Add ability to save data in char_scanner

3 files changed, 61 insertions, 8 deletions

diff --git a/libbutl/char-scanner.cxx b/libbutl/char-scanner.cxx
index 2e680b6..607d887 100644
--- a/libbutl/char-scanner.cxx
+++ b/libbutl/char-scanner.cxx
@@ -35,8 +35,10 @@ using namespace std;
 namespace butl
 {
   char_scanner::
-  char_scanner (istream& is, bool crlf)
-      : is_ (is),
+  char_scanner (istream& is, bool crlf, uint64_t l)
+      : line (l),
+        column (1),
+        is_ (is),
         buf_ (dynamic_cast<fdbuf*> (is.rdbuf ())),
         gptr_ (nullptr),
         egptr_ (nullptr),
diff --git a/libbutl/char-scanner.ixx b/libbutl/char-scanner.ixx
index 0e00dfd..c32b034 100644
--- a/libbutl/char-scanner.ixx
+++ b/libbutl/char-scanner.ixx
@@ -52,12 +52,31 @@ namespace butl
   inline void char_scanner::
   get_ ()
   {
+    int_type c;
+
     if (gptr_ != egptr_)
     {
       buf_->gbump (1);
-      ++gptr_;
+      c = *gptr_++;
     }
     else
-      is_.get (); // About as fast as ignore() and way faster than tellg().
+      c = is_.get (); // About as fast as ignore() and way faster than tellg().
+
+    if (save_ != nullptr && c != xchar::traits_type::eof ())
+      save_->push_back (static_cast<char_type> (c));
+  }
+
+  inline void char_scanner::
+  save_start (std::string& b)
+  {
+    assert (save_ == nullptr);
+    save_ = &b;
+  }
+
+  inline void char_scanner::
+  save_stop ()
+  {
+    assert (save_ != nullptr);
+    save_ = nullptr;
   }
 }
diff --git a/libbutl/char-scanner.mxx b/libbutl/char-scanner.mxx
index af4dad9..9bfdeec 100644
--- a/libbutl/char-scanner.mxx
+++ b/libbutl/char-scanner.mxx
@@ -6,7 +6,7 @@
 #pragma once
 #endif
 
-// C includes.
+#include <cassert>
 
 #ifndef __cpp_lib_modules
 #include <string>  // char_traits
@@ -45,7 +45,10 @@ LIBBUTL_MODEXPORT namespace butl
     // a number of optimizations that assume nobody else is messing with the
     // stream.
     //
-    char_scanner (std::istream& is, bool crlf = true);
+    // The line argument can be used to override the start line in the stream
+    // (useful when re-scanning data saved with the save_* facility).
+    //
+    char_scanner (std::istream& is, bool crlf = true, std::uint64_t line = 1);
 
     char_scanner (const char_scanner&) = delete;
     char_scanner& operator= (const char_scanner&) = delete;
@@ -107,8 +110,32 @@ LIBBUTL_MODEXPORT namespace butl
     // Line and column of the next character to be extracted from the stream
     // by peek() or get().
     //
-    std::uint64_t line = 1;
-    std::uint64_t column = 1;
+    std::uint64_t line;
+    std::uint64_t column;
+
+    // Ability to save raw data as it is being scanned. Note that the
+    // character is only saved when it is got, not peeked.
+    //
+  public:
+    void
+    save_start (std::string&);
+
+    void
+    save_stop ();
+
+    struct save_guard
+    {
+      explicit
+      save_guard (char_scanner& s, std::string& b): s_ (&s) {s.save_start (b);}
+
+      void
+      stop () {if (s_ != nullptr) {s_->save_stop (); s_ = nullptr;}}
+
+      ~save_guard () {stop ();}
+
+    private:
+      char_scanner* s_;
+    };
 
   protected:
     using int_type = xchar::int_type;
@@ -123,10 +150,15 @@ LIBBUTL_MODEXPORT namespace butl
   protected:
     std::istream& is_;
 
+    // Note that if you are reading from the buffer directly, then it is
+    // also your responsibility to save the data.
+    //
     fdbuf* buf_; // NULL if not ifdstream.
     const char_type* gptr_;
     const char_type* egptr_;
 
+    std::string* save_ = nullptr;
+
     bool crlf_;
     bool eos_ = false;