belka/elna
1
0
Fork 0
Code Issues 5 Pull Requests Releases Activity

Tokenize the input with flex

Browse Source
This commit is contained in:
Eugen Wissner
2024-04-06 16:10:07 +02:00
parent 4251c361c7
commit aa3df7f4d3
11 changed files with 274 additions and 369 deletions
Download Patch File Download Diff File Expand all files Collapse all files

260
source/lexer.cpp
View File

@ -1,79 +1,12 @@
#include "elna/source/lexer.hpp"
#include <cassert>
#include <variant>
#include <sstream>
namespace elna::source
{
using source_position = elna::source::position;
using source_error = elna::source::error;
std::pair<text_iterator, text_iterator> text_iterators(const std::string &buffer)
{
return std::make_pair<>(text_iterator(std::cbegin(buffer)),
text_iterator(std::cend(buffer), position{0, 0}));
}
text_iterator::text_iterator(std::string::const_iterator buffer,
const source_position start_position)
: m_buffer(buffer), m_position(start_position)
{
}
const source_position& text_iterator::position() const noexcept
{
return this->m_position;
}
text_iterator::reference text_iterator::operator*() const noexcept
{
return *m_buffer;
}
text_iterator::pointer text_iterator::operator->() const noexcept
{
return m_buffer.base();
}
text_iterator& text_iterator::operator++()
{
if (*this->m_buffer == '\n')
{
this->m_position.column = 1;
++this->m_position.line;
}
else
{
++this->m_position.column;
}
std::advance(this->m_buffer, 1);
return *this;
}
text_iterator& text_iterator::operator++(int)
{
auto tmp = *this;
++(*this);
return *this;
}
text_iterator text_iterator::operator+(std::size_t step)
{
auto result = *this;
return ++result;
}
bool text_iterator::operator==(const text_iterator& that) const noexcept
{
return this->m_buffer == that.m_buffer;
}
bool text_iterator::operator!=(const text_iterator& that) const noexcept
{
return !(*this == that);
}
token::value::value()
: nil(nullptr)
{
@ -103,6 +36,23 @@ namespace elna::source
{
}
token::token(type of, value&& value, const elna::source::position position)
: m_type(of), m_position(position)
{
if (has_identifier())
{
new((void *) &m_value.identifier) std::string(std::move(value.identifier));
}
else if (is_numeric())
{
m_value.number = value.number;
}
else
{
m_value.nil = nullptr;
}
}
token::token(const type of, source_position position)
: m_type(of), m_position(position)
{
@ -278,11 +228,12 @@ namespace elna::source
std::string unexpected_character::what() const
{
std::stringstream ss{ "Unexpected character '" };
std::string ss{ "Unexpected character '" };
ss << character << "'";
ss.insert(ss.cend(), character.cbegin(), character.cend());
ss.push_back('\'');
return ss.str();
return ss;
}
unexpected_token::unexpected_token(const token& token, const std::filesystem::path& path)
@ -371,173 +322,4 @@ namespace elna::source
{
return m_errors;
}
result<lexer> lex(const std::string& buffer, const std::filesystem::path& path)
{
std::vector<token> tokens;
auto [iterator, text_end] = text_iterators(buffer);
while (iterator != text_end)
{
if (*iterator == ' ' || *iterator == '\n')
{
}
else if (std::isdigit(*iterator))
{
tokens.emplace_back(
token::type::number,
static_cast<std::int32_t>(*iterator - '0'),
iterator.position()
);
}
else if (*iterator == '=')
{
tokens.emplace_back(token::type::equals, iterator.position());
}
else if (*iterator == '(')
{
tokens.emplace_back(token::type::left_paren, iterator.position());
}
else if (*iterator == ')')
{
tokens.emplace_back(token::type::right_paren, iterator.position());
}
else if (*iterator == ';')
{
tokens.emplace_back(token::type::semicolon, iterator.position());
}
else if (*iterator == ',')
{
tokens.emplace_back(token::type::comma, iterator.position());
}
else if (*iterator == '.')
{
tokens.emplace_back(token::type::dot, iterator.position());
}
else if (std::isalpha(*iterator))
{
std::string word;
auto i = iterator;
while (i != text_end && std::isalpha(*i))
{
word.push_back(*i);
++i;
}
if (word == "const")
{
tokens.emplace_back(token::type::let, iterator.position());
}
else if (word == "var")
{
tokens.emplace_back(token::type::var, iterator.position());
}
else if (word == "begin")
{
tokens.emplace_back(token::type::begin, iterator.position());
}
else if (word == "end")
{
tokens.emplace_back(token::type::end, iterator.position());
}
else if (word == "if")
{
tokens.emplace_back(token::type::when, iterator.position());
}
else if (word == "then")
{
tokens.emplace_back(token::type::then, iterator.position());
}
else if (word == "while")
{
tokens.emplace_back(token::type::loop, iterator.position());
}
else if (word == "do")
{
tokens.emplace_back(token::type::_do, iterator.position());
}
else if (word == "True")
{
tokens.emplace_back(token::type::boolean, 1, iterator.position());
}
else if (word == "False")
{
tokens.emplace_back(token::type::boolean, 0, iterator.position());
}
else if (word == "proc")
{
tokens.emplace_back(token::type::procedure, 0, iterator.position());
}
else
{
tokens.emplace_back(token::type::identifier, word.c_str(), iterator.position());
}
iterator = i;
continue;
}
else if (*iterator == '+' || *iterator == '-')
{
std::string _operator{ *iterator };
tokens.emplace_back(token::type::term_operator, _operator.c_str(), iterator.position());
}
else if (*iterator == '/' && iterator + 1 != text_end && *(iterator + 1) == '=')
{
tokens.emplace_back(token::type::comparison_operator, "n", iterator.position());
++iterator;
}
else if (*iterator == '*' || *iterator == '/')
{
std::string _operator{ *iterator };
tokens.emplace_back(token::type::factor_operator, _operator.c_str(), iterator.position());
}
else if (*iterator == '<')
{
std::string _operator;
auto operator_position = iterator.position();
if (iterator + 1 == text_end || *(iterator + 1) != '=')
{
_operator.push_back(*iterator);
}
else
{
++iterator;
_operator.push_back('l');
}
tokens.emplace_back(token::type::comparison_operator, _operator.c_str(), operator_position);
}
else if (*iterator == '>')
{
std::string _operator;
auto operator_position = iterator.position();
if (iterator + 1 == text_end || *(iterator + 1) != '=')
{
_operator.push_back(*iterator);
}
else
{
++iterator;
_operator.push_back('g');
}
tokens.emplace_back(token::type::comparison_operator, _operator.c_str(), operator_position);
}
else if (*iterator == ':' && iterator + 1 != text_end && *(iterator + 1) == '=')
{
tokens.emplace_back(token::type::assignment, iterator.position());
++iterator;
}
else if (*iterator == ':')
{
tokens.emplace_back(token::type::colon, iterator.position());
}
else
{
return result<lexer>(unexpected_character{ std::string{ *iterator }, path, iterator.position() });
}
++iterator;
}
return result<lexer>(std::in_place, std::move(tokens), iterator.position(), path);
}
}

