Add Solidity to programming languages

CMakeLists.txt

@@ -55,6 +55,12 @@ if(NOT EXISTS ${PROJECT_SOURCE_DIR}/src/tree-sitter/tree-sitter-php)
     )
 endif()
 
+if(NOT EXISTS ${PROJECT_SOURCE_DIR}/src/tree-sitter/tree-sitter-solidity)
+    execute_process(
+        COMMAND git clone https://github.com/JoranHonig/tree-sitter-solidity.git ${PROJECT_SOURCE_DIR}/src/tree-sitter/tree-sitter-solidity
+    )
+endif()
+
 if(NOT EXISTS ${PROJECT_SOURCE_DIR}/src/tree-sitter/tree-sitter-cpp)
     execute_process(
         COMMAND git clone https://github.com/tree-sitter/tree-sitter-cpp.git ${PROJECT_SOURCE_DIR}/src/tree-sitter/tree-sitter-cpp

scripts/mine_patterns.sh

@@ -10,7 +10,7 @@ function print_usage() {
   else
     echo "[-n number_of_processes_to_use_for_mining]  (default: num_cpus_on_system)"
   fi
-  echo "[-l source_language_number] (default: 1 (C), supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++)"
+  echo "[-l source_language_number] (default: 1 (C), supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (Solidity)"
   echo "[-g github_repo_id] (default: 0) A unique identifier for GitHub repository, if any"
   exit
 }
@@ -43,9 +43,9 @@ then
   print_usage $0
 fi
 
-if (( ${LANGUAGE} < 1  || ${LANGUAGE} > 4 ));
+if (( ${LANGUAGE} < 1  || ${LANGUAGE} > 5 ));
 then
-  echo "ERROR: Only 1 (C), 2 (Verilog), 3 (PHP), and 4 (C++) are supported languages; received ${LANGUAGE}"
+  echo "ERROR: Only 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (Solidity) are supported languages; received ${LANGUAGE}"
   print_usage $0
 fi
 
@@ -69,6 +69,9 @@ then
 elif [ "${LANGUAGE}" = "4" ];
 then
   find "${TRAIN_DIR}" -iname "*.cpp" -o -iname "*.cc" -o -iname "*.cxx" -o -iname "*.h" -o -iname "*.hpp" -o -iname "*.hxx" -type f > ${FILE_LIST}
+elif [ "${LANGUAGE}" = "5" ];
+then
+  find "${TRAIN_DIR}" -iname "*.sol" -type f > ${FILE_LIST}
 fi
 
 SCRIPTS_DIR=`dirname $0`

scripts/scan_for_anomalies.sh

@@ -14,7 +14,7 @@ function print_usage() {
   fi
   echo " [-o output_log_dir]                        (default: /tmp)"
   echo " [-a anomaly_threshold]                     (default: 3.0)"
-  echo " [-l source_language_number]                (default: 1 (C), supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++)"
+  echo " [-l source_language_number]                (default: 1 (C), supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (Solidity)"
 
   exit
 }
@@ -58,9 +58,9 @@ then
   print_usage $0
 fi
 
-if (( ${LANGUAGE} < 1  || ${LANGUAGE} > 4 ));
+if (( ${LANGUAGE} < 1  || ${LANGUAGE} > 5 ));
 then
-  echo "ERROR: Only 1 (C), 2 (Verilog), 3 (PHP), and 4 (C++) are supported languages; received ${LANGUAGE}"
+  echo "ERROR: Only 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (Solidity) are supported languages; received ${LANGUAGE}"
   print_usage $0
 fi 
 
@@ -77,6 +77,9 @@ then
 elif [ "${LANGUAGE}" = "4" ];
 then
   find "${SCAN_DIR}" -iname "*.cpp" -o -iname "*.cc" -o -iname "*.cxx" -o -iname "*.h" -o -iname "*.hpp" -o -iname "*.hxx" -type f > ${SCAN_FILE_LIST}
+elif [ "${LANGUAGE}" = "5" ];
+then
+  find "${SCAN_DIR}" -iname "*.sol" -type f > ${SCAN_FILE_LIST}
 fi
 
 SCRIPTS_DIR=`dirname $0`

src/CMakeLists.txt

@@ -27,6 +27,7 @@ set(COMMON_LINK_LIBRARIES
   tree-sitter
   tree-sitter-c
   tree-sitter-php
+  tree-sitter-solidity
   tree-sitter-cpp
   tree-sitter-verilog
   pthread

src/cf_dump_code_blocks.cpp

@@ -93,7 +93,7 @@ int handle_command_args(int argc, char* argv[], CFDumpArgs& command_args) {
                 << std::endl
                 << "  [-l source_language_number]   (default: "
                 << LANGUAGE_C << ")"
-                << ", supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++)"
+                << ", supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (SOLIDITY)"
                 << std::endl;
   };
 
