Solidity parser, can not parse much yet.

libsolidity/AST.cpp

@@ -0,0 +1,23 @@
+/*
+    This file is part of cpp-ethereum.
+
+    cpp-ethereum is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    cpp-ethereum is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with cpp-ethereum.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * @author Christian <c@ethdev.com>
+ * @date 2014
+ * Solidity abstract syntax tree.
+ */
+
+#include <libsolidity/AST.h>

libsolidity/AST.h

@@ -0,0 +1,287 @@
+/*
+    This file is part of cpp-ethereum.
+
+    cpp-ethereum is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    cpp-ethereum is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with cpp-ethereum.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * @author Christian <c@ethdev.com>
+ * @date 2014
+ * Solidity abstract syntax tree.
+ */
+
+#pragma once
+
+#include <string>
+#include <vector>
+#include <memory>
+
+#include <libsolidity/BaseTypes.h>
+#include <libsolidity/Token.h>
+
+namespace dev {
+namespace solidity {
+
+template <class T>
+using ptr = std::shared_ptr<T>;
+template <class T>
+using vecptr = std::vector<ptr<T>>;
+
+class VariableDeclaration;
+class StructDefinition;
+class FunctionDefinition;
+class TypeName;
+class Block;
+class Expression;
+
+class ASTNode
+{
+public:
+    explicit ASTNode(const Location& _location)
+        : m_location(_location)
+    {}
+private:
+    Location m_location;
+};
+
+class ContractDefinition : public ASTNode
+{
+public:
+    ContractDefinition(const Location& _location,
+                       const std::string& _name,
+                       const vecptr<StructDefinition>& _definedStructs,
+                       const vecptr<VariableDeclaration>& _stateVariables,
+                       const vecptr<FunctionDefinition>& _definedFunctions)
+        : ASTNode(_location),
+          m_name(_name),
+          m_definedStructs(_definedStructs),
+          m_stateVariables(_stateVariables),
+          m_definedFunctions(_definedFunctions)
+    {}
+
+private:
+    std::string m_name;
+    vecptr<StructDefinition> m_definedStructs;
+    vecptr<VariableDeclaration> m_stateVariables;
+    vecptr<FunctionDefinition> m_definedFunctions;
+};
+
+class StructDefinition : public ASTNode
+{
+private:
+    std::string m_name;
+    vecptr<VariableDeclaration> m_members;
+};
+
+class FunctionDefinition : public ASTNode
+{
+private:
+    std::string m_name;
+    vecptr<VariableDeclaration> m_arguments;
+    bool m_isDeclaredConst;
+    vecptr<VariableDeclaration> m_returns;
+    ptr<Block> m_body;
+};
+
+class VariableDeclaration : public ASTNode
+{
+public:
+    VariableDeclaration(const Location& _location,
+                        const ptr<TypeName>& _type,
+                        const std::string& _name)
+        : ASTNode(_location),
+          m_type(_type),
+          m_name(_name)
+    {}
+private:
+    ptr<TypeName> m_type; ///<s can be empty ("var")
+    std::string m_name;
+};
+
+/// types
+/// @{
+
