endings/BatchTester.php

@@ -0,0 +1,81 @@
+<?php
+
+/**
+ * Description of BathTester
+ *
+ * @author seagull
+ */
+require 'EndingsCompiler.php';
+
+class BatchTester
+{
+    private $compiler;
+    private $mode;
+    
+    public $epsilon;
+    
+    public function __construct($mode)
+    {
+        copy('../../../Library/NCL.NameCase.core.php','compiled/NCL.NameCase.core.php');
+        $testPHP = file_get_contents('../../Library/NameSplitUAname.php');
+        $testPHP = str_replace("require_once dirname(__FILE__) . '/../../Library/NCL.NameCase.ua.php';", "require_once dirname(__FILE__) . '/NCL.NameCase.ua.php';", $testPHP);
+        file_put_contents('compiled/Test.php', $testPHP);
+        
+        $this->compiler = new EndingsCompiler($mode);
+        $this->mode = $mode;
+        $this->compiler->compile();
+        
+        $this->epsilon = $this->runTest();
+    }
+    
+    public function testText($ending)
+    {
+        $this->compiler->compile($ending);
+        
+        return $this->runTest();
+    }
+    
+    public function runTest()
+    {
+        exec("phpunit compiled/Test.php", $output);
+        
+        preg_match_all('#Tests: ([\d]+), Assertions: ([\d]+), Failures: ([\d]+).#is', $output[count($output)-1], $found);
+        $tests = (float)$found[2][0];
+        $bad = (float)$found[3][0];
+        
+        //var_dump($tests);
+        //var_dump(($bad/$tests));
+       // print_r($found);
+        //echo $bad." ".$tests;
+        return (1-($bad/$tests));
+    }
+}
+
+$good = array();
+$bad = array();
+$tester = new BatchTester('endings3');
+
+$epsilon = $tester->epsilon;
+
+$endings = file('endings3.txt');
+
+foreach ($endings as $ending)
+{
+    $persent = $tester->testText(trim($ending));
+    echo trim($ending)." - ".$persent."\n";
+    
+    if($persent==$epsilon)
+    {
+        $good[] = $ending;
+    }
+    else
+    {
+        $bad[] = $ending;
+    }
+}
+
+echo "\n\n GOOD";
+print_r($good);
+echo "\n\n BAD";
+print_r($bad);
+?>

Tests/AutoTest/Endings/EndingsCompiler.php

@@ -0,0 +1,41 @@
+<?php
+/**
+ * Description of EndingsCompiler
+ *
+ * @author seagull
+ */
+class EndingsCompiler
+{
+    private $template;
+    private $destination = 'compiled/NCL.NameCase.ua.php';
+    private $sourceFile = '../../../Library/NCL.NameCase.ua.php';
+    
+    private $compiled = '';
+
+    private $mode = 'endings3';
+    
+    public function __construct($mode)
+    {
+        $this->template = file_get_contents($this->sourceFile);
+    }
+    
+    private function save()
+    {
+        file_put_contents($this->destination, $this->compiled);
+    }
+    
+    public function compile($text = '')
+    {
+        if($text)
+        {
+            $text = ", '".$text."'";
+        }
+        
+        $this->compiled = str_replace('/*{'.$this->mode.'}*/', $text , $this->template);
+        //$this->compiled = $this->template;
+        $this->save();
+        
+    }
+}
+
+?>

Tests/Library/NameSplitUAfather.php

@@ -0,0 +1,66 @@
+<?php
+
+require_once dirname(__FILE__) . '/../../Library/NCL.NameCase.ua.php';
+
+
+class NCLNameCaseUaTest extends PHPUnit_Framework_TestCase
+{
+
+    /**
+     * @var NCLNameCaseRu
+     */
+    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 NCLNameCaseUa;
+    }
+
+    /**
+     * Tears down the fixture, for example, closes a network connection.
+     * This method is called after a test is executed.
+     */
+    protected function tearDown()
+    {
+        
+    }
+
+
+    public function testSplDetect0()
+    {
+        $this->assertEquals('F-Олександрович', $this->object->splitFullName('Олександрович').'-Олександрович');
+    }
+    public function testSplDetect1()
+    {
+        $this->assertEquals('F-Миколайович', $this->object->splitFullName('Миколайович').'-Миколайович');
+    }
+    public function testSplDetect2()
+    {
+        $this->assertEquals('F-Кузьмич', $this->object->splitFullName('Кузьмич').'-Кузьмич');
+    }
+    public function testSplDetect3()
+    {
+        $this->assertEquals('F-Петрович', $this->object->splitFullName('Петрович').'-Петрович');
+    }
+    public function testSplDetect4()
+    {
+        $this->assertEquals('F-Ілліч', $this->object->splitFullName('Ілліч').'-Ілліч');
+    }
+    public function testSplDetect5()
+    {
+        $this->assertEquals('F-Василівна', $this->object->splitFullName('Василівна').'-Василівна');
+    }
+    public function testSplDetect6()
+    {
+        $this->assertEquals('F-Антонівна', $this->object->splitFullName('Антонівна').'-Антонівна');
+    }
+    public function testSplDetect7()
+    {
+        $this->assertEquals('F-Яківна', $this->object->splitFullName('Яківна').'-Яківна');
+    }
+
+}

