1 files changed, 242 insertions, 0 deletions
diff --git a/sax/tst/test_decoder.cc b/sax/tst/test_decoder.cc
new file mode 100644
index 0000000..86f230b
--- /dev/null
+++ b/sax/tst/test_decoder.cc
@@ -0,0 +1,242 @@
+#include "sax_decoder.hh"
+#include "sax_decoder_factory.hh"
+#include "sax_processor.hh"
+#include "sax_delegate.hh"
+
+#include <memory>
+#include <gtest/gtest.h>
+
+namespace {
+
+class TestDelegate : public modxml::sax::Delegate {
+ public:
+  ~TestDelegate() override = default;
+
+  void empty_element(std::string_view name,
+                     modxml::sax::Attributes const&) override {
+    EXPECT_EQ(name, "root");
+    if (name == "root") {
+      EXPECT_FALSE(have_root_);
+      have_root_ = true;
+    }
+  }
+
+  void error(std::string_view message) override {
+    have_error_ = true;
+    FAIL() << message;
+  }
+
+  bool have_root() const { return have_root_; }
+
+  bool have_error() const { return have_error_; }
+
+ private:
+  bool have_root_{false};
+  bool have_error_{false};
+};
+
+bool process_all(modxml::sax::Processor& processor,
+                 TestDelegate& delegate,
+                 std::span<uint8_t const> data) {
+  std::size_t offset = 0;
+  while (offset < data.size()) {
+    auto consumed = processor.process(data, offset);
+    if (consumed == 0 || delegate.have_error())
+      return false;
+    offset += consumed;
+  }
+  return true;
+}
+
+}  // namespace
+
+TEST(sax, decoder_utf8) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::string input = R"(<?xml version="1.0" encoding="utf-8"?><root />)";
+  std::cerr << input << std::endl;
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(reinterpret_cast<uint8_t const*>(input.data()),
+                               input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf8_bom) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::string input =
+      "\xef\xbb\xbf" R"(<?xml version="1.0" encoding="utf-8"?><root />)";
+  std::cerr << input << std::endl;
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(reinterpret_cast<uint8_t const*>(input.data()),
+                               input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf16) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u16string input = uR"(<?xml version="1.0" encoding="utf-16"?><root />)";
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(reinterpret_cast<uint8_t const*>(input.data()),
+                               input.size() * sizeof(char16_t))));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf16be) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u16string str = uR"(<?xml version="1.0" encoding="utf-16"?><root />)";
+  std::vector<uint8_t> input;
+  for (char16_t c : str) {
+    input.push_back(c >> 8);
+    input.push_back(c & 0xff);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf16le) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u16string str = uR"(<?xml version="1.0" encoding="utf-16"?><root />)";
+  std::vector<uint8_t> input;
+  for (char16_t c : str) {
+    input.push_back(c & 0xff);
+    input.push_back(c >> 8);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf16be_bom) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u16string str =
+      u"\ufffe" uR"(<?xml version="1.0" encoding="utf-16"?><root />)";
+  std::vector<uint8_t> input;
+  for (char16_t c : str) {
+    input.push_back(c >> 8);
+    input.push_back(c & 0xff);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf16le_bom) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u16string str =
+      u"\ufffe" uR"(<?xml version="1.0" encoding="utf-16"?><root />)";
+  std::vector<uint8_t> input;
+  for (char16_t c : str) {
+    input.push_back(c & 0xff);
+    input.push_back(c >> 8);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf32) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u32string input = UR"(<?xml version="1.0" encoding="utf-32"?><root />)";
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(reinterpret_cast<uint8_t const*>(input.data()),
+                               input.size() * sizeof(char32_t))));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf32be) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u32string str = UR"(<?xml version="1.0" encoding="utf-32"?><root />)";
+  std::vector<uint8_t> input;
+  for (char32_t c : str) {
+    input.push_back(c >> 24);
+    input.push_back((c >> 16) & 0xff);
+    input.push_back((c >> 8) & 0xff);
+    input.push_back(c & 0xff);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf32le) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u32string str = UR"(<?xml version="1.0" encoding="utf-32"?><root />)";
+  std::vector<uint8_t> input;
+  for (char32_t c : str) {
+    input.push_back(c & 0xff);
+    input.push_back((c >> 8) & 0xff);
+    input.push_back((c >> 16) & 0xff);
+    input.push_back(c >> 24);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf32be_bom) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u32string str =
+      U"\ufffe" UR"(<?xml version="1.0" encoding="utf-32"?><root />)";
+  std::vector<uint8_t> input;
+  for (char32_t c : str) {
+    input.push_back(c >> 24);
+    input.push_back((c >> 16) & 0xff);
+    input.push_back((c >> 8) & 0xff);
+    input.push_back(c & 0xff);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}
+
+TEST(sax, decoder_utf32le_bom) {
+  auto delegate = std::make_shared<TestDelegate>();
+  auto processor = modxml::sax::Processor::create(delegate);
+  std::u32string str =
+      U"\ufffe" R"(<?xml version="1.0" encoding="utf-32"?><root />)";
+  std::vector<uint8_t> input;
+  for (char32_t c : str) {
+    input.push_back(c & 0xff);
+    input.push_back((c >> 8) & 0xff);
+    input.push_back((c >> 16) & 0xff);
+    input.push_back(c >> 24);
+  }
+  EXPECT_TRUE(process_all(
+      *processor.get(),
+      *delegate.get(),
+      std::span<uint8_t const>(input.data(), input.size())));
+  EXPECT_TRUE(delegate->have_root());
+}