+class TypeName : public ASTNode
+{
+public:
+    explicit TypeName(const Location& _location)
+        : ASTNode(_location)
+    {}
+};
+
+/// any pre-defined type that is not a mapping
+class ElementaryTypeName : public TypeName
+{
+public:
+    explicit ElementaryTypeName(const Location& _location, Token::Value _type)
+        : TypeName(_location), m_type(_type)
+    {}
+private:
+    Token::Value m_type;
+};
+
+class UserDefinedTypeName : public TypeName
+{
+public:
+    UserDefinedTypeName(const Location& _location, const std::string& _name)
+        : TypeName(_location), m_name(_name)
+    {}
+private:
+    std::string m_name;
+};
+
+class MappingTypeName : public TypeName
+{
+public:
+    explicit MappingTypeName(const Location& _location)
+        : TypeName(_location)
+    {}
+private:
+    ptr<ElementaryTypeName> m_keyType;
+    ptr<TypeName> m_valueType;
+};
+
+/// @}
+
+/// Statements
+/// @{
+
+class Statement : public ASTNode
+{
+};
+
+class Block : public Statement
+{
+private:
+    vecptr<Statement> m_statements;
+};
+
+class IfStatement : public Statement
+{
+
+private:
+    ptr<Expression> m_condition;
+    ptr<Statement> m_trueBody;
+    ptr<Statement> m_falseBody;
+};
+
+class BreakableStatement : public Statement
+{
+
+};
+
+class WhileStatement : public BreakableStatement
+{
+private:
+    ptr<Expression> m_condition;
+    ptr<Statement> m_body;
+};
+
+class Continue : public Statement
+{
+
+};
+
+class Break : public Statement
+{
+
+};
+
+class Return : public Statement
+{
+private:
+    ptr<Expression> m_expression;
+};
+
+class VariableAssignment : public Statement
+{
+private:
+    ptr<VariableDeclaration> m_variable;
+    Token::Value m_assigmentOperator;
+    ptr<Expression> m_rightHandSide; ///< can be missing
+};
+
+class Expression : public Statement
+{
+private:
+};
+
+/// @}
+
+/// Expressions
+/// @{
+
+class Assignment : public Expression
+{
+private:
+    ptr<Expression> m_leftHandSide;
+    Token::Value m_assigmentOperator;
+    ptr<Expression> m_rightHandSide;
+};
+
+class UnaryOperation : public Expression
+{
+private:
+    Token::Value m_operator;
+    ptr<Expression> m_subExpression;
+    bool isPrefix;
+};
+
+class BinaryOperation : public Expression
+{
+private:
+    ptr<Expression> m_left;
+    ptr<Expression> m_right;
+    Token::Value m_operator;
+};
+
+class FunctionCall : public Expression
+{
+private:
+    std::string m_functionName; // TODO only calls to fixed, named functions for now
+    vecptr<Expression> m_arguments;
+};
+
+class MemberAccess : public Expression
+{
+private:
+    ptr<Expression> m_expression;
+    std::string m_memberName;
+};
+
+class IndexAccess : public Expression
+{
+    ptr<Expression> m_base;
+    ptr<Expression> m_index;
+};
+
+class PrimaryExpression : public Expression
+{
+};
+
+class Identifier : public PrimaryExpression
+{
+private:
+    std::string m_name;
+};
+
+class Literal : public PrimaryExpression
+{
+private:
+    std::string m_value;
+};
+
+/// @}
+
+
+} }

