Generate a bison parser

CMakeLists.txt

@@ -5,8 +5,9 @@ set(CMAKE_EXPORT_COMPILE_COMMANDS 1)
 set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)
 set(CMAKE_CXX_STANDARD 17)
 
-find_package(Boost COMPONENTS program_options REQUIRED)
+find_package(Boost COMPONENTS process program_options REQUIRED)
-find_package(FLEX)
+find_package(FLEX REQUIRED)
+find_package(BISON REQUIRED)
 include_directories(${Boost_INCLUDE_DIR})
 
 FLEX_TARGET(scanner source/scanner.l ${CMAKE_CURRENT_BINARY_DIR}/scanner.cpp)
@@ -26,3 +27,20 @@ add_executable(elna cli/main.cpp
 )
 target_include_directories(elna PRIVATE include)
 target_link_libraries(elna LINK_PUBLIC ${Boost_LIBRARIES})
+
+FLEX_TARGET(lexer parser/lexer.ll ${CMAKE_CURRENT_BINARY_DIR}/lexer.cpp)
+BISON_TARGET(parser parser/parser.yy ${CMAKE_CURRENT_BINARY_DIR}/parser.cpp)
+add_flex_bison_dependency(lexer parser)
+
+add_executable(test parser/main.cpp
+	source/lexer.cpp include/elna/source/lexer.hpp
+	source/parser.cpp include/elna/source/parser.hpp
+	source/types.cpp include/elna/source/types.hpp
+	source/symbol_table.cpp include/elna/source/symbol_table.hpp
+	source/result.cpp include/elna/source/result.hpp
+	source/semantic.cpp include/elna/source/semantic.hpp
+	source/optimizer.cpp include/elna/source/optimizer.hpp
+	${BISON_parser_OUTPUTS} ${FLEX_lexer_OUTPUTS}
+)
+target_include_directories(test PRIVATE ${CMAKE_CURRENT_BINARY_DIR} parser include)
+# target_link_libraries(test ${FLEX_LIBRARIES})

backend/target.cpp

@@ -4,87 +4,9 @@
 
 namespace elna::riscv
 {
-    elfio_section_writer::iterator::reference elfio_section_writer::iterator::operator*() const noexcept
+    elfio_writer::elfio_writer(ELFIO::section *text,
-    {
-        return payload;
-    }
-
-    elfio_section_writer::iterator::pointer elfio_section_writer::iterator::operator->() const noexcept
-    {
-        return &payload;
-    }
-
-    elfio_section_writer::iterator& elfio_section_writer::iterator::operator++()
-    {
-        this->payload.data += *this->sizes;
-        this->payload.label = *(++this->labels);
-        this->payload.size = *(++this->sizes);
-
-        return *this;
-    }
-
-    elfio_section_writer::iterator& elfio_section_writer::iterator::operator++(int)
-    {
-        auto tmp = *this;
-        ++(*this);
-        return *this;
-    }
-
-    bool elfio_section_writer::iterator::operator==(const iterator& that) const
-    {
-        return this->labels == that.labels;
-    }
-
-    bool elfio_section_writer::iterator::operator!=(const iterator& that) const
-    {
-        return !(*this == that);
-    }
-
-    elfio_section_writer::elfio_section_writer(ELFIO::section *section)
-        : m_section(section)
-    {
-    }
-
-    void elfio_section_writer::operator()(const std::string& label, const std::byte *data, std::size_t size)
-    {
-        labels.push_back(label);
-        sizes.push_back(size);
-        m_section->append_data(reinterpret_cast<const char *>(data), size);
-    }
-
-    std::pair<std::string_view, bool> elfio_section_writer::operator()(const std::byte *data, std::size_t size)
-    {
-        auto found = std::find_if(begin(), end(),
-                [data, size](elfio_section_writer::entry entry) {
-                    return size == entry.size && std::memcmp(entry.data, data, size) == 0;
-                });
-        if (found == end())
-        {
-            (*this)(".CL" + std::to_string(labels.size()), data, size);
-            return std::pair<std::string_view, bool>(labels.back(), true);
-        }
-        return std::pair<std::string_view, bool>(found->label, false);
-    }
-
-    elfio_section_writer::iterator elfio_section_writer::begin() const
-    {
-        return elfio_section_writer::iterator(labels.cbegin(), sizes.cbegin(),
-                reinterpret_cast<const std::byte *>(m_section->get_data()));
-    }
-
-    elfio_section_writer::iterator elfio_section_writer::end() const
-    {
-        return elfio_section_writer::iterator(labels.cend(), sizes.cend());
-    }
-
-    ELFIO::section *elfio_section_writer::section() noexcept
-    {
-        return m_section;
-    }
-
-    elfio_writer::elfio_writer(ELFIO::section *text, ELFIO::section *read_only,
             ELFIO::symbol_section_accessor symbol_accessor, ELFIO::string_section_accessor string_accessor)
-        : text(text), read_only(elfio_section_writer(read_only)),
+        : text(text),
         symbol_accessor(symbol_accessor), string_accessor(string_accessor)
     {
     }
@@ -100,20 +22,6 @@ namespace elna::riscv
         return text->get_size();
     }
 
