Fix issue where nowdocs with whitespace before labels are not detected

src/Filter/EvaluateStrings.php

@@ -6,7 +6,6 @@
 use Lkrms\PrettyPHP\Contract\Filter;
 use Lkrms\PrettyPHP\TokenUtil;
 use Salient\Utility\Exception\ShouldNotHappenException;
-use Salient\Utility\Regex;
 
 /**
  * Evaluate strings for comparison
@@ -31,8 +30,16 @@ public function filterTokens(array $tokens): array
                 if (!$lastString) {
                     $stack[] = $token;
                     $lastString = $token;
+                    // `b"` -> `"`
+                    if ($token->id === \T_DOUBLE_QUOTE) {
+                        $token->text = '"';
+                    }
                     continue;
                 }
+                // `b<<< "EOF"` -> `<<<EOF`
+                if ($lastString->id === \T_START_HEREDOC) {
+                    $lastString->text = '<<<' . trim(ltrim($lastString->text, 'bB'), "< \t\"\n\r") . "\n";
+                }
                 array_pop($stack);
                 $lastString = null;
                 continue;
@@ -67,17 +74,12 @@ public function filterTokens(array $tokens): array
                 $text = TokenUtil::unescapeBackticks($token->text);
                 eval("\$string = \"{$text}\";");
             } elseif ($lastString->id === \T_START_HEREDOC) {
-                $start = trim($lastString->text);
-                // Remove prefix if present, e.g. `b<<<EOF`
-                if ($start[0] !== '<') {
-                    /** @var string */
-                    $start = substr($start, 1);
-                }
+                $start = rtrim($lastString->text);
                 // Ignore nowdocs
-                if (substr($start, 0, 4) === "<<<'") {
+                if ($start[-1] === "'") {
                     continue;
                 }
-                $end = Regex::replace('/[^a-zA-Z0-9_]+/', '', $start);
+                $end = trim(ltrim($start, 'bB'), "< \t\"'");
                 eval("\$string = {$start}\n{$token->text}\n{$end};");
             } else {
                 // @codeCoverageIgnoreStart

src/Rule/NormaliseStrings.php

@@ -10,7 +10,6 @@
 use Lkrms\PrettyPHP\TokenUtil;
 use Salient\Utility\Exception\ShouldNotHappenException;
 use Salient\Utility\Regex;
-use Salient\Utility\Str;
 
 /**
  * Normalise strings
@@ -39,6 +38,8 @@ public static function getPriority(string $method): ?int
     public static function getTokens(AbstractTokenIndex $idx): array
     {
         return [
+            \T_DOUBLE_QUOTE => true,
+            \T_START_HEREDOC => true,
             \T_CONSTANT_ENCAPSED_STRING => true,
             \T_ENCAPSED_AND_WHITESPACE => true,
         ];
@@ -49,12 +50,15 @@ public static function getTokens(AbstractTokenIndex $idx): array
      */
     public static function needsSortedTokens(): bool
     {
-        return false;
+        return true;
     }
 
     /**
      * Apply the rule to the given tokens
      *
+     * Double quotes and leading whitespace are removed from heredoc and nowdoc
+     * labels. Binary prefixes are removed from all strings.
+     *
      * Strings other than nowdocs are normalised as follows:
      *
      * - Single- and double-quoted strings are replaced with the most readable
@@ -72,15 +76,30 @@ public function processTokens(array $tokens): void
     {
         $string = '';
         foreach ($tokens as $token) {
+            // `b"` -> `"`
+            if ($token->id === \T_DOUBLE_QUOTE) {
+                if ($token->text !== '"') {
+                    $token->setText('"');
+                }
+                continue;
+            }
+
+            // `b<<< "EOF"` -> `<<<EOF`
+            if ($token->id === \T_START_HEREDOC) {
+                $text = '<<<' . trim(ltrim($token->text, 'bB'), "< \t\"\n\r") . "\n";
+                if ($token->text !== $text) {
+                    $token->setText($text);
+                }
+                continue;
+            }
+
             if ($token->id === \T_ENCAPSED_AND_WHITESPACE) {
                 /** @var Token */
                 $openedBy = $token->String;
-                if ($openedBy->id === \T_START_HEREDOC && (
-                    Str::startsWith($openedBy->text, "<<<'") || (
-                        $openedBy->text[0] !== '<'
-                        && Str::startsWith(substr($openedBy->text, 1), "<<<'")
-                    )
-                )) {
+                if (
+                    $openedBy->id === \T_START_HEREDOC
+                    && rtrim($openedBy->text)[-1] === "'"
+                ) {
                     continue;
                 }
             } else {
@@ -132,10 +151,9 @@ public function processTokens(array $tokens): void
                     break;
 
                 case \T_START_HEREDOC:
-                    $closedBy = $openedBy->Data[Data::END_STRING];
-                    $start = trim($openedBy->text);
+                    $start = rtrim($openedBy->text);
+                    $end = trim($start, "<'");
                     $text = $token->text;
-                    $end = trim($closedBy->text);
                     if ($next->id === \T_END_HEREDOC) {
                         $text = substr($text, 0, -1);
                         $suffix = "\n";

src/Rule/StandardSpacing.php

@@ -354,7 +354,11 @@ static function () use ($idx, $innerIndent, $next, $last) {
                 continue;
             }
 
-            if ($token->id === \T_START_HEREDOC && $this->Formatter->Psr12) {
+            if (
+                $token->id === \T_START_HEREDOC
+                && $this->Formatter->Psr12
+                && $token !== $token->Statement
+            ) {
                 $token->Whitespace |= Space::NO_BLANK_BEFORE | Space::NO_LINE_BEFORE | Space::SPACE_BEFORE;
             }
         }

tests/fixtures/Formatter/in/issues/0195-nowdoc-with-space.php

@@ -0,0 +1,7 @@
+<?php
+$str = sprintf(
+    <<< 'EOP'
+        Lorem ipsum... %1$s
+        EOP,
+    'You know'
+);