From 1b57e247b8d1a7a41a8ee45d6d524c71edd63a81 Mon Sep 17 00:00:00 2001
From: Boris Kolpackov <boris@codesynthesis.com>
Date: Fri, 26 May 2017 17:20:30 +0200
Subject: Optimize char_scanner if used with ifdstream

---
 libbutl/char-scanner.cxx | 20 ++++++++++++++------
 libbutl/char-scanner.hxx | 31 +++++++++++++++++++++++++------
 libbutl/char-scanner.ixx | 31 +++++++++++++++++++++++++++++++
 libbutl/fdstream.hxx     | 12 ++++++++++--
 4 files changed, 80 insertions(+), 14 deletions(-)

diff --git a/libbutl/char-scanner.cxx b/libbutl/char-scanner.cxx
index 42a72cc..a2baaf3 100644
--- a/libbutl/char-scanner.cxx
+++ b/libbutl/char-scanner.cxx
@@ -4,12 +4,20 @@
 
 #include <libbutl/char-scanner.hxx>
 
-#include <istream>
-
 using namespace std;
 
 namespace butl
 {
+  char_scanner::
+  char_scanner (istream& is, bool crlf)
+      : is_ (is),
+        buf_ (dynamic_cast<fdbuf*> (is.rdbuf ())),
+        gptr_ (nullptr),
+        egptr_ (nullptr),
+        crlf_ (crlf)
+  {
+  }
+
   auto char_scanner::
   peek () -> xchar
   {
@@ -22,14 +30,14 @@ namespace butl
     if (eos_)
       return xchar (xchar::traits_type::eof (), line, column);
 
-    xchar::int_type v (is_.peek ());
+    int_type v (peek_ ());
 
     if (v == xchar::traits_type::eof ())
       eos_ = true;
     else if (crlf_ && v == 0x0D)
     {
-      is_.get ();
-      xchar::int_type v1 (is_.peek ());
+      get_ ();
+      int_type v1 (peek_ ());
 
       if (v1 != '\n')
       {
@@ -62,7 +70,7 @@ namespace butl
       //
       if (!eos (c))
       {
-        is_.get ();
+        get_ ();
 
         if (c == '\n')
         {
diff --git a/libbutl/char-scanner.hxx b/libbutl/char-scanner.hxx
index e71f286..80a1f2a 100644
--- a/libbutl/char-scanner.hxx
+++ b/libbutl/char-scanner.hxx
@@ -6,8 +6,10 @@
 #define LIBBUTL_CHAR_SCANNER_HXX
 
 #include <string>  // char_traits
-#include <iosfwd>
 #include <cstdint> // uint64_t
+#include <istream>
+
+#include <libbutl/fdstream.hxx>
 
 #include <libbutl/export.hxx>
 
@@ -23,8 +25,11 @@ namespace butl
     // 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) {}
+    // Note also that if the stream happens to be ifdstream, then it includes
+    // a number of optimizations that assume nobody else is messing with the
+    // stream.
+    //
+    char_scanner (std::istream& is, bool crlf = true);
 
     char_scanner (const char_scanner&) = delete;
     char_scanner& operator= (const char_scanner&) = delete;
@@ -43,9 +48,9 @@ namespace butl
     class xchar
     {
     public:
-      typedef std::char_traits<char> traits_type;
-      typedef traits_type::int_type int_type;
-      typedef traits_type::char_type char_type;
+      using traits_type = std::char_traits<char>;
+      using int_type = traits_type::int_type;
+      using char_type = traits_type::char_type;
 
       int_type value;
       std::uint64_t line;
@@ -90,8 +95,22 @@ namespace butl
     std::uint64_t column = 1;
 
   protected:
+    using int_type = xchar::int_type;
+    using char_type = xchar::char_type;
+
+    int_type
+    peek_ ();
+
+    void
+    get_ ();
+
+  protected:
     std::istream& is_;
 
+    fdbuf* buf_; // NULL if not ifdstream.
+    const char_type* gptr_;
+    const char_type* egptr_;
+
     bool crlf_;
     bool eos_ = false;
 
diff --git a/libbutl/char-scanner.ixx b/libbutl/char-scanner.ixx
index 2d96207..0e00dfd 100644
--- a/libbutl/char-scanner.ixx
+++ b/libbutl/char-scanner.ixx
@@ -29,4 +29,35 @@ namespace butl
     unget_ = true;
     ungetc_ = c;
   }
+
+  inline auto char_scanner::
+  peek_ () -> int_type
+  {
+    if (gptr_ != egptr_)
+      return *gptr_;
+
+    int_type r (is_.peek ());
+
+    // Update buffer pointers for the next chunk.
+    //
+    if (buf_ != nullptr)
+    {
+      gptr_ = buf_->gptr ();
+      egptr_ = buf_->egptr ();
+    }
+
+    return r;
+  }
+
+  inline void char_scanner::
+  get_ ()
+  {
+    if (gptr_ != egptr_)
+    {
+      buf_->gbump (1);
+      ++gptr_;
+    }
+    else
+      is_.get (); // About as fast as ignore() and way faster than tellg().
+  }
 }
diff --git a/libbutl/fdstream.hxx b/libbutl/fdstream.hxx
index e1f5790..dd33263 100644
--- a/libbutl/fdstream.hxx
+++ b/libbutl/fdstream.hxx
@@ -122,8 +122,10 @@ namespace butl
     fd () const {return fd_.get ();}
 
   public:
-    using int_type = std::basic_streambuf<char>::int_type;
-    using traits_type = std::basic_streambuf<char>::traits_type;
+    using base = std::basic_streambuf<char>;
+
+    using int_type = base::int_type;
+    using traits_type = base::traits_type;
 
     // basic_streambuf input interface.
     //
@@ -134,6 +136,12 @@ namespace butl
     virtual int_type
     underflow ();
 
+    // Direct access to the get area. Use with caution.
+    //
+    using base::gptr;
+    using base::egptr;
+    using base::gbump;
+
   private:
     bool
     load ();
-- 
cgit v1.1