From 888e1967c46516b6b0681bec35cc534ea045ebcb Mon Sep 17 00:00:00 2001
From: Selim Mustafaev <selim@fastmail.fm>
Date: Sun, 9 May 2021 08:07:38 +0300
Subject: [PATCH] Basic support of parsing scripts

---
 CMakeLists.txt         |  1 +
 Models/Block.cpp       |  2 +-
 Models/Script.cpp      | 55 +++++++++++++++++++++++++++++++++++++++---
 Models/Script.h        | 27 +++++++++++++++++++--
 Models/Transaction.cpp |  4 +--
 Models/TxInput.cpp     |  9 ++++---
 Models/TxInput.h       |  6 +++++
 Models/TxOutput.cpp    |  6 ++++-
 Models/TxOutput.h      |  4 +++
 main.cpp               | 15 +++++++++---
 10 files changed, 112 insertions(+), 17 deletions(-)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 394c768..691a592 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -3,5 +3,6 @@ project(btcexplorer)
 
 set(CMAKE_CXX_STANDARD 20)
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address")
+#set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS}")
 
 add_executable(btcexplorer main.cpp Models/Block.cpp Models/Block.h Models/Transaction.cpp Models/Transaction.h Models/VarInt.cpp Models/VarInt.h Models/TxInput.cpp Models/TxInput.h Models/TxOutput.cpp Models/TxOutput.h Models/Script.cpp Models/Script.h)
\ No newline at end of file
diff --git a/Models/Block.cpp b/Models/Block.cpp
index 4561dd9..8860089 100644
--- a/Models/Block.cpp
+++ b/Models/Block.cpp
@@ -9,8 +9,8 @@ Block::Block(const std::byte* pBlock, size_t size): _header{} {
     auto pTxData = pBlock + sizeof(BlockHeader) + txCount.size();
     for(size_t i = 0; i < txCount.value(); ++i) {
         Transaction transaction(pTxData);
-        _transactions.push_back(transaction);
         pTxData += transaction.size();
+        _transactions.emplace_back(std::move(transaction));
     }
 }
 
diff --git a/Models/Script.cpp b/Models/Script.cpp
index 08d6f4c..6312341 100644
--- a/Models/Script.cpp
+++ b/Models/Script.cpp
@@ -1,5 +1,52 @@
-//
-// Created by Selim Mustafaev on 05.05.2021.
-//
-
 #include "Script.h"
+
+#include <iostream>
+#include <iomanip>
+
+Script::Script(std::span<uint8_t> data) {
+    for(auto iter = data.begin(); iter != data.end();) {
+        ScriptOperation operation;
+        uint8_t* ptr = data.data();
+        size_t bytes = data.size_bytes();
+        operation.opCode = OpCode(*iter++);
+        //std::cout << "found opcode: " << std::showbase << std::hex << std::setw(4) << (int)operation.opCode << std::dec << std::endl;
+        if(operation.opCode <= OpCode::OP_PUSHDATA4) {
+            size_t dataSize = 0;
+            if(operation.opCode < OpCode::OP_PUSHDATA1) {
+                dataSize = operation.opCode;
+            } else if(operation.opCode == OpCode::OP_PUSHDATA1) {
+                dataSize = *iter++;
+            } else if(operation.opCode == OpCode::OP_PUSHDATA2) {
+                dataSize = *((uint16_t*)&iter[0]);
+                iter += 2;
+            } else if(operation.opCode == OpCode::OP_PUSHDATA4) {
+                dataSize = *((uint32_t *) &iter[0]);
+                iter += 4;
+            }
+            operation.input = std::vector(iter, iter + dataSize);
+            iter += dataSize;
+        }
+        _operations.emplace_back(operation);
+    }
+
+    _type = getType();
+}
+
+Script::Script() {
+
+}
+
+ScriptType Script::getType() {
+    if(_operations[0].opCode <= OpCode::OP_PUSHDATA4 && _operations[1].opCode == OpCode::OP_CHECKSIG) {
+        return ScriptType::P2PK;
+    } else if(_operations[0].opCode == OpCode::OP_DUP
+                && _operations[1].opCode == OpCode::OP_HASH160
+                && _operations[2].opCode <= OpCode::OP_PUSHDATA4
+                && _operations[3].opCode == OpCode::OP_EQUALVERIFY
+                && _operations[4].opCode == OpCode::OP_CHECKSIG)
+    {
+        return ScriptType::P2PKH;
+    } else {
+        throw std::runtime_error("Unsupported script type");
+    }
+}
diff --git a/Models/Script.h b/Models/Script.h
index 3e3d9c8..96e36b0 100644
--- a/Models/Script.h
+++ b/Models/Script.h
@@ -1,7 +1,12 @@
 #ifndef BTCEXPLORER_SCRIPT_H
 #define BTCEXPLORER_SCRIPT_H
 
