js_regex/

validator.rs

// Copyright (C) 2020 Quentin M. Kniep <hello@quentinkniep.com>
// Distributed under terms of the MIT license.

use std::collections::HashSet;
use std::ops::{Deref, DerefMut};

use crate::reader::Reader;
use crate::unicode::*;

fn is_syntax_character(cp: char) -> bool {
    return cp == '^'
        || cp == '$'
        || cp == '\\'
        || cp == '.'
        || cp == '*'
        || cp == '+'
        || cp == '?'
        || cp == '('
        || cp == ')'
        || cp == '['
        || cp == ']'
        || cp == '{'
        || cp == '}'
        || cp == '|';
}

fn is_unicode_property_name_character(cp: char) -> bool {
    cp.is_ascii_alphabetic() || cp == '_'
}

fn is_unicode_property_value_character(cp: char) -> bool {
    is_unicode_property_name_character(cp) || cp.is_digit(10)
}

fn is_regexp_identifier_start(cp: char) -> bool {
    is_id_start(cp) || cp == '$' || cp == '_'
}

fn is_regexp_identifier_part(cp: char) -> bool {
    is_id_continue(cp) ||
    cp == '$' ||
    cp == '_' ||
    cp == '\u{200c}' ||  // unicode zero-width non-joiner
    cp == '\u{200d}' // unicode zero-width joiner
}

fn is_id_start(cp: char) -> bool {
    if (cp as u32) < 0x41 {
        false
    } else if (cp as u32) < 0x5b {
        true
    } else if (cp as u32) < 0x61 {
        false
    } else if (cp as u32) < 0x7b {
        true
    } else {
        is_large_id_start(cp)
    }
}

fn is_id_continue(cp: char) -> bool {
    if (cp as u32) < 0x30 {
        false
    } else if (cp as u32) < 0x3a {
        true
    } else if (cp as u32) < 0x41 {
        false
    } else if (cp as u32) < 0x5b {
        true
    } else if (cp as u32) == 0x5f {
        true
    } else if (cp as u32) < 0x61 {
        false
    } else if (cp as u32) < 0x7b {
        true
    } else {
        is_large_id_start(cp) || is_large_id_continue(cp)
    }
}

fn is_valid_unicode(cp: i64) -> bool {
    cp <= 0x10ffff
}

fn is_lead_surrogate(cp: i64) -> bool {
    cp >= 0xd800 && cp <= 0xdbff
}

fn is_trail_surrogate(cp: i64) -> bool {
    cp >= 0xdc00 && cp <= 0xdfff
}

fn combine_surrogate_pair(lead: i64, trail: i64) -> i64 {
    (lead - 0xd800) * 0x400 + (trail - 0xdc00) + 0x10000
}

#[derive(Clone, Copy, Debug, Ord, PartialOrd, Eq, PartialEq, Hash)]
pub enum EcmaVersion {
    ES5,
    ES2015,
    ES2016,
    ES2017,
    ES2018,
    ES2019,
    ES2020,
    ES2021,
}

#[derive(Debug)]
pub struct EcmaRegexValidator {
    reader: Reader,
    strict: bool,
    ecma_version: EcmaVersion,
    u_flag: bool,
    n_flag: bool,
    last_int_value: i64,
    last_min_value: i64,
    last_max_value: i64,
    last_str_value: String,
    last_key_value: String,
    last_val_value: String,
    last_assertion_is_quantifiable: bool,
    num_capturing_parens: u32,
    group_names: HashSet<String>,
    backreference_names: HashSet<String>,
}

impl Deref for EcmaRegexValidator {
    type Target = Reader;

    fn deref(&self) -> &Self::Target {
        &self.reader
    }
}

impl DerefMut for EcmaRegexValidator {
    fn deref_mut(&mut self) -> &mut Self::Target {
        &mut self.reader
    }
}

impl EcmaRegexValidator {
    pub fn new(ecma_version: EcmaVersion) -> Self {
        EcmaRegexValidator {
            reader: Reader::new(),
            strict: false,
            ecma_version,
            u_flag: false,
            n_flag: false,
            last_int_value: 0,
            last_min_value: 0,
            last_max_value: 0,
            last_str_value: "".to_string(),
            last_key_value: "".to_string(),
            last_val_value: "".to_string(),
            last_assertion_is_quantifiable: false,
            num_capturing_parens: 0,
            group_names: HashSet::new(),
            backreference_names: HashSet::new(),
        }
    }

    /// Validates flags of a EcmaScript regular expression.
    pub fn validate_flags(&self, flags: &str) -> Result<(), String> {
        let mut existing_flags = HashSet::<char>::new();

        for flag in flags.chars() {
            if existing_flags.contains(&flag) {
                return Err(format!("Duplicated flag {}", flag));
            }
            existing_flags.insert(flag);

            if flag == 'g'
                || flag == 'i'
                || flag == 'm'
                || (flag == 'u' && self.ecma_version >= EcmaVersion::ES2015)
                || (flag == 'y' && self.ecma_version >= EcmaVersion::ES2015)
                || (flag == 's' && self.ecma_version >= EcmaVersion::ES2018)
            {
                // do nothing
            } else {
                return Err(format!("Invalid flag {}", flag));
            }
        }
        Ok(())
    }

