From eb4810893eb4c8379c3455f1e8a75ccd3b911aa6 Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Sat, 22 Apr 2017 11:52:44 +0200
Subject: Fix Windows CR/LF handling

---
 butl/char-scanner     |  4 ++--
 butl/char-scanner.cxx | 24 +++++++++++++-----------
 2 files changed, 15 insertions(+), 13 deletions(-)

diff --git a/butl/char-scanner b/butl/char-scanner
index 7dc56fa..cbeca96 100644
--- a/butl/char-scanner
+++ b/butl/char-scanner
@@ -20,8 +20,8 @@ namespace butl
   {
   public:
     // If the crlf argument is true, then recognize Windows newlines (0x0D
-    // 0x0A) and convert them to just '\n' (0x0A). Note that standalone 0x0D
-    // that are not followed by 0x0A are returned as is.
+    // 0x0A) and convert them to just '\n' (0x0A). Note that a standalone
+    // 0x0D is treated "as if" it was followed by 0x0A.
     //
     char_scanner (std::istream& is, bool crlf = true)
         : is_ (is), crlf_ (crlf) {}
diff --git a/butl/char-scanner.cxx b/butl/char-scanner.cxx
index b923aa6..8be7520 100644
--- a/butl/char-scanner.cxx
+++ b/butl/char-scanner.cxx
@@ -25,6 +25,19 @@ namespace butl
 
         if (v == xchar::traits_type::eof ())
           eos_ = true;
+        else if (crlf_ && v == 0x0D)
+        {
+          is_.get ();
+          xchar::int_type v1 (is_.peek ());
+
+          if (v1 != '\n')
+          {
+            unget_ = true;
+            buf_ = '\n';
+          }
+
+          v = '\n';
+        }
 
         return xchar (v, line, column);
       }
@@ -53,17 +66,6 @@ namespace butl
       {
         is_.get ();
 
-        if (crlf_ && c == 0x0D)
-        {
-          xchar c1 (peek ());
-
-          if (c1 == '\n')
-          {
-            is_.get ();
-            c = c1;
-          }
-        }
-
         if (c == '\n')
         {
           line++;
-- 
cgit v1.1