XssOutputValidator.php

Go to the documentation of this file.

<?php
namespace Magento\TestFramework\Utility;

class XssOutputValidator
{
    const ESCAPE_NOT_VERIFIED_PATTERN = '/\* @escapeNotVerified \*/';

    const ESCAPED_PATTERN = '/\* @noEscape \*/';

    private $origins = [];

    private $replacements = [];

    private $escapeFunctions = ['escapeHtml', 'escapeHtmlAttr', 'escapeUrl', 'escapeJs', 'escapeCss'];

    public function getLinesWithXssSensitiveOutput($file)
    {
        $fileContent = file_get_contents($file);
        $xssUnsafeBlocks = $this->getXssUnsafeBlocks($fileContent);

        $lines = [];
        foreach ($xssUnsafeBlocks as $block) {
            $lines = array_merge($lines, $this->findBlockLineNumbers($block, $fileContent));
        }

        if (count($lines)) {
            $lines = array_unique($lines);
            sort($lines);
            return implode(',', $lines);
        }

        return '';
    }

    private function findBlockLineNumbers($block, $content)
    {
        $results = [];
        $pos = strpos($content, $block, 0);
        while ($pos !== false) {
            $contentBeforeString = substr($content, 0, $pos);
            if ($this->isNotEscapeMarkedBlock($contentBeforeString)
                && $this->isNotInCommentBlock($contentBeforeString)
            ) {
                $results[] = count(explode(PHP_EOL, $contentBeforeString));
            }
            $pos = strpos($content, $block, $pos + 1);
        }

        return $results;
    }

    public function getXssUnsafeBlocks($fileContent)
    {
        $results = [];

        $fileContent = $this->replacePhpQuoteWithPlaceholders($fileContent);
        $fileContent = $this->replacePhpCommentsWithPlaceholders($fileContent);

        $this->addOriginReplacement('\'\'', "'-*=single=*-'");
        $this->addOriginReplacement('""', '"-*=double=*-"');

        if (preg_match_all('/<[?](php|=)(.*?)[?]>/sm', $fileContent, $phpBlockMatches)) {
            foreach ($phpBlockMatches[2] as $index => $phpBlock) {
                $phpCommands = explode(';', $phpBlock);
                if ($phpBlockMatches[1][$index] == 'php') {
                    $echoCommands = preg_grep('#( |^|/\*.*?\*/)echo[\s(]+.*#sm', $phpCommands);
                } else {
                    $echoCommands[] = $phpBlockMatches[0][$index];
                }
                $results = array_merge(
                    $results,
                    $this->getEchoUnsafeCommands($echoCommands)
                );
            }
        }

        $this->clearOriginReplacements();
        $results = array_unique($results);

        return $results;
    }

    private function getEchoUnsafeCommands(array $echoCommands)
    {
        $results = [];
        foreach ($echoCommands as $echoCommand) {
            if ($this->isNotEscapeMarkedCommand($echoCommand)) {
                $echoCommand = preg_replace('/^(.*?)echo/sim', 'echo', $echoCommand);
                $preparedEchoCommand = $this->prepareEchoCommand($echoCommand);
                $isEscapeFunctionArgument = preg_match(
                    '/->(' . implode('|', $this->escapeFunctions) . ')\(.*?\)$/sim',
                    $preparedEchoCommand
                );
                $xssUnsafeCommands = array_filter(
                    $isEscapeFunctionArgument ? [$preparedEchoCommand] : explode('.', $preparedEchoCommand),
                    [$this, 'isXssUnsafeCommand']
                );
                if (count($xssUnsafeCommands)) {
                    $results[] = str_replace(
                        $this->getReplacements(),
                        $this->getOrigins(),
                        $echoCommand
                    );
                }
            }
        }

        return $results;
    }

    private function prepareEchoCommand($command)
    {
        $command = preg_replace('/<[?]=(.*?)[?]>/sim', '\1', $command);
        return trim(ltrim(explode(';', $command)[0], 'echo'));
    }

    private function isNotEscapeMarkedBlock($contentBeforeString)
    {
        return !preg_match(
            '%(' . self::ESCAPE_NOT_VERIFIED_PATTERN . '|' . self::ESCAPED_PATTERN . ')$%sim',
            trim($contentBeforeString)
        );
    }

    private function isNotInCommentBlock($contentBeforeString)
    {
        $contentBeforeString = explode('<?php', $contentBeforeString);
        $contentBeforeString = preg_replace(
            '%/\*.*?\*/%si',
            '',
            end($contentBeforeString)
        );

        return (strpos($contentBeforeString, '/*') === false);
    }

    private function isNotEscapeMarkedCommand($command)
    {
        return !preg_match(
            '%' . self::ESCAPE_NOT_VERIFIED_PATTERN . '|' . self::ESCAPED_PATTERN . '%sim',
            $command
        );
    }

    public function isXssUnsafeCommand($command)
    {
        $command = trim($command);

