Merge #11630: Simplify Base32 and Base64 conversions

b3ea8cc Simplify Base32 and Base64 conversions (Pieter Wuille)
3296a3b Generalize ConvertBits (Pieter Wuille)

Pull request description:

  Generalize `ConvertBits` a bit to also be usable for the existing Base32 and Base64 convertions (rather than just for Bech32).

Tree-SHA512: 3858247f9b14ca4766c08ea040a09b1d6d70caaccc75c2436a54102d6d526f499ec07f5bdfcbbe16cbde5aae521cd16e9aa693e688a97e6c5e74b8e58ee55a13

Author: laanwj

src/key_io.cpp

@@ -43,14 +43,16 @@ class DestinationEncoder : public boost::static_visitor<std::string>
     std::string operator()(const WitnessV0KeyHash& id) const
     {
         std::vector<unsigned char> data = {0};
-        ConvertBits<8, 5, true>(data, id.begin(), id.end());
+        data.reserve(33);
+        ConvertBits<8, 5, true>([&](unsigned char c) { data.push_back(c); }, id.begin(), id.end());
         return bech32::Encode(m_params.Bech32HRP(), data);
     }
 
     std::string operator()(const WitnessV0ScriptHash& id) const
     {
         std::vector<unsigned char> data = {0};
-        ConvertBits<8, 5, true>(data, id.begin(), id.end());
+        data.reserve(53);
+        ConvertBits<8, 5, true>([&](unsigned char c) { data.push_back(c); }, id.begin(), id.end());
         return bech32::Encode(m_params.Bech32HRP(), data);
     }
 
@@ -60,7 +62,8 @@ class DestinationEncoder : public boost::static_visitor<std::string>
             return {};
         }
         std::vector<unsigned char> data = {(unsigned char)id.version};
-        ConvertBits<8, 5, true>(data, id.program, id.program + id.length);
+        data.reserve(1 + (id.length * 8 + 4) / 5);
+        ConvertBits<8, 5, true>([&](unsigned char c) { data.push_back(c); }, id.program, id.program + id.length);
         return bech32::Encode(m_params.Bech32HRP(), data);
     }
 
@@ -94,7 +97,8 @@ CTxDestination DecodeDestination(const std::string& str, const CChainParams& par
         // Bech32 decoding
         int version = bech.second[0]; // The first 5 bit symbol is the witness version (0-16)
         // The rest of the symbols are converted witness program bytes.
-        if (ConvertBits<5, 8, false>(data, bech.second.begin() + 1, bech.second.end())) {
+        data.reserve(((bech.second.size() - 1) * 5) / 8);
+        if (ConvertBits<5, 8, false>([&](unsigned char c) { data.push_back(c); }, bech.second.begin() + 1, bech.second.end())) {
             if (version == 0) {
                 {
                     WitnessV0KeyHash keyid;

src/test/base32_tests.cpp

@@ -16,9 +16,9 @@ BOOST_AUTO_TEST_CASE(base32_testvectors)
     for (unsigned int i=0; i<sizeof(vstrIn)/sizeof(vstrIn[0]); i++)
     {
         std::string strEnc = EncodeBase32(vstrIn[i]);
-        BOOST_CHECK(strEnc == vstrOut[i]);
+        BOOST_CHECK_EQUAL(strEnc, vstrOut[i]);
         std::string strDec = DecodeBase32(vstrOut[i]);
-        BOOST_CHECK(strDec == vstrIn[i]);
+        BOOST_CHECK_EQUAL(strDec, vstrIn[i]);
     }
 }
 

src/test/base64_tests.cpp

@@ -16,9 +16,9 @@ BOOST_AUTO_TEST_CASE(base64_testvectors)
     for (unsigned int i=0; i<sizeof(vstrIn)/sizeof(vstrIn[0]); i++)
     {
         std::string strEnc = EncodeBase64(vstrIn[i]);
-        BOOST_CHECK(strEnc == vstrOut[i]);
+        BOOST_CHECK_EQUAL(strEnc, vstrOut[i]);
         std::string strDec = DecodeBase64(strEnc);
-        BOOST_CHECK(strDec == vstrIn[i]);
+        BOOST_CHECK_EQUAL(strDec, vstrIn[i]);
     }
 }
 

src/utilstrencodings.cpp

@@ -127,46 +127,11 @@ std::string EncodeBase64(const unsigned char* pch, size_t len)
 {
     static const char *pbase64 = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
 