-    std::string_view elfio_writer::sink(const std::byte *data, std::size_t size)
-    {
-        auto offset = read_only.section()->get_size();
-        auto [result, inserted] = read_only(data, size);
-
-        if (inserted)
-        {
-            this->symbol_accessor.add_symbol(this->string_accessor, result.data(), offset,
-                result.size(), ELFIO::STB_LOCAL, ELFIO::STT_NOTYPE, 0,
-                read_only.section()->get_index());
-        }
-        return result;
-    }
-
     void elfio_writer::sink(const std::string& label)
     {
         this->symbol_accessor.add_symbol(this->string_accessor, "printf", 0x00000000, 0,
@@ -188,17 +96,11 @@ namespace elna::riscv
         rel_sec->set_link(sym_sec->get_index());
         rel_sec->set_flags(ELFIO::SHF_ALLOC);
 
-        // Create read only data section
-        ELFIO::section* ro_sec = writer.sections.add(".rodata");
-        ro_sec->set_type(ELFIO::SHT_PROGBITS);
-        ro_sec->set_flags(ELFIO::SHF_ALLOC);
-        ro_sec->set_addr_align(0x4);
-
         // Create symbol relocation table writers
         ELFIO::symbol_section_accessor syma(writer, sym_sec);
         ELFIO::relocation_section_accessor rela(writer, rel_sec);
-        auto _writer = std::make_shared<elfio_writer>(text_sec, ro_sec, syma, stra);
 
+        auto _writer = std::make_shared<elfio_writer>(text_sec, syma, stra);
         auto references = generate(intermediate_code_generator, table, _writer);
 
         syma.arrange_local_symbols();