libsolidity/BaseTypes.h

@@ -0,0 +1,22 @@
+#pragma once
+
+
+namespace dev {
+namespace solidity {
+
+// Representation of an interval of source positions.
+struct Location {
+    Location(int b, int e) : beg_pos(b), end_pos(e) { }
+    Location() : beg_pos(0), end_pos(0) { }
+
+    bool IsValid() const {
+        return beg_pos >= 0 && end_pos >= beg_pos;
+    }
+
+    static Location invalid() { return Location(-1, -1); }
+
+    int beg_pos;
+    int end_pos;
+};
+
+} }

libsolidity/Parser.cpp

@@ -0,0 +1,180 @@
+/*
+    This file is part of cpp-ethereum.
+
+    cpp-ethereum is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    cpp-ethereum is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with cpp-ethereum.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * @author Christian <c@ethdev.com>
+ * @date 2014
+ * Solidity parser.
+ */
+
+#include "libsolidity/BaseTypes.h"
+#include "libsolidity/Parser.h"
+#include "libsolidity/Scanner.h"
+
+namespace dev {
+namespace solidity {
+
+ptr<ASTNode> Parser::parse(Scanner& _scanner)
+{
+    m_scanner = &_scanner;
+
+    return parseContractDefinition();
+}
+
+
+/// AST node factory that also tracks the begin and end position of an AST node
+/// while it is being parsed
+class Parser::ASTNodeFactory
+{
+public:
+    ASTNodeFactory(const Parser& _parser)
+        : m_parser(_parser),
+          m_location(_parser.getPosition(), -1)
+    {}
+
+    void markEndPosition()
+    {
+        m_location.end_pos = m_parser.getEndPosition();
+    }
+
+    /// @todo: check that this actually uses perfect forwarding
+    template <class NodeType, typename... Args>
+    ptr<NodeType> createNode(Args&&... _args)
+    {
+        if (m_location.end_pos < 0) markEndPosition();
+        return std::make_shared<NodeType>(m_location, std::forward<Args>(_args)...);
+    }
+
+private:
+    const Parser& m_parser;
+    Location m_location;
+};
+
+int Parser::getPosition() const
+{
+    return m_scanner->getCurrentLocation().beg_pos;
+}
+
+int Parser::getEndPosition() const
+{
+    return m_scanner->getCurrentLocation().end_pos;
+}
+
+
+ptr<ContractDefinition> Parser::parseContractDefinition()
+{
+    ASTNodeFactory nodeFactory(*this);
+
+    expectToken(Token::CONTRACT);
+    std::string name = expectIdentifier();
+    expectToken(Token::LBRACE);
+
+    vecptr<StructDefinition> structs;
+    vecptr<VariableDeclaration> stateVariables;
+    vecptr<FunctionDefinition> functions;
+    bool visibilityIsPublic = true;
+    while (true) {
+        Token::Value currentToken = m_scanner->getCurrentToken();
+        if (currentToken == Token::RBRACE) {
+            break;
+        } else if (currentToken == Token::PUBLIC || currentToken == Token::PRIVATE) {
+            visibilityIsPublic = (m_scanner->getCurrentToken() == Token::PUBLIC);
+            m_scanner->next();
+            expectToken(Token::COLON);
+        } else if (currentToken == Token::FUNCTION) {
+            functions.push_back(parseFunctionDefinition(visibilityIsPublic));
+        } else if (currentToken == Token::STRUCT) {
+            structs.push_back(parseStructDefinition());
+            expectToken(Token::SEMICOLON);
+        } else if (currentToken == Token::IDENTIFIER || currentToken == Token::MAPPING ||
+                   Token::IsElementaryTypeName(currentToken)) {
+            stateVariables.push_back(parseVariableDeclaration());
+            expectToken(Token::SEMICOLON);
+        } else {
+            throwExpectationError("Function, variable or struct declaration expected.");
+        }
+    }
+    nodeFactory.markEndPosition();
+
+    m_scanner->next();
+    expectToken(Token::EOS);
+
+    return nodeFactory.createNode<ContractDefinition>(name, structs, stateVariables, functions);
+}
+
+ptr<FunctionDefinition> Parser::parseFunctionDefinition(bool _isPublic)
+{
+    (void) _isPublic;
+    throwExpectationError("Function parsing is not yet implemented.");
+}
+
+ptr<StructDefinition> Parser::parseStructDefinition()
+{
+    throwExpectationError("Struct definition parsing is not yet implemented.");
+}
+
+ptr<VariableDeclaration> Parser::parseVariableDeclaration()
+{
+    ASTNodeFactory nodeFactory(*this);
+
+    ptr<TypeName> type;
+    Token::Value token = m_scanner->getCurrentToken();
+    if (Token::IsElementaryTypeName(token)) {
+        type = ASTNodeFactory(*this).createNode<ElementaryTypeName>(token);
+        m_scanner->next();
+    } else if (token == Token::VAR) {
+        type = ASTNodeFactory(*this).createNode<TypeName>();
+        m_scanner->next();
+    } else if (token == Token::MAPPING) {
+        // TODO
+        throwExpectationError("mappings are not yet implemented");
+    } else if (token == Token::IDENTIFIER) {
+        type = ASTNodeFactory(*this).createNode<UserDefinedTypeName>(m_scanner->getCurrentLiteral());
+        m_scanner->next();
+    } else {
+        throwExpectationError("Expected variable declaration");
+    }
+    nodeFactory.markEndPosition();
+    std::string name = expectIdentifier();
+    return nodeFactory.createNode<VariableDeclaration>(type, name);
+}
+
+void Parser::expectToken(Token::Value _value)
+{
+    if (m_scanner->getCurrentToken() != _value)
+        throwExpectationError(std::string("Expected token ") + std::string(Token::Name(_value)));
+    m_scanner->next();
+}
+
+std::string Parser::expectIdentifier()
+{
+    if (m_scanner->getCurrentToken() != Token::IDENTIFIER)
+        throwExpectationError("Expected identifier");
+
+    std::string literal = m_scanner->getCurrentLiteral();
+    m_scanner->next();
+    return literal;
+}
+
+void Parser::throwExpectationError(const std::string& _description)
+{
+    (void) _description;
+    /// @todo make a proper exception hierarchy
+    throw std::exception();//_description);
+}
+
+
+} }

