Add unicode general category lookup

Generate the lookup tables from UnicodeData.txt, do to that,
add gen_ugc, which uses csv, buffers, line, io and other modules
to do the job.

1 files changed, 72 insertions, 0 deletions

diff --git a/test/decompress.cc b/test/decompress.cc
new file mode 100644
index 0000000..35c4477
--- /dev/null
+++ b/test/decompress.cc
@@ -0,0 +1,72 @@
+#include <gtest/gtest.h>
+
+#include "decompress.hh"
+
+TEST(z_decompress, empty) {
+  static const unsigned char data[] = {
+    0x1f, 0x8b, 0x08, 0x08, 0x33, 0xd4, 0xbd, 0x68,
+    0x02, 0x03, 0x65, 0x6d, 0x70, 0x74, 0x79, 0x00,
+    0x03, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00
+  };
+  auto reader = decompress::gzip(io::memory(std::string(
+      reinterpret_cast<const char*>(data), sizeof(data))));
+  char buf[10];
+  auto got = reader->read(buf, sizeof(buf));
+  ASSERT_TRUE(got.has_value());
+  EXPECT_EQ(0, got.value());
+}
+
+TEST(z_decompress, hello) {
+  static const unsigned char data[] = {
+    0x1f, 0x8b, 0x08, 0x08, 0xf7, 0xd5, 0xbd, 0x68,
+    0x02, 0x03, 0x68, 0x65, 0x6c, 0x6c, 0x6f, 0x00,
+    0xf3, 0x48, 0xcd, 0xc9, 0xc9, 0x07, 0x00, 0x82,
+    0x89, 0xd1, 0xf7, 0x05, 0x00, 0x00, 0x00,
+  };
+  auto reader = decompress::gzip(io::memory(std::string(
+      reinterpret_cast<const char*>(data), sizeof(data))));
+  char buf[10];
+  auto got = reader->read(buf, sizeof(buf));
+  ASSERT_TRUE(got.has_value());
+  EXPECT_EQ(5, got.value());
+  buf[5] = '\0';
+  EXPECT_STREQ("Hello", buf);
+}
+
+TEST(xz_decompress, empty) {
+  static const unsigned char data[] = {
+    0xfd, 0x37, 0x7a, 0x58, 0x5a, 0x00, 0x00, 0x04,
+    0xe6, 0xd6, 0xb4, 0x46, 0x00, 0x00, 0x00, 0x00,
+    0x1c, 0xdf, 0x44, 0x21, 0x1f, 0xb6, 0xf3, 0x7d,
+    0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x59, 0x5a
+  };
+  auto reader = decompress::xz(io::memory(std::string(
+      reinterpret_cast<const char*>(data), sizeof(data))));
+  char buf[10];
+  auto got = reader->read(buf, sizeof(buf));
+  ASSERT_TRUE(got.has_value());
+  EXPECT_EQ(0, got.value());
+}
+
+TEST(xz_decompress, hello) {
+  static const unsigned char data[] = {
+    0xfd, 0x37, 0x7a, 0x58, 0x5a, 0x00, 0x00, 0x04,
+    0xe6, 0xd6, 0xb4, 0x46, 0x04, 0xc0, 0x09, 0x05,
+    0x21, 0x01, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00,
+    0x00, 0x00, 0x00, 0x00, 0x76, 0xe9, 0x07, 0x70,
+    0x01, 0x00, 0x04, 0x48, 0x65, 0x6c, 0x6c, 0x6f,
+    0x00, 0x00, 0x00, 0x00, 0xc8, 0xac, 0x7b, 0xc8,
+    0x3b, 0x5c, 0xcf, 0x51, 0x00, 0x01, 0x25, 0x05,
+    0x43, 0x91, 0x1f, 0xb8, 0x1f, 0xb6, 0xf3, 0x7d,
+    0x01, 0x00, 0x00, 0x00, 0x00, 0x04, 0x59, 0x5a,
+  };
+  auto reader = decompress::xz(io::memory(std::string(
+      reinterpret_cast<const char*>(data), sizeof(data))));
+  char buf[10];
+  auto got = reader->read(buf, sizeof(buf));
+  ASSERT_TRUE(got.has_value());
+  EXPECT_EQ(5, got.value());
+  buf[5] = '\0';
+  EXPECT_STREQ("Hello", buf);
+}