FreshRSS/cli/i18n/PluralFormsCompiler.php

<?php
declare(strict_types=1);

/**
 * Plural form inspired by GNU gettext plural forms, converted into PHP lambdas.
 * https://www.gnu.org/software/gettext/manual/html_node/Plural-forms.html
 */
final class PluralFormsCompiler {
	private const FORMULA_PATTERN = '/^\s*nplurals\s*=\s*(\d+)\s*;\s*plural\s*=\s*(.+?)\s*;\s*$/';

	private const COMMENT_PATTERN = '/^\s*\/\/\s*Plural-Forms:\s*(?P<formula>.+?)\s*$/mi';

	private const COMMENT_PREFIX_PATTERN = '/^\s*\/\/\s*/';

	private const ALLOWED_EXPRESSION_PATTERN = '/^[0-9n\s!<>=&|?:()%+*\/-]+$/';

	/**
	 * @return array{formula:string,nplurals:int,lambda:string}
	 */
	public function compileFormula(string $pluralForms): array {
		['formula' => $formula, 'nplurals' => $pluralCount, 'expression' => $expression] =
			$this->parsePluralHeader($pluralForms);
		$this->validatePluralExpression($expression, $formula);

		$lambdaExpression = $pluralCount === 2 && !str_contains($expression, '?')
			? '((' . $this->transpileLeafExpression($expression) . ') ? 1 : 0)'
			: $this->transpileExpression($expression);

		return [
			'formula' => $formula,
			'nplurals' => $pluralCount,
			'lambda' => 'static fn (int $n): int => ' . $lambdaExpression,
		];
	}

	public function compileFormulaToLambda(string $pluralForms): string {
		return $this->compileFormula($pluralForms)['lambda'];
	}

	public function compileFile(string $filePath, bool $throwOnError = true): bool {
		try {
			if (!is_file($filePath)) {
				throw new InvalidArgumentException('Plural file not found: ' . $filePath);
			}

			$compiled = $this->compileFormula($this->extractPluralFormsFromFile($filePath));
			$newContent = $this->renderCompiledFile($compiled);
			$currentContent = file_get_contents($filePath);
			if (!is_string($currentContent)) {
				throw new RuntimeException('Unable to read plural file: ' . $filePath);
			}

			if ($currentContent === $newContent) {
				return false;
			}

			if (file_put_contents($filePath, $newContent) === false) {
				throw new RuntimeException('Unable to write plural file: ' . $filePath);
			}
		} catch (Throwable $e) {
			if ($throwOnError) {
				throw $e;
			}
			$message = 'Error compiling plural file `' . $filePath . '`: ' . $e->getMessage() . "\n";
			if (defined('STDERR')) {
				fwrite(STDERR, $message);
			} else {
				echo $message;
			}
			return false;
		}

		return true;
	}

	public function compileAll(string $globPattern = I18N_PATH . '/*/plurals.php'): int {
		$files = glob($globPattern) ?: [];
		sort($files, SORT_NATURAL);

		$changed = 0;
		foreach ($files as $filePath) {
			if ($this->compileFile($filePath, throwOnError: false)) {
				$changed++;
			}
		}

		return $changed;
	}

	private function extractPluralFormsFromFile(string $filePath): string {
		$fileContent = file_get_contents($filePath);
		if (!is_string($fileContent)) {
			throw new RuntimeException('Unable to read plural file: ' . $filePath);
		}

		if (preg_match(self::COMMENT_PATTERN, $fileContent, $matches) === 1) {
			return $this->normalisePluralForms($matches['formula']);
		}

		return $this->extractGetTextPluralFormsFromFile($filePath);
	}

	/**
	 * @param array{formula:string,nplurals:int,lambda:string} $compiled
	 */
	private function renderCompiledFile(array $compiled): string {
		return <<<PHP
			<?php

			// Plural-Forms: {$compiled['formula']}
			// This file is generated by cli/compile.plurals.php.
			// Edit the formula comment and run `make fix-all`.

			return array(
				'nplurals' => {$compiled['nplurals']},
				'plural' => {$compiled['lambda']},
			);

			PHP;
	}

	/**
	 * @return array{formula:string,nplurals:int,expression:string}
	 */
	private function parsePluralHeader(string $pluralForms): array {
		$formula = $this->normalisePluralForms($pluralForms);
		if (!preg_match(self::FORMULA_PATTERN, $formula, $matches)) {
			throw new InvalidArgumentException('Invalid plural formula: ' . $formula);
		}

		return [
			'formula' => $formula,
			'nplurals' => max(1, (int)$matches[1]),
			'expression' => $matches[2],
		];
	}

	private function normalisePluralForms(string $pluralForms): string {
		$pluralForms = trim($pluralForms);
		$pluralForms = preg_replace(self::COMMENT_PREFIX_PATTERN, '', $pluralForms) ?? $pluralForms;
		if (preg_match('/^\s*Plural-Forms:\s*(?P<formula>.+?)\s*$/i', $pluralForms, $matches) === 1) {
			$pluralForms = $matches['formula'];
		}

