From 2613a9387e8a41c7b0f8d724575b58ea5bed621c Mon Sep 17 00:00:00 2001
From: Justin King <jcking@google.com>
Date: Mon, 20 Dec 2021 06:51:13 -0800
Subject: [PATCH] [C++] Fix bugs in UnbufferedCharStream

---
 .../Cpp/runtime/src/UnbufferedCharStream.cpp  | 25 ++++++----------
 .../Cpp/runtime/src/UnbufferedCharStream.h    | 30 ++++++++-----------
 2 files changed, 21 insertions(+), 34 deletions(-)

diff --git a/runtime/Cpp/runtime/src/UnbufferedCharStream.cpp b/runtime/Cpp/runtime/src/UnbufferedCharStream.cpp
index 3993938032..fd234de099 100755
--- a/runtime/Cpp/runtime/src/UnbufferedCharStream.cpp
+++ b/runtime/Cpp/runtime/src/UnbufferedCharStream.cpp
@@ -13,9 +13,8 @@ using namespace antlrcpp;
 using namespace antlr4;
 using namespace antlr4::misc;
 
-UnbufferedCharStream::UnbufferedCharStream(std::wistream &input) : _input(input) {
-  InitializeInstanceFields();
-
+UnbufferedCharStream::UnbufferedCharStream(std::wistream &input)
+    : _p(0), _numMarkers(0), _lastChar(0), _lastCharBufferStart(0), _currentCharIndex(0), _input(input) {
   // The vector's size is what used to be n in Java code.
   fill(1); // prime
 }
@@ -74,9 +73,7 @@ size_t UnbufferedCharStream::fill(size_t n) {
 }
 
 char32_t UnbufferedCharStream::nextChar()  {
-  wchar_t result = 0;
-  _input >> result;
-  return result;
+  return _input.get();
 }
 
 void UnbufferedCharStream::add(char32_t c) {
@@ -101,7 +98,7 @@ size_t UnbufferedCharStream::LA(ssize_t i) {
     return EOF;
   }
 
-  if (_data[static_cast<size_t>(index)] == 0xFFFF) {
+  if (_data[static_cast<size_t>(index)] == std::char_traits<wchar_t>::eof()) {
     return EOF;
   }
 
@@ -178,7 +175,7 @@ std::string UnbufferedCharStream::getSourceName() const {
 }
 
 std::string UnbufferedCharStream::getText(const misc::Interval &interval) {
-  if (interval.a < 0 || interval.b >= interval.a - 1) {
+  if (interval.a < 0 || interval.b < interval.a - 1) {
     throw IllegalArgumentException("invalid interval");
   }
 
@@ -202,14 +199,10 @@ std::string UnbufferedCharStream::getText(const misc::Interval &interval) {
   return std::move(maybeUtf8).value();
 }
 
-size_t UnbufferedCharStream::getBufferStartIndex() const {
-  return _currentCharIndex - _p;
+std::string UnbufferedCharStream::toString() const {
+  throw UnsupportedOperationException("Unbuffered stream cannot be materialized to a string");
 }
 
-void UnbufferedCharStream::InitializeInstanceFields() {
-  _p = 0;
-  _numMarkers = 0;
-  _lastChar = 0;
-  _lastCharBufferStart = 0;
-  _currentCharIndex = 0;
+size_t UnbufferedCharStream::getBufferStartIndex() const {
+  return _currentCharIndex - _p;
 }
diff --git a/runtime/Cpp/runtime/src/UnbufferedCharStream.h b/runtime/Cpp/runtime/src/UnbufferedCharStream.h
index 98cdcc6142..a65c6ca054 100755
--- a/runtime/Cpp/runtime/src/UnbufferedCharStream.h
+++ b/runtime/Cpp/runtime/src/UnbufferedCharStream.h
@@ -18,10 +18,10 @@ namespace antlr4 {
     /// The name or source of this char stream.
     std::string name;
 
-    UnbufferedCharStream(std::wistream &input);
+    explicit UnbufferedCharStream(std::wistream &input);
 
-    virtual void consume() override;
-    virtual size_t LA(ssize_t i) override;
+    void consume() override;
+    size_t LA(ssize_t i) override;
 
     /// <summary>
     /// Return a marker that we can release later.
@@ -30,35 +30,32 @@ namespace antlr4 {
     /// protection against misuse where {@code seek()} is called on a mark or
     /// {@code release()} is called in the wrong order.
     /// </summary>
-    virtual ssize_t mark() override;
+    ssize_t mark() override;
 
     /// <summary>
     /// Decrement number of markers, resetting buffer if we hit 0. </summary>
     /// <param name="marker"> </param>
-    virtual void release(ssize_t marker) override;
-    virtual size_t index() override;
+    void release(ssize_t marker) override;
+    size_t index() override;
 
     /// <summary>
     /// Seek to absolute character index, which might not be in the current
     ///  sliding window.  Move {@code p} to {@code index-bufferStartIndex}.
     /// </summary>
-    virtual void seek(size_t index) override;
-    virtual size_t size() override;
-    virtual std::string getSourceName() const override;
-    virtual std::string getText(const misc::Interval &interval) override;
+    void seek(size_t index) override;
+    size_t size() override;
+    std::string getSourceName() const override;
+    std::string getText(const misc::Interval &interval) override;
+
+    std::string toString() const override;
 
   protected:
     /// A moving window buffer of the data being scanned. While there's a marker,
     /// we keep adding to buffer. Otherwise, <seealso cref="#consume consume()"/> resets so
     /// we start filling at index 0 again.
     // UTF-32 encoded.
-#if defined(_MSC_VER) && _MSC_VER == 1900
-    i32string _data; // Custom type for VS 2015.
-    typedef __int32 storage_type;
-#else
     std::u32string _data;
     typedef char32_t storage_type;
-#endif
 
     /// <summary>
     /// 0..n-1 index into <seealso cref="#data data"/> of next character.
@@ -115,9 +112,6 @@ namespace antlr4 {
     virtual char32_t nextChar();
     virtual void add(char32_t c);
     size_t getBufferStartIndex() const;
-
-  private:
-    void InitializeInstanceFields();
   };
 
 } // namespace antlr4