-    std::string strRet;
-    strRet.reserve((len+2)/3*4);
-
-    int mode=0, left=0;
-    const unsigned char *pchEnd = pch+len;
-
-    while (pch<pchEnd)
-    {
-        int enc = *(pch++);
-        switch (mode)
-        {
-            case 0: // we have no bits
-                strRet += pbase64[enc >> 2];
-                left = (enc & 3) << 4;
-                mode = 1;
-                break;
-
-            case 1: // we have two bits
-                strRet += pbase64[left | (enc >> 4)];
-                left = (enc & 15) << 2;
-                mode = 2;
-                break;
-
-            case 2: // we have four bits
-                strRet += pbase64[left | (enc >> 6)];
-                strRet += pbase64[enc & 63];
-                mode = 0;
-                break;
-        }
-    }
-
-    if (mode)
-    {
-        strRet += pbase64[left];
-        strRet += '=';
-        if (mode == 1)
-            strRet += '=';
-    }
-
-    return strRet;
+    std::string str;
+    str.reserve(((len + 2) / 3) * 4);
+    ConvertBits<8, 6, true>([&](int v) { str += pbase64[v]; }, pch, pch + len);
+    while (str.size() % 4) str += '=';
+    return str;
 }
 
 std::string EncodeBase64(const std::string& str)
@@ -193,68 +158,32 @@ std::vector<unsigned char> DecodeBase64(const char* p, bool* pfInvalid)
         -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
     };
 
-    if (pfInvalid)
-        *pfInvalid = false;
-
-    std::vector<unsigned char> vchRet;
-    vchRet.reserve(strlen(p)*3/4);
-
-    int mode = 0;
-    int left = 0;
-
-    while (1)
-    {
-         int dec = decode64_table[(unsigned char)*p];
-         if (dec == -1) break;
-         p++;
-         switch (mode)
-         {
-             case 0: // we have no bits and get 6
-                 left = dec;
-                 mode = 1;
-                 break;
-
-              case 1: // we have 6 bits and keep 4
-                  vchRet.push_back((left<<2) | (dec>>4));
-                  left = dec & 15;
-                  mode = 2;
-                  break;
-
-             case 2: // we have 4 bits and get 6, we keep 2
-                 vchRet.push_back((left<<4) | (dec>>2));
-                 left = dec & 3;
-                 mode = 3;
-                 break;
-
-             case 3: // we have 2 bits and get 6
-                 vchRet.push_back((left<<6) | dec);
-                 mode = 0;
-                 break;
-         }
+    const char* e = p;
+    std::vector<uint8_t> val;
+    val.reserve(strlen(p));
+    while (*p != 0) {
+        int x = decode64_table[(unsigned char)*p];
+        if (x == -1) break;
+        val.push_back(x);
+        ++p;
     }
 
-    if (pfInvalid)
-        switch (mode)
-        {
-            case 0: // 4n base64 characters processed: ok
-                break;
-
-            case 1: // 4n+1 base64 character processed: impossible
-                *pfInvalid = true;
-                break;
-
-            case 2: // 4n+2 base64 characters processed: require '=='
-                if (left || p[0] != '=' || p[1] != '=' || decode64_table[(unsigned char)p[2]] != -1)
-                    *pfInvalid = true;
-                break;
-
-            case 3: // 4n+3 base64 characters processed: require '='
-                if (left || p[0] != '=' || decode64_table[(unsigned char)p[1]] != -1)
-                    *pfInvalid = true;
-                break;
+    std::vector<unsigned char> ret;
+    ret.reserve((val.size() * 3) / 4);
+    bool valid = ConvertBits<6, 8, false>([&](unsigned char c) { ret.push_back(c); }, val.begin(), val.end());
+
+    const char* q = p;
+    while (valid && *p != 0) {
+        if (*p != '=') {
+            valid = false;
+            break;
         }
+        ++p;
+    }
+    valid = valid && (p - e) % 4 == 0 && p - q < 4;
+    if (pfInvalid) *pfInvalid = !valid;
 
-    return vchRet;
+    return ret;
 }
 
 std::string DecodeBase64(const std::string& str)
@@ -267,59 +196,11 @@ std::string EncodeBase32(const unsigned char* pch, size_t len)
 {
     static const char *pbase32 = "abcdefghijklmnopqrstuvwxyz234567";
 
-    std::string strRet;
-    strRet.reserve((len+4)/5*8);
-
-    int mode=0, left=0;
-    const unsigned char *pchEnd = pch+len;
-
-    while (pch<pchEnd)
-    {
-        int enc = *(pch++);
-        switch (mode)
-        {
-            case 0: // we have no bits
-                strRet += pbase32[enc >> 3];
-                left = (enc & 7) << 2;
-                mode = 1;
-                break;
-
-            case 1: // we have three bits
-                strRet += pbase32[left | (enc >> 6)];
-                strRet += pbase32[(enc >> 1) & 31];
-                left = (enc & 1) << 4;
-                mode = 2;
-                break;
-
-            case 2: // we have one bit
-                strRet += pbase32[left | (enc >> 4)];
-                left = (enc & 15) << 1;
-                mode = 3;
-                break;
-
-            case 3: // we have four bits
-                strRet += pbase32[left | (enc >> 7)];
-                strRet += pbase32[(enc >> 2) & 31];
-                left = (enc & 3) << 3;
-                mode = 4;
-                break;
-
-            case 4: // we have two bits
-                strRet += pbase32[left | (enc >> 5)];
-                strRet += pbase32[enc & 31];
-                mode = 0;
-        }
-    }
-
-    static const int nPadding[5] = {0, 6, 4, 3, 1};
-    if (mode)
-    {
-        strRet += pbase32[left];
-        for (int n=0; n<nPadding[mode]; n++)
-             strRet += '=';
-    }
-
-    return strRet;
+    std::string str;
+    str.reserve(((len + 4) / 5) * 8);
+    ConvertBits<8, 5, true>([&](int v) { str += pbase32[v]; }, pch, pch + len);
+    while (str.size() % 8) str += '=';
+    return str;
 }
 