include/elna/backend/target.hpp

@@ -7,87 +7,17 @@
 
 namespace elna::riscv
 {
-    /**
-     * Writer to a single label.
-     */
-    struct elfio_section_writer
-    {
-        struct entry
-        {
-            std::string_view label;
-            const std::byte *data{ nullptr };
-            std::size_t size{ 0 };
-        };
-
-        /**
-         * An iterator over label and string pairs.
-         */
-        struct iterator
-        {
-            using iterator_category = std::forward_iterator_tag;
-            using difference_type = ptrdiff_t;
-            using value_type = entry;
-            using pointer = const value_type *;
-            using reference = const value_type&;
-
-            reference operator*() const noexcept;
-            pointer operator->() const noexcept;
-            iterator& operator++();
-            iterator& operator++(int);
-            bool operator==(const iterator& that) const;
-            bool operator!=(const iterator& that) const;
-
-        private:
-            std::vector<std::string>::const_iterator labels;
-            std::vector<std::size_t>::const_iterator sizes;
-            value_type payload;
-
-            iterator(std::vector<std::string>::const_iterator labels, std::vector<std::size_t>::const_iterator sizes,
-                    const std::byte *data)
-                : labels(labels), sizes(sizes)
-            {
-                if (data != nullptr)
-                {
-                    payload = { *this->labels, data, *this->sizes};
-                }
-            }
-
-            iterator(std::vector<std::string>::const_iterator labels, std::vector<std::size_t>::const_iterator sizes)
-                : labels(labels), sizes(sizes), payload{}
-            {
-            }
-
-            friend elfio_section_writer;
-        };
-        explicit elfio_section_writer(ELFIO::section *section);
-
-        void operator()(const std::string& label, const std::byte *data, std::size_t size);
-        std::pair<std::string_view, bool> operator()(const std::byte *data, std::size_t size);
-
-        iterator begin() const;
-        iterator end() const;
-
-        ELFIO::section *section() noexcept;
-
-    private:
-        std::vector<std::string> labels;
-        std::vector<std::size_t> sizes;
-        ELFIO::section *m_section;
-    };
-
     class elfio_writer final : public source::writer<std::byte>
     {
         ELFIO::section *text;
-        elfio_section_writer read_only;
         ELFIO::symbol_section_accessor symbol_accessor;
         ELFIO::string_section_accessor string_accessor;
 
     public:
-        elfio_writer(ELFIO::section *text, ELFIO::section *read_only, ELFIO::symbol_section_accessor symbol_accessor,
+        elfio_writer(ELFIO::section *text, ELFIO::symbol_section_accessor symbol_accessor,
                 ELFIO::string_section_accessor string_accessor);
 
         std::size_t sink(const std::string& label, const std::byte *data, std::size_t size) override;
-        std::string_view sink(const std::byte *data, std::size_t size) override;
         void sink(const std::string& label) override;
         std::size_t size() const override;
     };

include/elna/source/result.hpp

@@ -171,16 +171,6 @@ namespace elna::source
          */
         virtual std::size_t sink(const std::string& label, const T *data, std::size_t size) = 0;
 
-        /**
-         * Writes data and returns a label under that the data can be accessed.
-         *
-         * \param data Data to be written.
-         * \param size Data size.
-         *
-         * \return Label for the symbol.
-         */
-        virtual std::string_view sink(const T *data, std::size_t size) = 0;
-
         /**
          * Creates an external symbol.
          */

