From 6b3eaa683391360bfd559db80027bb307b590f2e Mon Sep 17 00:00:00 2001
From: Adam Pippin <adam@label-engine.com>
Date: Mon, 15 Feb 2021 00:05:12 -0800
Subject: [PATCH] First pass at expression parser/evaluator

---
 app/Engine/Cfnpp/Expression/Expression.php    | 124 ++++++++++++++++++
 app/Engine/Cfnpp/Expression/Parser.php        |  44 +++++++
 app/Engine/Cfnpp/Expression/Token.php         |  27 ++++
 .../Cfnpp/Expression/TokenNumericLiteral.php  |  45 +++++++
 app/Engine/Cfnpp/Expression/TokenOperator.php |  58 ++++++++
 .../Cfnpp/Expression/TokenStringLiteral.php   |  61 +++++++++
 app/Engine/Cfnpp/Expression/TokenVariable.php |  46 +++++++
 7 files changed, 405 insertions(+)
 create mode 100644 app/Engine/Cfnpp/Expression/Expression.php
 create mode 100644 app/Engine/Cfnpp/Expression/Parser.php
 create mode 100644 app/Engine/Cfnpp/Expression/Token.php
 create mode 100644 app/Engine/Cfnpp/Expression/TokenNumericLiteral.php
 create mode 100644 app/Engine/Cfnpp/Expression/TokenOperator.php
 create mode 100644 app/Engine/Cfnpp/Expression/TokenStringLiteral.php
 create mode 100644 app/Engine/Cfnpp/Expression/TokenVariable.php

diff --git a/app/Engine/Cfnpp/Expression/Expression.php b/app/Engine/Cfnpp/Expression/Expression.php
new file mode 100644
index 0000000..dbd64e5
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/Expression.php
@@ -0,0 +1,124 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class Expression
+{
+	protected $tokens;
+
+	protected $stack;
+
+	protected $variables;
+
+	public function __construct(array $tokens)
+	{
+		$this->tokens = $tokens;
+	}
+
+	public function evaluate(array $variables = [])
+	{
+		$this->variables = $variables;
+		$this->stack = [];
+		$tokens = $this->tokens;
+
+		while (sizeof($tokens))
+		{
+			$token = array_shift($tokens);
+
+			if ($token instanceof TokenNumericLiteral)
+			{
+				$this->evaluateTokenNumericLiteral($token);
+			}
+			elseif ($token instanceof TokenStringLiteral)
+			{
+				$this->evaluateTokenStringLiteral($token);
+			}
+			elseif ($token instanceof TokenOperator)
+			{
+				$this->evaluateTokenOperator($token);
+			}
+			elseif ($token instanceof TokenVariable)
+			{
+				$this->evaluateTokenVariable($token);
+			}
+			else
+			{
+				throw new \Exception('Unhandled expression token type: '.basename(get_class($token)));
+			}
+		}
+
+		if (sizeof($this->stack) != 1)
+		{
+			throw new \Exception('Expression did not evaluate down to a single value');
+		}
+
+		return $this->stack[0];
+	}
+
+	protected function evaluateTokenNumericLiteral(TokenNumericLiteral $token)
+	{
+		$this->push($token->getValue());
+	}
+
+	protected function evaluateTokenStringLiteral(TokenStringLiteral $token)
+	{
+		$this->push($token->getValue());
+	}
+
+	protected function evaluateTokenOperator(TokenOperator $token)
+	{
+		switch ($token->getOperator())
+		{
+			case 'eq':
+				$this->push($this->pop() == $this->pop());
+				break;
+			case 'neq':
+				$this->push($this->pop() != $this->pop());
+				break;
+			case 'gt':
+				$this->push($this->pop(1) > $this->pop());
+				break;
+			case 'gte':
+				$this->push($this->pop(1) >= $this->pop());
+				break;
+			case 'lt':
+				$this->push($this->pop(1) < $this->pop());
+				break;
+			case 'lte':
+				$this->push($this->pop(1) <= $this->pop());
+				break;
+			case 'and':
+				$var1 = $this->pop();
+				$var2 = $this->pop();
+				$this->push($var1 && $var2);
+				break;
+			case 'or':
