Fix tab handling

This commit mirrors several changes from jgm/commonmark.js: - Fix tabs in list indentation (commonmark/commonmark.js#86) - Fix handling of partially-consumed tabs - Proper tab handling with blockquotes, fenced code, lists
thephpleague · Mar 27, 2016 · 0c816dd · 0c816dd
1 parent b3198bb
commit 0c816dd
Show file tree

Hide file tree

Showing 5 changed files with 58 additions and 33 deletions.
diff --git a/src/Block/Element/BlockQuote.php b/src/Block/Element/BlockQuote.php
@@ -55,9 +55,7 @@ public function matchesNextLine(Cursor $cursor)
         if (!$cursor->isIndented() && $cursor->getFirstNonSpaceCharacter() === '>') {
             $cursor->advanceToFirstNonSpace();
             $cursor->advance();
-            if ($cursor->getCharacter() === ' ') {
-                $cursor->advance();
-            }
+            $cursor->advanceBySpaceOrTab();
 
             return true;
         }

diff --git a/src/Block/Parser/BlockQuoteParser.php b/src/Block/Parser/BlockQuoteParser.php
@@ -38,9 +38,7 @@ public function parse(ContextInterface $context, Cursor $cursor)
 
         $cursor->advanceToFirstNonSpace();
         $cursor->advance();
-        if ($cursor->getCharacter() === ' ') {
-            $cursor->advance();
-        }
+        $cursor->advanceBySpaceOrTab();
 
         $context->addBlock(new BlockQuote());
 

diff --git a/src/Block/Parser/ListParser.php b/src/Block/Parser/ListParser.php
@@ -91,19 +91,18 @@ private function calculateListMarkerPadding(Cursor $cursor, $markerLength)
         $start = $cursor->saveState();
         $spacesStartCol = $cursor->getColumn();
 
-        do {
-            $cursor->advanceBy(1, true);
-            $nextChar = $cursor->getCharacter();
-        } while ($cursor->getColumn() - $spacesStartCol < 5 && ($nextChar === ' ' || $nextChar === "\t"));
+        while ($cursor->getColumn() - $spacesStartCol < 5) {
+            if (!$cursor->advanceBySpaceOrTab()) {
+                break;
+            }
+        }
 
         $blankItem = $cursor->peek() === null;
         $spacesAfterMarker = $cursor->getColumn() - $spacesStartCol;
 
         if ($spacesAfterMarker >= 5 || $spacesAfterMarker < 1 || $blankItem) {
             $cursor->restoreState($start);
-            if ($cursor->peek() === ' ') {
-                $cursor->advanceBy(1, true);
-            }
+            $cursor->advanceBySpaceOrTab();
 
             return $markerLength + 1;
         }

diff --git a/src/Cursor.php b/src/Cursor.php
@@ -53,6 +53,11 @@ class Cursor
      */
     private $firstNonSpaceCache;
 
+    /**
+     * @var bool
+     */
+    private $partiallyConsumedTab = false;
+
     /**
      * @param string $line
      */
@@ -186,30 +191,42 @@ public function advanceBy($characters, $advanceByColumns = false)
             return;
         }
 
+        $this->previousPosition = $this->currentPosition;
         $this->firstNonSpaceCache = null;
 
-        $i = 0;
-        $cols = 0;
-        while ($advanceByColumns ? ($cols < $characters) : ($i < $characters)) {
-            if ($this->peek($i) === "\t") {
-                $cols += (4 - (($this->column + $cols) % 4));
+        while ($characters > 0 && ($c = $this->getCharacter()) !== null) {
+            if ($c === "\t") {
+                $charsToTab = 4 - ($this->column % 4);
+                $this->partiallyConsumedTab = $advanceByColumns && $charsToTab > $characters;
+                $charsToAdvance = $charsToTab > $characters ? $characters : $charsToTab;
+                $this->column += $charsToAdvance;
+                $this->currentPosition += $this->partiallyConsumedTab ? 0 : 1;
+                $characters -= ($advanceByColumns ? $charsToAdvance : 1);
             } else {
-                $cols++;
+                $this->partiallyConsumedTab = false;
+                $this->currentPosition++;
+                $this->column++;
+                $characters--;
             }
-
-            $i++;
         }
+    }
 
-        $this->previousPosition = $this->currentPosition;
-        $newPosition = $this->currentPosition + $i;
+    /**
+     * Advances the cursor by a single space or tab, if present
+     *
+     * @return bool
+     */
+    public function advanceBySpaceOrTab()
+    {
+        $character = $this->getCharacter();
 
-        $this->column += $cols;
+        if ($character === ' ' || $character === "\t") {
+            $this->advanceBy(1, true);
 
-        if ($newPosition >= $this->length) {
-            $this->currentPosition = $this->length;
-        } else {
-            $this->currentPosition = $newPosition;
+            return true;
         }
+
+        return false;
     }
 
     /**
@@ -274,9 +291,17 @@ public function getRemainder()
     {
         if ($this->isAtEnd()) {
             return '';
-        } else {
-            return mb_substr($this->line, $this->currentPosition, $this->length, 'utf-8');
         }
+
+        $prefix = '';
+        $position = $this->currentPosition;
+        if ($this->partiallyConsumedTab) {
+            $position++;
+            $charsToTab = 4 - ($this->column % 4);
+            $prefix = str_repeat(' ', $charsToTab);
+        }
+
+        return $prefix . mb_substr($this->line, $position, $this->length, 'utf-8');
     }
 
     /**

diff --git a/tests/unit/CursorTest.php b/tests/unit/CursorTest.php
@@ -272,11 +272,16 @@ public function testAdvanceByColumnOffset()
     {
         $cursor = new Cursor("1. \t\tthere");
         $cursor->advanceBy(3);
+
+        $this->assertEquals(5, $cursor->getIndent());
+        $this->assertEquals(3, $cursor->getPosition());
+        $this->assertEquals(3, $cursor->getColumn());
+
         $cursor->advanceBy(4, true);
 
-        $this->assertEquals(0, $cursor->getIndent());
-        $this->assertEquals(5, $cursor->getPosition());
-        $this->assertEquals(8, $cursor->getColumn());
+        $this->assertEquals(1, $cursor->getIndent());
+        $this->assertEquals(4, $cursor->getPosition());
+        $this->assertEquals(7, $cursor->getColumn());
     }
 
     /**