From 155057d8ca8ae29024caf01081caf7d4f4667fd4 Mon Sep 17 00:00:00 2001
From: bbbgan <2893129936@qq.com>
Date: Sun, 9 Jul 2023 16:39:21 +0800
Subject: [PATCH] improve skip_till[xml]

---
 iguana/xml_reader.hpp |  39 +++++++-------
 iguana/xml_util.hpp   | 119 ++++++++++++++++++++++++++++++++++++++++++
 test/test_xml.cpp     |   2 +-
 3 files changed, 139 insertions(+), 21 deletions(-)

diff --git a/iguana/xml_reader.hpp b/iguana/xml_reader.hpp
index 5e37e971..f9acc0d5 100644
--- a/iguana/xml_reader.hpp
+++ b/iguana/xml_reader.hpp
@@ -74,11 +74,12 @@ IGUANA_INLINE void parse_attr(U &&value, It &&it, It &&end) {
 template <plain_t U, typename It>
 IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
                               std::string_view name) {
-  skip_till<'>'>(it, end);
+  skip_till_greater(it, end);
   ++it;
   skip_sapces_and_newline(it, end);
   auto value_begin = it;
-  auto value_end = skip_pass<'<'>(it, end);
+  auto value_end = skip_pass_smaller(it, end);
+  ;
   parse_value(value, value_begin, value_end);
   match_close_tag(it, end, name);
 }
@@ -96,14 +97,14 @@ IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
         --it;
         return;
       } else {
-        skip_till<'>'>(it, end);
+        skip_till_greater(it, end);
         ++it;
         skip_sapces_and_newline(it, end);
         continue;
       }
     }
     auto start = it;
-    skip_till<' ', '>'>(it, end);
+    skip_till_greater_or_space(it, end);
     std::string_view key = std::string_view{
         &*start, static_cast<size_t>(std::distance(start, it))};
     if (key != name) [[unlikely]] {
@@ -119,8 +120,8 @@ template <optional_t U, typename It>
 IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
                               std::string_view name) {
   using value_type = typename std::remove_reference_t<U>::value_type;
-  skip_till<'>'>(it, end);
-  if (*(it - 1) == '/') {
+  skip_till_greater(it, end);
+  if (*(it - 1) == '/') [[likely]] {
     ++it;
     return;
   }
@@ -128,11 +129,12 @@ IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
   if constexpr (plain_t<value_type>) {
     // The following code is for option not to be emplaced
     // when parse  "...> <..."
-    skip_till<'>'>(it, end);
+    skip_till_greater(it, end);
     ++it;
     skip_sapces_and_newline(it, end);
     auto value_begin = it;
-    auto value_end = skip_pass<'<'>(it, end);
+    auto value_end = skip_pass_smaller(it, end);
+    ;
     if (value_begin == value_end) {
       match_close_tag(it, end, name);
       return;
@@ -162,7 +164,7 @@ template <refletable T, typename It>
 IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
                               std::string_view name) {
   constexpr auto cdata_idx = get_type_index<is_cdata_t, std::decay_t<T>>();
-  skip_till<'>'>(it, end);
+  skip_till_greater(it, end);
   ++it;
   skip_sapces_and_newline(it, end);
   while (it != end) {
@@ -174,7 +176,7 @@ IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
       return; // reach the close tag
     } else if (*it == '?') [[unlikely]] {
       // <? ... ?>
-      skip_till<'>'>(it, end);
+      skip_till_greater(it, end);
       ++it;
       skip_sapces_and_newline(it, end);
       continue;
@@ -212,14 +214,14 @@ IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
       } else if (*it == 'D') {
         // <!D
         ++it;
-        skip_till<'>'>(it, end);
+        skip_till_greater(it, end);
         ++it;
         skip_sapces_and_newline(it, end);
         continue;
       } else {
         // <!-- -->
         ++it;
-        skip_till<'>'>(it, end);
+        skip_till_greater(it, end);
         ++it;
         skip_sapces_and_newline(it, end);
         continue;
@@ -227,7 +229,7 @@ IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
     }
 
     auto start = it;
-    skip_till<' ', '>'>(it, end);
+    skip_till_greater_or_space(it, end);
     std::string_view key = std::string_view{
         &*start, static_cast<size_t>(std::distance(start, it))};
     static constexpr auto frozen_map = get_iguana_struct_map<T>();
@@ -259,14 +261,14 @@ IGUANA_INLINE void from_xml(U &value, It &&it, It &&end) {
     skip_sapces_and_newline(it, end);
     match<'<'>(it, end);
     if (*it == '?') {
-      skip_till<'>'>(it, end);
+      skip_till_greater(it, end);
       ++it;
     } else {
       break;
     }
   }
   auto start = it;
-  skip_till<' ', '>'>(it, end);
+  skip_till_greater_or_space(it, end);
   std::string_view key =
       std::string_view{&*start, static_cast<size_t>(std::distance(start, it))};
   detail::parse_attr(value.attr(), it, end);
@@ -279,17 +281,14 @@ IGUANA_INLINE void from_xml(U &value, It &&it, It &&end) {
     skip_sapces_and_newline(it, end);
     match<'<'>(it, end);
     if (*it == '?') {
-      skip_till<'>'>(it, end);
+      skip_till_greater(it, end);
       ++it; // skip >
-      // '<?xml ' - xml declaration
-      // or
-      // Parse PI
     } else {
       break;
     }
   }
   auto start = it;
-  skip_till<' ', '>'>(it, end);
+  skip_till_greater_or_space(it, end);
   std::string_view key =
       std::string_view{&*start, static_cast<size_t>(std::distance(start, it))};
   detail::parse_item(value, it, end, key);
diff --git a/iguana/xml_util.hpp b/iguana/xml_util.hpp
index 9b98dd21..5da49dd6 100644
--- a/iguana/xml_util.hpp
+++ b/iguana/xml_util.hpp
@@ -253,4 +253,123 @@ IGUANA_INLINE void match_close_tag(auto &&it, auto &&end,
   // ++it;
 }
 