		return trim($pluralForms);
	}

	private function extractGetTextPluralFormsFromFile(string $filePath): string {
		$pluralData = include $filePath;
		$pluralForms = is_array($pluralData) ? ($pluralData['plural-forms'] ?? null) : null;
		if (!is_string($pluralForms) || $pluralForms === '') {
			throw new RuntimeException('No plural formula found in `' . $filePath . '`');
		}

		return $this->normalisePluralForms($pluralForms);
	}

	/**
	 * Lightweight validation only. The compiler transpiles real shipped formulas heuristically.
	 */
	private function validatePluralExpression(string $expression, string $pluralForms): void {
		if (!preg_match(self::ALLOWED_EXPRESSION_PATTERN, $expression)) {
			throw new RuntimeException('Unsupported token in plural expression `' . $pluralForms . '`');
		}

		$depth = 0;
		$length = strlen($expression);
		for ($index = 0; $index < $length; $index++) {
			$character = $expression[$index];
			if ($character === '(') {
				$depth++;
			} elseif ($character === ')') {
				$depth--;
				if ($depth < 0) {
					throw new RuntimeException('Unbalanced parentheses in plural expression `' . $pluralForms . '`');
				}
			}
		}

		if ($depth !== 0) {
			throw new RuntimeException('Unbalanced parentheses in plural expression `' . $pluralForms . '`');
		}

		if (substr_count($expression, '?') !== substr_count($expression, ':')) {
			throw new RuntimeException('Unbalanced ternary operators in plural expression `' . $pluralForms . '`');
		}

		if (str_contains($expression, '/')) {
			throw new RuntimeException('Operator `/` is not supported in plural expression `' . $pluralForms . '`');
		}
	}

	private function transpileExpression(string $expression): string {
		$expression = $this->stripOuterParentheses(trim($expression));
		[$condition, $ifTrue, $ifFalse] = $this->splitTopLevelTernary($expression);
		if ($condition === null) {
			return $this->transpileLeafExpression($expression);
		}

		return '(' . $this->transpileLeafExpression($condition) . ' ? ' . $this->transpileExpression($ifTrue) . ' : '
			. $this->transpileExpression($ifFalse) . ')';
	}

	private function transpileLeafExpression(string $expression): string {
		$expression = $this->stripOuterParentheses(trim($expression));
		// Convert gettext variable name to PHP variable syntax
		$expression = preg_replace('/\bn\b/', '$n', $expression) ?? $expression;
		// Enforce strict equality
		$expression = preg_replace('/(?<![=!<>])==(?!=)/', '===', $expression) ?? $expression;
		// Enforce strict inequality
		$expression = preg_replace('/!=(?!=)/', '!==', $expression) ?? $expression;
		// Normalise operator spacing
		$expression = preg_replace('/\s*(===|!==|==|!=|<=|>=|\|\||&&|[%*+\-<>])\s*/', ' $1 ', $expression) ?? $expression;
		// Collapse repeated whitespace
		$expression = preg_replace('/\s+/', ' ', trim($expression)) ?? trim($expression);
		return $expression;
	}

	/**
	 * @return array{0:?string,1:string,2:string}
	 */
	private function splitTopLevelTernary(string $expression): array {
		$questionPosition = null;
		$depth = 0;
		$ternaryDepth = 0;
		$length = strlen($expression);

		for ($index = 0; $index < $length; $index++) {
			$character = $expression[$index];
			if ($character === '(') {
				$depth++;
				continue;
			}

			if ($character === ')') {
				$depth--;
				continue;
			}

			if ($depth !== 0) {
				continue;
			}

			if ($character === '?') {
				$questionPosition ??= $index;
				$ternaryDepth++;
				continue;
			}

			if ($character === ':' && $questionPosition !== null) {
				$ternaryDepth--;
				if ($ternaryDepth === 0) {
					return [
						trim(substr($expression, 0, $questionPosition)),
						trim(substr($expression, $questionPosition + 1, $index - $questionPosition - 1)),
						trim(substr($expression, $index + 1)),
					];
				}
			}
		}

		return [null, '', ''];
	}

	private function stripOuterParentheses(string $expression): string {
		$expression = trim($expression);
		while (str_starts_with($expression, '(') && str_ends_with($expression, ')')) {
			$depth = 0;
			$isWrapped = true;
			$length = strlen($expression);
			for ($index = 0; $index < $length; $index++) {
				$character = $expression[$index];
				if ($character === '(') {
					$depth++;
				} elseif ($character === ')') {
					$depth--;
				}

				if ($depth === 0 && $index < $length - 1) {
					$isWrapped = false;
					break;
				}
			}

			if (!$isWrapped) {
				break;
			}

			$expression = trim(substr($expression, 1, -1));
		}

		return $expression;
	}
}