Tests/TestGenerator/UAGenderDetector.both.php.php

@@ -0,0 +1,99 @@
+<?php
+
+class TestGenerator
+{
+    public $maintemplate;
+    public $testtemplate;
+    
+    public $resultArrB=array();
+    public $resultArrG=array();
+    
+    private $gender;
+    
+    private $tests;
+    private $count=0;
+    
+    
+    public function __construct()
+    {
+        mysql_connect('localhost', 'root', '');
+        mysql_select_db('names');
+        mysql_set_charset('utf8');
+        //$this->resultArr[]=file('Names/boy_full_result.txt');
+        //$this->resultArr[]=file('Names/girl_full_result.txt');
+    }
+
+    private function make($gender)
+    {
+        $res = mysql_query("SELECT * FROM {$gender}second");
+        $secondArr = array();
+        while($row = mysql_fetch_assoc($res))
+        {
+            $secondArr[] = $row['name'];
+        }
+        $res = mysql_query("SELECT * FROM {$gender}names");
+        $firstArr = array();
+        while($row = mysql_fetch_assoc($res))
+        {
+            $firstArr[] = $row['name'];
+        }
+        
+        $countFirst = count($firstArr);
+        $countSecond = count($secondArr);
+        
+        $max = max(array($countFirst, $countSecond));
+        
+        $result = array();
+        
+        for($i=0; $i<$max; $i++)
+        {
+            $indexFirst = $i % $countFirst;
+            $indexSecond = $i % $countSecond;
+            
+            $this->generateTest($secondArr[$indexSecond],$firstArr[$indexFirst]);
+        }
+        
+        return $result;
+    }
+    
+    public function generate()
+    {
+        $this->maintemplate = file_get_contents('Template/NCLNameCaseUaTest.main');
+        $this->testtemplate = file_get_contents('Template/GenderGenBoth.test');
+
+        $this->count=0;
+        
+
+        $this->gender = 1;
+        $this->make('boy');
+        
+        
+        $this->gender = 2;
+        $this->make('girl');
+        
+        
+        $res=str_replace('{% tests %}', $this->tests, $this->maintemplate);
+        file_put_contents('../Library/GenderDetectUABoth.php',$res);
+    }
+    
+    private function generateTest($sname, $fname)
+    {
+        $tpl=$this->testtemplate;
+        $tpl=str_replace('{% id %}', $this->count, $tpl);
+        $tpl=str_replace('{% name %}', $fname, $tpl);
+        $tpl=str_replace('{% sname %}', $sname, $tpl);
+        $tpl=str_replace('{% gender %}', $this->gender, $tpl);
+
+        $this->tests.=$tpl;
+        $this->count++;
+    }
+    
+    
+}
+
+
+$p= new TestGenerator;
+$p->generate('Both');
+
+
+?>