-enum OpCode {
+#include <vector>
+#include <cstddef>
+#include <cstdint>
+#include <span>
+
+enum OpCode: uint8_t {
     // push value
     OP_0 = 0x00,
     OP_FALSE = OP_0,
@@ -142,8 +147,26 @@ enum OpCode {
     OP_INVALIDOPCODE = 0xff,
 };
 
-class Script {
+enum ScriptType {
+    P2PK, P2PKH
+};
 
+struct ScriptOperation {
+    OpCode opCode;
+    std::vector<uint8_t> input;
+};
+
+class Script {
+private:
+    std::vector<ScriptOperation> _operations;
+    ScriptType _type;
+
+private:
+    ScriptType getType();
+
+public:
+    Script();
+    explicit Script(std::span<uint8_t> data);
 };
 
 #endif //BTCEXPLORER_SCRIPT_H
diff --git a/Models/Transaction.cpp b/Models/Transaction.cpp
index 470a464..8ae0015 100644
--- a/Models/Transaction.cpp
+++ b/Models/Transaction.cpp
@@ -9,7 +9,7 @@ Transaction::Transaction(const std::byte *data) {
     for(size_t i = 0; i < inputCount.value(); ++i) {
         TxInput input(curPtr);
         curPtr += input.size();
-        _inputs.push_back(input);
+        _inputs.emplace_back(std::move(input));
     }
 
     VarInt outputCount(curPtr);
@@ -17,7 +17,7 @@ Transaction::Transaction(const std::byte *data) {
     for(size_t i = 0; i < outputCount.value(); ++i) {
         TxOutput output(curPtr);
         curPtr += output.size();
-        _outputs.push_back(output);
+        _outputs.emplace_back(std::move(output));
     }
 
     _lockTime = *((uint32_t*)curPtr);
diff --git a/Models/TxInput.cpp b/Models/TxInput.cpp
index 386fc63..671d5ef 100644
--- a/Models/TxInput.cpp
+++ b/Models/TxInput.cpp
@@ -1,6 +1,7 @@
 #include "TxInput.h"
 #include "VarInt.h"
 #include <algorithm>
+#include <iostream>
 
 TxInput::TxInput(const std::byte *data) {
     std::copy_n(data, _txId.size(), _txId.begin());
@@ -8,10 +9,12 @@ TxInput::TxInput(const std::byte *data) {
     _vOut = *reinterpret_cast<const uint32_t*>(data);
     data += sizeof(_vOut);
     VarInt scriptSigSize(data);
-
-    // TODO: save ScriptSig
-
     data += scriptSigSize.size();
+
+    //std::cout << "=== Creating signature script of size: " << scriptSigSize.value() << std::endl;
+    _signatureScript = std::make_unique<Script>(std::span((uint8_t*)data, scriptSigSize.value()));
+    data += scriptSigSize.value();
+
     _sequence = *reinterpret_cast<const uint32_t*>(data);
     _size = _txId.size() + sizeof(_vOut) + scriptSigSize.size() + scriptSigSize.value() + sizeof(_sequence);
 }
diff --git a/Models/TxInput.h b/Models/TxInput.h
index fa3091c..d5c5c9d 100644
--- a/Models/TxInput.h
+++ b/Models/TxInput.h
@@ -1,7 +1,12 @@
 #ifndef BTCEXPLORER_TXINPUT_H
 #define BTCEXPLORER_TXINPUT_H
 