    /// Validates the pattern of a EcmaScript regular expression.
    pub fn validate_pattern(&mut self, source: &str, u_flag: bool) -> Result<(), String> {
        self.strict = u_flag; // TODO: allow toggling strict independently of u flag
        self.u_flag = u_flag && self.ecma_version >= EcmaVersion::ES2015;
        self.n_flag = u_flag && self.ecma_version >= EcmaVersion::ES2018;
        //self.reset(source, 0, source.len(), u_flag);
        self.reset(source, 0, source.chars().count(), u_flag);
        self.consume_pattern()?;

        if !self.n_flag && self.ecma_version >= EcmaVersion::ES2018 && self.group_names.len() > 0 {
            self.n_flag = true;
            self.rewind(0);
            self.consume_pattern()?;
        }

        return Ok(());
    }

    /// Validate the next characters as a RegExp `Pattern` production.
    /// ```grammar
    /// Pattern[U, N]::
    ///     Disjunction[?U, ?N]
    /// ```
    fn consume_pattern(&mut self) -> Result<(), String> {
        self.num_capturing_parens = self.count_capturing_parens();
        self.group_names.clear();
        self.backreference_names.clear();

        self.consume_disjunction()?;

        if let Some(cp) = self.code_point_with_offset(0) {
            if cp == ')' {
                return Err("Unmatched ')'".to_string());
            } else if cp == '\\' {
                return Err("\\ at end of pattern".to_string());
            } else if cp == ']' || cp == '}' {
                return Err("Lone quantifier brackets".to_string());
            }
            return Err(format!("Unexpected character {}", cp));
        }

        for name in self.backreference_names.difference(&self.group_names) {
            return Err(format!("Invalid named capture referenced: {}", name));
        }
        return Ok(());
    }

    /// Validate the next characters as a RegExp `Disjunction` production.
    /// ```grammar
    /// Disjunction[U, N]::
    ///      Alternative[?U, ?N]
    ///      Alternative[?U, ?N] `|` Disjunction[?U, ?N]
    /// ```
    fn consume_disjunction(&mut self) -> Result<(), String> {
        self.consume_alternative()?;
        while self.eat('|') {
            self.consume_alternative()?;
        }

        if self.consume_quantifier(true)? {
            return Err("Nothing to repeat".to_string());
        } else if self.eat('{') {
            return Err("Lone quantifier brackets".to_string());
        }
        return Ok(());
    }

    /// Validate the next characters as a RegExp `Alternative` production.
    /// ```grammar
    /// Alternative[U, N]::
    ///      ε
    ///      Alternative[?U, ?N] Term[?U, ?N]
    /// ```
    fn consume_alternative(&mut self) -> Result<(), String> {
        while self.code_point_with_offset(0).is_some() && self.consume_term()? {
            // do nothing
        }
        Ok(())
    }

    /// Validate the next characters as a RegExp `Term` production if possible.
    /// ```grammar
    /// Term[U, N]::
    ///      [strict] Assertion[+U, ?N]
    ///      [strict] Atom[+U, ?N]
    ///      [strict] Atom[+U, ?N] Quantifier
    ///      [annexB][+U] Assertion[+U, ?N]
    ///      [annexB][+U] Atom[+U, ?N]
    ///      [annexB][+U] Atom[+U, ?N] Quantifier
    ///      [annexB][~U] QuantifiableAssertion[?N] Quantifier
    ///      [annexB][~U] Assertion[~U, ?N]
    ///      [annexB][~U] ExtendedAtom[?N] Quantifier
    ///      [annexB][~U] ExtendedAtom[?N]
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_term(&mut self) -> Result<bool, String> {
        if self.u_flag || self.strict {
            return Ok(self.consume_assertion()?
                || (self.consume_atom()? && self.consume_optional_quantifier()?));
        }
        return Ok((self.consume_assertion()?
            && (!self.last_assertion_is_quantifiable || self.consume_optional_quantifier()?))
            || (self.consume_extended_atom()? && self.consume_optional_quantifier()?));
    }

    fn consume_optional_quantifier(&mut self) -> Result<bool, String> {
        self.consume_quantifier(false)?;
        Ok(true)
    }

    /// Validate the next characters as a RegExp `Term` production if possible.
    /// Set `self.last_assertion_is_quantifiable` if the consumed assertion was a
    /// `QuantifiableAssertion` production.
    /// ```grammar
    /// Assertion[U, N]::
    ///      `^`
    ///      `$`
    ///      `\b`
    ///      `\B`
    ///      [strict] `(?=` Disjunction[+U, ?N] `)`
    ///      [strict] `(?!` Disjunction[+U, ?N] `)`
    ///      [annexB][+U] `(?=` Disjunction[+U, ?N] `)`
    ///      [annexB][+U] `(?!` Disjunction[+U, ?N] `)`
    ///      [annexB][~U] QuantifiableAssertion[?N]
    ///      `(?<=` Disjunction[?U, ?N] `)`
    ///      `(?<!` Disjunction[?U, ?N] `)`
    /// QuantifiableAssertion[N]::
    ///      `(?=` Disjunction[~U, ?N] `)`
    ///      `(?!` Disjunction[~U, ?N] `)`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_assertion(&mut self) -> Result<bool, String> {
        let start = self.index();
        self.last_assertion_is_quantifiable = false;

        if self.eat('^') || self.eat('$') || self.eat2('\\', 'B') || self.eat2('\\', 'b') {
            return Ok(true);
        }

        // Lookahead / Lookbehind
        if self.eat2('(', '?') {
            let lookbehind = self.ecma_version >= EcmaVersion::ES2018 && self.eat('<');
            let mut flag = self.eat('=');
            if !flag {
                flag = self.eat('!');
            }
            if flag {
                self.consume_disjunction()?;
                if !self.eat(')') {
                    return Err("Unterminated group".to_string());
                }
                self.last_assertion_is_quantifiable = !lookbehind && !self.strict;
                return Ok(true);
            }
            self.rewind(start);
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `Quantifier` production if possible.
    /// ```grammar
    /// Quantifier::
    ///      QuantifierPrefix
    ///      QuantifierPrefix `?`
    /// QuantifierPrefix::
    ///      `*`
    ///      `+`
    ///      `?`
    ///      `{` DecimalDigits `}`
    ///      `{` DecimalDigits `,}`
    ///      `{` DecimalDigits `,` DecimalDigits `}`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_quantifier(&mut self, no_consume: bool) -> Result<bool, String> {
        // QuantifierPrefix
        if !self.eat('*')
            && !self.eat('+')
            && !self.eat('?')
            && !self.eat_braced_quantifier(no_consume)?
        {
            return Ok(false);
        }

