tiny.smg.tree.scanner.typescript

1pub const typescript = @import("typescript.zig");

1const std = @import("std");2const ecma = @import("ecma/root.zig");34pub const Token = enum(u16) {5    automatic_semicolon,6    template_chars,7    ternary_qmark,8    html_comment,9    logical_or,10    escape_sequence,11    regex_pattern,12    jsx_text,13    function_signature_automatic_semicolon,14    error_recovery,15};1617pub const Valid = packed struct(u16) {18    automatic_semicolon: bool = false,19    template_chars: bool = false,20    ternary_qmark: bool = false,21    html_comment: bool = false,22    logical_or: bool = false,23    escape_sequence: bool = false,24    regex_pattern: bool = false,25    jsx_text: bool = false,26    function_signature_automatic_semicolon: bool = false,27    error_recovery: bool = false,28    _: u6 = 0,29};3031pub const Scanner = struct {32    pub fn scan(_: *Scanner, lexer: anytype, valid: Valid) bool {33        if (valid.template_chars) {34            if (valid.automatic_semicolon) return false;35            return ecma.scanTemplateChars(lexer, @backingInt(Token.template_chars));36        }3738        if (valid.jsx_text and ecma.scanJsxText(lexer, @backingInt(Token.jsx_text))) return true;3940        if (valid.automatic_semicolon or valid.function_signature_automatic_semicolon) {41            var scanned_comment = false;42            const result = scanAutomaticSemicolon(lexer, valid, &scanned_comment);43            if (!result and !scanned_comment and valid.ternary_qmark and lexer.lookahead() == '?') {44                return scanTernaryQmark(lexer);45            }46            return result;47        }4849        if (valid.ternary_qmark) return scanTernaryQmark(lexer);5051        if (valid.html_comment and !valid.logical_or and !valid.escape_sequence and !valid.regex_pattern) {52            return ecma.scanHtmlComment(lexer, @backingInt(Token.html_comment));53        }5455        return false;56    }5758    pub fn serialize(_: Scanner, _: []u8) usize {59        return 0;60    }6162    pub fn deserialize(_: *Scanner, serialized: []const u8) void {63        std.debug.assert(serialized.len == 0);64    }65};6667fn scanWhitespaceAndComments(lexer: anytype, scanned_comment: *bool) bool {68    while (true) {69        while (ecma.isSpace(lexer.lookahead())) lexer.advance(true);7071        if (lexer.lookahead() != '/') return true;72        lexer.advance(true);7374        if (lexer.lookahead() == '/') {75            lexer.advance(true);76            while (lexer.lookahead() != 0 and lexer.lookahead() != '\n') lexer.advance(true);77            scanned_comment.* = true;78        } else if (lexer.lookahead() == '*') {79            lexer.advance(true);80            while (lexer.lookahead() != 0) {81                if (lexer.lookahead() == '*') {82                    lexer.advance(true);83                    if (lexer.lookahead() == '/') {84                        lexer.advance(true);85                        break;86                    }87                } else {88                    lexer.advance(true);89                }90            }91        } else {92            return false;93        }94    }95}9697fn scanAutomaticSemicolon(lexer: anytype, valid: Valid, scanned_comment: *bool) bool {98    lexer.setResult(@backingInt(Token.automatic_semicolon));99    lexer.markEnd();100101    while (true) {102        if (lexer.lookahead() == 0) return true;103        if (lexer.lookahead() == '}') {104            while (true) {105                lexer.advance(true);106                if (!ecma.isSpace(lexer.lookahead())) break;107            }108            if (lexer.lookahead() == ':') return valid.logical_or;109            return true;110        }111        if (!ecma.isSpace(lexer.lookahead())) return false;112        if (lexer.lookahead() == '\n') break;113        lexer.advance(true);114    }115116    lexer.advance(true);117    if (!scanWhitespaceAndComments(lexer, scanned_comment)) return false;118119    switch (lexer.lookahead()) {120        '`', ',', '.', ';', '*', '%', '>', '<', '=', '?', '^', '|', '&', '/', ':' => return false,121        '{' => {122            if (valid.function_signature_automatic_semicolon) return false;123        },124        '(', '[' => {125            if (valid.logical_or) return false;126        },127        '+' => {128            lexer.advance(true);129            return lexer.lookahead() == '+';130        },131        '-' => {132            lexer.advance(true);133            return lexer.lookahead() == '-';134        },135        '!' => {136            lexer.advance(true);137            return lexer.lookahead() != '=';138        },139        'i' => {140            lexer.advance(true);141            if (lexer.lookahead() != 'n') return true;142            lexer.advance(true);143            if (!ecma.isAlpha(lexer.lookahead())) return false;144            for ("stanceof") |letter| {145                if (lexer.lookahead() != letter) return true;146                lexer.advance(true);147            }148            if (!ecma.isAlpha(lexer.lookahead())) return false;149        },150        else => {},151    }152153    return true;154}155156fn scanTernaryQmark(lexer: anytype) bool {157    while (ecma.isSpace(lexer.lookahead())) lexer.advance(true);158    if (lexer.lookahead() != '?') return false;159160    lexer.advance(false);161    if (lexer.lookahead() == '?' or lexer.lookahead() == '.') return false;162163    lexer.markEnd();164    lexer.setResult(@backingInt(Token.ternary_qmark));165166    while (ecma.isSpace(lexer.lookahead())) lexer.advance(false);167168    if (lexer.lookahead() == ':' or lexer.lookahead() == ')' or lexer.lookahead() == ',') return false;169170    if (lexer.lookahead() == '.') {171        lexer.advance(false);172        return ecma.isDigit(lexer.lookahead());173    }174175    return true;176}