@@ -137,6 +137,9 @@ int main(int argc, char* argv[]) {
       case LANGUAGE_PHP:
         DumpCodeBlocksFromSourceFile<LANGUAGE_PHP>(command_args);
         break;
+      case LANGUAGE_SOLIDITY:
+        DumpCodeBlocksFromSourceFile<LANGUAGE_SOLIDITY>(command_args);
+        break;
       case LANGUAGE_CPP:
         DumpCodeBlocksFromSourceFile<LANGUAGE_CPP>(command_args);
         break;

src/cf_file_scanner.cpp

@@ -55,7 +55,7 @@ static int handle_command_args(int argc, char* argv[], FileScannerArgs& args) {
            << "  [-a anomaly_threshold]                     (default: 3.0)"
            << std::endl
            << "  [-l source_language_number]                (default: 1 (C), "
-           << "supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++))"
+           << "supported: 1 (C), 2 (Verilog), 3 (PHP), 4 (C++), 5 (SOLIDITY) )"
            << std::endl
            << "  [-v log_level ]                            (default: 0, "
            << "{ERROR, 0}, {INFO, 1}, {DEBUG, 2})"
@@ -162,6 +162,10 @@ int main(int argc, char* argv[]) {
             status = train_and_scan_util.ScanFile<LANGUAGE_PHP>(eval_file,
                                                                 log_file);
             break;
+          case LANGUAGE_SOLIDITY:
+            status = train_and_scan_util.ScanFile<LANGUAGE_SOLIDITY>(eval_file,
+                                                                log_file);
+            break;
           case LANGUAGE_CPP:
             status = train_and_scan_util.ScanFile<LANGUAGE_CPP>(eval_file,
                                                                 log_file);

src/common_util.cpp

@@ -127,6 +127,8 @@ ManagedTSTree GetTSTree<LANGUAGE_VERILOG>(const std::string&, bool);
 template
 ManagedTSTree GetTSTree<LANGUAGE_PHP>(const std::string&, bool);
 template
+ManagedTSTree GetTSTree<LANGUAGE_SOLIDITY>(const std::string&, bool);
+template
 ManagedTSTree GetTSTree<LANGUAGE_CPP>(const std::string&, bool);
 template
 ManagedTSTree GetTSTree<LANGUAGE_C>(const std::string&, std::string&);
@@ -135,6 +137,8 @@ ManagedTSTree GetTSTree<LANGUAGE_VERILOG>(const std::string&, std::string&);
 template
 ManagedTSTree GetTSTree<LANGUAGE_PHP>(const std::string&, std::string&);
 template
+ManagedTSTree GetTSTree<LANGUAGE_SOLIDITY>(const std::string&, std::string&);
+template
 ManagedTSTree GetTSTree<LANGUAGE_CPP>(const std::string&, std::string&);
 template
 void CollectCodeBlocksOfInterest<LANGUAGE_C>(const ManagedTSTree&,
@@ -146,5 +150,8 @@ template
 void CollectCodeBlocksOfInterest<LANGUAGE_PHP>(const ManagedTSTree &,
                                                code_blocks_t&);
 template
+void CollectCodeBlocksOfInterest<LANGUAGE_SOLIDITY>(const ManagedTSTree &,
+                                               code_blocks_t&);
+template
 void CollectCodeBlocksOfInterest<LANGUAGE_CPP>(const ManagedTSTree &,
                                                code_blocks_t&);

src/parser.h

@@ -32,13 +32,15 @@
 extern "C" const TSLanguage *tree_sitter_c();
 extern "C" const TSLanguage *tree_sitter_verilog();
 extern "C" const TSLanguage *tree_sitter_php();
+extern "C" const TSLanguage *tree_sitter_solidity();
 extern "C" const TSLanguage *tree_sitter_cpp();
 
 enum Language {
   LANGUAGE_C = 1,
   LANGUAGE_VERILOG = 2,
   LANGUAGE_PHP = 3,
-  LANGUAGE_CPP = 4
+  LANGUAGE_CPP = 4,
+  LANGUAGE_SOLIDITY = 5
 };
 
 #define LANGUAGE_MIN LANGUAGE_C
@@ -68,6 +70,10 @@ template <> inline const TSLanguage* GetTSLanguage<LANGUAGE_PHP> () {
   return tree_sitter_php();
 }
 
+template <> inline const TSLanguage* GetTSLanguage<LANGUAGE_SOLIDITY> () {
+  return tree_sitter_solidity();
+}
+
 template<Language L>
 class ParserBase {
  public:
@@ -126,6 +132,9 @@ template <> inline bool IsIfStatement<LANGUAGE_VERILOG>(const TSNode& node) {
 template <> inline bool IsIfStatement<LANGUAGE_PHP>(const TSNode& node) {
   return IsTSNodeofType(node, "if_statement");
 }
+template <> inline bool IsIfStatement<LANGUAGE_SOLIDITY>(const TSNode& node) {
+  return IsTSNodeofType(node, "if_statement");
+}
 template <> inline bool IsIfStatement<LANGUAGE_CPP>(const TSNode& node) {
   return IsTSNodeofType(node, "if_statement");
 }
@@ -214,6 +223,13 @@ inline TSNode GetIfConditionNode<LANGUAGE_PHP>(const TSNode& if_statement) {
                       kIfCondition.c_str(), kIfCondition.length());
 }
 
