Add unary expression parsing

Signed-off-by: CHatingPython <chatingpython@gmail.com>

furc/include/furc/ast/expression.hpp

@@ -9,7 +9,8 @@ namespace ast {
 
 enum class expression_node_t {
     Literal,
-    Binop
+    Unaryop,
+    Binop,
 };
 
 class expression_node : public statement_node {
@@ -25,6 +26,37 @@ protected:
 
 using expression_node_h = node_handle<expression_node>;
 
+enum class unaryop_expression_node_t {
+    Positive,
+    Negative,
+    PrefixIncrement,
+    PostfixIncrement,
+    PrefixDecrement,
+    PostfixDecrement,
+};
+
+class unaryop_expression_node : public expression_node {
+public:
+    unaryop_expression_node(unaryop_expression_node_t type, expression_node_h&& node)
+      : m_type(type), m_node(std::move(node)) {}
+
+    void set_node(expression_node_h&& node) { m_node = std::move(node); }
+
+    unaryop_expression_node_t type() const { return m_type; }
+    const expression_node_h&  get_node() const { return m_node; }
+    expression_node_h&        get_node() { return m_node; }
+    expression_node_h&&       move_node() { return std::move(m_node); }
+public:
+    expression_node_t expression_type() const override { return expression_node_t::Unaryop; }
+
+    std::ostream& print(std::ostream& os) const override;
+private:
+    unaryop_expression_node_t m_type;
+    expression_node_h         m_node;
+};
+
+using unaryop_expression_node_h = node_handle<unaryop_expression_node>;
+
 enum class binop_expression_node_t {
     Add,
     Sub,
@@ -40,7 +72,11 @@ public:
 
     binop_expression_node_t  type() const { return m_type; };
     const expression_node_h& lhs() const { return m_lhs; };
+    expression_node_h&       lhs() { return m_lhs; };
+    expression_node_h&&      move_lhs() { return std::move(m_lhs); };
     const expression_node_h& rhs() const { return m_rhs; };
+    expression_node_h&       rhs() { return m_rhs; };
+    expression_node_h&&      move_rhs() { return std::move(m_rhs); };
 public:
     expression_node_t expression_type() const override { return expression_node_t::Binop; }
 

furc/include/furc/front/parser.hpp

@@ -33,7 +33,8 @@ private:
     ast::literal_node_h     parse_literal();
 
     ast::expression_node_h parse_expression_primary();
-    ast::expression_node_h parse_expression_rhs(const ast::expression_node_h& init, std::uint32_t precedence);
+    ast::expression_node_h parse_expression_unary(std::uint32_t precedence);
+    ast::expression_node_h parse_expression_rhs(ast::expression_node_h&& init, std::uint32_t precedence);
 
     ast::function_body_h parse_body();
 private:

furc/include/furc/front/token.hpp

@@ -32,6 +32,8 @@ enum class token_t {
     Star,
     Slash,
     Percent,
+    DPlus,
+    DMinus,
 };
 
 static inline std::ostream& operator<<(std::ostream& os, token_t type) {
@@ -56,6 +58,8 @@ static inline std::ostream& operator<<(std::ostream& os, token_t type) {
     case token_t::Star: return os << "'*'";
     case token_t::Slash: return os << "'/'";
     case token_t::Percent: return os << "'%'";
+    case token_t::DPlus: return os << "++";
+    case token_t::DMinus: return os << "--";
     }
     return os;
 }
@@ -82,6 +86,8 @@ static inline std::string operator+(const std::string& str, token_t type) {
     case token_t::Star: return str + "'*'";
     case token_t::Slash: return str + "'/'";
     case token_t::Percent: return str + "'%'";
+    case token_t::DPlus: return str + "++";
+    case token_t::DMinus: return str + "--";
     }
     return str;
 }
@@ -167,7 +173,9 @@ struct token {
         case token_t::Minus:
         case token_t::Star:
         case token_t::Slash:
-        case token_t::Percent: return true;
+        case token_t::Percent:
+        case token_t::DPlus:
+        case token_t::DMinus: return true;
         }
     }
 };

furc/src/ast.cpp

@@ -14,7 +14,7 @@ bool literal_node::equal(const node& rhs) const {
 
 std::ostream& integer_literal_node::print(std::ostream& os) const {
     if (m_value.has_error()) return os << m_value.error();
-    return os << "integer literal (" << *m_value << ")";
+    return os << *m_value;
 }
 
 bool integer_literal_node::equal(const node& rhs) const {
@@ -23,7 +23,7 @@ bool integer_literal_node::equal(const node& rhs) const {
 
 std::ostream& string_literal_node::print(std::ostream& os) const {
     if (m_value.has_error()) return os << m_value.error();
-    return os << "string literal (" << *m_value << ")";
+    return os << '"' << *m_value << '"';
 }
 
 bool string_literal_node::equal(const node& rhs) const {
@@ -34,6 +34,30 @@ bool expression_node::equal(const node& rhs) const {
     return expression_type() == reinterpret_cast<const expression_node&>(rhs).expression_type();
 }
 
