[String_] Add rawValue attribute (#831)

nikic · May 15, 2022 · 5d83adc · 5d83adc
1 parent 3bf0082
commit 5d83adc
Show file tree

Hide file tree

Showing 8 changed files with 46 additions and 25 deletions.
diff --git a/grammar/php5.y b/grammar/php5.y
@@ -689,9 +689,7 @@ array_expr:
 
 scalar_dereference:
       array_expr '[' dim_offset ']'                         { $$ = Expr\ArrayDimFetch[$1, $3]; }
-    | T_CONSTANT_ENCAPSED_STRING '[' dim_offset ']'
-          { $attrs = attributes(); $attrs['kind'] = strKind($1);
-            $$ = Expr\ArrayDimFetch[new Scalar\String_(Scalar\String_::parse($1), $attrs), $3]; }
+    | T_CONSTANT_ENCAPSED_STRING '[' dim_offset ']'         { $$ = Expr\ArrayDimFetch[Scalar\String_::fromString($1, attributes()), $3]; }
     | constant '[' dim_offset ']'                           { $$ = Expr\ArrayDimFetch[$1, $3]; }
     | scalar_dereference '[' dim_offset ']'                 { $$ = Expr\ArrayDimFetch[$1, $3]; }
     /* alternative array syntax missing intentionally */
@@ -794,9 +792,7 @@ ctor_arguments:
 common_scalar:
       T_LNUMBER                                             { $$ = $this->parseLNumber($1, attributes(), true); }
     | T_DNUMBER                                             { $$ = Scalar\DNumber::fromString($1, attributes()); }
-    | T_CONSTANT_ENCAPSED_STRING
-          { $attrs = attributes(); $attrs['kind'] = strKind($1);
-            $$ = new Scalar\String_(Scalar\String_::parse($1, false), $attrs); }
+    | T_CONSTANT_ENCAPSED_STRING                            { $$ = Scalar\String_::fromString($1, attributes(), false); }
     | T_LINE                                                { $$ = Scalar\MagicConst\Line[]; }
     | T_FILE                                                { $$ = Scalar\MagicConst\File[]; }
     | T_DIR                                                 { $$ = Scalar\MagicConst\Dir[]; }

diff --git a/grammar/php7.y b/grammar/php7.y
@@ -1014,9 +1014,7 @@ dereferencable_scalar:
           { $attrs = attributes(); $attrs['kind'] = Expr\Array_::KIND_LONG;
             $$ = new Expr\Array_($3, $attrs); }
     | array_short_syntax                                    { $$ = $1; }
-    | T_CONSTANT_ENCAPSED_STRING
-          { $attrs = attributes(); $attrs['kind'] = strKind($1);
-            $$ = new Scalar\String_(Scalar\String_::parse($1), $attrs); }
+    | T_CONSTANT_ENCAPSED_STRING                            { $$ = Scalar\String_::fromString($1, attributes()); }
     | '"' encaps_list '"'
           { $attrs = attributes(); $attrs['kind'] = Scalar\String_::KIND_DOUBLE_QUOTED;
             parseEncapsed($2, '"', true); $$ = new Scalar\Encapsed($2, $attrs); }

diff --git a/grammar/phpyLang.php b/grammar/phpyLang.php
@@ -128,14 +128,6 @@ function($matches) {
                        . ' else { ' . $args[0] . ' = null; }';
             }
 
-            if ('strKind' === $name) {
-                assertArgs(1, $args, $name);
-
-                return '(' . $args[0] . '[0] === "\'" || (' . $args[0] . '[1] === "\'" && '
-                       . '(' . $args[0] . '[0] === \'b\' || ' . $args[0] . '[0] === \'B\')) '
-                       . '? Scalar\String_::KIND_SINGLE_QUOTED : Scalar\String_::KIND_DOUBLE_QUOTED)';
-            }
-
             if ('prependLeadingComments' === $name) {
                 assertArgs(1, $args, $name);
 

diff --git a/lib/PhpParser/Node/Scalar/String_.php b/lib/PhpParser/Node/Scalar/String_.php
@@ -42,6 +42,22 @@ public function getSubNodeNames() : array {
         return ['value'];
     }
 
+    /**
+     * @param bool $parseUnicodeEscape Whether to parse PHP 7 \u escapes
+     */
+    public static function fromString(string $str, array $attributes = [], bool $parseUnicodeEscape = true): self
+    {
+        $attributes['kind'] = ($str[0] === "'" || ($str[1] === "'" && ($str[0] === 'b' || $str[0] === 'B')))
+            ? Scalar\String_::KIND_SINGLE_QUOTED
+            : Scalar\String_::KIND_DOUBLE_QUOTED;
+
+        $attributes['rawValue'] = $str;
+
+        $string = self::parse($str, $parseUnicodeEscape);
+
+        return new self($string, $attributes);
+    }
+
     /**
      * @internal
      *

