fix(lexer): add scientific notation numbers (#253)

xensik · Feb 16, 2025 · 1df75e8 · 1df75e8
1 parent 79a240c
commit 1df75e8
Show file tree

Hide file tree

Showing 2 changed files with 42 additions and 6 deletions.
diff --git a/src/arc/lexer.cpp b/src/arc/lexer.cpp
@@ -427,16 +427,17 @@ auto lexer::lex() -> token
 
             auto dot = last == '.' ? 1 : 0;
             auto flt = 0;
+            auto exp = 0;
 
             while (true)
             {
                 if (reader_.ended())
                     break;
 
-                if (curr == '\'' && (last == '\'' || last == 'f' || last == '.'))
+                if (curr == '\'' && (last == '\'' || last == 'f' || last == '.' || last == 'e' || last == 'E'))
                     throw comp_error(loc_, "invalid number literal");
 
-                if ((curr == '.' || curr == 'f') && last == '\'')
+                if ((curr == '.' || curr == 'f' || curr == 'e' || curr == 'E') && last == '\'')
                     throw comp_error(loc_, "invalid number literal");
 
                 if (curr == '\'')
@@ -449,6 +450,22 @@ auto lexer::lex() -> token
                     flt++;
                 else if (curr == '.')
                     dot++;
+                else if (curr == 'e' || curr == 'E')
+                {
+                    exp++;
+                    if (exp > 1)
+                        throw comp_error(loc_, "invalid number literal");
+                    push(curr);
+                    advance();
+
+                    // TODO: check stream end
+                    if (curr == '+' || curr == '-')
+                    {
+                        push(curr);
+                        advance();
+                    }
+                    continue;
+                }
                 else if (!(curr > 47 && curr < 58))
                     break;
 
@@ -462,7 +479,8 @@ auto lexer::lex() -> token
             if (dot > 1 || flt > 1 || (flt && buffer_[buflen_ - 1] != 'f'))
                 throw comp_error(loc_, "invalid number literal");
 
-            if (dot || flt)
+            // TODO: exp can be int or float
+            if (dot || flt || exp)
                 return token{ token::FLT, spacing_, loc_, std::string{ &buffer_[0], buflen_ } };
 
             return token{ token::INT, spacing_, loc_, std::string{ &buffer_[0], buflen_ } };

diff --git a/src/gsc/lexer.cpp b/src/gsc/lexer.cpp
@@ -427,16 +427,17 @@ auto lexer::lex() -> token
 
             auto dot = last == '.' ? 1 : 0;
             auto flt = 0;
+            auto exp = 0;
 
             while (true)
             {
                 if (reader_.ended())
                     break;
 
-                if (curr == '\'' && (last == '\'' || last == 'f' || last == '.'))
+                if (curr == '\'' && (last == '\'' || last == 'f' || last == '.' || last == 'e' || last == 'E'))
                     throw comp_error(loc_, "invalid number literal");
 
-                if ((curr == '.' || curr == 'f') && last == '\'')
+                if ((curr == '.' || curr == 'f' || curr == 'e' || curr == 'E') && last == '\'')
                     throw comp_error(loc_, "invalid number literal");
 
                 if (curr == '\'')
@@ -449,6 +450,22 @@ auto lexer::lex() -> token
                     flt++;
                 else if (curr == '.')
                     dot++;
+                else if (curr == 'e' || curr == 'E')
+                {
+                    exp++;
+                    if (exp > 1)
+                        throw comp_error(loc_, "invalid number literal");
+                    push(curr);
+                    advance();
+
+                    // TODO: check stream end
+                    if (curr == '+' || curr == '-')
+                    {
+                        push(curr);
+                        advance();
+                    }
+                    continue;
+                }
                 else if (!(curr > 47 && curr < 58))
                     break;
 
@@ -462,7 +479,8 @@ auto lexer::lex() -> token
             if (dot > 1 || flt > 1 || (flt && buffer_[buflen_ - 1] != 'f'))
                 throw comp_error(loc_, "invalid number literal");
 
-            if (dot || flt)
+            // TODO: exp can be int or float
+            if (dot || flt || exp)
                 return token{ token::FLT, spacing_, loc_, std::string{ &buffer_[0], buflen_ } };
 
             return token{ token::INT, spacing_, loc_, std::string{ &buffer_[0], buflen_ } };