Added line counting to lexer.
This commit is contained in:
@@ -11,19 +11,24 @@ Lexer Lexer::operator<<(const Pattern &p) {
|
||||
|
||||
std::vector<Token> Lexer::lex(const std::string &input) {
|
||||
std::vector<Pattern>::iterator iter, end;
|
||||
//std::string::const_iterator str_end = input.end();
|
||||
std::vector<Token> retval;
|
||||
bool found;
|
||||
std::string::const_iterator input_iter = input.begin();
|
||||
|
||||
int current_col = 0;
|
||||
int current_line = 0;
|
||||
|
||||
while (input_iter != input.end()) {
|
||||
found = false;
|
||||
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);
|
||||
t.start.column = input_iter - input.begin();
|
||||
t.end.column = t.start.column + t.text.size();
|
||||
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();
|
||||
found = true;
|
||||
@@ -36,15 +41,30 @@ std::vector<Token> Lexer::lex(const std::string &input) {
|
||||
if (regex_search(input_iter, input.end(), what, iter->regex, boost::match_continuous)) {
|
||||
std::string whitespace(what[0]);
|
||||
input_iter += whitespace.size();
|
||||
current_col += whitespace.size();
|
||||
found = true;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (!found) {
|
||||
const std::string err(input_iter, input.end());
|
||||
std::cout << "Unknown string at: " << err << std::endl;
|
||||
return retval;
|
||||
for (iter = line_sep_patterns.begin(), end = line_sep_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)) {
|
||||
std::string cr(what[0]);
|
||||
input_iter += cr.size();
|
||||
found = true;
|
||||
++current_line;
|
||||
current_col = 0;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
if (!found) {
|
||||
const std::string err(input_iter, input.end());
|
||||
std::cout << "Unknown string at: " << err << std::endl;
|
||||
return retval;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -54,3 +74,11 @@ std::vector<Token> Lexer::lex(const std::string &input) {
|
||||
void Lexer::set_skip(const Pattern &p) {
|
||||
skip_patterns.push_back(p);
|
||||
}
|
||||
|
||||
void Lexer::set_line_sep(const Pattern &p) {
|
||||
line_sep_patterns.push_back(p);
|
||||
}
|
||||
|
||||
void Lexer::set_command_sep(const Pattern &p) {
|
||||
command_sep_patterns.push_back(p);
|
||||
}
|
||||
|
Reference in New Issue
Block a user