finslide

Author: archibate

15/08/a.cpp

@@ -0,0 +1,28 @@
+#include <string>
+#include <vector>
+#include <string_view>
+#include <iostream>
+
+using namespace std;
+
+struct Empty {
+};
+
+struct NonEmpty {
+    char c;
+};
+
+struct DerivedEmpty : Empty {
+    int i;
+};
+
+struct DerivedNonEmpty : NonEmpty {
+    int i;
+};
+
+int main() {
+    cout << sizeof(Empty) << endl;            // 1 = 1(empty)
+    cout << sizeof(NonEmpty) << endl;         // 1 = 1(char c)
+    cout << sizeof(DerivedEmpty) << endl;     // 4 = 0(empty) + 4(int i)
+    cout << sizeof(DerivedNonEmpty) << endl;  // 8 = 1(char c) + 3(padding) + 4(int i)
+}

15/09/a.cpp

@@ -0,0 +1,55 @@
+#include <string>
+#include <iostream>
+#include <iomanip>
+#include <cstring>
+
+using namespace std;
+
+int main() {
+    cout << "\x7a\u00df\u6c34\U0001f34c" << endl;
+    {
+        cout << "char: ";
+        string s = "\x7a\u00df\u6c34\U0001f34c";
+        for (size_t i = 0; i < s.size(); i++) {
+            cout << setw(2) << setfill('0') << hex;
+            cout << hex << (uint32_t)(uint8_t)s[i] << ' ';
+        }
+        cout << endl;
+    }
+    {
+        cout << "wchar_t: ";
+        wstring s = L"\x7a\u00df\u6c34\U0001f34c";
+        for (size_t i = 0; i < s.size(); i++) {
+            cout << setw(sizeof(wchar_t) * 2) << setfill('0') << hex;
+            cout << (uint32_t)s[i] << ' ';
+        }
+        cout << endl;
+    }
+    //{
+        //cout << "char8_t: ";
+        //u8string s = u8"\x7a\u00df\u6c34\U0001f34c";
+        //for (size_t i = 0; i < s.size(); i++) {
+            //cout << setw(2) << setfill('0') << hex;
+            //cout << (uint16_t)s[i] << ' ';
+        //}
+        //cout << endl;
+    //}
+    {
+        cout << "char16_t: ";
+        u16string s = u"\x7a\u00df\u6c34\U0001f34c";
+        for (size_t i = 0; i < s.size(); i++) {
+            cout << setw(4) << setfill('0') << hex;
+            cout << (uint16_t)s[i] << ' ';
+        }
+        cout << endl;
+    }
+    {
+        cout << "char32_t: ";
+        u32string s = U"\x7a\u00df\u6c34\U0001f34c";
+        for (size_t i = 0; i < s.size(); i++) {
+            cout << setw(8) << setfill('0') << hex;
+            cout << (uint32_t)s[i] << ' ';
+        }
+        cout << endl;
+    }
+}

15/09/b.cpp

@@ -0,0 +1,71 @@
+#include <string>
+#include <string_view>
+#include <stdexcept>
+#include <iostream>
+#include <iomanip>
+#include <locale>
+
+using namespace std;
+
+template <class Internal = wchar_t>
+string convcodec(string_view src, locale const &srcLoc, locale const &dstLoc) {
+    auto strCodecvtError = [] (codecvt_base::result res) -> string {
+        if (res == codecvt_base::result::noconv)
+            return "noconv";
+        if (res == codecvt_base::result::error)
+            return "error";
+        if (res == codecvt_base::result::partial)
+            return "partial";
+        return "ok";
+    };
+
+    string dst;
+    basic_string<Internal> tmp;
+    mbstate_t mbstate = {};
+
+    auto &srcFacet = use_facet<codecvt<Internal, char, mbstate_t>>(srcLoc);
+    tmp.resize(src.size());
+
+    const char *src_begin = src.data();
+    const char *src_end = src.data() + src.size();
+    const char *src_next;
+    Internal *tmp_begin = tmp.data();
+    Internal *tmp_end = tmp.data() + tmp.size();
+    Internal *tmp_next;
+    auto resIn = srcFacet.in(mbstate, src_begin, src_end, src_next, tmp_begin, tmp_end, tmp_next);
+    if (resIn != codecvt_base::result::ok)
+        throw runtime_error("cannot decode input sequence: " + strCodecvtError(resIn));
+    size_t tmp_size = tmp_next - tmp_begin;
+    tmp_end = tmp_next;
+
+    auto &dstFacet = use_facet<codecvt<Internal, char, mbstate_t>>(dstLoc);
+    dst.resize(tmp_size * dstFacet.max_length());
+
+    char *dst_begin = dst.data();
+    char *dst_end = dst.data() + dst.size();
+    char *dst_next;
+    const Internal *tmp_next_next;
+    auto resOut = dstFacet.out(mbstate, tmp_begin, tmp_end, tmp_next_next, dst_begin, dst_end, dst_next);
+    if (resOut != codecvt_base::result::ok)
+        throw runtime_error("cannot encode output sequence: " + strCodecvtError(resOut));
+    size_t dst_size = dst_next - dst_begin;
+
+    dst.resize(dst_size);
+    return dst;
+}
+
+void showstr(string_view s) {
+    for (char c: s) {
+        cout << hex << setfill('0') << setw(2);
+        cout << (uint32_t)(uint8_t)c << ' ';
+    }
+    cout << s << endl;
+}
+
+int main() {
+    string s = "你好";
+    //string s = "你好\U0001f34c";  // error: cannot encode U+1F34C
+    showstr(s);
+    s = convcodec(s, locale("en_US.UTF-8"), locale("zh_CN.GBK"));
+    showstr(s);
+}