+#include "Script.h"
+
+#include <cstdint>
+#include <cstddef>
 #include <array>
+#include <memory>
 
 class TxInput {
 private:
@@ -9,6 +14,7 @@ private:
     uint32_t _vOut;
     uint32_t _sequence;
     size_t _size;
+    std::unique_ptr<Script> _signatureScript;
 
 public:
     explicit TxInput(const std::byte* data);
diff --git a/Models/TxOutput.cpp b/Models/TxOutput.cpp
index 7513e01..99c1186 100644
--- a/Models/TxOutput.cpp
+++ b/Models/TxOutput.cpp
@@ -1,12 +1,16 @@
 #include "TxOutput.h"
 #include "VarInt.h"
 
+#include <iostream>
+
 TxOutput::TxOutput(const std::byte *data) {
     _value = *reinterpret_cast<const uint64_t*>(data);
     data += sizeof(_value);
     VarInt scriptPubKeySize(data);
+    data += scriptPubKeySize.size();
 
-    // TODO: Read ScriptSizeKey
+    //std::cout << "=== Creating PubKey script of size: " << scriptPubKeySize.value() << std::endl;
+    _pubKeyScript = std::make_unique<Script>(std::span((uint8_t*)data, scriptPubKeySize.value()));
 
     _size = sizeof(_value) + scriptPubKeySize.size() + scriptPubKeySize.value();
 }
diff --git a/Models/TxOutput.h b/Models/TxOutput.h
index c0309de..7ff4d0c 100644
--- a/Models/TxOutput.h
+++ b/Models/TxOutput.h
@@ -1,13 +1,17 @@
 #ifndef BTCEXPLORER_TXOUTPUT_H
 #define BTCEXPLORER_TXOUTPUT_H
 
+#include "Script.h"
+
 #include <cstdint>
 #include <cstddef>
+#include <memory>
 
 class TxOutput {
 private:
     uint64_t _value;
     size_t  _size;
+    std::unique_ptr<Script> _pubKeyScript;
 
 public:
     explicit TxOutput(const std::byte* data);
diff --git a/main.cpp b/main.cpp
index 14f22f0..0fab9f6 100644
--- a/main.cpp
+++ b/main.cpp
@@ -2,15 +2,18 @@
 #include <fstream>
 #include <memory>
 #include <vector>
+#include <chrono>
 
 #include "Models/Block.h"
 
 int main() {
-    std::string path = "/Users/selim/Documents/blk00000.dat";
+    std::string path = "/home/selim/dl/blk00000.dat";
     std::fstream file(path);
 
     std::vector<Block> blocks;
 
+    auto start = std::chrono::high_resolution_clock::now();
+
     PreBlockHeader header{};
     size_t index = 0;
     while (true) {
@@ -24,13 +27,17 @@ int main() {
         auto blockData = std::make_unique<std::byte[]>(header.blockSize);
         file.read((char*)blockData.get(), header.blockSize);
 
-        Block block(blockData.get(), header.blockSize);
-        blocks.push_back(block);
+        blocks.emplace_back(blockData.get(), header.blockSize);
 
-        //std::cout << "Parsed new block " << index++ << " with size: " << header.blockSize << std::endl;
+        std::cout << "Parsed new block " << index++ << " with size: " << header.blockSize << std::endl;
+        break;
     }
 
+    auto stop = std::chrono::high_resolution_clock::now();
+    auto duration = duration_cast<std::chrono::microseconds>(stop - start);
+
     std::cout << "Blocks found: " << blocks.size() << std::endl;
+    std::cout << "Duration: " << double(duration.count())/1000000 << " seconds" << std::endl;
 //    for(size_t i = 0; i < 100; ++i) {
 //        std::cout << "version: " << blocks[i].time() << std::endl;
 //    }