+				$var1 = $this->pop();
+				$var2 = $this->pop();
+				$this->push($var1 || $var2);
+				break;
+		}
+	}
+
+	protected function evaluateTokenVariable(TokenVariable $token)
+	{
+		$name = $token->getName();
+		if (!isset($this->variables[$name]))
+		{
+			throw new \Exception('Undefined variable: '.$name);
+		}
+		$this->push($this->variables[$name]);
+	}
+
+	protected function pop(int $offset = 0)
+	{
+		return array_splice($this->stack, -1 * ($offset + 1), 1)[0];
+	}
+
+	protected function push($value)
+	{
+		array_push($this->stack, $value);
+	}
+}
diff --git a/app/Engine/Cfnpp/Expression/Parser.php b/app/Engine/Cfnpp/Expression/Parser.php
new file mode 100644
index 0000000..f483438
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/Parser.php
@@ -0,0 +1,44 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class Parser
+{
+	public const TOKEN_TYPES = [
+		TokenNumericLiteral::class,
+		TokenOperator::class,
+		TokenStringLiteral::class,
+		TokenVariable::class
+	];
+
+	public function parse(string $value): Expression
+	{
+		$original_value = $value;
+		$value = $value.' ';
+		$tokens = [];
+		while (strlen($value) > 0)
+		{
+			foreach (static::TOKEN_TYPES as $token_class)
+			{
+				if ($token_class::isToken($value))
+				{
+					$tokens[] = $token_class::getToken($value);
+
+					if (substr($value, 0, 1) != ' ')
+					{
+						throw new \Exception('Incompletely consumed token at offset '.(strlen($original_value) - strlen($value)).' in expression '.$original_value);
+					}
+
+					$value = substr($value, 1);
+					continue 2;
+				}
+			}
+
+			throw new \Exception('Unparseable value at offset '.(strlen($original_value) - strlen($value)).' in expression '.$original_value);
+		}
+
+		return new Expression($tokens);
+	}
+}
diff --git a/app/Engine/Cfnpp/Expression/Token.php b/app/Engine/Cfnpp/Expression/Token.php
new file mode 100644
index 0000000..e5f4c1e
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/Token.php
@@ -0,0 +1,27 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+abstract class Token
+{
+	/**
+	 * Check whether the stream passed in contains a valid token of this
+	 * type.
+	 *
+	 * @param  string $stream
+	 * @return bool
+	 */
+	abstract public static function isToken(string $stream): bool;
+
+	/**
+	 * Consume a token from the stream.
+	 *
+	 * Only valid if isToken is true.
+	 *
+	 * @param  string $stream
+	 * @return Token
+	 */
+	abstract public static function getToken(string &$stream): Token;
+}
diff --git a/app/Engine/Cfnpp/Expression/TokenNumericLiteral.php b/app/Engine/Cfnpp/Expression/TokenNumericLiteral.php
new file mode 100644
index 0000000..1a4127d
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/TokenNumericLiteral.php
@@ -0,0 +1,45 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class TokenNumericLiteral extends Token
+{
+	protected $value;
+
+	public function __construct($value)
+	{
+		$this->value = $value;
+	}
+
+	public function getValue()
+	{
+		return $this->value;
+	}
+
+	public static function isToken(string $stream): bool
+	{
+		return is_numeric($stream[0]);
+	}
+
+	public static function getToken(string &$stream): Token
+	{
+		$buffer = '';
+		for ($i = 0; $i < strlen($stream); $i++)
+		{
+			if (preg_match('/^[0-9]$/', $stream[$i]))
+			{
+				$buffer .= $stream[$i];
+			}
+			else
+			{
+				break;
+			}
+		}
+
+		$stream = substr($stream, strlen($buffer));
+
+		return new TokenNumericLiteral($buffer);
