@devaloop/devalang 0.0.1-beta.1 → 0.0.1-beta.2

package/rust/core/lexer/handler/driver.rs

@@ -1,337 +1,337 @@
-use crate::core::lexer::{
-    handler::{
-        arrow::handle_arrow_lexer,
-        at::handle_at_lexer,
-        brace::{handle_lbrace_lexer, handle_rbrace_lexer},
-        colon::handle_colon_lexer,
-        comment::handle_comment_lexer,
-        dot::handle_dot_lexer,
-        identifier::handle_identifier_lexer,
-        indent::handle_indent_lexer,
-        newline::handle_newline_lexer,
-        number::handle_number_lexer,
-        operator::handle_operator_lexer,
-        parenthesis::{handle_lparen_lexer, handle_rparen_lexer},
-        slash::handle_slash_lexer,
-        string::handle_string_lexer,
-    },
-    token::{Token, TokenKind},
-};
-
-fn advance_char<I: Iterator<Item = char>>(
-    chars: &mut std::iter::Peekable<I>,
-    _line: &mut usize,
-    column: &mut usize,
-) -> Option<char> {
-    for c in chars.by_ref() {
-        if c == '\r' {
-            continue;
-        } else if c == '\n' {
-            // newline: don't increment column
-        } else {
-            *column += 1;
-        }
-        return Some(c);
-    }
-    None
-}
-
-pub fn handle_content_lexing(content: String) -> Result<Vec<Token>, String> {
-    let mut tokens = Vec::new();
-
-    let mut line = 1;
-    let mut column = 1;
-
-    let mut indent_stack: Vec<usize> = vec![0];
-    let mut current_indent = 0;
-    let mut at_line_start = true;
-
-    let mut chars = content.chars().peekable();
-
-    while chars.peek().is_some() {
-        if at_line_start {
-            handle_indent_lexer(
-                &mut chars,
-                &mut current_indent,
-                &mut indent_stack,
-                &mut tokens,
-                &mut line,
-                &mut column,
-            );
-
-            at_line_start = false;
-        }
-
-        let Some(ch) = advance_char(&mut chars, &mut line, &mut column) else {
-            break;
-        };
-
-        match ch {
-            '\n' => {
-                handle_newline_lexer(
-                    ch,
-                    &mut chars,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                    &mut at_line_start,
-                    &mut current_indent,
-                );
-            }
-            ' ' | '\t' => {
-                // Already handled by indent_lexer
-            }
-            '#' => {
-                handle_comment_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            ':' => {
-                handle_colon_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '=' | '!' | '<' | '>' | '+' | '*' => {
-                handle_operator_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '/' => {
-                handle_slash_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '-' => {
-                handle_arrow_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-                // If not parsed as arrow or number, fallback as Minus token
-                if let Some(last) = tokens.last() {
-                    if last.kind == TokenKind::Unknown && last.lexeme == "-" {
-                        // replace last with Minus
-                        let _ = tokens.pop();
-                        tokens.push(Token {
-                            kind: TokenKind::Minus,
-                            lexeme: "-".to_string(),
-                            line,
-                            column,
-                            indent: current_indent,
-                        });
-                    }
-                }
-            }
-            '{' => {
-                handle_lbrace_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '}' => {
-                handle_rbrace_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '[' => {
-                tokens.push(Token {
-                    kind: TokenKind::LBracket,
-                    lexeme: "[".to_string(),
-                    line,
-                    column,
-                    indent: current_indent,
-                });
-            }
-            ']' => {
-                tokens.push(Token {
-                    kind: TokenKind::RBracket,
-                    lexeme: "]".to_string(),
-                    line,
-                    column,
-                    indent: current_indent,
-                });
-            }
-            ',' => {
-                tokens.push(Token {
-                    kind: TokenKind::Comma,
-                    lexeme: ",".to_string(),
-                    line,
-                    column,
-                    indent: current_indent,
-                });
-            }
-            '(' => {
-                handle_lparen_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            ')' => {