+// skip_till<'>'>(it, end);
+IGUANA_INLINE void skip_till_greater(auto &&it, auto &&end) {
+  static_assert(std::contiguous_iterator<std::decay_t<decltype(it)>>);
+
+  auto has_zero = [](uint64_t chunk) {
+    return (((chunk - 0x0101010101010101) & ~chunk) & 0x8080808080808080);
+  };
+  auto has_greater = [&](uint64_t chunk) {
+    return has_zero(
+        chunk ^
+        0b0011111000111110001111100011111000111110001111100011111000111110);
+  };
+
+  if (std::distance(it, end) >= 7) [[likely]] {
+    const auto end_m7 = end - 7;
+    for (; it < end_m7; it += 8) {
+      const auto chunk = *reinterpret_cast<const uint64_t *>(&*it);
+      uint64_t test = has_greater(chunk);
+      if (test != 0) {
+        it += (std::countr_zero(test) >> 3);
+        return;
+      }
+    }
+  }
+
+  // Tail end of buffer. Should be rare we even get here
+  while (it < end) {
+    switch (*it) {
+    case '>':
+      return;
+    }
+    ++it;
+  }
+  throw std::runtime_error("Expected >");
+}
+
+IGUANA_INLINE void skip_till_greater_or_space(auto &&it, auto &&end) {
+  static_assert(std::contiguous_iterator<std::decay_t<decltype(it)>>);
+
+  auto has_zero = [](uint64_t chunk) {
+    return (((chunk - 0x0101010101010101) & ~chunk) & 0x8080808080808080);
+  };
+  auto has_greater = [&](uint64_t chunk) {
+    return has_zero(
+        chunk ^
+        0b0011111000111110001111100011111000111110001111100011111000111110);
+  };
+  auto has_space = [&](uint64_t chunk) {
+    return has_zero(
+        chunk ^
+        0b0010000000100000001000000010000000100000001000000010000000100000);
+  };
+  if (std::distance(it, end) >= 7) [[likely]] {
+    const auto end_m7 = end - 7;
+    for (; it < end_m7; it += 8) {
+      const auto chunk = *reinterpret_cast<const uint64_t *>(&*it);
+      uint64_t test = has_greater(chunk) | has_space(chunk);
+      if (test != 0) {
+        it += (std::countr_zero(test) >> 3);
+        return;
+      }
+    }
+  }
+
+  // Tail end of buffer. Should be rare we even get here
+  while (it < end) {
+    switch (*it) {
+    case '>':
+    case ' ':
+      return;
+    }
+    ++it;
+  }
+  throw std::runtime_error("Expected > or space");
+}
+
+IGUANA_INLINE void skip_till_smaller(auto &&it, auto &&end) {
+  static_assert(std::contiguous_iterator<std::decay_t<decltype(it)>>);
+
+  auto has_zero = [](uint64_t chunk) {
+    return (((chunk - 0x0101010101010101) & ~chunk) & 0x8080808080808080);
+  };
+  auto has_smaller = [&](uint64_t chunk) {
+    return has_zero(
+        chunk ^
+        0b0011110000111100001111000011110000111100001111000011110000111100);
+  };
+
+  if (std::distance(it, end) >= 7) [[likely]] {
+    const auto end_m7 = end - 7;
+    for (; it < end_m7; it += 8) {
+      const auto chunk = *reinterpret_cast<const uint64_t *>(&*it);
+      uint64_t test = has_smaller(chunk);
+      if (test != 0) {
+        it += (std::countr_zero(test) >> 3);
+        return;
+      }
+    }
+  }
+
+  // Tail end of buffer. Should be rare we even get here
+  while (it < end) {
+    switch (*it) {
+    case '>':
+      return;
+    }
+    ++it;
+  }
+  throw std::runtime_error("Expected >");
+}
+
+IGUANA_INLINE auto skip_pass_smaller(auto &&it, auto &&end) {
+  skip_till_smaller(it, end);
+  auto res = it++ - 1;
+  while (*res == ' ') {
+    --res;
+  }
+  return res + 1;
+}
 } // namespace iguana
\ No newline at end of file
diff --git a/test/test_xml.cpp b/test/test_xml.cpp
index 0295dcbf..89cb9266 100644
--- a/test/test_xml.cpp
+++ b/test/test_xml.cpp
@@ -108,7 +108,7 @@ TEST_CASE("test vector with attr") {
     <item index="0">1</item>
     <item index="1">2</item>
     <item index="2">3</item>
-    <item index="3">4</item>
+    <item index="3">4 </item>
   </test_arr_t>
   )";
   test_arr_t arr;