diff --git a/lib/PhpParser/Parser/Php5.php b/lib/PhpParser/Parser/Php5.php
@@ -2147,8 +2147,7 @@ protected function initReduceCallbacks() {
                  $this->semValue = new Expr\ArrayDimFetch($this->semStack[$stackPos-(4-1)], $this->semStack[$stackPos-(4-3)], $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes);
             },
             392 => function ($stackPos) {
-                 $attrs = $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes; $attrs['kind'] = ($this->semStack[$stackPos-(4-1)][0] === "'" || ($this->semStack[$stackPos-(4-1)][1] === "'" && ($this->semStack[$stackPos-(4-1)][0] === 'b' || $this->semStack[$stackPos-(4-1)][0] === 'B')) ? Scalar\String_::KIND_SINGLE_QUOTED : Scalar\String_::KIND_DOUBLE_QUOTED);
-            $this->semValue = new Expr\ArrayDimFetch(new Scalar\String_(Scalar\String_::parse($this->semStack[$stackPos-(4-1)]), $attrs), $this->semStack[$stackPos-(4-3)], $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes);
+                 $this->semValue = new Expr\ArrayDimFetch(Scalar\String_::fromString($this->semStack[$stackPos-(4-1)], $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes), $this->semStack[$stackPos-(4-3)], $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes);
             },
             393 => function ($stackPos) {
                  $this->semValue = new Expr\ArrayDimFetch($this->semStack[$stackPos-(4-1)], $this->semStack[$stackPos-(4-3)], $this->startAttributeStack[$stackPos-(4-1)] + $this->endAttributes);
@@ -2278,8 +2277,7 @@ protected function initReduceCallbacks() {
                  $this->semValue = Scalar\DNumber::fromString($this->semStack[$stackPos-(1-1)], $this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes);
             },
             435 => function ($stackPos) {
-                 $attrs = $this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes; $attrs['kind'] = ($this->semStack[$stackPos-(1-1)][0] === "'" || ($this->semStack[$stackPos-(1-1)][1] === "'" && ($this->semStack[$stackPos-(1-1)][0] === 'b' || $this->semStack[$stackPos-(1-1)][0] === 'B')) ? Scalar\String_::KIND_SINGLE_QUOTED : Scalar\String_::KIND_DOUBLE_QUOTED);
-            $this->semValue = new Scalar\String_(Scalar\String_::parse($this->semStack[$stackPos-(1-1)], false), $attrs);
+                 $this->semValue = Scalar\String_::fromString($this->semStack[$stackPos-(1-1)], $this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes, false);
             },
             436 => function ($stackPos) {
                  $this->semValue = new Scalar\MagicConst\Line($this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes);

diff --git a/lib/PhpParser/Parser/Php7.php b/lib/PhpParser/Parser/Php7.php
@@ -2543,8 +2543,7 @@ protected function initReduceCallbacks() {
                  $this->semValue = $this->semStack[$stackPos-(1-1)];
             },
             512 => function ($stackPos) {
-                 $attrs = $this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes; $attrs['kind'] = ($this->semStack[$stackPos-(1-1)][0] === "'" || ($this->semStack[$stackPos-(1-1)][1] === "'" && ($this->semStack[$stackPos-(1-1)][0] === 'b' || $this->semStack[$stackPos-(1-1)][0] === 'B')) ? Scalar\String_::KIND_SINGLE_QUOTED : Scalar\String_::KIND_DOUBLE_QUOTED);
-            $this->semValue = new Scalar\String_(Scalar\String_::parse($this->semStack[$stackPos-(1-1)]), $attrs);
+                 $this->semValue = Scalar\String_::fromString($this->semStack[$stackPos-(1-1)], $this->startAttributeStack[$stackPos-(1-1)] + $this->endAttributes);
             },
             513 => function ($stackPos) {
                  $attrs = $this->startAttributeStack[$stackPos-(3-1)] + $this->endAttributes; $attrs['kind'] = Scalar\String_::KIND_DOUBLE_QUOTED;

diff --git a/test/PhpParser/Node/Scalar/StringTest.php b/test/PhpParser/Node/Scalar/StringTest.php
@@ -2,8 +2,28 @@
 
 namespace PhpParser\Node\Scalar;
 
+use PhpParser\Node\Stmt\Echo_;
+use PhpParser\ParserFactory;
+
 class StringTest extends \PHPUnit\Framework\TestCase
 {
+    public function testRawValue()
+    {
+        $parser = (new ParserFactory())->create(ParserFactory::PREFER_PHP7);
+        $nodes = $parser->parse('<?php echo "sequence \x41";');
+
+        $echo = $nodes[0];
+        $this->assertInstanceOf(Echo_::class, $echo);
+
+        /** @var Echo_ $echo */
+        $string = $echo->exprs[0];
+        $this->assertInstanceOf(String_::class, $string);
+
+        /** @var String_ $string */
+        $this->assertSame('sequence A', $string->value);
+        $this->assertSame('"sequence \\x41"', $string->getAttribute('rawValue'));
+    }
+
     /**
      * @dataProvider provideTestParseEscapeSequences
      */

diff --git a/test/PhpParser/NodeAbstractTest.php b/test/PhpParser/NodeAbstractTest.php
@@ -297,7 +297,8 @@ function functionName(&$a = 0, $b = 1.0) {
                         "attributes": {
                             "startLine": 5,
                             "endLine": 5,
-                            "kind": 1
+                            "kind": 1,
+                            "rawValue": "'Foo'"
                         }
                     }
                 ],
@@ -452,7 +453,8 @@ function functionName(&$a = 0, $b = 1.0) {
                         "attributes": {
                             "startLine": 5,
                             "endLine": 5,
-                            "kind": 1
+                            "kind": 1,
+                            "rawValue": "'Foo'"
                         },
                         "value": "Foo"
                     }