python2ruby/src/lexer.l
2025-11-25 13:53:50 +01:00

68 lines
1.9 KiB
Text

%option c++
%option noyywrap
%option yyclass="Scanner"
%x STRING
%{
#include "scanner.hpp"
#include "indent_helper.hpp"
IndentHelper indent_helper;
bool at_line_start = true;
%}
%%
^[ \t]* {
if (!at_line_start) {
REJECT;
}
at_line_start = false;
int spaces = 0;
for (int i = 0; i < yyleng; i++)
spaces += (yytext[i] == '\t') ? 8 : 1;
int result = indent_helper.processLine(spaces);
if (result > 0) return yy::parser::token::TOK_INDENT;
if (result < 0) return yy::parser::token::TOK_DEDENT;
}
\n {
at_line_start = true;
return yy::parser::token::TOK_NEWLINE;
}
def { return yy::parser::token::TOK_KW_DEF; }
if { return yy::parser::token::TOK_KW_IF; }
return { return yy::parser::token::TOK_KW_RETURN; }
for { return yy::parser::token::TOK_KW_FOR; }
in { return yy::parser::token::TOK_KW_IN; }
\( { return yy::parser::token::TOK_L_PAREN; }
\) { return yy::parser::token::TOK_R_PAREN; }
\: { return yy::parser::token::TOK_COLON; }
\= { return yy::parser::token::TOK_ASSIGN; }
\+ { return yy::parser::token::TOK_PLUS; }
\- { return yy::parser::token::TOK_MINUS; }
\" { yymore(); BEGIN(STRING); }
<STRING>\"\" { yymore(); }
<STRING>\" { BEGIN(INITIAL); yylval->as<std::string>() = yytext; return yy::parser::token::TOK_STRING; }
<STRING>. { yymore(); }
[0-9]+ { yylval->as<std::string>() = yytext; return yy::parser::token::TOK_NUMBER; }
[A-Za-z_][A-Za-z0-9_]* { yylval->as<std::string>() = yytext; return yy::parser::token::TOK_IDENTIFIER; }
<<EOF>> {
if (indent_helper.hasPendingDedents()) {
indent_helper.consumeDedent();
return yy::parser::token::TOK_DEDENT;
}
int d;
if ((d = indent_helper.flushDedent()) != 0) return yy::parser::token::TOK_DEDENT;
return yy::parser::token::TOK_EOF;
}
. { /* ignore other characters */ }