indention start

src/indent_helper.hpp

@@ -0,0 +1,65 @@
+#pragma once
+#include <stack>
+
+class IndentHelper
+{
+    std::stack<int> indent_stack;
+    int pending_dedents = 0;
+
+public:
+    IndentHelper() { indent_stack.push(0); }
+
+    int processLine(int spaces)
+    {
+        int current = indent_stack.top();
+
+        if (spaces > current)
+        {
+            indent_stack.push(spaces);
+            return 1;
+        }
+
+        if (spaces == current)
+        {
+            return 0;
+        }
+
+        pending_dedents = 0;
+        while (indent_stack.top() > spaces)
+        {
+            indent_stack.pop();
+            pending_dedents++;
+        }
+
+        if (indent_stack.top() != spaces)
+        {
+            throw std::runtime_error("Indentation error");
+        }
+
+        return -1;
+    }
+
+    bool hasPendingDedents() const
+    {
+        return pending_dedents > 0;
+    }
+
+    void consumePendingDedent()
+    {
+        if (pending_dedents > 0)
+        {
+            pending_dedents--;
+        }
+    }
+
+    int flushAllDedents()
+    {
+        int total_dedents = 0;
+        while (indent_stack.size() > 1)
+        {
+            indent_stack.pop();
+            total_dedents++;
+        }
+        return total_dedents;
+    }
+};

src/lexer.l

@@ -5,28 +5,64 @@
 
 %{
 #include "scanner.hpp"
+#include "indent_helper.hpp"
+
+IndentHelper indent_helper;
+bool at_line_start = true;
 %}
 
 %%
-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_IF;}
-\n              { return yy::parser::token::TOK_NEWLINE; }
-\(              { return yy::parser::token::TOK_L_PAREN; }
-\)              { 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); }
-[0-9]+          { yylval->as<std::string>() = yytext; return yy::parser::token::TOK_NUMBER; }
-[_a-zA-Z]+      { yylval->as<std::string>() = yytext; return yy::parser::token::TOK_IDENTIFIER; }
-<<EOF>>         { return yy::parser::token::TOK_EOF; }
-.               { /* the last rule, ignore anything else (there should be nothing) */}
 
-<STRING>\"\"    { yymore(); }
-<STRING>\"      { BEGIN(INITIAL); yylval->as<std::string>() = yytext; return yy::parser::token::TOK_STRING; }
-<STRING>.       { yymore(); }
-%%
+^[ \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 */ }

src/parser.y

@@ -29,7 +29,7 @@
 %token <int> NUMBER
 %token ASSIGN PLUS MINUS L_PAREN R_PAREN COLON
 %token KW_DEF KW_IF KW_RETURN KW_FOR KW_IN
-%token EOF NEWLINE INDENT_UP INDENT_DOWN
+%token EOF NEWLINE INDENT DEDENT
 
 
 %%