parser/lexer.ll

@@ -0,0 +1,131 @@
+%{
+#define YY_NO_UNISTD_H
+#define YY_USER_ACTION this->location.columns(yyleng);
+
+#include <sstream>
+#include "parser.hpp"
+
+#undef YY_DECL
+#define YY_DECL yy::parser::symbol_type elna::syntax::FooLexer::lex()
+#define yyterminate() return yy::parser::make_YYEOF(this->location)
+%}
+
+%option c++ noyywrap never-interactive
+%option yyclass="elna::syntax::FooLexer"
+
+%%
+%{
+    this->location.step();
+%}
+
+\-\-.*                  {
+                            /* Skip the comment */
+                        }
+[\ \t\r]                ; /* Skip the whitespaces */
+\n+                     {
+                            this->location.lines(yyleng);
+                            this->location.step();
+                        }
+if                      {
+                            return yy::parser::make_IF(this->location);
+                        }
+then                    {
+                            return yy::parser::make_THEN(this->location);
+                        }
+while                   {
+                            return yy::parser::make_WHILE(this->location);
+                        }
+do                      {
+                            return yy::parser::make_DO(this->location);
+                        }
+proc                    {
+                            return yy::parser::make_PROCEDURE(this->location);
+                        }
+begin                   {
+                            return yy::parser::make_BEGIN_BLOCK(this->location);
+                        }
+end                     {
+                            return yy::parser::make_END_BLOCK(this->location);
+                        }
+const                   {
+                            return yy::parser::make_CONST(this->location);
+                        }
+var                     {
+                            return yy::parser::make_VAR(this->location);
+                        }
+True                    {
+                            return yy::parser::make_BOOLEAN(true, this->location);
+                        }
+False                   {
+                            return yy::parser::make_BOOLEAN(false, this->location);
+                        }
+[A-Za-z_][A-Za-z0-9_]*  {
+                            return yy::parser::make_IDENTIFIER(yytext, this->location);
+                        }
+[0-9]+                  {
+                            return yy::parser::make_NUMBER(strtol(yytext, NULL, 10), this->location);
+                        }
+\(                      {
+                            return yy::parser::make_LEFT_PAREN(this->location);
+                        }
+\)                      {
+                            return yy::parser::make_RIGHT_PAREN(this->location);
+                        }
+\>=                     {
+                            return yy::parser::make_GREATER_EQUAL(this->location);
+                        }
+\<=                     {
+                            return yy::parser::make_LESS_EQUAL(this->location);
+                        }
+\>                      {
+                            return yy::parser::make_GREATER_THAN(this->location);
+                        }
+\<                      {
+                            return yy::parser::make_LESS_THAN(this->location);
+                        }
+\/=                     {
+                            return yy::parser::make_NOT_EQUAL(this->location);
+                        }
+=                       {
+                            return yy::parser::make_EQUALS(this->location);
+                        }
+;                       {
+                            return yy::parser::make_SEMICOLON(this->location);
+                        }
+\.                      {
+                            return yy::parser::make_DOT(this->location);
+                        }
+,                       {
+                            return yy::parser::make_COMMA(this->location);
+                        }
+\+                      {
+                            return yy::parser::make_PLUS(this->location);
+                        }
+\-                      {
+                            return yy::parser::make_MINUS(this->location);
+                        }
+\*                      {
+                            return yy::parser::make_MULTIPLICATION(this->location);
+                        }
+\/                      {
+                            return yy::parser::make_DIVISION(this->location);
+                        }
+:=                      {
+                            return yy::parser::make_ASSIGNMENT(this->location);
+                        }
+:                       {
+                            return yy::parser::make_COLON(this->location);
+                        }
+\^                      {
+                            return yy::parser::make_HAT(this->location);
+                        }
+@                       {
+                            return yy::parser::make_AT(this->location);
+                        }
+.                       {
+                            std::stringstream ss;
+
+                            ss << "Illegal character 0x" << std::hex << static_cast<unsigned char>(yytext[0]);
+                            throw yy::parser::syntax_error(this->location, ss.str());
+                        }
+%%

parser/main.cpp

@@ -0,0 +1,22 @@
+#include "parser.hpp"
+#include <sstream>
+
+int main()
+{
+    std::istringstream inp("const world = 5, hello = 7;");
+
+    std::unique_ptr<elna::source::program> program;
+
+    elna::syntax::FooLexer lexer(inp);
+    yy::parser parser(lexer, program);
+    auto result = parser();
+
+    for (auto& definition : program->definitions())
+    {
+        auto const_definition = dynamic_cast<elna::source::constant_definition *>(definition.get());
+
+        std::cout << "const " << const_definition->identifier() << " = "
+            << const_definition->body().number() << std::endl;
+    }
+    return result;
+}