        self.eat('?');
        return Ok(true);
    }

    /// Eats the next characters as the following alternatives if possible.
    /// Sets `self.last_min_value` and `self.last_max_value` if it consumed the next characters
    /// successfully.
    /// ```grammar
    ///      `{` DecimalDigits `}`
    ///      `{` DecimalDigits `,}`
    ///      `{` DecimalDigits `,` DecimalDigits `}`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn eat_braced_quantifier(&mut self, no_error: bool) -> Result<bool, &str> {
        let start = self.index();
        if self.eat('{') {
            self.last_min_value = 0;
            self.last_max_value = i64::MAX;
            if self.eat_decimal_digits() {
                self.last_min_value = self.last_int_value;
                self.last_max_value = self.last_int_value;
                if self.eat(',') {
                    self.last_max_value = if self.eat_decimal_digits() {
                        self.last_int_value
                    } else {
                        i64::MAX
                    }
                }
                if self.eat('}') {
                    if !no_error && self.last_max_value < self.last_min_value {
                        return Err("numbers out of order in {} quantifier");
                    }
                    return Ok(true);
                }
            }
            if !no_error && (self.u_flag || self.strict) {
                return Err("Incomplete quantifier");
            }
            self.rewind(start);
        }
        return Ok(false);
    }

    /// Validate the next characters as a RegExp `Atom` production if possible.
    /// ```grammar
    /// Atom[U, N]::
    ///      PatternCharacter
    ///      `.`
    ///      `\\` AtomEscape[?U, ?N]
    ///      CharacterClass[?U]
    ///      `(?:` Disjunction[?U, ?N] )
    ///      `(` GroupSpecifier[?U] Disjunction[?U, ?N] `)`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_atom(&mut self) -> Result<bool, String> {
        Ok(self.consume_pattern_character()
            || self.consume_dot()
            || self.consume_reverse_solidus_atom_escape()?
            || self.consume_character_class()?
            || self.consume_uncapturing_group()?
            || self.consume_capturing_group()?)
    }

    /// Validate the next characters as the following alternatives if possible.
    /// ```grammar
    ///      `.`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_dot(&mut self) -> bool {
        if self.eat('.') {
            return true;
        }
        return false;
    }

    /// Validate the next characters as the following alternatives if possible.
    /// ```grammar
    ///      `\\` AtomEscape[?U, ?N]
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_reverse_solidus_atom_escape(&mut self) -> Result<bool, String> {
        let start = self.index();
        if self.eat('\\') {
            if self.consume_atom_escape()? {
                return Ok(true);
            }
            self.rewind(start);
        }
        return Ok(false);
    }

    /// Validate the next characters as the following alternatives if possible.
    /// ```grammar
    ///      `(?:` Disjunction[?U, ?N] )
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_uncapturing_group(&mut self) -> Result<bool, String> {
        if self.eat3('(', '?', ':') {
            self.consume_disjunction()?;
            if !self.eat(')') {
                return Err("Unterminated group".to_string());
            }
            return Ok(true);
        }
        return Ok(false);
    }

    /// Validate the next characters as the following alternatives if possible.
    /// ```grammar
    ///      `(` GroupSpecifier[?U] Disjunction[?U, ?N] `)`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_capturing_group(&mut self) -> Result<bool, String> {
        if !self.eat('(') {
            return Ok(false);
        }

        if self.ecma_version >= EcmaVersion::ES2018 {
            self.consume_group_specifier()?;
        } else if self.code_point_with_offset(0) == Some('?') {
            return Err("Invalid group".to_string());
        }

        self.consume_disjunction()?;
        if !self.eat(')') {
            return Err("Unterminated group".to_string());
        }
        Ok(true)
    }

    /// Validate the next characters as a RegExp `ExtendedAtom` production if possible.
    /// ```grammar
    /// ExtendedAtom[N]::
    ///      `.`
    ///      `\` AtomEscape[~U, ?N]
    ///      `\` [lookahead = c]
    ///      CharacterClass[~U]
    ///      `(?:` Disjunction[~U, ?N] `)`
    ///      `(` Disjunction[~U, ?N] `)`
    ///      InvalidBracedQuantifier
    ///      ExtendedPatternCharacter
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_extended_atom(&mut self) -> Result<bool, String> {
        Ok(self.eat('.')
            || self.consume_reverse_solidus_atom_escape()?
            || self.consume_reverse_solidus_followed_by_c()
            || self.consume_character_class()?
            || self.consume_uncapturing_group()?
            || self.consume_capturing_group()?
            || self.consume_invalid_braced_quantifier()?
            || self.consume_extended_pattern_character())
    }

    /// Validate the next characters as the following alternatives if possible.
    /// ```grammar
    ///      `\` [lookahead = c]
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_reverse_solidus_followed_by_c(&mut self) -> bool {
        if self.code_point_with_offset(0) == Some('\\')
            && self.code_point_with_offset(1) == Some('c')
        {
            self.last_int_value = '\\' as i64;
            self.advance();
            return true;
        }
        return false;
    }

    /// Validate the next characters as a RegExp `InvalidBracedQuantifier`
    /// production if possible.
    /// ```grammar
    /// InvalidBracedQuantifier::
    ///      `{` DecimalDigits `}`
    ///      `{` DecimalDigits `,}`
    ///      `{` DecimalDigits `,` DecimalDigits `}`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_invalid_braced_quantifier(&mut self) -> Result<bool, &str> {
        if self.eat_braced_quantifier(true)? {
            return Err("Nothing to repeat");
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `PatternCharacter` production if
    /// possible.
    /// ```grammar
    /// PatternCharacter::
    ///      SourceCharacter but not SyntaxCharacter
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_pattern_character(&mut self) -> bool {
        if let Some(cp) = self.code_point_with_offset(0) {
            if !is_syntax_character(cp) {
                self.advance();
                return true;
            }
        }
        return false;
    }

