tiny.smg.tree.scanner.javascript

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};1415pub const Valid = packed struct(u8) {16    automatic_semicolon: bool = false,17    template_chars: bool = false,18    ternary_qmark: bool = false,19    html_comment: bool = false,20    logical_or: bool = false,21    escape_sequence: bool = false,22    regex_pattern: bool = false,23    jsx_text: bool = false,24};2526const WhitespaceResult = enum {27    reject,28    no_newline,29    accept,30};3132pub const Scanner = struct {33    pub fn scan(_: *Scanner, lexer: anytype, valid: Valid) bool {34        if (valid.template_chars) {35            if (valid.automatic_semicolon) return false;36            return ecma.scanTemplateChars(lexer, @backingInt(Token.template_chars));37        }3839        if (valid.jsx_text and ecma.scanJsxText(lexer, @backingInt(Token.jsx_text))) return true;4041        if (valid.automatic_semicolon) {42            var scanned_comment = false;43            const result = scanAutomaticSemicolon(lexer, !valid.logical_or, &scanned_comment);44            if (!result and !scanned_comment and valid.ternary_qmark and lexer.lookahead() == '?') {45                return scanTernaryQmark(lexer);46            }47            return result;48        }4950        if (valid.ternary_qmark) return scanTernaryQmark(lexer);5152        if (valid.html_comment and !valid.logical_or and !valid.escape_sequence and !valid.regex_pattern) {53            return ecma.scanHtmlComment(lexer, @backingInt(Token.html_comment));54        }5556        return false;57    }5859    pub fn serialize(_: Scanner, _: []u8) usize {60        return 0;61    }6263    pub fn deserialize(_: *Scanner, serialized: []const u8) void {64        std.debug.assert(serialized.len == 0);65    }66};6768fn scanWhitespaceAndComments(lexer: anytype, scanned_comment: *bool, consume: bool) WhitespaceResult {69    var saw_block_newline = false;7071    while (true) {72        while (ecma.isSpace(lexer.lookahead())) lexer.advance(true);7374        if (lexer.lookahead() != '/') return .accept;75        lexer.advance(true);7677        if (lexer.lookahead() == '/') {78            lexer.advance(true);79            while (lexer.lookahead() != 0 and !ecma.isNewline(lexer.lookahead())) lexer.advance(true);80            scanned_comment.* = true;81        } else if (lexer.lookahead() == '*') {82            lexer.advance(true);83            while (lexer.lookahead() != 0) {84                if (lexer.lookahead() == '*') {85                    lexer.advance(true);86                    if (lexer.lookahead() == '/') {87                        lexer.advance(true);88                        scanned_comment.* = true;89                        if (lexer.lookahead() != '/' and !consume) {90                            return if (saw_block_newline) .accept else .no_newline;91                        }92                        break;93                    }94                } else if (ecma.isNewline(lexer.lookahead())) {95                    saw_block_newline = true;96                    lexer.advance(true);97                } else {98                    lexer.advance(true);99                }100            }101        } else {102            return .reject;103        }104    }105}106107fn scanAutomaticSemicolon(lexer: anytype, comment_condition: bool, scanned_comment: *bool) bool {108    lexer.setResult(@backingInt(Token.automatic_semicolon));109    lexer.markEnd();110111    while (true) {112        if (lexer.lookahead() == 0) return true;113114        if (lexer.lookahead() == '/') {115            const result = scanWhitespaceAndComments(lexer, scanned_comment, false);116            if (result == .reject) return false;117            if (result == .accept and comment_condition and lexer.lookahead() != ',' and lexer.lookahead() != '=') {118                return true;119            }120        }121122        if (lexer.lookahead() == '}') return true;123        if (lexer.includedRangeStart()) return true;124        if (ecma.isNewline(lexer.lookahead())) break;125        if (!ecma.isSpace(lexer.lookahead())) return false;126        lexer.advance(true);127    }128129    lexer.advance(true);130    if (scanWhitespaceAndComments(lexer, scanned_comment, true) == .reject) return false;131132    switch (lexer.lookahead()) {133        '`', ',', ':', ';', '*', '%', '>', '<', '=', '[', '(', '?', '^', '|', '&', '/' => return false,134        '.' => {135            lexer.advance(true);136            return ecma.isDigit(lexer.lookahead());137        },138        '+' => {139            lexer.advance(true);140            return lexer.lookahead() == '+';141        },142        '-' => {143            lexer.advance(true);144            return lexer.lookahead() == '-';145        },146        '!' => {147            lexer.advance(true);148            return lexer.lookahead() != '=';149        },150        'i' => {151            lexer.advance(true);152            if (lexer.lookahead() != 'n') return true;153            lexer.advance(true);154            if (!ecma.isAlpha(lexer.lookahead())) return false;155            for ("stanceof") |letter| {156                if (lexer.lookahead() != letter) return true;157                lexer.advance(true);158            }159            if (!ecma.isAlpha(lexer.lookahead())) return false;160        },161        else => {},162    }163164    return true;165}166167fn scanTernaryQmark(lexer: anytype) bool {168    while (ecma.isSpace(lexer.lookahead())) lexer.advance(true);169    if (lexer.lookahead() != '?') return false;170171    lexer.advance(false);172    if (lexer.lookahead() == '?') return false;173174    lexer.markEnd();175    lexer.setResult(@backingInt(Token.ternary_qmark));176177    if (lexer.lookahead() == '.') {178        lexer.advance(false);179        return ecma.isDigit(lexer.lookahead());180    }181    return true;182}

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