Merge pull request #188 from qicosmos/improve_xml

.github/workflows/clang-format.yml

@@ -2,9 +2,9 @@ name: Clang Format Diff
 
 on:
   push:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
   pull_request:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
 
 jobs:
   build:

.github/workflows/linux-clang.yml

@@ -2,9 +2,9 @@ name: Ubuntu (clang)
 
 on:
   push:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
   pull_request:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
 
 jobs:
   build:

.github/workflows/linux-gcc.yml

@@ -2,9 +2,9 @@ name: Ubuntu (gcc)
 
 on:
   push:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
   pull_request:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
 
 jobs:
   build:

.github/workflows/mac.yml

@@ -2,9 +2,9 @@ name:  macOS Monterey 12
 
 on:
   push:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
   pull_request:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
 
 jobs:
   build:

.github/workflows/windows.yml

@@ -2,9 +2,9 @@ name: Windows Server 2022
 
 on: 
   push:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
   pull_request:
-    branches: [ master, fullsupport_xml]
+    branches: [ master, improve_xml]
 
 jobs:
   build:

CMakeLists.txt

@@ -53,6 +53,7 @@ set(TEST_YAML test/test_yaml.cpp test/test_yaml_bech.cpp)
 set(YAMLBENCH  benchmark/yaml_benchmark.cpp)
 set(TEST_NOTHROW test/test_yaml_nothrow.cpp)
 set(TEST_UTIL test/test_util.cpp)
+set(TEST_XMLNOTHROW test/test_xml_nothrow.cpp)
 
 add_executable(json_example 	${JSON_EXAMPLE})
 add_executable(xml_example 		${XML_EXAMPLE})
@@ -67,6 +68,7 @@ add_executable(test_yaml ${TEST_YAML})
 add_executable(yaml_benchmark ${YAMLBENCH})
 add_executable(test_nothrow ${TEST_NOTHROW})
 add_executable(test_util ${TEST_UTIL})
+add_executable(test_xml_nothrow ${TEST_XMLNOTHROW})
 
 # unit test
 option(BUILD_UNIT_TESTS "Build unit tests" ON)
@@ -91,3 +93,4 @@ add_test(NAME test_xml COMMAND test_xml)
 add_test(NAME test_yaml COMMAND test_yaml)
 add_test(NAME test_nothrow COMMAND test_nothrow)
 add_test(NAME test_util COMMAND test_util)
+add_test(NAME test_xml_nothrow COMMAND test_xml_nothrow)

README.md

@@ -96,14 +96,10 @@ person p = {"admin", 20};
 iguana::string_stream ss;  // here use std::string is also ok
 iguana::to_xml(ss, p);
 std::cout << ss.str() << std::endl;
-// you can also call iguana::to_xml_pretty to get pretty string
-iguana::string_stream s;
-iguana::to_xml_pretty(s, p);
-std::cout << s.str() << std::endl;
 
 // deserialization the structure from the string
 std::string xml = "<?xml version=\"1.0\" encoding=\"UTF-8\">  <name>buke</name> <age>30</name>";
-iguana::from_xml(p, xml.data());
+iguana::from_xml(p, xml);
 ```
 #### Serialization of yaml
 
@@ -212,7 +208,7 @@ std::string str = R"(
 </library>
 )";
 library_t lib;
-iguana::from_xml(lib, str.data());
+iguana::from_xml(lib, str);
 ```
 #### yaml
 

benchmark/xml_bench.hpp

@@ -96,13 +96,13 @@ REFLECTION(rss_t, channel);
 // ************ struct for bench_num.xml ****************
 
 struct goods_t {
-  int id;
-  int sales;
-  int inventory;
-  double weight;
-  double price;
-  double rating;
-  double discount;
+  std::string_view id;
+  std::string_view sales;
+  std::string_view inventory;
+  std::string_view weight;
+  std::string_view price;
+  std::string_view rating;
+  std::string_view discount;
 };
 REFLECTION(goods_t, id, sales, inventory, weight, price, rating, discount);
 struct storeowner_t {

benchmark/xml_benchmark.cpp

@@ -1,5 +1,6 @@
 #include "../rapidxml/rapidxml.hpp"
 #include "xml_bench.hpp"
+#include <cstdio>
 
 class ScopedTimer {
 public:
@@ -44,16 +45,18 @@ const int iterations = 1000;
 void bench_de_sample_filelists() {
 
   std::string xmlfilelist = xml_file_content("../data/rpm_filelists.xml");
+  rapidxml::xml_document<> doc;
   {
     ScopedTimer timer("rapidxml fastest parse rpm_filelists.xml");
     for (int i = 0; i < iterations; ++i) {
-      rapidxml::xml_document<> doc;
       doc.parse<rapidxml::parse_fastest>(xmlfilelist.data());
+      doc.clear();
     }
   }
+
+  filelists_t filelist;
   {
     ScopedTimer timer("iguana_xml deserialize rpm_filelists.xml");
-    filelists_t filelist;
     for (int i = 0; i < iterations; ++i) {
       iguana::from_xml(filelist, xmlfilelist.begin(), xmlfilelist.end());
       filelist.package.clear();
@@ -63,16 +66,18 @@ void bench_de_sample_filelists() {
 