    /// Validate the next characters as a RegExp `ExtendedPatternCharacter`
    /// production if possible.
    /// ```grammar
    /// ExtendedPatternCharacter::
    ///      SourceCharacter but not one of ^ $ \ . * + ? ( ) [ |
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_extended_pattern_character(&mut self) -> bool {
        if let Some(cp) = self.code_point_with_offset(0) {
            if cp != '^'
                && cp != '$'
                && cp != '\\'
                && cp != '.'
                && cp != '*'
                && cp != '+'
                && cp != '?'
                && cp != '('
                && cp != ')'
                && cp != '['
                && cp != '|'
            {
                self.advance();
                return true;
            }
        }
        return false;
    }

    /// Validate the next characters as a RegExp `GroupSpecifier` production.
    /// Set `self.last_str_value` if the group name existed.
    /// ```grammar
    /// GroupSpecifier[U]::
    ///      ε
    ///      `?` GroupName[?U]
    /// ```
    /// Returns `true` if the group name existed.
    fn consume_group_specifier(&mut self) -> Result<bool, String> {
        if self.eat('?') {
            if self.eat_group_name()? {
                if !self.group_names.contains(&self.last_str_value) {
                    self.group_names.insert(self.last_str_value.clone());
                    return Ok(true);
                }
                return Err("Duplicate capture group name".to_string());
            }
            return Err("Invalid group".to_string());
        }
        return Ok(false);
    }

    /// Validate the next characters as a RegExp `AtomEscape` production if possible.
    /// ```grammar
    /// AtomEscape[U, N]::
    ///      [strict] DecimalEscape
    ///      [annexB][+U] DecimalEscape
    ///      [annexB][~U] DecimalEscape but only if the CapturingGroupNumber of DecimalEscape is <= NcapturingParens
    ///      CharacterClassEscape[?U]
    ///      [strict] CharacterEscape[?U]
    ///      [annexB] CharacterEscape[?U, ?N]
    ///      [+N] `k` GroupName[?U]
    /// ```
    /// Returns `Ok(true)` if it consumed the next characters successfully.
    fn consume_atom_escape(&mut self) -> Result<bool, String> {
        if self.consume_backreference()?
            || self.consume_character_class_escape()?
            || self.consume_character_escape()?
            || (self.n_flag && self.consume_k_group_name()?)
        {
            return Ok(true);
        }
        if self.strict || self.u_flag {
            return Err("Invalid escape".to_string());
        }
        return Ok(false);
    }

    /// Validate the next characters as the follwoing alternatives if possible.
    /// ```grammar
    ///      [strict] DecimalEscape
    ///      [annexB][+U] DecimalEscape
    ///      [annexB][~U] DecimalEscape but only if the CapturingGroupNumber of DecimalEscape is <= NcapturingParens
    /// ```
    /// Returns `Ok(true)` if it consumed the next characters successfully.
    fn consume_backreference(&mut self) -> Result<bool, &str> {
        let start = self.index();
        if self.eat_decimal_escape() {
            if self.last_int_value <= self.num_capturing_parens as i64 {
                return Ok(true);
            } else if self.strict || self.u_flag {
                return Err("Invalid escape");
            }
            self.rewind(start);
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `DecimalEscape` production if possible.
    /// Set `-1` to `self.last_int_value` as meaning of a character set if it ate the next
    /// characters successfully.
    /// ```grammar
    /// CharacterClassEscape[U]::
    ///      `d`
    ///      `D`
    ///      `s`
    ///      `S`
    ///      `w`
    ///      `W`
    ///      [+U] `p{` UnicodePropertyValueExpression `}`
    ///      [+U] `P{` UnicodePropertyValueExpression `}`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_character_class_escape(&mut self) -> Result<bool, String> {
        if self.eat('d')
            || self.eat('D')
            || self.eat('s')
            || self.eat('S')
            || self.eat('w')
            || self.eat('W')
        {
            self.last_int_value = -1;
            return Ok(true);
        }

        if self.u_flag
            && self.ecma_version >= EcmaVersion::ES2018
            && (self.eat('p') || self.eat('P'))
        {
            self.last_int_value = -1;
            if self.eat('{') && self.eat_unicode_property_value_expression()? && self.eat('}') {
                return Ok(true);
            }
            return Err("Invalid property name".to_string());
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `CharacterEscape` production if possible.
    /// ```grammar
    /// CharacterEscape[U, N]::
    ///      ControlEscape
    ///      `c` ControlLetter
    ///      `0` [lookahead ∉ DecimalDigit]
    ///      HexEscapeSequence
    ///      RegExpUnicodeEscapeSequence[?U]
    ///      [annexB][~U] LegacyOctalEscapeSequence
    ///      IdentityEscape[?U, ?N]
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_character_escape(&mut self) -> Result<bool, String> {
        Ok(self.eat_control_escape()
            || self.eat_c_control_letter()
            || self.eat_zero()
            || self.eat_hex_escape_sequence()?
            || self.eat_regexp_unicode_escape_sequence(false)?
            || (!self.strict && !self.u_flag && self.eat_legacy_octal_escape_sequence())
            || self.eat_identity_escape())
    }

    /// Validate the next characters as the follwoing alternatives if possible.
    /// ```grammar
    ///      `k` GroupName[?U]
    /// ```
    /// Returns `Ok(true)` if it consumed the next characters successfully.
    fn consume_k_group_name(&mut self) -> Result<bool, String> {
        if self.eat('k') {
            if self.eat_group_name()? {
                let group_name = self.last_str_value.clone();
                self.backreference_names.insert(group_name);
                return Ok(true);
            }
            return Err("Invalid named reference".to_string());
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `CharacterClass` production if possible.
    /// ```grammar
    /// CharacterClass[U]::
    ///      `[` [lookahead ≠ ^] ClassRanges[?U] `]`
    ///      `[^` ClassRanges[?U] `]`
    /// ```
    /// Returns `true` if it consumed the next characters successfully.
    fn consume_character_class(&mut self) -> Result<bool, String> {
        if !self.eat('[') {
            return Ok(false);
        }
        self.consume_class_ranges()?;
        if !self.eat(']') {
            return Err("Unterminated character class".to_string());
        }
        Ok(true)
    }