+std::ostream& operator<<(std::ostream& os, unaryop_expression_node_t type) {
+    switch (type) {
+    case unaryop_expression_node_t::Positive: return os << "+";
+    case unaryop_expression_node_t::Negative: return os << "-";
+    case unaryop_expression_node_t::PrefixIncrement:
+    case unaryop_expression_node_t::PostfixIncrement: return os << "++";
+    case unaryop_expression_node_t::PrefixDecrement:
+    case unaryop_expression_node_t::PostfixDecrement: return os << "--";
+    }
+    return os;
+}
+
+std::ostream& unaryop_expression_node::print(std::ostream& os) const {
+    switch (m_type) {
+    case unaryop_expression_node_t::Positive:
+    case unaryop_expression_node_t::Negative:
+    case unaryop_expression_node_t::PrefixIncrement:
+    case unaryop_expression_node_t::PrefixDecrement: return os << '(' << m_type << *m_node << ')';
+    case unaryop_expression_node_t::PostfixIncrement:
+    case unaryop_expression_node_t::PostfixDecrement: return os << '(' << *m_node << m_type << ')';
+    }
+    return os;
+}
+
 std::ostream& operator<<(std::ostream& os, binop_expression_node_t type) {
     switch (type) {
     case binop_expression_node_t::Add: return os << '+';
@@ -45,7 +69,7 @@ std::ostream& operator<<(std::ostream& os, binop_expression_node_t type) {
 }
 
 std::ostream& binop_expression_node::print(std::ostream& os) const {
-    return os << *m_lhs << ' ' << m_type << ' ' << *m_rhs;
+    return os << '(' << *m_lhs << ' ' << m_type << ' ' << *m_rhs << ')';
 }
 
 bool binop_expression_node::equal(const node& rhsNode) const {
@@ -87,7 +111,7 @@ bool statement_node::equal(const node& rhs) const {
 
 std::ostream& return_statement_node::print(std::ostream& os) const {
     os << "return statement";
-    if (m_value.present()) return os << " (" << *m_value << ')';
+    if (m_value.present()) return os << ' ' << *m_value;
     return os;
 }
 

furc/src/front/lexer.cpp

@@ -2,6 +2,7 @@
 
 #include <cctype>
 #include <limits>
+#include <map>
 #include <string>
 #include <unordered_map>
 
@@ -45,23 +46,7 @@ token_handle<> lexer::next_token() {
 
     location location = current_location();
 
-    char ch = get();
+    switch (get()) {
-    switch (ch) {
-    case '(': ++m_cursor; return { location, token_t::Lparen };
-    case ')': ++m_cursor; return { location, token_t::Rparen };
-    case '{': ++m_cursor; return { location, token_t::Lbrace };
-    case '}': ++m_cursor; return { location, token_t::Rbrace };
-    case '[': ++m_cursor; return { location, token_t::Lbracket };
-    case ']': ++m_cursor; return { location, token_t::Rbracket };
-    case ';': ++m_cursor; return { location, token_t::Semicolon };
-    case ':': ++m_cursor; return { location, token_t::Colon };
-    case ',': ++m_cursor; return { location, token_t::Comma };
-    case '.': ++m_cursor; return { location, token_t::Dot };
-    case '+': ++m_cursor; return { location, token_t::Plus };
-    case '-': ++m_cursor; return { location, token_t::Minus };
-    case '*': ++m_cursor; return { location, token_t::Star };
-    case '/': ++m_cursor; return { location, token_t::Slash };
-    case '%': ++m_cursor; return { location, token_t::Percent };
     case '"': {
         std::size_t begin = ++m_cursor;
         while (m_cursor < m_content.size() && m_content[m_cursor] != '"')
@@ -73,7 +58,7 @@ token_handle<> lexer::next_token() {
     }
     case std::char_traits<char>::eof(): return { location, token_t::None };
     default: {
-        if (std::isdigit(ch) != 0) {
+        if (std::isdigit(get()) != 0) {
             integer_token integer    = 0;
             integer_token max        = std::numeric_limits<integer_token>::max();
             integer_token upperBound = max / 10;
@@ -95,7 +80,7 @@ token_handle<> lexer::next_token() {
             return { location, integer };
         }
 
