generic-library/ChaiScript
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Switch Token to smart pointer

Browse Source
This commit is contained in:
Jonathan Turner
2009-05-27 15:07:29 +00:00
parent fbb9534601
commit 332fdf0c8e
5 changed files with 43 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
langkit/langkit_lexer.cpp
View File

@@ -9,9 +9,9 @@ Lexer Lexer::operator<<(const Pattern &p) {
return *this;
}
std::vector<Token> Lexer::lex(const std::string &input, const char *filename) {
std::vector<TokenPtr> Lexer::lex(const std::string &input, const char *filename) {
std::vector<Pattern>::iterator iter, end;
std::vector<Token> retval;
std::vector<TokenPtr> retval;
bool found;
std::string::const_iterator input_iter = input.begin();
@@ -23,14 +23,14 @@ std::vector<Token> Lexer::lex(const std::string &input, const char *filename) {
for (iter = lex_patterns.begin(), end = lex_patterns.end(); iter != end; ++iter) {
boost::match_results<std::string::const_iterator> what;
if (regex_search(input_iter, input.end(), what, iter->regex, boost::match_continuous)) {
Token t(what[0], iter->identifier, filename);
t.start.column = current_col;
t.start.line = current_line;
current_col += t.text.size();
t.end.column = current_col;
t.end.line = current_line;
TokenPtr t(new Token(what[0], iter->identifier, filename));
t->start.column = current_col;
t->start.line = current_line;
current_col += t->text.size();
t->end.column = current_col;
t->end.line = current_line;
retval.push_back(t);
input_iter += t.text.size();
input_iter += t->text.size();
found = true;
break;
}

9
langkit/langkit_lexer.hpp
View File

@@ -4,8 +4,9 @@
#ifndef LANGKIT_LEXER_HPP_
#define LANGKIT_LEXER_HPP_
#include <string>
#include <boost/regex.hpp>
#include <tr1/memory>
#include <string>
struct File_Position {
int line;
@@ -24,13 +25,15 @@ struct Pattern {
Pattern(const std::string &regexp, int id) : regex(regexp), identifier(id) { }
};
typedef std::tr1::shared_ptr<struct Token> TokenPtr;
struct Token {
std::string text;
int identifier;
const char *filename;
File_Position start, end;
std::vector<Token> children;
std::vector<TokenPtr> children;
Token(const std::string &token_text, int id, const char *fname) : text(token_text), identifier(id), filename(fname) { }
};
@@ -42,7 +45,7 @@ struct Lexer {
std::vector<Pattern> line_sep_patterns;
Lexer operator<<(const Pattern &p);
std::vector<Token> lex(const std::string &input, const char *fname);
std::vector<TokenPtr> lex(const std::string &input, const char *fname);
void set_skip(const Pattern &p);
void set_line_sep(const Pattern &p);

26
langkit/langkit_parser.cpp
View File

@@ -6,43 +6,43 @@
#include "langkit_lexer.hpp"
#include "langkit_parser.hpp"
std::pair<Token_Iterator, Token> String_Rule(Token_Iterator iter, Token_Iterator end, const std::string &val) {
std::pair<Token_Iterator, TokenPtr> String_Rule(Token_Iterator iter, Token_Iterator end, const std::string &val) {
if (iter != end) {
if (iter->text == val) {
return std::pair<Token_Iterator, Token>(++iter, *iter);
if ((*iter)->text == val) {
return std::pair<Token_Iterator, TokenPtr>(++iter, *iter);
}
}
return std::pair<Token_Iterator, Token>(iter, *iter);
return std::pair<Token_Iterator, TokenPtr>(iter, *iter);
}
std::pair<Token_Iterator, Token> Type_Rule(Token_Iterator iter, Token_Iterator end, const int val) {
std::pair<Token_Iterator, TokenPtr> Type_Rule(Token_Iterator iter, Token_Iterator end, const int val) {
if (iter != end) {
if (iter->identifier == val) {
return std::pair<Token_Iterator, Token>(++iter, *iter);
if ((*iter)->identifier == val) {
return std::pair<Token_Iterator, TokenPtr>(++iter, *iter);
}
}
return std::pair<Token_Iterator, Token>(iter, *iter);
return std::pair<Token_Iterator, TokenPtr>(iter, *iter);
}
std::pair<Token_Iterator, Token> Or_Rule(Token_Iterator iter, Token_Iterator end, const Rule &lhs, const Rule &rhs) {
std::pair<Token_Iterator, TokenPtr> Or_Rule(Token_Iterator iter, Token_Iterator end, const Rule &lhs, const Rule &rhs) {
Token_Iterator new_iter;
if (iter != end) {
new_iter = lhs.rule(iter, end).first;
if (new_iter != iter) {
return std::pair<Token_Iterator, Token>(new_iter, *iter);
return std::pair<Token_Iterator, TokenPtr>(new_iter, *iter);
}
else {
new_iter = rhs.rule(iter, end).first;
if (new_iter != iter) {
return std::pair<Token_Iterator, Token>(new_iter, *iter);
return std::pair<Token_Iterator, TokenPtr>(new_iter, *iter);
}
}
}
return std::pair<Token_Iterator, Token>(iter, *iter);
return std::pair<Token_Iterator, TokenPtr>(iter, *iter);
}
/*
@@ -64,6 +64,6 @@ std::pair<Token_Iterator, Token> And_Rule(Token_Iterator iter, Token_Iterator en
}
*/
std::pair<Token_Iterator, Token> Rule::operator()(Token_Iterator iter, Token_Iterator end) {
std::pair<Token_Iterator, TokenPtr> Rule::operator()(Token_Iterator iter, Token_Iterator end) {
return this->rule(iter, end);
}