libsolidity/Parser.h

@@ -0,0 +1,64 @@
+/*
+    This file is part of cpp-ethereum.
+
+    cpp-ethereum is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    cpp-ethereum is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with cpp-ethereum.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * @author Christian <c@ethdev.com>
+ * @date 2014
+ * Solidity parser.
+ */
+
+#pragma once
+
+#include "libsolidity/AST.h"
+
+namespace dev {
+namespace solidity {
+
+class Scanner;
+
+class Parser
+{
+public:
+    ptr<ASTNode> parse(Scanner& _scanner);
+
+private:
+    class ASTNodeFactory;
+
+    /// Start position of the current token
+    int getPosition() const;
+    /// End position of the current token
+    int getEndPosition() const;
+
+    /// Parsing functions for the AST nodes
+    /// @{
+    ptr<ContractDefinition> parseContractDefinition();
+    ptr<FunctionDefinition> parseFunctionDefinition(bool _isPublic);
+    ptr<StructDefinition> parseStructDefinition();
+    ptr<VariableDeclaration> parseVariableDeclaration();
+    /// @}
+
+    /// Helper functions
+    /// @{
+    /// If current token value is not _value, throw exception otherwise advance token.
+    void expectToken(Token::Value _value);
+    std::string expectIdentifier();
+    void throwExpectationError(const std::string& _description);
+    /// @}
+
+    Scanner* m_scanner;
+};
+
+} }

libsolidity/Scanner.cpp

@@ -82,16 +82,10 @@ void Scanner::reset(const CharStream& _source)
 {
     m_source = _source;
 
-    // Initialize current_ to not refer to a literal.
-    m_current_token.token = Token::ILLEGAL;
-    m_current_token.literal.clear();
-
-    m_hasLineTerminatorBeforeNext = true;
-    m_hasMultilineCommentBeforeNext = false;
-
     m_char = m_source.get();
     skipWhitespace();
     scanToken();
+    next();
 }
 
 
@@ -466,7 +460,7 @@ Token::Value Scanner::scanString()
     literal.Complete();
 
     advance();  // consume quote
-    return Token::STRING;
+    return Token::STRING_LITERAL;
 }
 
 
@@ -551,13 +545,17 @@ Token::Value Scanner::scanNumber(bool _periodSeen)
 // Keyword Matcher
 
