Handle Windows CR/LF style line-ending in char_scanner

2 files changed, 18 insertions, 1 deletions

diff --git a/butl/char-scanner b/butl/char-scanner
index e070e45..7dc56fa 100644
--- a/butl/char-scanner
+++ b/butl/char-scanner
@@ -19,7 +19,12 @@ namespace butl
   class LIBBUTL_EXPORT char_scanner
   {
   public:
-    char_scanner (std::istream& is): is_ (is) {}
+    // 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.
+    //
+    char_scanner (std::istream& is, bool crlf = true)
+        : is_ (is), crlf_ (crlf) {}
 
     char_scanner (const char_scanner&) = delete;
     char_scanner& operator= (const char_scanner&) = delete;
@@ -74,6 +79,7 @@ namespace butl
 
   protected:
     std::istream& is_;
+    bool crlf_;
 
     bool unget_ {false};
     xchar buf_ = '\0';
diff --git a/butl/char-scanner.cxx b/butl/char-scanner.cxx
index a54c20d..b923aa6 100644
--- a/butl/char-scanner.cxx
+++ b/butl/char-scanner.cxx
@@ -53,6 +53,17 @@ namespace butl
       {
         is_.get ();
 
+        if (crlf_ && c == 0x0D)
+        {
+          xchar c1 (peek ());
+
+          if (c1 == '\n')
+          {
+            is_.get ();
+            c = c1;
+          }
+        }
+
         if (c == '\n')
         {
           line++;