-        if (std::isalnum(ch) != 0 || ch == '_') {
+        if (std::isalnum(get()) != 0 || get() == '_') {
             std::size_t start = m_cursor++;
             while (std::isalnum(get()) != 0 || get() == '_')
                 next();
@@ -111,7 +96,47 @@ token_handle<> lexer::next_token() {
             return { location, token_t::Identifier, value };
         }
 
-        return { location, "unexpected character '"s.append(m_content.substr(m_cursor, 1)) + "'" };
+        struct compare {
+            bool operator()(const std::string_view& lhs, const std::string_view& rhs) const {
+                if (lhs.size() != rhs.size()) return lhs.size() > rhs.size();
+                return lhs < rhs;
+            }
+        };
+
+        static std::map<std::string_view, token_t, compare> s_tokens = {
+            { "(", token_t::Lparen },
+            { ")", token_t::Rparen },
+            { "{", token_t::Lbrace },
+            { "}", token_t::Rbrace },
+            { "[", token_t::Lbracket },
+            { "]", token_t::Rbracket },
+            { ";", token_t::Semicolon },
+            { ":", token_t::Colon },
+            { ",", token_t::Comma },
+            { ".", token_t::Dot },
+            { "+", token_t::Plus },
+            { "-", token_t::Minus },
+            { "*", token_t::Star },
+            { "/", token_t::Slash },
+            { "%", token_t::Percent },
+            { "++", token_t::DPlus },
+            { "--", token_t::DMinus },
+        };
+
+        token_t     type   = token_t::None;
+        std::size_t length = 1;
+        while (m_cursor + length <= m_content.size()) {
+            auto it = s_tokens.find(m_content.substr(m_cursor, length));
+            if (it == s_tokens.end()) break;
+            type = it->second;
+            ++length;
+        }
+        if (type != token_t::None) {
+            m_cursor += length - 1;
+            return { location, type };
+        }
+
+        return { location, "unexpected character '"s.append(m_content.substr(m_cursor, length)) + "'" };
     }
     }
 }

furc/src/front/parser.cpp

@@ -29,7 +29,7 @@ parser::parser(std::string_view filename)
 ast::program_node_h parser::parse() & {
     auto program = m_arena.allocate_shared<ast::program_node>();
 
-    while (peek_token()->type != token_t::None && !m_lexer.empty()) {
+    while (peek_token().present() && peek_token()->type != token_t::None && !m_lexer.empty()) {
         program->push(std::move(parse_declaration()));
     }
 
@@ -119,7 +119,7 @@ ast::statement_node_h parser::parse_statement() {
 }
 
 ast::expression_node_h parser::parse_expression() {
-    return parse_expression_rhs(parse_expression_primary(), 16);
+    return parse_expression_rhs(parse_expression_unary(16), 16);
 }
 
 ast::literal_node_h parser::parse_literal() {
@@ -143,25 +143,6 @@ ast::literal_node_h parser::parse_literal() {
     return { tok.location(), "unexpected token "s + tok->type + ", expected literal" };
 }
 
-enum class binop_associativity {
-    Left,
-    Right,
-};
-
-struct binop_info {
-    ast::binop_expression_node_t type;
-    std::uint32_t                precedence;
-    binop_associativity          associativity;
-
-    static binop_info left(ast::binop_expression_node_t type, std::uint32_t precedence) {
-        return { type, precedence, binop_associativity::Left };
-    }
-
-    static binop_info right(ast::binop_expression_node_t type, std::uint32_t precedence) {
-        return { type, precedence, binop_associativity::Right };
-    }
-};
-
 ast::expression_node_h parser::parse_expression_primary() {
     const auto& tok = peek_token();
 
@@ -170,37 +151,132 @@ ast::expression_node_h parser::parse_expression_primary() {
     return { tok.location(), "unexpected token"s + tok->type + ", expected expression or literal" };
 }
 
-ast::expression_node_h parser::parse_expression_rhs(const ast::expression_node_h& init, std::uint32_t precedence) {
+struct unaryop_info {
-    static std::unordered_map<token_t, binop_info> s_binops = {
+    ast::unaryop_expression_node_t type;
-        { token_t::Plus, binop_info::left(ast::binop_expression_node_t::Add, 5) },
+    std::uint32_t                  precedence;
-        { token_t::Minus, binop_info::left(ast::binop_expression_node_t::Sub, 5) },
-        { token_t::Star, binop_info::left(ast::binop_expression_node_t::Mul, 4) },
-        { token_t::Slash, binop_info::left(ast::binop_expression_node_t::Div, 4) },
-        { token_t::Percent, binop_info::left(ast::binop_expression_node_t::Mod, 5) },
 };
 