 #define KEYWORDS(KEYWORD_GROUP, KEYWORD)                                     \
+  KEYWORD_GROUP('a')                                                         \
+  KEYWORD("address", Token::BREAK)                                           \
   KEYWORD_GROUP('b')                                                         \
   KEYWORD("break", Token::BREAK)                                             \
+  KEYWORD("bool", Token::BOOL)                                               \
   KEYWORD_GROUP('c')                                                         \
   KEYWORD("case", Token::CASE)                                               \
   KEYWORD("catch", Token::CATCH)                                             \
   KEYWORD("const", Token::CONST)                                             \
   KEYWORD("continue", Token::CONTINUE)                                       \
+  KEYWORD("contract", Token::CONTRACT)                                       \
   KEYWORD_GROUP('d')                                                         \
   KEYWORD("debugger", Token::DEBUGGER)                                       \
   KEYWORD("default", Token::DEFAULT)                                         \
@@ -571,31 +569,55 @@ Token::Value Scanner::scanNumber(bool _periodSeen)
   KEYWORD("finally", Token::FINALLY)                                         \
   KEYWORD("for", Token::FOR)                                                 \
   KEYWORD("function", Token::FUNCTION)                                       \
+  KEYWORD_GROUP('h')                                                         \
+  KEYWORD("hash", Token::HASH)                                               \
+  KEYWORD("hash32", Token::HASH32)                                           \
+  KEYWORD("hash64", Token::HASH64)                                           \
+  KEYWORD("hash128", Token::HASH128)                                         \
+  KEYWORD("hash256", Token::HASH256)                                         \
   KEYWORD_GROUP('i')                                                         \
   KEYWORD("if", Token::IF)                                                   \
   KEYWORD("implements", Token::FUTURE_STRICT_RESERVED_WORD)                  \
   KEYWORD("in", Token::IN)                                                   \
   KEYWORD("instanceof", Token::INSTANCEOF)                                   \
+  KEYWORD("int", Token::INT)                                                 \
+  KEYWORD("int32", Token::INT32)                                             \
+  KEYWORD("int64", Token::INT64)                                             \
+  KEYWORD("int128", Token::INT128)                                           \
+  KEYWORD("int256", Token::INT256)                                           \
   KEYWORD("interface", Token::FUTURE_STRICT_RESERVED_WORD)                   \
   KEYWORD_GROUP('l')                                                         \
+  KEYWORD_GROUP('m')                                                         \
   KEYWORD_GROUP('n')                                                         \
+  KEYWORD("mapping", Token::MAPPING)                                         \
   KEYWORD("new", Token::NEW)                                                 \
   KEYWORD("null", Token::NULL_LITERAL)                                       \
   KEYWORD_GROUP('p')                                                         \
   KEYWORD("package", Token::FUTURE_STRICT_RESERVED_WORD)                     \
-  KEYWORD("private", Token::FUTURE_STRICT_RESERVED_WORD)                     \
+  KEYWORD("private", Token::PRIVATE)                                         \
   KEYWORD("protected", Token::FUTURE_STRICT_RESERVED_WORD)                   \
-  KEYWORD("public", Token::FUTURE_STRICT_RESERVED_WORD)                      \
+  KEYWORD("public", Token::PUBLIC)                                           \
   KEYWORD_GROUP('r')                                                         \
+  KEYWORD("real", Token::REAL)                                               \
   KEYWORD("return", Token::RETURN)                                           \
   KEYWORD_GROUP('s')                                                         \
+  KEYWORD("string", Token::STRING_TYPE)                                           \
+  KEYWORD("struct", Token::STRUCT)                                           \
   KEYWORD("switch", Token::SWITCH)                                           \
   KEYWORD_GROUP('t')                                                         \
+  KEYWORD("text", Token::TEXT)                                               \
   KEYWORD("this", Token::THIS)                                               \
   KEYWORD("throw", Token::THROW)                                             \
   KEYWORD("true", Token::TRUE_LITERAL)                                       \
   KEYWORD("try", Token::TRY)                                                 \
   KEYWORD("typeof", Token::TYPEOF)                                           \
+  KEYWORD_GROUP('u')                                                         \
+  KEYWORD("uint", Token::UINT)                                               \
+  KEYWORD("uint32", Token::UINT32)                                           \
+  KEYWORD("uint64", Token::UINT64)                                           \
+  KEYWORD("uint128", Token::UINT128)                                         \
+  KEYWORD("uint256", Token::UINT256)                                         \
+  KEYWORD("ureal", Token::UREAL)                                             \
   KEYWORD_GROUP('v')                                                         \
   KEYWORD("var", Token::VAR)                                                 \
   KEYWORD("void", Token::VOID)                                               \

libsolidity/Scanner.h

@@ -47,6 +47,7 @@
 #include <libdevcore/Common.h>
 #include <libdevcore/Log.h>
 #include <libdevcore/CommonData.h>
+#include <libsolidity/BaseTypes.h>
 #include <libsolidity/Token.h>
 
 namespace dev {
@@ -111,21 +112,6 @@ public:
         bool complete_;
     };
 