 std::string EncodeBase32(const std::string& str)
@@ -346,102 +227,32 @@ std::vector<unsigned char> DecodeBase32(const char* p, bool* pfInvalid)
         -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1
     };
 
-    if (pfInvalid)
-        *pfInvalid = false;
-
-    std::vector<unsigned char> vchRet;
-    vchRet.reserve((strlen(p))*5/8);
-
-    int mode = 0;
-    int left = 0;
-
-    while (1)
-    {
-         int dec = decode32_table[(unsigned char)*p];
-         if (dec == -1) break;
-         p++;
-         switch (mode)
-         {
-             case 0: // we have no bits and get 5
-                 left = dec;
-                 mode = 1;
-                 break;
-
-              case 1: // we have 5 bits and keep 2
-                  vchRet.push_back((left<<3) | (dec>>2));
-                  left = dec & 3;
-                  mode = 2;
-                  break;
-
-             case 2: // we have 2 bits and keep 7
-                 left = left << 5 | dec;
-                 mode = 3;
-                 break;
-
-             case 3: // we have 7 bits and keep 4
-                 vchRet.push_back((left<<1) | (dec>>4));
-                 left = dec & 15;
-                 mode = 4;
-                 break;
-
-             case 4: // we have 4 bits, and keep 1
-                 vchRet.push_back((left<<4) | (dec>>1));
-                 left = dec & 1;
-                 mode = 5;
-                 break;
-
-             case 5: // we have 1 bit, and keep 6
-                 left = left << 5 | dec;
-                 mode = 6;
-                 break;
-
-             case 6: // we have 6 bits, and keep 3
-                 vchRet.push_back((left<<2) | (dec>>3));
-                 left = dec & 7;
-                 mode = 7;
-                 break;
-
-             case 7: // we have 3 bits, and keep 0
-                 vchRet.push_back((left<<5) | dec);
-                 mode = 0;
-                 break;
-         }
+    const char* e = p;
+    std::vector<uint8_t> val;
+    val.reserve(strlen(p));
+    while (*p != 0) {
+        int x = decode32_table[(unsigned char)*p];
+        if (x == -1) break;
+        val.push_back(x);
+        ++p;
     }
 
-    if (pfInvalid)
-        switch (mode)
-        {
-            case 0: // 8n base32 characters processed: ok
-                break;
-
-            case 1: // 8n+1 base32 characters processed: impossible
-            case 3: //   +3
-            case 6: //   +6
-                *pfInvalid = true;
-                break;
-
-            case 2: // 8n+2 base32 characters processed: require '======'
-                if (left || p[0] != '=' || p[1] != '=' || p[2] != '=' || p[3] != '=' || p[4] != '=' || p[5] != '=' || decode32_table[(unsigned char)p[6]] != -1)
-                    *pfInvalid = true;
-                break;
-
-            case 4: // 8n+4 base32 characters processed: require '===='
-                if (left || p[0] != '=' || p[1] != '=' || p[2] != '=' || p[3] != '=' || decode32_table[(unsigned char)p[4]] != -1)
-                    *pfInvalid = true;
-                break;
-
-            case 5: // 8n+5 base32 characters processed: require '==='
-                if (left || p[0] != '=' || p[1] != '=' || p[2] != '=' || decode32_table[(unsigned char)p[3]] != -1)
-                    *pfInvalid = true;
-                break;
-
-            case 7: // 8n+7 base32 characters processed: require '='
-                if (left || p[0] != '=' || decode32_table[(unsigned char)p[1]] != -1)
-                    *pfInvalid = true;
-                break;
+    std::vector<unsigned char> ret;
+    ret.reserve((val.size() * 5) / 8);
+    bool valid = ConvertBits<5, 8, false>([&](unsigned char c) { ret.push_back(c); }, val.begin(), val.end());
+
+    const char* q = p;
+    while (valid && *p != 0) {
+        if (*p != '=') {
+            valid = false;
+            break;
         }
+        ++p;
+    }
+    valid = valid && (p - e) % 8 == 0 && p - q < 8;
+    if (pfInvalid) *pfInvalid = !valid;
 
-    return vchRet;
+    return ret;
 }
 
 std::string DecodeBase32(const std::string& str)