[String_] Add rawValue attribute

[TomasVotruba]

In Rector we have to hack around tokens to get the original value, because the parse() method flattens original information:

PHP-Parser/lib/PhpParser/Node/Scalar/String_.php

Lines 55 to 72 in a6e3466

	public static function parse(string $str, bool $parseUnicodeEscape = true) : string {
	$bLength = 0;
	if ('b' === $str[0] || 'B' === $str[0]) {
	$bLength = 1;
	}
	if ('\'' === $str[$bLength]) {
	return str_replace(
	['\\\\', '\\\''],
	['\\', '\''],
	substr($str, $bLength + 1, -1)
	);
	} else {
	return self::parseEscapeSequences(
	substr($str, $bLength + 1, -1), '"', $parseUnicodeEscape
	);
	}
	}

We could use also attribute value, but it would require more grammar changes.

Closes #577

Also fixes #576

[dryabov]

Isn't getStartFilePos/getEndFilePos sufficient to get raw data? Note it requires to manually create the lexer and pass it to the parser, e.g.:

use PhpParser\Lexer\Emulative;
use PhpParser\ParserFactory;

$options = ['usedAttributes' => ['comments', 'startLine', 'endLine', 'startFilePos', 'endFilePos']];
$lexer = new Emulative($options);
$parser = (new ParserFactory)->create(ParserFactory::PREFER_PHP7, $lexer);

$code = '<?php
Foo::bar("This is a string with escape sequence \x41");
';
$ast = $parser->parse($code);
$node = $ast[0]->expr->args[0]->value;

$startFilePos = $node->getStartFilePos();
$endFilePos = $node->getEndFilePos();
var_dump(substr($code, $startFilePos, $endFilePos - $startFilePos + 1));

[TomasVotruba]

Avoiding work with tokens to get original value is exactly point. During Rector upgrade it might be crutial to know the original value, not to replace something deprecated.

[dryabov]

But get(Start|End)FilePos is a standard way to get access to raw data, e.g. well-known psalm works in this way (see Psalm\Internal\Provider\StatementsProvider::parseStatements for parser initialization there, and search its source code for getStartFilePos() and getAttribute('startFilePos') to get usage examples).

[TomasVotruba]

That seems like a work around. The change of value is the issue.

[TomasVotruba]

Added rawValue attribute ✔️

[TomasVotruba]

Updated and rebased 👍 Ready for review ✔️

[TomasVotruba]

Thanks 👍

[nikic]

Probably this should be added to EncapsedStringPart as well?

[TomasVotruba]

What exactly do you mean? Have fromString() method there? (I never used this node much, so I have no idea about it.)

[nikic]

I mean having a rawValue attribute. It seems inconsistent to only add it to non-interpolated strings.

[TomasVotruba]

I'm on it 😉

[TomasVotruba]

See #837