13
langkit/langkit_parser.hpp
View File

@@ -8,20 +8,21 @@
#include "langkit_lexer.hpp"
typedef std::vector<Token>::iterator Token_Iterator;
typedef std::vector<TokenPtr>::iterator Token_Iterator;
struct Rule {
int identifier;
boost::function<std::pair<Token_Iterator, Token>(Token_Iterator iter, Token_Iterator end)> rule;
std::pair<Token_Iterator, Token> operator()(Token_Iterator iter, Token_Iterator end);
boost::function<std::pair<Token_Iterator, TokenPtr>(Token_Iterator iter, Token_Iterator end)> rule;
std::pair<Token_Iterator, TokenPtr> operator()(Token_Iterator iter, Token_Iterator end);
Rule() : identifier(-1) {}
Rule(int id) : identifier(id) {}
};
std::pair<Token_Iterator, Token> String_Rule(Token_Iterator iter, Token_Iterator end, const std::string &val);
std::pair<Token_Iterator, Token> Type_Rule(Token_Iterator iter, Token_Iterator end, const int val);
std::pair<Token_Iterator, Token> Or_Rule(Token_Iterator iter, Token_Iterator end, const Rule &lhs, const Rule &rhs);
std::pair<Token_Iterator, TokenPtr> String_Rule(Token_Iterator iter, Token_Iterator end, const std::string &val);
std::pair<Token_Iterator, TokenPtr> Type_Rule(Token_Iterator iter, Token_Iterator end, const int val);
std::pair<Token_Iterator, TokenPtr> Or_Rule(Token_Iterator iter, Token_Iterator end, const Rule &lhs, const Rule &rhs);

17
langkit/main.cpp
View File

@@ -13,11 +13,11 @@
class TokenType { public: enum Type { Whitespace, Identifier, Number, Operator, Parens_Open, Parens_Close,
Square_Open, Square_Close, Curly_Open, Curly_Close, Comma, Quoted_String, Single_Quoted_String, Carriage_Return, Semicolon }; };
void debug_print(Token &token) {
std::cout << "Token: " << token.text << "(" << token.identifier << ") @ " << token.filename << ": (" << token.start.column
<< ", " << token.start.line << ") to (" << token.end.column << ", " << token.end.line << ") " << std::endl;
void debug_print(TokenPtr token) {
std::cout << "Token: " << token->text << "(" << token->identifier << ") @ " << token->filename << ": (" << token->start.column
<< ", " << token->start.line << ") to (" << token->end.column << ", " << token->end.line << ") " << std::endl;
}
void debug_print(std::vector<Token> &tokens) {
void debug_print(std::vector<TokenPtr> &tokens) {
for (unsigned int i = 0; i < tokens.size(); ++i) {
debug_print(tokens[i]);
}
@@ -42,14 +42,13 @@ std::string load_file(const char *filename) {
return ret_val;
}
void parse(std::vector<Token> &tokens) {
void parse(std::vector<TokenPtr> &tokens) {
Rule rule;
rule.rule = boost::bind(String_Rule, _1, _2, "def");
Token_Iterator iter = tokens.begin(), end = tokens.end();
std::pair<Token_Iterator, Token> results = rule(iter, end);
std::pair<Token_Iterator, TokenPtr> results = rule(iter, end);
if (results.first != iter) {
std::cout << "Parse successful: " << std::endl;
@@ -84,7 +83,7 @@ int main(int argc, char *argv[]) {
std::cout << "Expression> ";
std::getline(std::cin, input);
while (input != "quit") {
std::vector<Token> tokens = lexer.lex(input, "INPUT");
std::vector<TokenPtr> tokens = lexer.lex(input, "INPUT");
debug_print(tokens);
parse(tokens);
@@ -93,7 +92,7 @@ int main(int argc, char *argv[]) {
}
}
else {
std::vector<Token> tokens = lexer.lex(load_file(argv[1]), argv[1]);
std::vector<TokenPtr> tokens = lexer.lex(load_file(argv[1]), argv[1]);
debug_print(tokens);
}
}