-    // Representation of an interval of source positions.
-    struct Location {
-        Location(int b, int e) : beg_pos(b), end_pos(e) { }
-        Location() : beg_pos(0), end_pos(0) { }
-
-        bool IsValid() const {
-            return beg_pos >= 0 && end_pos >= beg_pos;
-        }
-
-        static Location invalid() { return Location(-1, -1); }
-
-        int beg_pos;
-        int end_pos;
-    };
-
     explicit Scanner(const CharStream& _source);
 
     // Resets the scanner as if newly constructed with _input as input.
@@ -163,8 +149,6 @@ private:
         std::string literal;
     };
 
-    static const int kCharacterLookaheadBufferSize = 1;
-
     // Literal buffer support
     inline void startNewLiteral() {
         m_next_token.literal.clear();

libsolidity/Token.h

@@ -152,6 +152,7 @@ namespace solidity {
   K(CASE, "case", 0)                                                 \
   K(CATCH, "catch", 0)                                               \
   K(CONTINUE, "continue", 0)                                         \
+  K(CONTRACT, "contract", 0)                                         \
   K(DEBUGGER, "debugger", 0)                                         \
   K(DEFAULT, "default", 0)                                           \
   /* DELETE */                                                       \
@@ -163,8 +164,12 @@ namespace solidity {
   K(IF, "if", 0)                                                     \
   /* IN */                                                           \
   /* INSTANCEOF */                                                   \
+  K(MAPPING, "mapping", 0)                                           \
   K(NEW, "new", 0)                                                   \
+  K(PUBLIC, "public", 0)                                             \
+  K(PRIVATE, "private", 0)                                           \
   K(RETURN, "return", 0)                                             \
+  K(STRUCT, "struct", 0)                                             \
   K(SWITCH, "switch", 0)                                             \
   K(THIS, "this", 0)                                                 \
   K(THROW, "throw", 0)                                               \
@@ -175,12 +180,36 @@ namespace solidity {
   K(WHILE, "while", 0)                                               \
   K(WITH, "with", 0)                                                 \
                                                                      \
+  /* type keywords, keep them in this order, keep int as first keyword TODO more to be added */ \
+  K(INT, "int", 0)                                                   \
+  K(INT32, "int32", 0)                                               \
+  K(INT64, "int64", 0)                                               \
+  K(INT128, "int128", 0)                                             \
+  K(INT256, "int256", 0)                                             \
+  K(UINT, "uint", 0)                                                 \
+  K(UINT32, "uint32", 0)                                             \
+  K(UINT64, "uint64", 0)                                             \
+  K(UINT128, "uint128", 0)                                           \
+  K(UINT256, "uint256", 0)                                           \
+  K(HASH, "hash", 0)                                                 \
+  K(HASH32, "hash32", 0)                                             \
+  K(HASH64, "hash64", 0)                                             \
+  K(HASH128, "hash128", 0)                                           \
+  K(HASH256, "hash256", 0)                                           \
+  K(ADDRESS, "address", 0)                                           \
+  K(BOOL, "bool", 0)                                                 \
+  K(STRING_TYPE, "string", 0)                                        \
+  K(TEXT, "text", 0)                                                 \
+  K(REAL, "real", 0)                                                 \
+  K(UREAL, "ureal", 0)                                               \
+  T(TYPES_END, NULL, 0) /* used as type enum end marker */           \
+                                                                     \
   /* Literals (ECMA-262, section 7.8, page 16). */                   \
   K(NULL_LITERAL, "null", 0)                                         \
   K(TRUE_LITERAL, "true", 0)                                         \
   K(FALSE_LITERAL, "false", 0)                                       \
   T(NUMBER, NULL, 0)                                                 \
-  T(STRING, NULL, 0)                                                 \
+  T(STRING_LITERAL, NULL, 0)                                         \
                                                                      \
   /* Identifiers (not keywords or future reserved words). */         \
   T(IDENTIFIER, NULL, 0)                                             \
@@ -231,6 +260,10 @@ class Token {
       return tok == IDENTIFIER;
   }
 