 void bench_de_sample_rss() {
   std::string xmlrss = xml_file_content("../data/sample_rss.xml");
+  rapidxml::xml_document<> doc;
   {
     ScopedTimer timer("rapidxml fastest parse sample_rss.xml");
     for (int i = 0; i < iterations; ++i) {
-      rapidxml::xml_document<> doc;
       doc.parse<rapidxml::parse_fastest>(xmlrss.data());
+      doc.clear();
     }
   }
+
+  rss_t rss;
   {
     ScopedTimer timer("iguana_xml deserialize sample_rss.xml");
-    rss_t rss;
     for (int i = 0; i < iterations; ++i) {
       iguana::from_xml(rss, xmlrss.begin(), xmlrss.end());
       rss.channel.item.clear();
@@ -82,28 +87,46 @@ void bench_de_sample_rss() {
 
 void bench_num() {
   std::string xmlnum = xml_file_content("../data/bench_num.xml");
+
+  store_t s;
+  iguana::from_xml(s, xmlnum);
+  auto good = s.goods[1];
+  std::cout << iguana::get_number<int>(good.sales) << ", "
+            << iguana::get_number<double>(good.weight) << "\n";
+  s.goods.clear();
+
   {
     ScopedTimer timer("iguana_xml deserialize bench_num.xml");
     for (int i = 0; i < iterations; ++i) {
-      store_t s;
       iguana::from_xml(s, xmlnum);
+      s.goods.clear();
     }
   }
-  store_t store;
-  iguana::from_xml(store, xmlnum);
-  std::string ss;
-  ss.reserve(xmlnum.size());
+
+  rapidxml::xml_document<> doc;
   {
-    ScopedTimer timer("iguana_xml serialize bench_num.xml");
+    ScopedTimer timer("rapidxml fastest parse bench_num.xml");
     for (int i = 0; i < iterations; ++i) {
-      iguana::to_xml(store, ss);
-      ss.clear();
+      doc.parse<rapidxml::parse_fastest>(xmlnum.data());
+      doc.clear();
     }
   }
+
+  // store_t store;
+  // iguana::from_xml(store, xmlnum);
+  // std::string ss;
+  // ss.reserve(xmlnum.size());
+  // {
+  //   ScopedTimer timer("iguana_xml serialize bench_num.xml");
+  //   for (int i = 0; i < iterations; ++i) {
+  //     iguana::to_xml(store, ss);
+  //     ss.clear();
+  //   }
+  // }
 }
 
 int main() {
   bench_de_sample_filelists();
   bench_de_sample_rss();
-  // bench_num();
+  bench_num();
 }

frozen/string.h

@@ -28,6 +28,7 @@
 #include "frozen/bits/hash_string.h"
 #include "frozen/bits/version.h"
 
+#include <cstring>
 #include <functional>
 
 #ifdef FROZEN_LETITGO_HAS_STRING_VIEW

iguana/json_writer.hpp

@@ -174,12 +174,12 @@ template <typename Stream, tuple_t T>
 IGUANA_INLINE void render_json_value(Stream &s, T &&t) {
   using U = typename std::decay_t<T>;
   s.push_back('[');
-  const size_t size = std::tuple_size_v<U>;
+  constexpr size_t size = std::tuple_size_v<U>;
   for_each(std::forward<T>(t),
            [&s, size](auto &v, auto i) IGUANA__INLINE_LAMBDA {
              render_json_value(s, v);
 
-             if (i != size - 1)
+             if (i != size - 1) [[likely]]
                s.push_back(',');
            });
   s.push_back(']');

iguana/xml_reader.hpp

@@ -134,7 +134,6 @@ IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
     skip_sapces_and_newline(it, end);
     auto value_begin = it;
     auto value_end = skip_pass_smaller(it, end);
-    ;
     if (value_begin == value_end) {
       match_close_tag(it, end, name);
       return;
@@ -160,6 +159,30 @@ IGUANA_INLINE void parse_item(U &value, It &&it, It &&end,
   parse_item(value.value(), it, end, name);
 }
 
+//  /> or skip <?>、 <!> and <tag></tag> until the </name>
+//  loose inspection here
+template <typename It>
+IGUANA_INLINE void skip_object_value(It &&it, It &&end, std::string_view name) {
+  skip_till_greater(it, end);
+  ++it;
+  if (*(it - 2) == '/') {
+    // .../>
+    return;
+  }
+  // </name>
+  size_t size = name.size();
+  while (it != end) {
+    skip_till_greater(it, end);
+    if (*(it - size - 2) == '<' && *(it - size - 1) == '/' &&
+        (std::string_view{&*(it - size), size} == name)) {
+      ++it;
+      return;
+    }
+    ++it;
+  }
+  throw std::runtime_error("unclosed tag: " + std::string(name));
+}
+
 template <refletable T, typename It>
 IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
                               std::string_view name) {
@@ -242,12 +265,16 @@ IGUANA_INLINE void parse_item(T &value, It &&it, It &&end,
                 "type must be memberptr");
             using V = std::remove_reference_t<decltype(value.*member_ptr)>;
             if constexpr (!cdata_t<V>) {
-              detail::parse_item(value.*member_ptr, it, end, key);
+              parse_item(value.*member_ptr, it, end, key);
             }
           },
           member_it->second);
     } else [[unlikely]] {
+#ifdef THROW_UNKNOWN_KEY
       throw std::runtime_error("Unknown key: " + std::string(key));
+#else
+      skip_object_value(it, end, key);
+#endif
     }
     skip_sapces_and_newline(it, end);
   }
@@ -299,4 +326,10 @@ IGUANA_INLINE void from_xml(U &value, const View &view) {
   from_xml(value, std::begin(view), std::end(view));
 }
 
+template <num_t Num> IGUANA_INLINE Num get_number(std::string_view str) {
+  Num num;
+  detail::parse_value(num, str.begin(), str.end());
+  return num;
+}
+
 } // namespace iguana