php-kilo/src/Tokens/PHP.php

<?php declare(strict_types=1);

namespace Aviat\Kilo\Tokens;

use function Aviat\Kilo\str_contains;
use function Aviat\Kilo\tabs_to_spaces;

class PHP {

	private string $code;

	private array $rawLines;

	private array $tokens;

	private int $lineNum = 1;

	private function __construct(string $code)
	{
		$lines = explode("\n", $code);
		array_unshift($lines, '');
		unset($lines[0]);

		$this->code = $code;
		$this->rawLines = $lines;
		$this->tokens = array_fill(1, count($lines), []);
	}

	/**
	 * Use 'token_get_all' to get the tokens for a file,
	 * organized by row number
	 *
	 * @param string $code
	 * @return array
	 */
	public static function getTokens(string $code): array
	{
		return (new self($code))->organizeTokens();
	}

	/**
	 * Return tokens for the current $filename, organized
	 * by row number
	 *
	 * @param string $filename
	 * @return array
	 */
	public static function getFileTokens(string $filename): array
	{
		$code = file_get_contents($filename);

		if ($code === FALSE)
		{
			return [];
		}

		return self::getTokens($code);
	}

	protected function organizeTokens(): array
	{
		$rawTokens = token_get_all($this->code);
		foreach ($rawTokens as $t)
		{
			if (is_array($t))
			{
				$this->processArrayToken($t);
			}
			else if (is_string($t))
			{
				$this->processStringToken($t);
			}
		}

		ksort($this->tokens);

		return $this->tokens;
	}

	protected function processArrayToken(array $token): void
	{
		[$type, $rawChar, $currentLine] = $token;
		$char = tabs_to_spaces($rawChar);

		if ($currentLine !== $this->lineNum)
		{
			$this->lineNum = $currentLine;
		}

		$current = [
			'type' => $type,
			'typeName' => token_name($type),
			'char' => $char,
			'line' => $currentLine,
		];

		// Single new line, or starts with a new line with other whitespace
		if (strpos($char, "\n") === 0 && trim($char) === '')
		{
			$this->tokens[$this->lineNum][] = $current;
			$this->lineNum++;

			if ( ! array_key_exists($this->lineNum, $this->tokens))
			{
				$this->tokens[$this->lineNum] = [];
			}

			return;
		}

		// Only return the first line of a multi-line token for this line array
		if (str_contains($char, "\n"))
		{
			$chars = explode("\n", $char);
			$current['original'] = [
				'string' => $char,
				'lines' => $chars,
			];
			$current['char'] = array_shift($chars);

			// Add new lines for additional newline characters
			$nextLine = $currentLine;
			foreach ($chars as $char)
			{
				$nextLine++;

				if ( ! array_key_exists($nextLine, $this->tokens))
				{
					$tokens[$nextLine] = [];
				}

				if ( ! empty($char))
				{
					$this->processStringToken($char, $nextLine);
				}
			}
		}

		$this->tokens[$this->lineNum][] = $current;
	}

	protected function processStringToken(string $token, ?int $startLine = NULL): void
	{
		$char = tabs_to_spaces($token);

		$startLine = $startLine ?? $this->lineNum;
		$lineNumber = $this->findCorrectLine($char, $startLine) ?? $startLine;

		// Simple characters, usually delimiters or single character operators
		$this->tokens[$lineNumber][] = [
			'type' => -1,
			'typeName' => 'RAW',
			'char' => tabs_to_spaces($token),
		];
	}

	private function findCorrectLine(string $search, int $rowOffset, int $searchLength = 5): ?int
	{
		$end = $rowOffset + $searchLength;
		if ($end > count($this->rawLines))
		{
			$end = count($this->rawLines);
		}

		for ($i = $rowOffset; $i < $end; $i++)
		{
			if (str_contains($this->rawLines[$i], $search))
			{
				return $i;
			}
		}

		return NULL;
	}
}