Tests/TestGenerator/UAGenderDetector.generator.php.php

@@ -0,0 +1,87 @@
+<?php
+
+class TestGenerator
+{
+
+    public $maintemplate;
+    public $testtemplate;
+    public $resultArrB = array();
+    public $resultArrG = array();
+    private $gender;
+    private $tests;
+    private $count = 0;
+
+    public function __construct()
+    {
+        mysql_connect('localhost', 'root', '');
+        mysql_select_db('names');
+        mysql_set_charset('utf8');
+        //$this->resultArr[]=file('Names/boy_full_result.txt');
+        //$this->resultArr[]=file('Names/girl_full_result.txt');
+    }
+
+    public function generate($fname)
+    {
+        $this->maintemplate = file_get_contents('Template/NCLNameCaseUaTest.main');
+        $this->testtemplate = file_get_contents('Template/GenderGen' . $fname . '.test');
+//        $this->resultArrB=file('NamesAll/boy'.$fname.'.txt');
+//        $this->resultArrG=file('NamesAll/girl'.$fname.'.txt');
+        $this->count = 0;
+        
+        
+        $fnewname = $fname;
+        if ($fnewname == 'sirname')
+        {
+            $fnewname = 'second';
+        }
+        else
+        {
+            $fnewname = 'names';
+        }
+        
+        
+        $this->gender = 1;
+        $this->dbTable = 'boy' . strtolower($fnewname);
+        $this->resultArr = mysql_query("SELECT * FROM {$this->dbTable}");
+        //foreach ($this->resultArr as $key=>$value)
+        //{
+        //$this->gender=$key+1;
+        while ($row = mysql_fetch_array($this->resultArr))
+        {
+            $this->generateTest(trim($row['name']));
+        }
+        
+        $this->dbTable = 'girl' . strtolower($fnewname);
+        $this->gender = 2;
+        $this->resultArr = mysql_query("SELECT * FROM {$this->dbTable}");
+        //foreach ($this->resultArr as $key=>$value)
+        //{
+        //$this->gender=$key+1;
+        while ($row = mysql_fetch_array($this->resultArr))
+        {
+            $this->generateTest(trim($row['name']));
+        }
+        //}
+        $res = str_replace('{% tests %}', $this->tests, $this->maintemplate);
+        $res = str_replace('{% name %}', $fname, $res);
+        file_put_contents('../Library/GenderDetectUA' . $fname . '.php', $res);
+    }
+
+    private function generateTest($name)
+    {
+        $tpl = $this->testtemplate;
+        $tpl = str_replace('{% id %}', $this->count, $tpl);
+        $tpl = str_replace('{% name %}', $name, $tpl);
+        $tpl = str_replace('{% gender %}', $this->gender, $tpl);
+
+        $this->tests.=$tpl;
+        $this->count++;
+    }
+
+}
+
+$p = new TestGenerator;
+$p->generate('name');
+$p = new TestGenerator;
+$p->generate('sirname');
+?>

Tests/TestGenerator/UAnameSplit.generator.php.php

@@ -0,0 +1,96 @@
+<?php
+
+class TestGenerator
+{
+
+    public $maintemplate;
+    public $testtemplate;
+    public $resultArrB = array();
+    public $resultArrG = array();
+    private $gender;
+    private $tests;
+    private $count = 0;
+
+    public function __construct()
+    {
+        mysql_connect('localhost', 'root', '');
+        mysql_select_db('names');
+        mysql_set_charset('utf8');
+    }
+
+    private function getList($gender, $type)
+    {
+        $res = mysql_query("SELECT * FROM {$gender}{$type}");
+        $resultArr = array();
+        while($row = mysql_fetch_assoc($res))
+        {
+            $resultArr[] = $row['name'];
+        }
+        return $resultArr;
+    }
+
+    public function generate($fname)
+    {
+        $this->maintemplate = file_get_contents('Template/NCLNameCaseUaTest.main');
+        $this->testtemplate = file_get_contents('Template/NameSplit.test');
+
+        $this->count = 0;
+        if ($fname == 'name')
+        {
+            $this->gender = 'N';
+            $tableName = 'names';
+        }
+        elseif ($fname == 'sirname')
+        {
+            $this->gender = 'S';
+            $tableName = 'second';
+        }
+        else
+        {
+            $this->gender = 'F';
+            $tableName = 'father';
+        }
+        
+        $names = array_merge($this->getList('boy', $tableName), $this->getList('girl', $tableName));
+        
+        
+        //foreach ($this->resultArr as $key=>$value)
+        //{
+        //$this->gender=$key+1;
+        foreach ($names as $id => $name)
+        {
+            $this->generateTest(trim($name));
+        }
+        //foreach ($this->resultArr as $key=>$value)
+        //{
+        //$this->gender=$key+1;
+//        foreach ($this->resultArrG as $id => $name)
+//        {
+//            $this->generateTest(trim($name));
+//        }
+        //}
+        $res = str_replace('{% tests %}', $this->tests, $this->maintemplate);
+        $res = str_replace('{% name %}', $fname, $res);
+        file_put_contents('../Library/NameSplitUA' . $fname . '.php', $res);
+    }
+
+    private function generateTest($name)
+    {
+        $tpl = $this->testtemplate;
+        $tpl = str_replace('{% id %}', $this->count, $tpl);
+        $tpl = str_replace('{% name %}', $name, $tpl);
+        $tpl = str_replace('{% gender %}', $this->gender, $tpl);
+
+        $this->tests.=$tpl;
+        $this->count++;
+    }
+
+}
+
+$p = new TestGenerator;
+$p->generate('name');
+$p = new TestGenerator;
+$p->generate('sirname');
+$p = new TestGenerator;
+$p->generate('father');
+?>

Tests/TestGenerator/nameSplit.generator.php.php

@@ -16,7 +16,6 @@ class TestGenerator
     
     public function __construct()
     {
-        
         //$this->resultArr[]=file('Names/boy_full_result.txt');
         //$this->resultArr[]=file('Names/girl_full_result.txt');
     }