-                handle_rparen_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '.' => {
-                handle_dot_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '@' => {
-                handle_at_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '$' => {
-                // Treat `$` as start of a special identifier like `$env` or `$math`
-                let mut ident = String::from("$");
-                while let Some(&c) = chars.peek() {
-                    if c.is_ascii_alphanumeric() || c == '_' {
-                        ident.push(c);
-                        chars.next();
-                        column += 1;
-                    } else {
-                        break;
-                    }
-                }
-                tokens.push(Token {
-                    kind: TokenKind::Identifier,
-                    lexeme: ident,
-                    line,
-                    column,
-                    indent: current_indent,
-                });
-            }
-            '0'..='9' => {
-                handle_number_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            'a'..='z' | 'A'..='Z' | '_' => {
-                handle_identifier_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            '"' | '\'' => {
-                handle_string_lexer(
-                    ch,
-                    &mut chars,
-                    &mut current_indent,
-                    &mut indent_stack,
-                    &mut tokens,
-                    &mut line,
-                    &mut column,
-                );
-            }
-            _ => {
-                // Ignore unknown char
-            }
-        }
-    }
-
-    while indent_stack.len() > 1 {
-        indent_stack.pop();
-        current_indent = *indent_stack.last().unwrap();
-        tokens.push(Token {
-            kind: TokenKind::Dedent,
-            lexeme: String::new(),
-            line,
-            column,
-            indent: current_indent,
-        });
-    }
-
-    tokens.push(Token {
-        kind: TokenKind::EOF,
-        lexeme: String::new(),
-        line: line + 1,
-        column: 0,
-        indent: 0,
-    });
-
-    // NOTE: Debug only
-    // for token in &tokens {
-    //     println!(
-    //         "{:?} @ line {}, col {}, indent {}",
-    //         token.kind,
-    //         token.line,
-    //         token.column,
-    //         token.indent
-    //     );
-    // }
-
-    Ok(tokens)
-}
+use crate::core::lexer::{
+    handler::{
+        arrow::handle_arrow_lexer,
+        at::handle_at_lexer,
+        brace::{handle_lbrace_lexer, handle_rbrace_lexer},
+        colon::handle_colon_lexer,
+        comment::handle_comment_lexer,
+        dot::handle_dot_lexer,
+        identifier::handle_identifier_lexer,
+        indent::handle_indent_lexer,
+        newline::handle_newline_lexer,
+        number::handle_number_lexer,
+        operator::handle_operator_lexer,
+        parenthesis::{handle_lparen_lexer, handle_rparen_lexer},
+        slash::handle_slash_lexer,
+        string::handle_string_lexer,
+    },
+    token::{Token, TokenKind},
+};
+
+fn advance_char<I: Iterator<Item = char>>(
+    chars: &mut std::iter::Peekable<I>,
+    _line: &mut usize,
+    column: &mut usize,
+) -> Option<char> {
+    for c in chars.by_ref() {
+        if c == '\r' {
+            continue;
+        } else if c == '\n' {
+            // newline: don't increment column
+        } else {
+            *column += 1;
+        }
+        return Some(c);
+    }
+    None
+}
+
+pub fn handle_content_lexing(content: String) -> Result<Vec<Token>, String> {
+    let mut tokens = Vec::new();
+
+    let mut line = 1;
+    let mut column = 1;
+
+    let mut indent_stack: Vec<usize> = vec![0];
+    let mut current_indent = 0;
+    let mut at_line_start = true;
+
+    let mut chars = content.chars().peekable();
+
+    while chars.peek().is_some() {
+        if at_line_start {
+            handle_indent_lexer(
+                &mut chars,
+                &mut current_indent,
+                &mut indent_stack,
+                &mut tokens,
+                &mut line,
+                &mut column,
+            );
+
+            at_line_start = false;
+        }
+
+        let Some(ch) = advance_char(&mut chars, &mut line, &mut column) else {
+            break;
+        };
+
+        match ch {
+            '\n' => {
+                handle_newline_lexer(
+                    ch,
+                    &mut chars,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                    &mut at_line_start,
+                    &mut current_indent,
+                );
+            }
+            ' ' | '\t' => {
+                // Already handled by indent_lexer
+            }
+            '#' => {
