Create a tokenizer interface for different implementations

src/Parser.php

@@ -25,18 +25,28 @@ final class Parser implements ParserInterface
      */
     private $factory;
 
+    /**
+     * @var TokenizerInterface $tokenizer
+     */
+    private $tokenizer;
+
 
     /**
      * Create a new instance.
      *
      * @param FactoryInterface $factory A segment factory for creating segments
      */
-    public function __construct(FactoryInterface $factory = null)
+    public function __construct(FactoryInterface $factory = null, TokenizerInterface $tokenizer = null)
     {
         if ($factory === null) {
             $factory = new Factory();
         }
         $this->factory = $factory;
+
+        if ($tokenizer === null) {
+            $tokenizer = new Tokenizer();
+        }
+        $this->tokenizer = $tokenizer;
     }
 
 
@@ -51,11 +61,9 @@ public function __construct(FactoryInterface $factory = null)
      */
     public function parse(string $message, ControlCharactersInterface $characters = null): iterable
     {
-        $tokenizer = new Tokenizer();
-
         $characters = $this->getControlCharacters($message, $characters);
 
-        $tokens = $tokenizer->getTokens($message, $characters);
+        $tokens = $this->tokenizer->getTokens($message, $characters);
 
         $segments = $this->convertTokensToSegments($tokens, $characters);
 

src/Tokenizer.php

@@ -12,7 +12,7 @@
 /**
  * Convert EDI messages into tokens for parsing.
  */
-final class Tokenizer
+final class Tokenizer implements TokenizerInterface
 {
     /**
      * @var string $message The message that we are tokenizing.

src/TokenizerInterface.php

@@ -0,0 +1,20 @@
+<?php
+
+namespace Estrato\Edifact;
+
+use Estrato\Edifact\Control\CharactersInterface as ControlCharactersInterface;
+use Estrato\Edifact\Exceptions\ParseException;
+
+/**
+ * Convert EDI messages into tokens for parsing.
+ */
+interface TokenizerInterface
+{
+    /**
+     * Convert the passed message into tokens.
+     *
+     * @return Token[]
+     * @throws ParseException
+     */
+    public function getTokens(string $message, ControlCharactersInterface $characters): array;
+}

tests/ParserTest.php

@@ -5,8 +5,11 @@
 use duncan3dc\ObjectIntruder\Intruder;
 use Estrato\Edifact\Control\CharactersInterface as ControlCharactersInterface;
 use Estrato\Edifact\Parser;
+use Estrato\Edifact\Segments\Factory;
 use Estrato\Edifact\Segments\Segment;
 use Estrato\Edifact\Segments\SegmentInterface;
+use Estrato\Edifact\Token;
+use Estrato\Edifact\TokenizerInterface;
 use Mockery;
 use PHPUnit\Framework\TestCase;
 
@@ -168,4 +171,27 @@ public function testEscapeSequence(): void
             new Segment("ERC", ["10", ":+?' - :+?' - :+?'"]),
         ]);
     }
+
+
+    public function testCustomTokenizer(): void
+    {
+        $tokenizer = Mockery::mock(TokenizerInterface::class);
+        $tokens = [
+            new Token(Token::CONTENT, "QTY"),
+            new Token(Token::DATA_SEPARATOR, "+"),
+            new Token(Token::CONTENT, "136"),
+            new Token(Token::COMPONENT_SEPARATOR, ":"),
+            new Token(Token::CONTENT, "999"),
+            new Token(Token::TERMINATOR, "'"),
+        ];
+        $tokenizer->shouldReceive('getTokens')->once()->with("QTY+136:12,235", Mockery::type(ControlCharactersInterface::class))->andReturn($tokens);
+
+        $parser = new Parser(new Factory(), $tokenizer);
+        $segments = $parser->parse("UNA:+,? '\nQTY+136:12,235");
+        $segments = is_array($segments) ? $segments : iterator_to_array($segments);
+
+        $this->assertEquals([
+            new Segment("QTY", ["136", "999"]),
+        ], $segments);
+    }
 }