wasm2js.cpp

#include <fstream>
#include <array>
#include <vector>

using namespace std;

constexpr size_t calcEncodedSize(size_t size)
{
    return (size * 8 + 5) / 6;
}

static constexpr char btoa[] = "+-0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz";

void encode(char *dst, const uint8_t *ptr, size_t in_len)
{
    auto end = ptr + in_len;
    // 3byteずつ4文字に詰め込む
    for (; ptr + 3 <= end; ptr += 3)
    {
        uint32_t c = (ptr[0] << 16) | (ptr[1] << 8) | ptr[2];
        *dst++ = btoa[c >> 18];
        *dst++ = btoa[(c >> 12) & 63];
        *dst++ = btoa[(c >> 6) & 63];
        *dst++ = btoa[c & 63];
    }
    // 残り1byteの場合2文字目は末尾2bitを先頭に持って行ってXX0000の形にする
    if (ptr + 1 == end)
    {
        *dst++ = btoa[ptr[0] >> 2];
        *dst++ = btoa[(ptr[0] << 4) & 63];
    }
    // 残り2byteの場合3文字目はXXXX00の形
    else if (ptr + 2 == end)
    {
        uint32_t c = (ptr[0] << 8) | ptr[1];
        *dst++ = btoa[c >> 10];
        *dst++ = btoa[(c >> 4) & 63];
        *dst++ = btoa[(c << 2) & 63];
    }
}

// wasmをbase64化しArrayBufferに復元するjsコードもセットで埋め込む
int main(int argc, const char **argv)
{
    ifstream ifs(argv[1], ios::binary);
    ifs.seekg(0, ios::end);
    size_t size = ifs.tellg();
    ifs.seekg(0, ios::beg);
    vector<char> buf(size);
    ifs.read(buf.data(), size);

    vector<char> result(calcEncodedSize(size));
    encode(result.data(), reinterpret_cast<const uint8_t*>(buf.data()), size);

    ofstream ofs(argv[2], ios::binary);
    ofs << R"(
const decoder = new Uint8Array(123);
decoder.set([)";
    char8_t pre = 0;
    int count = 0;
    for (auto c : btoa)
    {
        if (c == 0)break;
        while (pre != c)
        {
            ofs << "0,";
            pre++;
        }
        ofs << count++ << ',';
        pre++;
    }
    ofs << R"(]);
const encoded = ')";
    // 長すぎる文字列リテラルはminifierで失敗するので8kbごとに区切る
    for (size_t blocks = result.size() >> 13, i = 0; i <= blocks; ++i)
    {
        ofs.write(result.data() + (i << 13), std::min<size_t>(8192, result.size() - (i << 13)));
        ofs << "'+'";
    }
    ofs << R"(';
const decode_bin = new Uint8Array()" << size << R"();
let dst = 0;
let ptr = 0;
const end = encoded.length;
const read = i => decoder[encoded.charCodeAt(i)];
for (; ptr + 4 <= end; ptr += 4) {
    const c = (read(ptr) << 18) | (read(ptr + 1) << 12) | (read(ptr + 2) << 6) | read(ptr + 3);
    decode_bin[dst++] = c >> 16;
    decode_bin[dst++] = (c >> 8) & 255;
    decode_bin[dst++] = c & 255;
}
switch (end - ptr) {
case 2:
decode_bin[dst] = (read(ptr) << 2) | (read(ptr + 1) >> 4);
break;
case 3:
{
const c = (read(ptr) << 10) | (read(ptr + 1) << 4) | (read(ptr + 2) >> 2);
decode_bin[dst++] = c >> 8;
decode_bin[dst] = c & 255;
}break;
}
var Module = { wasmBinary: decode_bin };
module.exports = Module;
const process = undefined;
const Buffer = undefined;
const setImmediate = undefined;
)";
    return 0;
}