-    ast::expression_node_h lhs = init;
+ast::expression_node_h parser::parse_expression_unary(std::uint32_t precedence) {
+    static std::unordered_map<token_t, unaryop_info> s_prefixes = {
+        { token_t::Plus, unaryop_info{ ast::unaryop_expression_node_t::Positive, 3 } },
+        { token_t::Minus, unaryop_info{ ast::unaryop_expression_node_t::Negative, 3 } },
+        { token_t::DPlus, unaryop_info{ ast::unaryop_expression_node_t::PrefixIncrement, 2 } },
+        { token_t::DMinus, unaryop_info{ ast::unaryop_expression_node_t::PrefixDecrement, 2 } },
+    };
+
+    ast::unaryop_expression_node_h result;
     while (true) {
-        auto it = s_binops.find(peek_token()->type);
+        auto it = s_prefixes.find(peek_token()->type);
-        if (it == s_binops.end()) return lhs;
+        if (it == s_prefixes.end()) break;
-        binop_info current = it->second;
+        auto current = it->second;
+        if (current.precedence >= precedence) break;
+        auto token = next_token();
+
+        ast::expression_node_h expression;
+
+        auto nextIt = s_prefixes.find(peek_token()->type);
+        if (nextIt != s_prefixes.end()) {
+            auto next  = nextIt->second;
+            expression = parse_expression_unary(current.precedence + 1);
+        }
+
+        result = { token.location(), m_arena, current.type, std::move(expression) };
+    }
+
+    if (!result.present()) return parse_expression_primary();
+    if (!result->get_node().present()) result->set_node(parse_expression_primary());
+    return result;
+}
+
+enum class associativity {
+    Left,
+    Right,
+};
+
+enum class rhsop_info_t {
+    Unaryop,
+    Binop,
+};
+
+struct rhsop_info {
+    rhsop_info_t  type;
+    std::uint32_t precedence;
+    associativity associativity;
+    union {
+        ast::unaryop_expression_node_t unary;
+        ast::binop_expression_node_t   binary;
+    };
+
+    static rhsop_info create(ast::unaryop_expression_node_t type, std::uint32_t precedence) {
+        rhsop_info info{};
+        info.type          = rhsop_info_t::Unaryop;
+        info.precedence    = precedence;
+        info.associativity = associativity::Left;
+        info.unary         = type;
+        return info;
+    }
+
+    static rhsop_info create(ast::binop_expression_node_t type,
+        std::uint32_t                                     precedence,
+        enum associativity                                associativity) {
+        rhsop_info info{};
+        info.type          = rhsop_info_t::Binop;
+        info.precedence    = precedence;
+        info.associativity = associativity;
+        info.binary        = type;
+        return info;
+    }
+};
+
+ast::expression_node_h parser::parse_expression_rhs(ast::expression_node_h&& init, std::uint32_t precedence) {
+    static std::unordered_map<token_t, rhsop_info> s_rhsops = {
+        { token_t::Plus, rhsop_info::create(ast::binop_expression_node_t::Add, 5, associativity::Left) },
+        { token_t::Minus, rhsop_info::create(ast::binop_expression_node_t::Sub, 5, associativity::Left) },
+        { token_t::Star, rhsop_info::create(ast::binop_expression_node_t::Mul, 4, associativity::Left) },
+        { token_t::Slash, rhsop_info::create(ast::binop_expression_node_t::Div, 4, associativity::Left) },
+        { token_t::Percent, rhsop_info::create(ast::binop_expression_node_t::Mod, 5, associativity::Left) },
+        { token_t::DPlus, rhsop_info::create(ast::unaryop_expression_node_t::PostfixIncrement, 1) },
+        { token_t::DMinus, rhsop_info::create(ast::unaryop_expression_node_t::PostfixDecrement, 1) },
+    };
+
+    ast::expression_node_h lhs = std::move(init);
+    while (peek_token().present()) {
+        auto it = s_rhsops.find(peek_token()->type);
+        if (it == s_rhsops.end()) return lhs;
+
+        rhsop_info current = it->second;
         if (current.precedence >= precedence) return lhs;
         auto opToken = next_token();
 