        switch (true) {
            case preg_match(
                '/->(' . implode('|', $this->escapeFunctions) . '|.*html.*)\(/simU',
                $this->getLastMethod($command)
            ):
                return false;
            case preg_match('/^\((int|bool|float)\)/sim', $command):
                return false;
            case preg_match('/^count\(/sim', $command):
                return false;
            case preg_match("/^'.*'$/sim", $command):
                return false;
            case preg_match('/^".*?"$/sim', $command, $matches):
                return $this->isContainPhpVariables($this->getOrigin($matches[0]));
            default:
                return true;
        }
    }

    private function getLastMethod($command)
    {
        if (preg_match_all(
            '/->.*?\(.*?\)/sim',
            $this->clearMethodBracketContent($command),
            $matches
        )) {
            $command = end($matches[0]);
            $command = substr($command, 0, strpos($command, '(') + 1);
        }

        return $command;
    }

    private function clearMethodBracketContent($command)
    {
        $bracketInterval = [];
        $bracketOpenPos = [];
        $command = str_split($command);
        foreach ($command as $index => $character) {
            if ($character == '(') {
                array_push($bracketOpenPos, $index);
            }
            if (count($bracketOpenPos)) {
                if ($character == ')') {
                    $lastOpenPos = array_pop($bracketOpenPos);
                    if (count($bracketOpenPos) == 0) {
                        $bracketInterval[] = [$lastOpenPos, $index];
                    }
                }
            }
        }
        foreach ($bracketInterval as $interval) {
            for ($i = $interval[0] + 1; $i < $interval[1]; $i++) {
                unset($command[$i]);
            }
        }
        $command = implode('', $command);

        return $command;
    }

    private function isContainPhpVariables($content)
    {
        return preg_match('/[^\\\\]\$[a-z_\x7f-\xff]/sim', $content);
    }

    private function replacePhpQuoteWithPlaceholders($fileContent)
    {
        $origins = [];
        $replacements = [];
        if (preg_match_all('/<[?](php|=)(.*?)[?]>/sm', $fileContent, $phpBlockMatches)) {
            foreach ($phpBlockMatches[2] as $phpBlock) {
                $phpBlockQuoteReplaced = preg_replace(
                    ['/([^\\\\])\'\'/si', '/([^\\\\])""/si'],
                    ["\1'-*=single=*-'", '\1"-*=double=*-"'],
                    $phpBlock
                );

                $this->addQuoteOriginsReplacements(
                    $phpBlockQuoteReplaced,
                    [
                        '/([^\\\\])([\'])(.*?)([^\\\\])([\'])/sim'
                    ]
                );
                $this->addQuoteOriginsReplacements(
                    $phpBlockQuoteReplaced,
                    [
                        '/([^\\\\])(["])(.*?)([^\\\\])(["])/sim',
                    ]
                );

                $origins[] = $phpBlock;
                $replacements[]  = str_replace(
                    $this->getOrigins(),
                    $this->getReplacements(),
                    $phpBlockQuoteReplaced
                );
            }
        }

        return str_replace($origins, $replacements, $fileContent);
    }

    private function replacePhpCommentsWithPlaceholders($fileContent)
    {
        $origins= [];
        $replacements = [];
        if (preg_match_all('%/\*.*?\*/%simu', $fileContent, $docCommentMatches, PREG_SET_ORDER)) {
            foreach ($docCommentMatches as $docCommentMatch) {
                if ($this->isNotEscapeMarkedCommand($docCommentMatch[0])
                    && !$this->issetOrigin($docCommentMatch[0])) {
                    $origin = $docCommentMatch[0];
                    $replacement = '-*!' . count($this->getOrigins()) . '!*-';
                    $origins[] = $origin;
                    $replacements[] = $replacement;
                    $this->addOriginReplacement(
                        $origin,
                        $replacement
                    );
                }
            }
        }

        return str_replace($origins, $replacements, $fileContent);
    }

    private function addQuoteOriginsReplacements($phpBlock, array $patterns)
    {
        foreach ($patterns as $pattern) {
            if (preg_match_all($pattern, $phpBlock, $quoteMatches, PREG_SET_ORDER)) {
                foreach ($quoteMatches as $quoteMatch) {
                    $origin = $quoteMatch[2] . $quoteMatch[3] . $quoteMatch[4] . $quoteMatch[5];
                    if (!$this->issetOrigin($origin)) {
                        $this->addOriginReplacement(
                            $origin,
                            $quoteMatch[2] . '-*=' . count($this->getOrigins()) . '=*-' . $quoteMatch[5]
                        );
                    }
                }
            }
        }
    }

    private function addOriginReplacement($origin, $replacement)
    {
        $this->origins[$replacement] = $origin;
        $this->replacements[$replacement] = $replacement;
    }

    private function clearOriginReplacements()
    {
        $this->origins = [];
        $this->replacements = [];
    }

    private function getOrigins()
    {
        return $this->origins;
    }

    private function getOrigin($key)
    {
        return array_key_exists($key, $this->origins) ? $this->origins[$key] : null;
    }

    private function issetOrigin($origin)
    {
        return in_array($origin, $this->origins);
    }

    private function getReplacements()
    {
        return $this->replacements;
    }
}