+template <>
+inline TSNode GetIfConditionNode<LANGUAGE_SOLIDITY>(const TSNode& if_statement) {
+  const std::string& kIfCondition = "condition";
+  return ts_node_child_by_field_name(if_statement,
+                      kIfCondition.c_str(), kIfCondition.length());
+}
+
 std::string OriginalSourceExpression(const TSNode&, const std::string&);
 
 template <Language L>

src/train_and_scan_util.cpp

@@ -30,6 +30,8 @@ template int TrainAndScanUtil::ScanFile<LANGUAGE_VERILOG>(
   const std::string& test_file, std::ostream& log_file) const;
 template int TrainAndScanUtil::ScanFile<LANGUAGE_PHP>(
   const std::string& test_file, std::ostream& log_file) const;
+template int TrainAndScanUtil::ScanFile<LANGUAGE_SOLIDITY>(
+  const std::string& test_file, std::ostream& log_file) const;
 template int TrainAndScanUtil::ScanFile<LANGUAGE_CPP>(
   const std::string& test_file, std::ostream& log_file) const;
 template int TrainAndScanUtil::ScanExpression<LANGUAGE_C>(
@@ -38,6 +40,8 @@ template int TrainAndScanUtil::ScanExpression<LANGUAGE_VERILOG>(
   const std::string& expression, std::ostream& log_file) const;
 template int TrainAndScanUtil::ScanExpression<LANGUAGE_PHP>(
   const std::string& expression, std::ostream& log_file) const;
+template int TrainAndScanUtil::ScanExpression<LANGUAGE_SOLIDITY>(
+  const std::string& expression, std::ostream& log_file) const;
 template int TrainAndScanUtil::ScanExpression<LANGUAGE_CPP>(
   const std::string& expression, std::ostream& log_file) const;
 

src/tree-sitter/CMakeLists.txt

@@ -36,6 +36,15 @@ target_include_directories(tree-sitter-php
   tree-sitter-php/src
 )
 
+add_library(tree-sitter-solidity STATIC
+  tree-sitter-solidity/src/parser.c
+)
+
+target_include_directories(tree-sitter-solidity
+  PRIVATE
+  tree-sitter-solidity/src
+)
+
 add_library(tree-sitter-cpp STATIC
   tree-sitter-cpp/src/parser.c
   tree-sitter-cpp/src/scanner.cc

src/tree_abstraction.h

@@ -147,6 +147,12 @@ inline std::string NodeToString<LEVEL_MIN, LANGUAGE_PHP>(
   return NodeToString<LEVEL_MIN, LANGUAGE_C>(conditional_expression);
 }
 
+template <>
+inline std::string NodeToString<LEVEL_MIN, LANGUAGE_SOLIDITY>(
+    const TSNode& conditional_expression) {
+  return NodeToString<LEVEL_MIN, LANGUAGE_C>(conditional_expression);
+}
+
 // ---------------------------------------------------------------------------
 
 inline std::string AbstractTerminalString(const TSNode& node) {
@@ -405,6 +411,12 @@ inline std::string NodeToString<LEVEL_TWO, LANGUAGE_PHP>(
   const TSNode& conditional_expression) {
     return NodeToString<LEVEL_MIN, LANGUAGE_PHP>(conditional_expression);
 }
+
+template <>
+inline std::string NodeToString<LEVEL_TWO, LANGUAGE_SOLIDITY>(
+  const TSNode& conditional_expression) {
+    return NodeToString<LEVEL_MIN, LANGUAGE_SOLIDITY>(conditional_expression);
+}
 // -----------------------------------------------------------------------
 // Close to full-detailed level with using Tree-sitter print. Only
 // difference is in printing operators for binary and unary ops.
@@ -456,6 +468,12 @@ inline std::string NodeToString<LEVEL_ONE, LANGUAGE_PHP>(
     return NodeToString<LEVEL_ONE, LANGUAGE_C>(conditional_expression);
 }
 
+template <>
+inline std::string NodeToString<LEVEL_ONE, LANGUAGE_SOLIDITY>(
+  const TSNode& conditional_expression) {
+    return NodeToString<LEVEL_ONE, LANGUAGE_C>(conditional_expression);
+}
+
 template <>
 inline std::string NodeToString<LEVEL_ONE, LANGUAGE_CPP>(
   const TSNode& conditional_expression) {

tests/CMakeLists.txt

@@ -43,6 +43,7 @@ foreach(file ${files})
       tree-sitter
       tree-sitter-c
       tree-sitter-php
+      tree-sitter-solidity
       tree-sitter-cpp
       tree-sitter-verilog
       pthread)