+  static bool IsElementaryTypeName(Value tok) {
+      return INT <= tok && tok < TYPES_END;
+  }
+
   static bool IsAssignmentOp(Value tok) {
     return INIT_VAR <= tok && tok <= ASSIGN_MOD;
   }

libsolidity/grammar.txt

@@ -0,0 +1,32 @@
+ContractDefinition = 'contract' Identifier '{' ContractPart* '}'
+ContractPart = VariableDeclaration ';' | StructDefinition ';' |
+               FunctionDefinition ';' | 'public:' | 'private:'
+
+StructDefinition = 'struct' Identifier '{'
+                     ( VariableDeclaration (';' VariableDeclaration)* )? '}
+
+FunctionDefinition = 'function' Identifier ArgumentList 'const'?
+                     'returns' ArgumentList Block
+ArgumentList = '(' ( VariableDeclaration (',' VariableDeclaration)* )? ')'
+// semantic restriction: mappings and structs (recursively) containing mappings
+// are not allowed in argument lists
+VariableDeclaration = TypeName Identifier
+TypeName = PredefinedType | Identifier | MappingType
+MappingType = 'mapping' '(' SimplePredefinedType '=>' TypeName ')'
+
+Block = '{' Statement* '}'
+Statement = IfStatement | WhileStatement | Continue | Break | Return | VariableAssignment | Expression ';' | Block
+
+IfStatement = 'if' '(' Expression ')' Statement ( 'else' Statement )?
+WhileStatement = 'while' '(' Expression ')' Statement
+Continue = 'continue' ';'
+Break = 'break' ';'
+Return = 'return' Expression? ';'
+VariableAssignment = VariableDeclaration ( AssignmentOp Expression )? ';'
+
+Expression = Assignment | UnaryOperation | BinaryOperation | FunctionCall | IndexAccess | MemberAccess | PrimaryExpression
+Assignment = Expression (AssignmentOp Expression)
+FunctionCall = Identifier '(' ( Expression ( ',' Expression )* ) ')'
+MemberAccess = Expression '.' Identifier
+IndexAccess = Expression '[' Expresison ']'
+PrimaryExpression = Identifier | NumberLiteral | StringLiteral | '(' Expression ')'

test/solidityParser.cpp

@@ -0,0 +1,52 @@
+/*
+    This file is part of cpp-ethereum.
+
+    cpp-ethereum is free software: you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation, either version 3 of the License, or
+    (at your option) any later version.
+
+    cpp-ethereum is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License
+    along with cpp-ethereum.  If not, see <http://www.gnu.org/licenses/>.
+*/
+/**
+ * @author Christian <c@ethdev.com>
+ * @date 2014
+ * Unit tests for the solidity parser.
+ */
+
+#include <string>
+
+#include <libsolidity/Scanner.h>
+#include <libsolidity/Parser.h>
+#include <boost/test/unit_test.hpp>
+
+namespace dev {
+namespace solidity {
+namespace test {
+
+BOOST_AUTO_TEST_SUITE(SolidityParser)
+
+BOOST_AUTO_TEST_CASE(smoke_test)
+{
+    std::string text = "contract test123 {\n"
+                       "\tuint256 stateVariable1;\n"
+                       "}\n";
+    Parser parser;
+    CharStream str(text);
+    // @todo: figure out why this does not compile
+    //Scanner scanner(CharStream(text));
+    Scanner scanner(str);
+    BOOST_CHECK_NO_THROW(parser.parse(scanner));
+}
+
+
+BOOST_AUTO_TEST_SUITE_END()
+
+} } } // end namespaces
+

test/solidityScanner.cpp

@@ -27,21 +27,26 @@ namespace dev {
 namespace solidity {
 namespace test {
 
-BOOST_AUTO_TEST_SUITE(solidity)
+BOOST_AUTO_TEST_SUITE(SolidityScanner)
+
+BOOST_AUTO_TEST_CASE(test_empty)
+{
+    Scanner scanner(CharStream(""));
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::EOS);
+}
 