parser/parser.yy

@@ -0,0 +1,141 @@
+%require "3.2"
+%language "c++"
+
+%code requires {
+    #include <cstdint>
+    #include <iostream>
+    #include "elna/source/parser.hpp"
+
+
+    #if ! defined(yyFlexLexerOnce)
+        #include <FlexLexer.h>
+    #endif
+
+    namespace elna::syntax
+    {
+    class FooLexer;
+    }
+}
+
+%code provides {
+    namespace elna::syntax
+    {
+
+    class FooLexer : public yyFlexLexer
+    {
+    public:
+        yy::location location;
+
+        FooLexer(std::istream& arg_yyin)
+            : yyFlexLexer(&arg_yyin)
+        {
+        }
+
+        yy::parser::symbol_type lex();
+    };
+
+    }
+}
+
+%define api.token.raw
+%define api.token.constructor
+%define api.value.type variant
+%define parse.assert
+
+%parse-param {elna::syntax::FooLexer& lexer}
+%parse-param {std::unique_ptr<elna::source::program>& program}
+%locations
+
+%header
+
+%code {
+    #define yylex lexer.lex
+}
+%start program;
+
+%token <std::string> IDENTIFIER "identifier"
+%token <std::int32_t> NUMBER "number"
+%token <bool> BOOLEAN
+%token IF THEN WHILE DO
+%token CONST VAR PROCEDURE
+%token BEGIN_BLOCK END_BLOCK
+%token TRUE FALSE
+%token LEFT_PAREN RIGHT_PAREN SEMICOLON DOT COMMA
+%token GREATER_EQUAL LESS_EQUAL LESS_THAN GREATER_THAN NOT_EQUAL EQUALS
+%token PLUS MINUS MULTIPLICATION DIVISION
+%token ASSIGNMENT COLON HAT AT
+
+%type <std::unique_ptr<elna::source::integer_literal>> integer_literal;
+%type <std::unique_ptr<elna::source::constant_definition>> constant_definition;
+%type <std::vector<std::unique_ptr<elna::source::constant_definition>>> constant_definition_part constant_definitions;
+%type <std::unique_ptr<elna::source::type_expression>> type_expression;
+%%
+program: constant_definition_part
+        {
+            elna::source::position position;
+            std::vector<std::unique_ptr<elna::source::declaration>> declarations;
+            std::vector<std::unique_ptr<elna::source::definition>> definitions($1.size());
+            std::vector<std::unique_ptr<elna::source::definition>>::iterator definition = definitions.begin();
+
+            for (auto& constant : $1)
+            {
+                *definition++ = std::move(constant);
+            }
+            program = std::make_unique<elna::source::program>(position,
+                 std::move(definitions), std::move(declarations),
+                 std::make_unique<elna::source::compound_statement>(position));
+        }
+integer_literal: NUMBER
+        {
+            elna::source::position position{
+                static_cast<std::size_t>(@1.begin.line),
+                static_cast<std::size_t>(@1.begin.column)
+            };
+            $$ = std::make_unique<elna::source::integer_literal>(position, $1); 
+        };
+type_expression:
+        HAT IDENTIFIER
+            {
+                elna::source::position position{
+                    static_cast<std::size_t>(@1.begin.line),
+                    static_cast<std::size_t>(@1.begin.column)
+                };
+                $$ = std::make_unique<elna::source::type_expression>(position, $2, true);
+            }
+        | IDENTIFIER
+            {
+                elna::source::position position{
+                    static_cast<std::size_t>(@1.begin.line),
+                    static_cast<std::size_t>(@1.begin.column)
+                };
+                $$ = std::make_unique<elna::source::type_expression>(position, $1, false);
+            }
+variable_declaration: IDENTIFIER COLON type_expression
+variable_declarations:
+        variable_declaration COMMA variable_declarations
+        | variable_declaration
+constant_definition: IDENTIFIER EQUALS integer_literal
+        {
+            elna::source::position position{
+                static_cast<std::size_t>(@1.begin.line),
+                static_cast<std::size_t>(@1.begin.column)
+            };
+            $$ = std::make_unique<elna::source::constant_definition>(position,
+                $1, std::move($3)); 
+        };
+constant_definitions:
+    constant_definition COMMA constant_definitions
+        {
+            std::swap($$, $3);
+            $$.emplace($$.cbegin(), std::move($1));
+        }
+    | constant_definition { $$.emplace_back(std::move($1)); }
+constant_definition_part:
+    /* no constant definitions */ {}
+    | CONST constant_definitions SEMICOLON { std::swap($$, $2); };
+%%
+
+void yy::parser::error(const location_type& loc, const std::string &message)
+{
+    std::cerr << "Error: " << message << std::endl;
+}