-        auto rhs = parse_expression_primary();
+        ast::expression_node_h rhs;
-
+        if (current.type == rhsop_info_t::Binop) {
-        auto nextIt = s_binops.find(peek_token()->type);
+            rhs = parse_expression_unary(current.precedence + 1); // unary prefix is always right-associative
-        if (nextIt != s_binops.end()) {
-            binop_info next = nextIt->second;
-
-            rhs = std::move(parse_expression_rhs(rhs,
-                current.precedence + static_cast<std::uint32_t>(current.associativity == binop_associativity::Right)));
         }
 
-        lhs = ast::binop_expression_node_h{ opToken.location(), m_arena, current.type, std::move(lhs), std::move(rhs) };
+        auto nextIt = s_rhsops.find(peek_token()->type);
+        if (nextIt != s_rhsops.end()) {
+            rhsop_info next = nextIt->second;
+
+            if (current.type == rhsop_info_t::Binop) {
+                rhs = std::move(parse_expression_rhs(std::move(rhs),
+                    current.precedence + static_cast<std::uint32_t>(current.associativity == associativity::Right)));
+            } else {
+                lhs = std::move(parse_expression_rhs(std::move(lhs), current.precedence));
             }
         }
 
+        lhs = current.type == rhsop_info_t::Binop
+                  ? ast::expression_node_h(ast::binop_expression_node_h{ opToken.location(),
+                        m_arena,
+                        current.binary,
+                        std::move(lhs),
+                        std::move(rhs) })
+                  : ast::expression_node_h(
+                        ast::unaryop_expression_node_h{ opToken.location(), m_arena, current.unary, std::move(lhs) });
+    }
+    return lhs;
+}
+
 ast::function_body_h parser::parse_body() {
     ast::function_body body;
 

furc/src/main.cpp

@@ -5,7 +5,7 @@
 #include <iostream>
 
 int main(void) {
-    furc::front::parser parser("<TEMP>", "func main() {\n    return 7 + 6 * 10; // 130\n}");
+    furc::front::parser parser("<TEMP>", "func main() {\n    return 6 - -7 * 9++;\n}");
     std::cout << parser.parse() << '\n';
 
     return 0;

furc/test/parser.cpp

@@ -133,4 +133,51 @@ TEST(Parser, OperatorPrecedence_Complex) {
     EXPECT_INTLIT(div->rhs(), 2);
 }
 
+TEST(Parser, UnaryOperator_Simple) {
+    parser parser("<TEMP>", "func main() { return -5; }");
+    auto   program = parser.parse();
+    EXPECT_TRUE(program.present());
+    EXPECT_EQ(program->declarations().size(), 1);
+    auto func = program->declarations()[0];
+    EXPECT_TRUE(func.present());
+    EXPECT_EQ(func->declaration_type(), declaration_node_t::FunctionDefinition);
+    function_definition_node_h funcDef = func;
+    EXPECT_EQ(funcDef->name()->string, "main");
+    EXPECT_EQ(funcDef->body()->statements.size(), 1);
+    return_statement_node_h ret = funcDef->body()->statements[0];
+
+    auto retVal = ret->value();
+    EXPECT_TRUE(retVal.present());
+
+    EXPECT_EQ(retVal->expression_type(), expression_node_t::Unaryop);
+    unaryop_expression_node_h neg = retVal;
+    EXPECT_EQ(neg->type(), unaryop_expression_node_t::Negative);
+    EXPECT_INTLIT(neg->get_node(), 5);
+}
+
+TEST(Parser, UnaryOperator_PrePost) {
+    parser parser("<TEMP>", "func main() { return --5++; }");
+    auto   program = parser.parse();
+    EXPECT_TRUE(program.present());
+    EXPECT_EQ(program->declarations().size(), 1);
+    auto func = program->declarations()[0];
+    EXPECT_TRUE(func.present());
+    EXPECT_EQ(func->declaration_type(), declaration_node_t::FunctionDefinition);
+    function_definition_node_h funcDef = func;
+    EXPECT_EQ(funcDef->name()->string, "main");
+    EXPECT_EQ(funcDef->body()->statements.size(), 1);
+    return_statement_node_h ret = funcDef->body()->statements[0];
+
+    auto retVal = ret->value();
+    EXPECT_TRUE(retVal.present());
+
+    EXPECT_EQ(retVal->expression_type(), expression_node_t::Unaryop);
+    unaryop_expression_node_h inc = retVal;
+    EXPECT_EQ(inc->type(), unaryop_expression_node_t::PostfixIncrement);
+
+    EXPECT_EQ(inc->get_node()->expression_type(), expression_node_t::Unaryop);
+    unaryop_expression_node_h dec = inc->get_node();
+    EXPECT_INTLIT(dec->get_node(), 5);
+}
+
 } // namespace