    /// Validate the next characters as a RegExp `ClassRanges` production.
    /// ```grammar
    /// ClassRanges[U]::
    ///      ε
    ///      NonemptyClassRanges[?U]
    /// NonemptyClassRanges[U]::
    ///      ClassAtom[?U]
    ///      ClassAtom[?U] NonemptyClassRangesNoDash[?U]
    ///      ClassAtom[?U] `-` ClassAtom[?U] ClassRanges[?U]
    /// NonemptyClassRangesNoDash[U]::
    ///      ClassAtom[?U]
    ///      ClassAtomNoDash[?U] NonemptyClassRangesNoDash[?U]
    ///      ClassAtomNoDash[?U] `-` ClassAtom[?U] ClassRanges[?U]
    /// ```
    fn consume_class_ranges(&mut self) -> Result<(), String> {
        loop {
            // Consume the first ClassAtom
            if !self.consume_class_atom()? {
                break;
            }
            let min = self.last_int_value;

            // Consume `-`
            if !self.eat('-') {
                continue;
            }

            // Consume the second ClassAtom
            if !self.consume_class_atom()? {
                break;
            }
            let max = self.last_int_value;

            // Validate
            if min == -1 || max == -1 {
                if self.strict {
                    return Err("Invalid character class".to_string());
                }
                continue;
            }

            println!("min: {},  max: {}", min, max);
            if min > max {
                return Err("Range out of order in character class".to_string());
            }
        }
        Ok(())
    }

    /// Validate the next characters as a RegExp `ClassAtom` production if possible.
    /// Set `self.last_int_value` if it consumed the next characters successfully.
    /// ```grammar
    /// ClassAtom[U, N]::
    ///      `-`
    ///      ClassAtomNoDash[?U, ?N]
    /// ClassAtomNoDash[U, N]::
    ///      SourceCharacter but not one of \ ] -
    ///      `\` ClassEscape[?U, ?N]
    ///      [annexB] `\` [lookahead = c]
    /// ```
    /// Returns `Ok(true)` if it consumed the next characters successfully.
    fn consume_class_atom(&mut self) -> Result<bool, String> {
        let start = self.index();

        if let Some(cp) = self.code_point_with_offset(0) {
            if cp != '\\' && cp != ']' {
                self.advance();
                self.last_int_value = cp as i64;
                return Ok(true);
            }
        }

        if self.eat('\\') {
            if self.consume_class_escape()? {
                return Ok(true);
            }
            if !self.strict && self.code_point_with_offset(0) == Some('c') {
                self.last_int_value = '\\' as i64;
                return Ok(true);
            }
            if self.strict || self.u_flag {
                return Err("Invalid escape".to_string());
            }
            self.rewind(start);
        }
        Ok(false)
    }

    /// Validate the next characters as a RegExp `ClassEscape` production if possible.
    /// Set `self.last_int_value` if it consumed the next characters successfully.
    /// ```grammar
    /// ClassEscape[U, N]::
    ///      `b`
    ///      [+U] `-`
    ///      [annexB][~U] `c` ClassControlLetter
    ///      CharacterClassEscape[?U]
    ///      CharacterEscape[?U, ?N]
    /// ClassControlLetter::
    ///      DecimalDigit
    ///      `_`
    /// ```
    /// Returns `Ok(true)` if it consumed the next characters successfully.
    fn consume_class_escape(&mut self) -> Result<bool, String> {
        if self.eat('b') {
            self.last_int_value = 0x08; // backspace
            return Ok(true);
        }

        // [+U] `-`
        if self.u_flag && self.eat('-') {
            self.last_int_value = '-' as i64;
            return Ok(true);
        }

        // [annexB][~U] `c` ClassControlLetter
        if !self.strict && !self.u_flag && self.code_point_with_offset(0) == Some('c') {
            if let Some(cp) = self.code_point_with_offset(1) {
                if cp.is_digit(10) || cp == '_' {
                    self.advance();
                    self.advance();
                    self.last_int_value = cp as i64 % 0x20;
                    return Ok(true);
                }
            }
        }

        Ok(self.consume_character_class_escape()? || self.consume_character_escape()?)
    }

    /// Eat the next characters as a RegExp `GroupName` production if possible.
    /// Set `self.last_str_value` if the group name existed.
    /// ```grammar
    /// GroupName[U]::
    ///      `<` RegExpIdentifierName[?U] `>`
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_group_name(&mut self) -> Result<bool, String> {
        if self.eat('<') {
            if self.eat_regexp_identifier_name()? && self.eat('>') {
                return Ok(true);
            }
            return Err("Invalid capture group name".to_string());
        }
        return Ok(false);
    }

    /// Eat the next characters as a RegExp `RegExpIdentifierName` production if
    /// possible.
    /// Set `self.last_str_value` if the identifier name existed.
    /// ```grammar
    /// RegExpIdentifierName[U]::
    ///      RegExpIdentifierStart[?U]
    ///      RegExpIdentifierName[?U] RegExpIdentifierPart[?U]
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_identifier_name(&mut self) -> Result<bool, String> {
        if self.eat_regexp_identifier_start()? {
            self.last_str_value = std::char::from_u32(self.last_int_value as u32)
                .unwrap()
                .to_string();
            while self.eat_regexp_identifier_part()? {
                self.last_str_value
                    .push(std::char::from_u32(self.last_int_value as u32).unwrap());
            }
            return Ok(true);
        }
        return Ok(false);
    }

