enhancement - add 'not' negated character classes (#40)

*add negated versions of methods generating ranges:
-notDigit,
-notLetter,
-notUppercateLetter
-notOneOf

Author: BartoszKolanko

src/Builder.php

@@ -165,12 +165,28 @@ public function oneOf(string $chars)
     {
         $this->validateAndAddMethodType(self::METHOD_TYPE_CHARACTER, self::METHOD_TYPES_ALLOWED_FOR_CHARACTERS);
 
-        $chars = implode('', array_map([$this, 'escape'], str_split($chars)));
-        $chars = str_replace(['-', ']'], ['\\-', '\\]'], $chars);
+        $chars = $this->escape($chars);
+        $chars = $this->escapeRangeSpecificChars($chars);
 
         return $this->add('[' . $chars . ']');
     }
 
+    /**
+     * Literally match anything but one of these characters.
+     *
+     * @param string $chars
+     * @return Builder
+     */
+    public function notOneOf(string $chars)
+    {
+        $this->validateAndAddMethodType(self::METHOD_TYPE_CHARACTER, self::METHOD_TYPES_ALLOWED_FOR_CHARACTERS);
+
+        $chars = $this->escape($chars);
+        $chars = $this->escapeRangeSpecificChars($chars);
+
+        return $this->add('[^' . $chars . ']');
+    }
+
     /**
      * Literally match all of these characters in that order.
      *
@@ -181,7 +197,7 @@ public function literally(string $chars) : self
     {
         $this->validateAndAddMethodType(self::METHOD_TYPE_CHARACTER, self::METHOD_TYPES_ALLOWED_FOR_CHARACTERS);
 
-        return $this->add('(?:' . implode('', array_map([$this, 'escape'], str_split($chars))) . ')');
+        return $this->add('(?:' . $this->escape($chars) . ')');
     }
 
     /**
@@ -212,6 +228,20 @@ public function digit(int $min = 0, int $max = 9) : self
         return $this->add("[$min-$max]");
     }
 
+    /**
+     * Match anything but digit (in given span). Default will be a digit between 0 and 9.
+     *
+     * @param int $min
+     * @param int $max
+     * @return Builder
+     */
+    public function notDigit(int $min = 0, int $max = 9) : self
+    {
+        $this->validateAndAddMethodType(self::METHOD_TYPE_CHARACTER, self::METHOD_TYPES_ALLOWED_FOR_CHARACTERS);
+
+        return $this->add("[^$min-$max]");
+    }
+
     /**
      * Match any uppercase letter (between A to Z).
      *
@@ -238,6 +268,32 @@ public function letter(string $min = 'a', string $max = 'z') : self
         return $this->add("[$min-$max]");
     }
 
+    /**
+     * Match anything but uppercase letter (between A to Z).
+     *
+     * @param string $min
+     * @param string $max
+     * @return Builder
+     */
+    public function notUppercaseLetter(string $min = 'A', string $max = 'Z') : self
+    {
+        return $this->notLetter($min, $max);
+    }
+
+    /**
+     * Match anything but lowercase letter (between a to z).
+     *
+     * @param string $min
+     * @param string $max
+     * @return Builder
+     */
+    public function notLetter(string $min = 'a', string $max = 'z') : self
+    {
+        $this->validateAndAddMethodType(self::METHOD_TYPE_CHARACTER, self::METHOD_TYPES_ALLOWED_FOR_CHARACTERS);
+
+        return $this->add("[^$min-$max]");
+    }
+
     /**********************************************************/
     /*                        GROUPS                          */
     /**********************************************************/
@@ -508,17 +564,38 @@ public function until($toCondition) : self
     /*                   INTERNAL METHODS                     */
     /**********************************************************/
 
+    /**
+     * Escape all characters in string.
+     *
+     * @param string $chars
+     * @return string
+     */
+    protected function escape(string $chars)
+    {
+        return implode('', array_map([$this, 'escapeChar'], str_split($chars)));
+    }
+
     /**
      * Escape specific character.
      *
      * @param string $char
      * @return string
      */
-    protected function escape(string $char)
+    protected function escapeChar(string $char)
     {
         return (strpos(static::NON_LITERAL_CHARACTERS, $char) !== false ? '\\' : '') . $char;
     }
 