+                handle_comment_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            ':' => {
+                handle_colon_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '=' | '!' | '<' | '>' | '+' | '*' => {
+                handle_operator_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '/' => {
+                handle_slash_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '-' => {
+                handle_arrow_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+                // If not parsed as arrow or number, fallback as Minus token
+                if let Some(last) = tokens.last() {
+                    if last.kind == TokenKind::Unknown && last.lexeme == "-" {
+                        // replace last with Minus
+                        let _ = tokens.pop();
+                        tokens.push(Token {
+                            kind: TokenKind::Minus,
+                            lexeme: "-".to_string(),
+                            line,
+                            column,
+                            indent: current_indent,
+                        });
+                    }
+                }
+            }
+            '{' => {
+                handle_lbrace_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '}' => {
+                handle_rbrace_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '[' => {
+                tokens.push(Token {
+                    kind: TokenKind::LBracket,
+                    lexeme: "[".to_string(),
+                    line,
+                    column,
+                    indent: current_indent,
+                });
+            }
+            ']' => {
+                tokens.push(Token {
+                    kind: TokenKind::RBracket,
+                    lexeme: "]".to_string(),
+                    line,
+                    column,
+                    indent: current_indent,
+                });
+            }
+            ',' => {
+                tokens.push(Token {
+                    kind: TokenKind::Comma,
+                    lexeme: ",".to_string(),
+                    line,
+                    column,
+                    indent: current_indent,
+                });
+            }
+            '(' => {
+                handle_lparen_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            ')' => {
+                handle_rparen_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '.' => {
+                handle_dot_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '@' => {
+                handle_at_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '$' => {
+                // Treat `$` as start of a special identifier like `$env` or `$math`
+                let mut ident = String::from("$");
+                while let Some(&c) = chars.peek() {
+                    if c.is_ascii_alphanumeric() || c == '_' {
+                        ident.push(c);
+                        chars.next();
+                        column += 1;
+                    } else {
+                        break;
+                    }
+                }
+                tokens.push(Token {
+                    kind: TokenKind::Identifier,
+                    lexeme: ident,
+                    line,
+                    column,
+                    indent: current_indent,
+                });
+            }
+            '0'..='9' => {
+                handle_number_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            'a'..='z' | 'A'..='Z' | '_' => {
+                handle_identifier_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            '"' | '\'' => {
+                handle_string_lexer(
+                    ch,
+                    &mut chars,
+                    &mut current_indent,
+                    &mut indent_stack,
+                    &mut tokens,
+                    &mut line,
+                    &mut column,
+                );
+            }
+            _ => {
+                // Ignore unknown char
+            }
+        }
+    }
+
+    while indent_stack.len() > 1 {
+        indent_stack.pop();
+        current_indent = *indent_stack.last().unwrap();
+        tokens.push(Token {
+            kind: TokenKind::Dedent,
+            lexeme: String::new(),
+            line,
+            column,
+            indent: current_indent,
+        });
+    }
+
+    tokens.push(Token {
+        kind: TokenKind::EOF,
+        lexeme: String::new(),
+        line: line + 1,
+        column: 0,
+        indent: 0,
+    });
+
+    // NOTE: Debug only
+    // for token in &tokens {
+    //     println!(
+    //         "{:?} @ line {}, col {}, indent {}",
+    //         token.kind,
+    //         token.line,
+    //         token.column,
+    //         token.indent
+    //     );
+    // }
+
+    Ok(tokens)
+}