    /// Eat the next characters as a RegExp `RegExpIdentifierStart` production if
    /// possible.
    /// Set `self.last_int_value` if the identifier start existed.
    /// ```grammar
    /// RegExpIdentifierStart[U] ::
    ///      UnicodeIDStart
    ///      `$`
    ///      `_`
    ///      `\` RegExpUnicodeEscapeSequence[+U]
    ///      [~U] UnicodeLeadSurrogate UnicodeTrailSurrogate
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_identifier_start(&mut self) -> Result<bool, String> {
        let start = self.index();
        let force_u_flag = !self.u_flag && self.ecma_version >= EcmaVersion::ES2020;

        if let Some(mut cp) = self.code_point_with_offset(0) {
            self.advance();
            let cp1 = self.code_point_with_offset(0);
            if cp == '\\' && self.eat_regexp_unicode_escape_sequence(force_u_flag)? {
                cp = std::char::from_u32(self.last_int_value as u32).unwrap();
            } else if force_u_flag
                && is_lead_surrogate(cp as i64)
                && cp1.is_some()
                && is_trail_surrogate(cp1.unwrap() as i64)
            {
                cp = std::char::from_u32(
                    combine_surrogate_pair(cp as i64, cp1.unwrap() as i64) as u32,
                )
                .unwrap();
                self.advance();
            }

            if is_regexp_identifier_start(cp) {
                self.last_int_value = cp as i64;
                return Ok(true);
            }
        }

        if self.index() != start {
            self.rewind(start);
        }
        return Ok(false);
    }

    /// Eat the next characters as a RegExp `RegExpIdentifierPart` production if
    /// possible.
    /// Set `self.last_int_value` if the identifier part existed.
    /// ```grammar
    /// RegExpIdentifierPart[U] ::
    ///      UnicodeIDContinue
    ///      `$`
    ///      `_`
    ///      `\` RegExpUnicodeEscapeSequence[+U]
    ///      [~U] UnicodeLeadSurrogate UnicodeTrailSurrogate
    ///      <ZWNJ>
    ///      <ZWJ>
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_identifier_part(&mut self) -> Result<bool, String> {
        let start = self.index();
        let force_u_flag = !self.u_flag && self.ecma_version >= EcmaVersion::ES2020;
        let mut cp = self.code_point_with_offset(0);
        self.advance();
        let cp1 = self.code_point_with_offset(0);

        if cp == Some('\\') && self.eat_regexp_unicode_escape_sequence(force_u_flag)? {
            // TODO: convert unicode code point to char
            cp = std::char::from_u32(self.last_int_value as u32);
        } else if force_u_flag
            && is_lead_surrogate(cp.unwrap() as i64)
            && is_trail_surrogate(cp1.unwrap() as i64)
        {
            cp = std::char::from_u32(combine_surrogate_pair(
                cp.unwrap() as i64,
                cp1.unwrap() as i64,
            ) as u32);
            self.advance();
        }

        if cp.is_some() && is_regexp_identifier_part(cp.unwrap()) {
            self.last_int_value = cp.unwrap() as i64;
            return Ok(true);
        }

        if self.index() != start {
            self.rewind(start);
        }
        Ok(false)
    }

    /// Eat the next characters as the follwoing alternatives if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    ///      `c` ControlLetter
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_c_control_letter(&mut self) -> bool {
        let start = self.index();
        if self.eat('c') {
            if self.eat_control_letter() {
                return true;
            }
            self.rewind(start);
        }
        false
    }

    /// Eat the next characters as the follwoing alternatives if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    ///      `0` [lookahead ∉ DecimalDigit]
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_zero(&mut self) -> bool {
        if self.code_point_with_offset(0) != Some('0') {
            return false;
        }
        if let Some(cp) = self.code_point_with_offset(1) {
            if cp.is_digit(10) {
                return false;
            }
        }
        self.last_int_value = 0;
        self.advance();
        return true;
    }

    /// Eat the next characters as a RegExp `ControlEscape` production if
    /// possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// ControlEscape:: one of
    ///      f n r t v
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_control_escape(&mut self) -> bool {
        if self.eat('f') {
            self.last_int_value = 0x0c; // formfeed
            return true;
        }
        if self.eat('n') {
            self.last_int_value = 0x0a; // linefeed
            return true;
        }
        if self.eat('r') {
            self.last_int_value = 0x0d; // carriage return
            return true;
        }
        if self.eat('t') {
            self.last_int_value = 0x09; // character tabulation
            return true;
        }
        if self.eat('v') {
            self.last_int_value = 0x0b; // line tabulation
            return true;
        }
        false
    }

    /// Eat the next characters as a RegExp `ControlLetter` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// ControlLetter:: one of
    ///      a b c d e f g h i j k l m n o p q r s t u v w x y z
    ///      A B C D E F G H I J K L M N O P Q R S T U V W X Y Z
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_control_letter(&mut self) -> bool {
        if let Some(cp) = self.code_point_with_offset(0) {
            if cp.is_ascii_alphabetic() {
                self.advance();
                self.last_int_value = cp as i64 % 0x20;
                return true;
            }
        }
        false
    }

    /// Eat the next characters as a RegExp `RegExpUnicodeEscapeSequence`
    /// production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// RegExpUnicodeEscapeSequence[U]::
    ///      [+U] `u` LeadSurrogate `\u` TrailSurrogate
    ///      [+U] `u` LeadSurrogate
    ///      [+U] `u` TrailSurrogate
    ///      [+U] `u` NonSurrogate
    ///      [~U] `u` Hex4Digits
    ///      [+U] `u{` CodePoint `}`
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_unicode_escape_sequence(&mut self, force_u_flag: bool) -> Result<bool, &str> {
        let start = self.index();
        let u_flag = force_u_flag || self.u_flag;

        if self.eat('u') {
            if (u_flag && self.eat_regexp_unicode_surrogate_pair_escape())
                || self.eat_fixed_hex_digits(4)
                || (u_flag && self.eat_regexp_unicode_codepoint_escape())
            {
                return Ok(true);
            }
            if self.strict || u_flag {
                return Err("Invalid unicode escape");
            }
            self.rewind(start);
        }