9
source/parser.cpp
View File

@ -95,7 +95,7 @@ namespace elna::source
}
declaration::declaration(const std::string& identifier, const std::string& type)
: m_identifier(identifier), m_type(type)
: definition(identifier), m_type(type)
{
}
@ -104,11 +104,6 @@ namespace elna::source
visitor->visit(this);
}
std::string& declaration::identifier() noexcept
{
return m_identifier;
}
std::string& declaration::type() noexcept
{
return m_type;
@ -710,7 +705,7 @@ namespace elna::source
else
{
iterator.add_error(*iterator);
break;
return nullptr;
}
}

194
source/scanner.l Normal file
View File

@ -0,0 +1,194 @@
%{
#define YY_NO_UNISTD_H
#define YY_USER_ACTION token_position = elna::source::position{ line_no, column_no }; column_no += yyleng;
#include <fstream>
#include "elna/source/lexer.hpp"
elna::source::token::value yylval{};
elna::source::position token_position{};
static std::size_t column_no = 1;
static std::size_t line_no = 1;
%}
%option noyywrap
%option never-interactive
%%
\-\-.* {
/* Skip the comment */
}
[\ \t\r] {
/* Skip the whitespaces */
}
\n {
++line_no;
column_no = 1;
}
if {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::when);
}
then {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::then);
}
while {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::loop);
}
do {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::_do);
}
proc {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::procedure);
}
begin {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::begin);
}
end {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::end);
}
const {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::let);
}
var {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::var);
}
True {
yylval.number = 1;
return static_cast<int>(elna::source::token::type::boolean);
}
False {
yylval.number = 0;
return static_cast<int>(elna::source::token::type::boolean);
}
[A-Za-z_][A-Za-z0-9_]* {
new((void *) &yylval.identifier) std::string(yytext);
return static_cast<int>(elna::source::token::type::identifier);
}
[0-9]+ {
yylval.number = strtol(yytext, NULL, 10);
return static_cast<int>(elna::source::token::type::number);
}
\( {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::left_paren);
}
\) {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::right_paren);
}
\>= {
new((void *) &yylval.identifier) std::string(1, 'g');
return static_cast<int>(elna::source::token::type::comparison_operator);
}
\<= {
new((void *) &yylval.identifier) std::string(1, 'l');
return static_cast<int>(elna::source::token::type::comparison_operator);
}
(>|<) {
new((void *) &yylval.identifier) std::string(yytext);
return static_cast<int>(elna::source::token::type::comparison_operator);
}
\/= {
new((void *) &yylval.identifier) std::string(1, 'n');
return static_cast<int>(elna::source::token::type::comparison_operator);
}
= {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::equals);
}
; {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::semicolon);
}
\. {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::dot);
}
, {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::comma);
}
(\+|\-) {
new((void *) &yylval.identifier) std::string(yytext);
return static_cast<int>(elna::source::token::type::term_operator);
}
(\*|\/) {
new((void *) &yylval.identifier) std::string(yytext);
return static_cast<int>(elna::source::token::type::factor_operator);
}
:= {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::assignment);
}
: {
yylval.nil = nullptr;
return static_cast<int>(elna::source::token::type::colon);
}
. {
return -1;
}
%%
namespace elna::source
{
result<lexer> tokenize(const std::filesystem::path& path)
{
int yytoken;
std::vector<token> tokens;
yyin = fopen(path.c_str(), "rb");
if (yyin == nullptr)
{
throw std::ios_base::failure("File does not exist");
}
do
{
yytoken = yylex();
if (yytoken < 0)
{
return result<lexer>(unexpected_character{ std::string{ yytext[0] }, path, token_position });
}
tokens.emplace_back(static_cast<token::type>(yytoken), std::move(yylval), token_position);
}
while (yytoken != 0);
return result<lexer>(std::in_place, std::move(tokens), position{ line_no, column_no }, path);
}
}