+	}
+}
diff --git a/app/Engine/Cfnpp/Expression/TokenOperator.php b/app/Engine/Cfnpp/Expression/TokenOperator.php
new file mode 100644
index 0000000..ca2c02e
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/TokenOperator.php
@@ -0,0 +1,58 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class TokenOperator extends Token
+{
+	public const OPERATORS = [
+		'eq',
+		'gt',
+		'gte',
+		'lt',
+		'lte',
+		'neq',
+		'and',
+		'or'
+	];
+
+	protected $operator;
+
+	public function __construct($operator)
+	{
+		$this->operator = $operator;
+	}
+
+	public function getOperator()
+	{
+		return $this->operator;
+	}
+
+	public static function isToken(string $stream): bool
+	{
+		foreach (static::OPERATORS as $operator)
+		{
+			if (strlen($stream) > strlen($operator) &&
+				substr($stream, 0, strlen($operator)) == $operator)
+			{
+				return true;
+			}
+		}
+		return false;
+	}
+
+	public static function getToken(string &$stream): Token
+	{
+		foreach (static::OPERATORS as $operator)
+		{
+			if (strlen($stream) > strlen($operator) &&
+				substr($stream, 0, strlen($operator)) == $operator)
+			{
+				$operator = substr($stream, 0, strlen($operator));
+				$stream = substr($stream, strlen($operator));
+				return new TokenOperator($operator);
+			}
+		}
+	}
+}
diff --git a/app/Engine/Cfnpp/Expression/TokenStringLiteral.php b/app/Engine/Cfnpp/Expression/TokenStringLiteral.php
new file mode 100644
index 0000000..30bec17
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/TokenStringLiteral.php
@@ -0,0 +1,61 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class TokenStringLiteral extends Token
+{
+	protected $value;
+
+	public function __construct($value)
+	{
+		$this->value = $value;
+	}
+
+	public function getValue()
+	{
+		return $this->value;
+	}
+
+	public static function isToken(string $stream): bool
+	{
+		return $stream[0] == '"';
+	}
+
+	public static function getToken(string &$stream): Token
+	{
+		$buffer = '';
+		$in_string = false;
+		$escaped = false;
+
+		for ($i = 0; $i < strlen($stream); $i++)
+		{
+			if ($escaped)
+			{
+				$buffer .= $stream[$i];
+			}
+			elseif ($stream[$i] == '"')
+			{
+				if ($in_string)
+				{
+					break;
+				}
+
+				$in_string = true;
+			}
+			elseif ($stream[$i] == '\\')
+			{
+				$escaped = true;
+			}
+			else
+			{
+				$buffer .= $stream[$i];
+			}
+		}
+
+		$stream = substr($stream, $i + 1);
+
+		return new TokenStringLiteral($buffer);
+	}
+}
diff --git a/app/Engine/Cfnpp/Expression/TokenVariable.php b/app/Engine/Cfnpp/Expression/TokenVariable.php
new file mode 100644
index 0000000..ea8cc1f
--- /dev/null
+++ b/app/Engine/Cfnpp/Expression/TokenVariable.php
@@ -0,0 +1,46 @@
+<?php
+
+declare(strict_types=1);
+
+namespace App\Engine\Cfnpp\Expression;
+
+class TokenVariable extends Token
+{
+	protected $name;
+
+	public function __construct($name)
+	{
+		$this->name = $name;
+	}
+
+	public function getName()
+	{
+		return $this->name;
+	}
+
+	public static function isToken(string $stream): bool
+	{
+		return preg_match('/^[A-Za-z]$/', $stream[0]);
+	}
+
+	public static function getToken(string &$stream): Token
+	{
+		$buffer = '';
+		$buffer = $stream[0];
+		for ($i = 1; $i < strlen($stream); $i++)
+		{
+			if (preg_match('/^[A-Za-z0-9]$/', $stream[$i]))
+			{
+				$buffer .= $stream[$i];
+			}
+			else
+			{
+				break;
+			}
+		}
+
+		$stream = substr($stream, strlen($buffer));
+
+		return new TokenVariable($buffer);
+	}
+}