From 1c25c679b2aba0d94abf1791d32b3450de8737c3 Mon Sep 17 00:00:00 2001
From: Nathan Bishop <nbish11@hotmail.com>
Date: Wed, 29 May 2019 13:38:00 +1000
Subject: [PATCH] Add locational information about a lexeme's position with a
 formatted json stream/file

---
 src/Lexer.php | 31 +++++++++++++++++++++++++++++--
 1 file changed, 29 insertions(+), 2 deletions(-)

diff --git a/src/Lexer.php b/src/Lexer.php
index dd3f981..c2df3bf 100644
--- a/src/Lexer.php
+++ b/src/Lexer.php
@@ -8,6 +8,8 @@ class Lexer implements \IteratorAggregate
     private $bytesIterator;
 
     private $position = 0;
+    private $line = 1;
+    private $column = 0;
 
     /**
      * @param \Traversable $bytesIterator
@@ -16,7 +18,7 @@ class Lexer implements \IteratorAggregate
     {
         $this->bytesIterator = $bytesIterator;
     }
-    
+
     /**
      * @return \Generator
      */
@@ -25,6 +27,7 @@ class Lexer implements \IteratorAggregate
         $inString = false;
         $tokenBuffer = '';
         $isEscaping = false;
+        $width = 0;
 
         ${' '} = 0;
         ${"\n"} = 0;
@@ -42,29 +45,37 @@ class Lexer implements \IteratorAggregate
             for ($i = 0; $i < $bytesLength; ++$i) {
                 $byte = $bytes[$i];
                 ++$this->position;
-
                 if ($inString) {
                     if ($byte === '"' && !$isEscaping) {
                         $inString = false;
                     }
                     $isEscaping = ($byte =='\\' && !$isEscaping);
                     $tokenBuffer .= $byte;
+                    $width++;
                     continue;
                 }
 
                 if (isset($$byte)) {
+                    $this->column++;
                     if ($tokenBuffer !== '') {
                         yield $tokenBuffer;
+                        $this->column += $width;
                         $tokenBuffer = '';
+                        $width = 0;
                     }
                     if ($$byte) { // is not whitespace
                         yield $byte;
+                    // track line number and reset column for each newline
+                    } elseif ($byte === "\r" || $byte === "\n") {
+                        $this->line++;
+                        $this->column = 0;
                     }
                 } else {
                     if ($byte === '"') {
                         $inString = true;
                     }
                     $tokenBuffer .= $byte;
+                    $width++;
                 }
             }
         }
@@ -80,4 +91,20 @@ class Lexer implements \IteratorAggregate
     {
         return $this->position;
     }
+
+    /**
+     * @return integer The line number of the lexeme currently being processed (index starts at one).
+     */
+    public function getLine()
+    {
+        return $this->line;
+    }
+
+    /**
+     * @return integer The, currently being processed, lexeme's position within the line (index starts at one).
+     */
+    public function getColumn()
+    {
+        return $this->column;
+    }
 }