+    /**
+     * Escape '-' and ']' in string to be used in range.
+     *
+     * @return string
+     */
+    protected function escapeRangeSpecificChars(string $chars)
+    {
+        return str_replace(['-', ']'], ['\\-', '\\]'], $chars);
+    }
+
     /**
      * Get the raw regular expression without delimiter or modifiers.
      *

src/Language/Helpers/Matcher.php

@@ -32,9 +32,13 @@ class Matcher
         'tab' => ['class' => Methods\SimpleMethod::class, 'method' => 'tab'],
         'unicode' => ['class' => Methods\SimpleMethod::class, 'method' => 'unicode'],
         'digit' => ['class' => Methods\SimpleMethod::class, 'method' => 'digit'],
+        'not digit' => ['class' => Methods\SimpleMethod::class, 'method' => 'notDigit'],
         'number' => ['class' => Methods\SimpleMethod::class, 'method' => 'digit'],
+        'not number' => ['class' => Methods\SimpleMethod::class, 'method' => 'notDigit'],
         'letter' => ['class' => Methods\SimpleMethod::class, 'method' => 'letter'],
+        'not letter' => ['class' => Methods\SimpleMethod::class, 'method' => 'notLetter'],
         'uppercase letter' => ['class' => Methods\SimpleMethod::class, 'method' => 'uppercaseLetter'],
+        'not uppercase letter' => ['class' => Methods\SimpleMethod::class, 'method' => 'notUppercaseLetter'],
         'once' => ['class' => Methods\SimpleMethod::class, 'method' => 'once'],
         'twice' => ['class' => Methods\SimpleMethod::class, 'method' => 'twice'],
         'first match' => ['class' => Methods\SimpleMethod::class, 'method' => 'firstMatch'],
@@ -50,11 +54,16 @@ class Matcher
         'until' => ['class' => Methods\DefaultMethod::class, 'method' => 'until'],
         'raw' => ['class' => Methods\DefaultMethod::class, 'method' => 'raw'],
         'one of' => ['class' => Methods\DefaultMethod::class, 'method' => 'oneOf'],
+        'not one of' => ['class' => Methods\DefaultMethod::class, 'method' => 'notOneOf'],
 
         'digit from' => ['class' => Methods\ToMethod::class, 'method' => 'digit'],
+        'not digit from' => ['class' => Methods\ToMethod::class, 'method' => 'notDigit'],
         'number from' => ['class' => Methods\ToMethod::class, 'method' => 'digit'],
+        'not number from' => ['class' => Methods\ToMethod::class, 'method' => 'notDigit'],
         'letter from' => ['class' => Methods\ToMethod::class, 'method' => 'letter'],
+        'not letter from' => ['class' => Methods\ToMethod::class, 'method' => 'notLetter'],
         'uppercase letter from' => ['class' => Methods\ToMethod::class, 'method' => 'uppercaseLetter'],
+        'not uppercase letter from' => ['class' => Methods\ToMethod::class, 'method' => 'notUppercaseLetter'],
         'exactly' => ['class' => Methods\TimesMethod::class, 'method' => 'exactly'],
         'at least' => ['class' => Methods\TimesMethod::class, 'method' => 'atLeast'],
         'between' => ['class' => Methods\AndMethod::class, 'method' => 'between'],

tests/BuilderTest.php

@@ -148,4 +148,29 @@ public function testRaw()
     {
         $this->assertTrue(SRL::literally('foo')->raw('b[a-z]r')->isValid());
     }
+
+    public function testNotDigit()
+    {
+        $regex = SRL::notDigit()
+            ->notDigit(2, 5)
+            ->get();
+
+        $this->assertEquals('/[^0-9][^2-5]/', $regex);
+    }
+
+    public function testNotLetter()
+    {
+        $regex = SRL::notLetter()
+            ->notLetter('a', 'd')
+            ->get();
+
+        $this->assertEquals('/[^a-z][^a-d]/', $regex);
+    }
+
+    public function testNotOneOf()
+    {
+        $regex = SRL::notOneOf('a$#')->get();
+
+        $this->assertEquals('/[^a\$#]/', $regex);
+    }
 }

tests/ExceptionsTest.php

@@ -128,7 +128,7 @@ public function testInvalidArgument()
      */
     public function testInvalidExactlyArgument()
     {
-        new SRL('letter exactly 2 times,not digit');
+        new SRL('letter exactly 2 times,foo digit');
     }
 
     /**

tests/LanguageInterpreterTest.php

@@ -44,6 +44,15 @@ public function testParser()
         $this->assertFalse($srl->isMatching('123abc'));
         $this->assertFalse($srl->isMatching('1a'));
         $this->assertFalse($srl->isMatching(''));
+
+        $srl = new SRL('starts with not digit, not digit from 0 to 2, not number, not number from 3 to 5  ');
+        $this->assertEquals('/^[^0-9][^0-2][^0-9][^3-5]/', $srl->get());
+
+        $srl = new SRL('starts with not letter, not uppercase letter, not letter from f to o, not uppercase letter from O to z');
+        $this->assertEquals('/^[^a-z][^A-Z][^f-o][^O-z]/', $srl->get());
+
+        $srl = new SRL('starts with not one of "!@#/"');
+        $this->assertEquals('/^[^!@#\/]/', $srl->get());
     }
 
     public function testEmail()