Support different type for iterators and sentinel in utf8 functions

3 files changed, 36 insertions, 30 deletions

diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index f5913f0f..883d7b72 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -163,7 +163,7 @@ private:
     };
     friend constexpr bool with_bit_ops(Meta::Type<Flags>) { return true; }
 
-    using Iterator = utf8::iterator<const char*, Codepoint, int, InvalidPolicy>;
+    using Iterator = utf8::iterator<const char*, const char*, Codepoint, int, InvalidPolicy>;
     using NodeIndex = ParsedRegex::NodeIndex;
 
     NodeIndex disjunction(unsigned capture = -1)
diff --git a/src/utf8.hh b/src/utf8.hh
index 4d4de274..9089d655 100644
--- a/src/utf8.hh
+++ b/src/utf8.hh
@@ -43,8 +43,8 @@ struct Pass
 // returns the codepoint of the character whose first byte
 // is pointed by it
 template<typename InvalidPolicy = utf8::InvalidPolicy::Pass,
-         typename Iterator>
-Codepoint read_codepoint(Iterator& it, const Iterator& end)
+         typename Iterator, typename Sentinel>
+Codepoint read_codepoint(Iterator& it, const Sentinel& end)
     noexcept(noexcept(InvalidPolicy{}(0)))
 {
     if (it == end)
@@ -83,8 +83,8 @@ Codepoint read_codepoint(Iterator& it, const Iterator& end)
 }
 
 template<typename InvalidPolicy = utf8::InvalidPolicy::Pass,
-         typename Iterator>
-Codepoint codepoint(Iterator it, const Iterator& end)
+         typename Iterator, typename Sentinel>
+Codepoint codepoint(Iterator it, const Sentinel& end)
     noexcept(noexcept(read_codepoint<InvalidPolicy>(it, end)))
 {
     return read_codepoint<InvalidPolicy>(it, end);
@@ -125,8 +125,8 @@ inline ByteCount codepoint_size(Codepoint cp)
         throw invalid_codepoint{};
 }
 
-template<typename Iterator>
-void to_next(Iterator& it, const Iterator& end) noexcept
+template<typename Iterator, typename Sentinel>
+void to_next(Iterator& it, const Sentinel& end) noexcept
 {
     if (it != end)
         ++it;
@@ -135,8 +135,8 @@ void to_next(Iterator& it, const Iterator& end) noexcept
 }
 
 // returns an iterator to next character first byte
-template<typename Iterator>
-Iterator next(Iterator it, const Iterator& end) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator next(Iterator it, const Sentinel& end) noexcept
 {
     to_next(it, end);
     return it;
@@ -144,16 +144,16 @@ Iterator next(Iterator it, const Iterator& end) noexcept
 
 // returns it's parameter if it points to a character first byte,
 // or else returns next character first byte
-template<typename Iterator>
-Iterator finish(Iterator it, const Iterator& end) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator finish(Iterator it, const Sentinel& end) noexcept
 {
     while (it != end and (*(it) & 0xC0) == 0x80)
         ++it;
     return it;
 }
 
-template<typename Iterator>
-void to_previous(Iterator& it, const Iterator& begin) noexcept
+template<typename Iterator, typename Sentinel>
+void to_previous(Iterator& it, const Sentinel& begin) noexcept
 {
     if (it != begin)
         --it;
@@ -161,8 +161,8 @@ void to_previous(Iterator& it, const Iterator& begin) noexcept
         --it;
 }
 // returns an iterator to the previous character first byte
-template<typename Iterator>
-Iterator previous(Iterator it, const Iterator& begin) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator previous(Iterator it, const Sentinel& begin) noexcept
 {
     to_previous(it, begin);
     return it;
@@ -171,8 +171,8 @@ Iterator previous(Iterator it, const Iterator& begin) noexcept
 // returns an iterator pointing to the first byte of the
 // dth character after (or before if d < 0) the character
 // pointed by it
-template<typename Iterator>
-Iterator advance(Iterator it, const Iterator& end, CharCount d) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator advance(Iterator it, const Sentinel& end, CharCount d) noexcept
 {
     if (it == end)
         return it;
@@ -193,8 +193,8 @@ Iterator advance(Iterator it, const Iterator& end, CharCount d) noexcept
 // returns an iterator pointing to the first byte of the
 // character at the dth column after (or before if d < 0)
 // the character pointed by it
-template<typename Iterator>
-Iterator advance(Iterator it, const Iterator& end, ColumnCount d) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator advance(Iterator it, const Sentinel& end, ColumnCount d) noexcept
 {
     if (it == end)
         return it;
@@ -222,8 +222,8 @@ Iterator advance(Iterator it, const Iterator& end, ColumnCount d) noexcept
 }
 