 BOOST_AUTO_TEST_CASE(smoke_test)
 {
     Scanner scanner(CharStream("function break;765  \t  \"string1\",'string2'\nidentifier1"));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::FUNCTION);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::FUNCTION);
     BOOST_CHECK_EQUAL(scanner.next(), Token::BREAK);
     BOOST_CHECK_EQUAL(scanner.next(), Token::SEMICOLON);
     BOOST_CHECK_EQUAL(scanner.next(), Token::NUMBER);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "765");
-    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING);
+    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING_LITERAL);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "string1");
     BOOST_CHECK_EQUAL(scanner.next(), Token::COMMA);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING);
+    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING_LITERAL);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "string2");
     BOOST_CHECK_EQUAL(scanner.next(), Token::IDENTIFIER);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "identifier1");
@@ -51,26 +56,23 @@ BOOST_AUTO_TEST_CASE(smoke_test)
 BOOST_AUTO_TEST_CASE(string_escapes)
 {
     Scanner scanner(CharStream("  { \"a\\x61\""));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::LBRACE);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::LBRACE);
+    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING_LITERAL);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "aa");
 }
 
 BOOST_AUTO_TEST_CASE(string_escapes_with_zero)
 {
     Scanner scanner(CharStream("  { \"a\\x61\\x00abc\""));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::LBRACE);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::LBRACE);
+    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING_LITERAL);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::STRING);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), std::string("aa\0abc", 6));
 }
 
 BOOST_AUTO_TEST_CASE(string_escape_illegal)
 {
     Scanner scanner(CharStream(" bla \"\\x6rf\" (illegalescape)"));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::IDENTIFIER);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::IDENTIFIER);
     BOOST_CHECK_EQUAL(scanner.next(), Token::ILLEGAL);
     BOOST_CHECK_EQUAL(scanner.getCurrentLiteral(), "");
     // TODO recovery from illegal tokens should be improved
@@ -83,8 +85,7 @@ BOOST_AUTO_TEST_CASE(string_escape_illegal)
 BOOST_AUTO_TEST_CASE(hex_numbers)
 {
     Scanner scanner(CharStream("var x = 0x765432536763762734623472346;"));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::VAR);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::VAR);
     BOOST_CHECK_EQUAL(scanner.next(), Token::IDENTIFIER);
     BOOST_CHECK_EQUAL(scanner.next(), Token::ASSIGN);
     BOOST_CHECK_EQUAL(scanner.next(), Token::NUMBER);
@@ -96,8 +97,7 @@ BOOST_AUTO_TEST_CASE(hex_numbers)
 BOOST_AUTO_TEST_CASE(locations)
 {
     Scanner scanner(CharStream("function_identifier has ; -0x743/*comment*/\n ident //comment"));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::IDENTIFIER);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::IDENTIFIER);
     BOOST_CHECK_EQUAL(scanner.getCurrentLocation().beg_pos, 0);
     BOOST_CHECK_EQUAL(scanner.getCurrentLocation().end_pos, 19);
     BOOST_CHECK_EQUAL(scanner.next(), Token::IDENTIFIER);
@@ -122,8 +122,7 @@ BOOST_AUTO_TEST_CASE(ambiguities)
 {
     // test scanning of some operators which need look-ahead
     Scanner scanner(CharStream("<=""<""+ +=a++ =>""<<"));
-    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::ILLEGAL);
+    BOOST_CHECK_EQUAL(scanner.getCurrentToken(), Token::LTE);
-    BOOST_CHECK_EQUAL(scanner.next(), Token::LTE);
     BOOST_CHECK_EQUAL(scanner.next(), Token::LT);
     BOOST_CHECK_EQUAL(scanner.next(), Token::ADD);
     BOOST_CHECK_EQUAL(scanner.next(), Token::ASSIGN_ADD);