        return Ok(false);
    }

    /// Eat the next characters as the following alternatives if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    ///      LeadSurrogate `\u` TrailSurrogate
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_unicode_surrogate_pair_escape(&mut self) -> bool {
        let start = self.index();

        if self.eat_fixed_hex_digits(4) {
            let lead = self.last_int_value;
            if is_lead_surrogate(lead)
                && self.eat('\\')
                && self.eat('u')
                && self.eat_fixed_hex_digits(4)
            {
                let trail = self.last_int_value;
                if is_trail_surrogate(trail) {
                    self.last_int_value = combine_surrogate_pair(lead, trail);
                    return true;
                }
            }

            self.rewind(start);
        }

        return false;
    }

    /// Eat the next characters as the following alternatives if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    ///      `{` CodePoint `}`
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_regexp_unicode_codepoint_escape(&mut self) -> bool {
        let start = self.index();

        if self.eat('{')
            && self.eat_hex_digits()
            && self.eat('}')
            && is_valid_unicode(self.last_int_value)
        {
            return true;
        }

        self.rewind(start);
        return false;
    }

    /// Eat the next characters as a RegExp `IdentityEscape` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// IdentityEscape[U, N]::
    ///      [+U] SyntaxCharacter
    ///      [+U] `/`
    ///      [strict][~U] SourceCharacter but not UnicodeIDContinue
    ///      [annexB][~U] SourceCharacterIdentityEscape[?N]
    /// SourceCharacterIdentityEscape[N]::
    ///      [~N] SourceCharacter but not c
    ///      [+N] SourceCharacter but not one of c k
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_identity_escape(&mut self) -> bool {
        if let Some(cp) = self.code_point_with_offset(0) {
            if self.is_valid_identity_escape(cp) {
                self.last_int_value = cp as i64;
                self.advance();
                return true;
            }
        }
        return false;
    }
    fn is_valid_identity_escape(&self, cp: char) -> bool {
        if self.u_flag {
            return is_syntax_character(cp) || cp == '/';
        } else if self.strict {
            return !is_id_continue(cp);
        } else if self.n_flag {
            return !(cp == 'c' || cp == 'k');
        }
        return cp != 'c';
    }

    /// Eat the next characters as a RegExp `DecimalEscape` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// DecimalEscape::
    ///      NonZeroDigit DecimalDigits(opt) [lookahead ∉ DecimalDigit]
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_decimal_escape(&mut self) -> bool {
        self.last_int_value = 0;
        if let Some(cp) = self.code_point_with_offset(0) {
            if cp.is_digit(10) {
                self.last_int_value = 10 * self.last_int_value + cp.to_digit(10).unwrap() as i64;
                self.advance();
                while let Some(cp) = self.code_point_with_offset(0) {
                    if !cp.is_digit(10) {
                        break;
                    }
                    self.last_int_value = 10 * self.last_int_value + cp.to_digit(10).unwrap() as i64;
                    self.advance();
                }
                return true;
            }
        }
        return false;
    }

    /// Eat the next characters as a RegExp `UnicodePropertyValueExpression` production if possible.
    /// Set `self.last_key_value` and `self.last_val_value` if it ate the next characters
    /// successfully.
    /// ```grammar
    /// UnicodePropertyValueExpression::
    ///      UnicodePropertyName `=` UnicodePropertyValue
    ///      LoneUnicodePropertyNameOrValue
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_unicode_property_value_expression(&mut self) -> Result<bool, &str> {
        let start = self.index();

        // UnicodePropertyName `=` UnicodePropertyValue
        if self.eat_unicode_property_name() && self.eat('=') {
            self.last_key_value = self.last_str_value.clone();
            if self.eat_unicode_property_value() {
                self.last_val_value = self.last_str_value.clone();
                if is_valid_unicode_property(
                    self.ecma_version,
                    &self.last_key_value,
                    &self.last_val_value,
                ) {
                    return Ok(true);
                }
                return Err("Invalid property name");
            }
        }
        self.rewind(start);

        // LoneUnicodePropertyNameOrValue
        if self.eat_lone_unicode_property_name_or_value() {
            let name_or_value = self.last_str_value.clone();
            if is_valid_unicode_property(self.ecma_version, "General_Category", &name_or_value) {
                self.last_key_value = "General_Category".to_string();
                self.last_val_value = name_or_value;
                return Ok(true);
            }
            if is_valid_lone_unicode_property(self.ecma_version, &name_or_value) {
                self.last_key_value = name_or_value;
                self.last_val_value = "".to_string();
                return Ok(true);
            }
            return Err("Invalid property name");
        }
        Ok(false)
    }

    /// Eat the next characters as a RegExp `UnicodePropertyName` production if possible.
    /// Set `self.last_str_value` if it ate the next characters successfully.
    /// ```grammar
    /// UnicodePropertyName::
    ///      UnicodePropertyNameCharacters
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_unicode_property_name(&mut self) -> bool {
        self.last_str_value = "".to_string();
        while let Some(cp) = self.code_point_with_offset(0) {
            if !is_unicode_property_name_character(cp) {
                break;
            }
            self.last_str_value.push(cp);
            self.advance();
        }
        self.last_str_value != ""
    }

    /// Eat the next characters as a RegExp `UnicodePropertyValue` production if possible.
    /// Set `self.last_str_value` if it ate the next characters successfully.
    /// ```grammar
    /// UnicodePropertyValue::
    ///      UnicodePropertyValueCharacters
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_unicode_property_value(&mut self) -> bool {
        self.last_str_value = "".to_string();
        while let Some(cp) = self.code_point_with_offset(0) {
            if !is_unicode_property_value_character(cp) {
                break;
            }
            self.last_str_value.push(cp);
            self.advance();
        }
        self.last_str_value != ""
    }