 // returns the character count between begin and end
-template<typename Iterator>
-CharCount distance(Iterator begin, const Iterator& end) noexcept
+template<typename Iterator, typename Sentinel>
+CharCount distance(Iterator begin, const Sentinel& end) noexcept
 {
     CharCount dist = 0;
 
@@ -236,8 +236,8 @@ CharCount distance(Iterator begin, const Iterator& end) noexcept
 }
 
 // returns the column count between begin and end
-template<typename Iterator>
-ColumnCount column_distance(Iterator begin, const Iterator& end) noexcept
+template<typename Iterator, typename Sentinel>
+ColumnCount column_distance(Iterator begin, const Sentinel& end) noexcept
 {
     ColumnCount dist = 0;
 
@@ -247,8 +247,8 @@ ColumnCount column_distance(Iterator begin, const Iterator& end) noexcept
 }
 
 // returns an iterator to the first byte of the character it is into
-template<typename Iterator>
-Iterator character_start(Iterator it, const Iterator& begin) noexcept
+template<typename Iterator, typename Sentinel>
+Iterator character_start(Iterator it, const Sentinel& begin) noexcept
 {
     while (it != begin and not is_character_start(*it))
         --it;
diff --git a/src/utf8_iterator.hh b/src/utf8_iterator.hh
index 20386069..c145f900 100644
--- a/src/utf8_iterator.hh
+++ b/src/utf8_iterator.hh
@@ -14,6 +14,7 @@ namespace utf8
 // adapter for an iterator on bytes which permits to iterate
 // on unicode codepoints instead.
 template<typename BaseIt,
+         typename Sentinel = BaseIt,
          typename CodepointType = Codepoint,
          typename DifferenceType = CharCount,
          typename InvalidPolicy = utf8::InvalidPolicy::Pass>
@@ -25,7 +26,7 @@ public:
     iterator() = default;
     constexpr static bool noexcept_policy = noexcept(InvalidPolicy{}(0));
 
-    iterator(BaseIt it, BaseIt begin, BaseIt end) noexcept
+    iterator(BaseIt it, Sentinel begin, Sentinel end) noexcept
         : m_it{std::move(it)}, m_begin{std::move(begin)}, m_end{std::move(end)}
     {}
 
@@ -105,8 +106,13 @@ public:
     bool operator> (const iterator& other) const noexcept { return m_it > other.m_it; }
     bool operator>= (const iterator& other) const noexcept { return m_it >= other.m_it; }
 
-    bool operator==(const BaseIt& other) const noexcept { return m_it == other; }
-    bool operator!=(const BaseIt& other) const noexcept { return m_it != other; }
+    template<typename T>
+    std::enable_if_t<std::is_same<T, BaseIt>::value or std::is_same<T, Sentinel>::value, bool>
+    operator==(const T& other) const noexcept { return m_it == other; }
+
+    template<typename T>
+    std::enable_if_t<std::is_same<T, BaseIt>::value or std::is_same<T, Sentinel>::value, bool>
+    operator!=(const T& other) const noexcept { return m_it != other; }
 
     bool operator< (const BaseIt& other) const noexcept { return m_it < other; }
     bool operator<= (const BaseIt& other) const noexcept { return m_it <= other; }
@@ -136,8 +142,8 @@ private:
     }
 
     BaseIt m_it;
-    BaseIt m_begin;
-    BaseIt m_end;
+    Sentinel m_begin;
+    Sentinel m_end;
     mutable CodepointType m_value = -1;
 };