From 112a83c346a537f1a5eac6fc17ee2ce3143d625b Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Thu, 18 Jun 2020 16:40:00 +0300
Subject: Fix lexer to fail on invalid UTF-8 sequences

---
 libbuild2/lexer.hxx | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

(limited to 'libbuild2/lexer.hxx')

diff --git a/libbuild2/lexer.hxx b/libbuild2/lexer.hxx
index 6c2b90b..cc42219 100644
--- a/libbuild2/lexer.hxx
+++ b/libbuild2/lexer.hxx
@@ -187,6 +187,23 @@ namespace build2
     pair<pair<char, char>, bool>
     peek_chars ();
 
+    // As base::get() but in case of an invalid character issue diagnostics
+    // and throw failed.
+    //
+    xchar
+    get ();
+
+    // Get previously peeked character (faster).
+    //
+    void
+    get (const xchar&);
+
+    // As base::peek() but in case of an invalid character issue diagnostics
+    // and throw failed.
+    //
+    xchar
+    peek ();
+
   protected:
     struct state
     {
@@ -243,6 +260,9 @@ namespace build2
   protected:
     fail_mark fail;
 
+    [[noreturn]] void
+    fail_char (const xchar&);
+
     // Lexer state.
     //
   protected:
@@ -266,6 +286,13 @@ namespace build2
     std::stack<state> state_;
 
     bool sep_; // True if we skipped spaces in peek().
+
+  private:
+    using base = char_scanner<butl::utf8_validator, 2>;
+
+    // Buffer for a get()/peek() potential error.
+    //
+    string ebuf_;
   };
 }
 
@@ -284,4 +311,6 @@ namespace butl // ADL
   }
 }
 
+#include <libbuild2/lexer.ixx>
+
 #endif // LIBBUILD2_LEXER_HXX
-- 
cgit v1.1