    /// Eat the next characters as a RegExp `UnicodePropertyValue` production if possible.
    /// Set `self.last_str_value` if it ate the next characters successfully.
    /// ```grammar
    /// LoneUnicodePropertyNameOrValue::
    ///      UnicodePropertyValueCharacters
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_lone_unicode_property_name_or_value(&mut self) -> bool {
        self.eat_unicode_property_value()
    }

    /// Eat the next characters as a `HexEscapeSequence` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// HexEscapeSequence::
    ///      `x` HexDigit HexDigit
    /// HexDigit:: one of
    ///      0 1 2 3 4 5 6 7 8 9 a b c d e f A B C D E F
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_hex_escape_sequence(&mut self) -> Result<bool, &str> {
        let start = self.index();
        if self.eat('x') {
            if self.eat_fixed_hex_digits(2) {
                return Ok(true);
            }
            if self.u_flag || self.strict {
                return Err("Invalid escape");
            }
            self.rewind(start);
        }
        Ok(false)
    }

    /// Eat the next characters as a `DecimalDigits` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// DecimalDigits::
    ///      DecimalDigit
    ///      DecimalDigits DecimalDigit
    /// DecimalDigit:: one of
    ///      0 1 2 3 4 5 6 7 8 9
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_decimal_digits(&mut self) -> bool {
        let start = self.index();

        self.last_int_value = 0;
        while let Some(cp) = self.code_point_with_offset(0) {
            if !cp.is_digit(10) {
                break;
            }
            self.last_int_value = 10 * self.last_int_value
                + self
                    .code_point_with_offset(0)
                    .unwrap()
                    .to_digit(10)
                    .unwrap() as i64;
            self.advance();
        }

        return self.index() != start;
    }

    /// Eat the next characters as a `HexDigits` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// HexDigits::
    ///      HexDigit
    ///      HexDigits HexDigit
    /// HexDigit:: one of
    ///      0 1 2 3 4 5 6 7 8 9 a b c d e f A B C D E F
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_hex_digits(&mut self) -> bool {
        let start = self.index();
        self.last_int_value = 0;
        while let Some(cp) = self.code_point_with_offset(0) {
            if !cp.is_digit(16) {
                break;
            }
            self.last_int_value = 16 * self.last_int_value + cp.to_digit(16).unwrap() as i64;
            self.advance();
        }
        return self.index() != start;
    }

    /// Eat the next characters as a `HexDigits` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// LegacyOctalEscapeSequence::
    ///      OctalDigit [lookahead ∉ OctalDigit]
    ///      ZeroToThree OctalDigit [lookahead ∉ OctalDigit]
    ///      FourToSeven OctalDigit
    ///      ZeroToThree OctalDigit OctalDigit
    /// OctalDigit:: one of
    ///      0 1 2 3 4 5 6 7
    /// ZeroToThree:: one of
    ///      0 1 2 3
    /// FourToSeven:: one of
    ///      4 5 6 7
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_legacy_octal_escape_sequence(&mut self) -> bool {
        if self.eat_octal_digit() {
            let n1 = self.last_int_value;
            if self.eat_octal_digit() {
                let n2 = self.last_int_value;
                if n1 <= 3 && self.eat_octal_digit() {
                    self.last_int_value = n1 * 64 + n2 * 8 + self.last_int_value
                } else {
                    self.last_int_value = n1 * 8 + n2;
                }
            } else {
                self.last_int_value = n1;
            }
            return true;
        }
        return false;
    }

    /// Eat the next characters as a `OctalDigit` production if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// OctalDigit:: one of
    ///      0 1 2 3 4 5 6 7
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_octal_digit(&mut self) -> bool {
        if let Some(cp) = self.code_point_with_offset(0) {
            if cp.is_digit(8) {
                self.advance();
                self.last_int_value = cp.to_digit(8).unwrap() as i64;
                return true;
            }
        }
        self.last_int_value = 0;
        return false;
    }

    /// Eat the next characters as the given number of `HexDigit` productions if possible.
    /// Set `self.last_int_value` if it ate the next characters successfully.
    /// ```grammar
    /// HexDigit:: one of
    ///      0 1 2 3 4 5 6 7 8 9 a b c d e f A B C D E F
    /// ```
    /// Returns `true` if it ate the next characters successfully.
    fn eat_fixed_hex_digits(&mut self, length: i64) -> bool {
        let start = self.index();
        self.last_int_value = 0;
        for _ in 0..length {
            let cp = self.code_point_with_offset(0);
            if cp.is_none() || !cp.unwrap().is_digit(16) {
                self.rewind(start);
                return false;
            }
            self.last_int_value =
                16 * self.last_int_value + cp.unwrap().to_digit(16).unwrap() as i64;
            self.advance();
        }
        return true;
    }

    fn count_capturing_parens(&mut self) -> u32 {
        let start = self.index();
        let mut in_class = false;
        let mut escaped = false;
        let mut count = 0;

        while let Some(cp) = self.code_point_with_offset(0) {
            if escaped {
                escaped = false;
            } else if cp == '\\' {
                escaped = true;
            } else if cp == '[' {
                in_class = true;
            } else if cp == ']' {
                in_class = false;
            } else if cp == '('
                && !in_class
                && (self.code_point_with_offset(1) != Some('?')
                    || (self.code_point_with_offset(2) == Some('<')
                        && self.code_point_with_offset(3) != Some('=')
                        && self.code_point_with_offset(3) != Some('!')))
            {
                count += 1
            }
            self.advance();
        }

        self.rewind(start);
        count
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn count_capturing_parens_test() {
        let mut validator = EcmaRegexValidator::new(EcmaVersion::ES2018);
        let source = "foo|(abc)de";
        validator.reset(source, 0, source.len(), false);
        assert_eq!(validator.count_capturing_parens(), 1);
        let source = "foo|(?:abc)de";
        validator.reset(source, 0, source.len(), false);
        assert_eq!(validator.count_capturing_parens(), 0);
        let source = "((foo)|(abc)de)";
        validator.reset(source, 0, source.len(), false);
        assert_eq!(validator.count_capturing_parens(), 3);
    }
}