Matthias Andreas Benkard | 12a5735 | 2021-12-28 18:02:04 +0100 | [diff] [blame] | 1 | Twig Internals |
| 2 | ============== |
| 3 | |
| 4 | Twig is very extensible and you can hack it. Keep in mind that you |
| 5 | should probably try to create an extension before hacking the core, as most |
| 6 | features and enhancements can be handled with extensions. This chapter is also |
| 7 | useful for people who want to understand how Twig works under the hood. |
| 8 | |
| 9 | How does Twig work? |
| 10 | ------------------- |
| 11 | |
| 12 | The rendering of a Twig template can be summarized into four key steps: |
| 13 | |
| 14 | * **Load** the template: If the template is already compiled, load it and go |
| 15 | to the *evaluation* step, otherwise: |
| 16 | |
| 17 | * First, the **lexer** tokenizes the template source code into small pieces |
| 18 | for easier processing; |
| 19 | |
| 20 | * Then, the **parser** converts the token stream into a meaningful tree |
| 21 | of nodes (the Abstract Syntax Tree); |
| 22 | |
| 23 | * Finally, the *compiler* transforms the AST into PHP code. |
| 24 | |
| 25 | * **Evaluate** the template: It means calling the ``display()`` |
| 26 | method of the compiled template and passing it the context. |
| 27 | |
| 28 | The Lexer |
| 29 | --------- |
| 30 | |
| 31 | The lexer tokenizes a template source code into a token stream (each token is |
| 32 | an instance of ``\Twig\Token``, and the stream is an instance of |
| 33 | ``\Twig\TokenStream``). The default lexer recognizes 13 different token types: |
| 34 | |
| 35 | * ``\Twig\Token::BLOCK_START_TYPE``, ``\Twig\Token::BLOCK_END_TYPE``: Delimiters for blocks (``{% %}``) |
| 36 | * ``\Twig\Token::VAR_START_TYPE``, ``\Twig\Token::VAR_END_TYPE``: Delimiters for variables (``{{ }}``) |
| 37 | * ``\Twig\Token::TEXT_TYPE``: A text outside an expression; |
| 38 | * ``\Twig\Token::NAME_TYPE``: A name in an expression; |
| 39 | * ``\Twig\Token::NUMBER_TYPE``: A number in an expression; |
| 40 | * ``\Twig\Token::STRING_TYPE``: A string in an expression; |
| 41 | * ``\Twig\Token::OPERATOR_TYPE``: An operator; |
| 42 | * ``\Twig\Token::PUNCTUATION_TYPE``: A punctuation sign; |
| 43 | * ``\Twig\Token::INTERPOLATION_START_TYPE``, ``\Twig\Token::INTERPOLATION_END_TYPE``: Delimiters for string interpolation; |
| 44 | * ``\Twig\Token::EOF_TYPE``: Ends of template. |
| 45 | |
| 46 | You can manually convert a source code into a token stream by calling the |
| 47 | ``tokenize()`` method of an environment:: |
| 48 | |
| 49 | $stream = $twig->tokenize(new \Twig\Source($source, $identifier)); |
| 50 | |
| 51 | As the stream has a ``__toString()`` method, you can have a textual |
| 52 | representation of it by echoing the object:: |
| 53 | |
| 54 | echo $stream."\n"; |
| 55 | |
| 56 | Here is the output for the ``Hello {{ name }}`` template: |
| 57 | |
| 58 | .. code-block:: text |
| 59 | |
| 60 | TEXT_TYPE(Hello ) |
| 61 | VAR_START_TYPE() |
| 62 | NAME_TYPE(name) |
| 63 | VAR_END_TYPE() |
| 64 | EOF_TYPE() |
| 65 | |
| 66 | .. note:: |
| 67 | |
| 68 | The default lexer (``\Twig\Lexer``) can be changed by calling |
| 69 | the ``setLexer()`` method:: |
| 70 | |
| 71 | $twig->setLexer($lexer); |
| 72 | |
| 73 | The Parser |
| 74 | ---------- |
| 75 | |
| 76 | The parser converts the token stream into an AST (Abstract Syntax Tree), or a |
| 77 | node tree (an instance of ``\Twig\Node\ModuleNode``). The core extension defines |
| 78 | the basic nodes like: ``for``, ``if``, ... and the expression nodes. |
| 79 | |
| 80 | You can manually convert a token stream into a node tree by calling the |
| 81 | ``parse()`` method of an environment:: |
| 82 | |
| 83 | $nodes = $twig->parse($stream); |
| 84 | |
| 85 | Echoing the node object gives you a nice representation of the tree:: |
| 86 | |
| 87 | echo $nodes."\n"; |
| 88 | |
| 89 | Here is the output for the ``Hello {{ name }}`` template: |
| 90 | |
| 91 | .. code-block:: text |
| 92 | |
| 93 | \Twig\Node\ModuleNode( |
| 94 | \Twig\Node\TextNode(Hello ) |
| 95 | \Twig\Node\PrintNode( |
| 96 | \Twig\Node\Expression\NameExpression(name) |
| 97 | ) |
| 98 | ) |
| 99 | |
| 100 | .. note:: |
| 101 | |
| 102 | The default parser (``\Twig\TokenParser\AbstractTokenParser``) can be changed by calling the |
| 103 | ``setParser()`` method:: |
| 104 | |
| 105 | $twig->setParser($parser); |
| 106 | |
| 107 | The Compiler |
| 108 | ------------ |
| 109 | |
| 110 | The last step is done by the compiler. It takes a node tree as an input and |
| 111 | generates PHP code usable for runtime execution of the template. |
| 112 | |
| 113 | You can manually compile a node tree to PHP code with the ``compile()`` method |
| 114 | of an environment:: |
| 115 | |
| 116 | $php = $twig->compile($nodes); |
| 117 | |
| 118 | The generated template for a ``Hello {{ name }}`` template reads as follows |
| 119 | (the actual output can differ depending on the version of Twig you are |
| 120 | using):: |
| 121 | |
| 122 | /* Hello {{ name }} */ |
| 123 | class __TwigTemplate_1121b6f109fe93ebe8c6e22e3712bceb extends Template |
| 124 | { |
| 125 | protected function doDisplay(array $context, array $blocks = []) |
| 126 | { |
| 127 | // line 1 |
| 128 | echo "Hello "; |
| 129 | echo twig_escape_filter($this->env, (isset($context["name"]) ? $context["name"] : null), "html", null, true); |
| 130 | } |
| 131 | |
| 132 | // some more code |
| 133 | } |
| 134 | |
| 135 | .. note:: |
| 136 | |
| 137 | The default compiler (``\Twig\Compiler``) can be changed by calling the |
| 138 | ``setCompiler()`` method:: |
| 139 | |
| 140 | $twig->setCompiler($compiler); |