Request: Distinguish quoted and unquoted heredoc strings

[JakeTunaley]

Running the parser (v4.2.1) against the following two code snippets produces identical ASTs.

Snippet 1

<?php
echo <<<FOO
hello world
FOO;

Snippet 2

<?php
echo <<<"FOO"
hello world
FOO;

Common AST

[
    {
        "nodeType": "Stmt_Echo",
        "exprs": [
            {
                "nodeType": "Scalar_String",
                "value": "hello world",
                "attributes": {
                    "startLine": 1,
                    "endLine": 3,
                    "kind": 3,
                    "docLabel": "FOO",
                    "docIndentation": ""
                }
            }
        ],
        "attributes": {
            "startLine": 1,
            "endLine": 3
        }
    }
]

Suggestion

Add a new boolean attribute which is true if and only if kind is PhpParser\Node\ScalarString_::KIND_HEREDOC and the heredoc identifier is wrapped in double-quotes. (This is preferable to a new kind, e.g. PhpParser\Node\ScalarString_::KIND_QUOTED_HEREDOC, to preserve backwards compatibility with scripts checking that attribute for either type of heredoc string).

[nikic]

What use case do you have in mind for this? Is it to throw a warning when it's quoted / not quoted, to preserve the format during pretty-printing, or something else?

[JakeTunaley]

I'm writing a tool that will detect PHP features not supported in some targeted version.

My co-workers and I were instructed that they could write PHP 7.1 before we tried to upgrade the servers, which was a mistake because one of the upgrades didn't take, so now we need to audit all the code we've written since then for things like this.

(And yes, I know the version of PHP just before quoted Heredoc strings were introduced in (5.2) is very, very unsupported, but the server's running Windows Server 2003...)

My current workaround is as follows:

// Node $node: The node being checked
// string $phpCode: The complete contents of the PHP file being checked

if ($node instanceof Node\Scalar\String_ && $node->getAttribute('kind') === Node\Scalar\String_::KIND_HEREDOC) {
    $docLabel = $node->getAttribute('docLabel');
    $sourceCode = substr($phpCode, $node->getStartFilePos(), $node->getEndFilePos() + 1 - $node->getStartFilePos());
    $heredocOpener = substr($sourceCode, 0, strpos($sourceCode, "\n"));
    // Regex matches a quoted heredoc opener
    if (preg_match('/^<<<\s*"' . preg_quote($docLabel) . '"/', $firstLine) === 1) {
        // $node is a quoted heredoc string
    }
}

[TomasVotruba]

I'm writing a tool that will detect PHP features not supported in some targeted version.

Just a tip for similar tool, only with tokens: https://github.com/PHPCompatibility/PHPCompatibility