12075733362

Committed 28 Nov 2024 09:40PM UTC coverage: 82.811% (-0.3%) from 83.091%

Build # 12075733362

Build Type

push

github

Committed by

Aloso

Commit Message

feat: Add RE2 and improve diagnostics

Run Details

69 of 91 new or added lines in 12 files covered. (75.82%)

1 existing line in 1 file now uncovered.

4307 of 5201 relevant lines covered (82.81%)

413156.17 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

92.18

/pomsky-lib/src/exprs/char_class.rs

//! Implements _character classes_. The analogue in the regex world are
//! [character classes](https://www.regular-expressions.info/charclass.html),
//! [shorthand character classes](https://www.regular-expressions.info/shorthand.html),
//! [non-printable characters](https://www.regular-expressions.info/nonprint.html),
//! [Unicode categories/scripts/blocks](https://www.regular-expressions.info/unicode.html#category),
//! [POSIX classes](https://www.regular-expressions.info/posixbrackets.html#class) and the
//! [dot](https://www.regular-expressions.info/dot.html).
//!
//! All kinds of character classes mentioned above require `[` square brackets
//! `]` in Pomsky. A character class can be negated by putting the keyword `not`
//! after the opening bracket. For example, `![.]` compiles to `\n`.
//!
//! ## Items
//!
//! A character class can contain multiple _items_, which can be
//!
//! - A __code point__, e.g. `['a']` or `[U+107]`
//!
//!   - This includes [non-printable characters](https://www.regular-expressions.info/nonprint.html).\
//!     Supported are `[n]`, `[r]`, `[t]`, `[a]`, `[e]` and `[f]`.
//!
//! - A __range of code points__. For example, `[U+10 - U+200]` matches any code
//!   point P where `U+10 ≤ P ≤ U+200`
//!
//! - A __named character class__, which can be one of
//!
//!   - a [shorthand character class](https://www.regular-expressions.info/shorthand.html).\
//!     Supported are `[w]`, `[d]`, `[s]`, `[h]`, `[v]` and `[R]`.
//!
//!   - a [POSIX class](https://www.regular-expressions.info/posixbrackets.html#class).\
//!     Supported are `[ascii_alnum]`, `[ascii_alpha]`, `[ascii]`,
//!     `[ascii_blank]`, `[ascii_cntrl]`, `[ascii_digit]`, `[ascii_graph]`,
//!     `[ascii_lower]`, `[ascii_print]`, `[ascii_punct]`, ´ `[ascii_space]`,
//!     `[ascii_upper]`, `[ascii_word]` and `[ascii_xdigit]`.\ _Note_: POSIX
//!     classes are not Unicode aware!\ _Note_: They're converted to ranges,
//!     e.g. `[ascii_alpha]` = `[a-zA-Z]`.
//!
//!   - a [Unicode category, script or block](https://www.regular-expressions.info/unicode.html#category).\
//!     For example: `[Letter]` compiles to `\p{Letter}`. Pomsky currently
//!     treats any uppercase identifier except `R` as Unicode class.
//!
//! ## Compilation
//!
//! When a character class contains only a single item (e.g. `[w]`), the
//! character class is "flattened":
//!
//! - `['a']` = `a`
//! - `[w]` = `\w`
//! - `[Letter]` = `\p{Letter}`
//!
//! When there is more than one item or a range (e.g. `['a'-'z' '!']`), a regex
//! character class is created:
//!
//! - `['a'-'z' '!']` = `[a-z!]`
//! - `[w e Punctuation]` = `[\w\e\p{Punctuation}]`
//!
//! ### Negation
//!
//! Negation is implemented as follows:
//!
//! - Ranges and chars such as `!['a'-'z' '!' e]` are wrapped in a negative
//!   character class, e.g. `[^a-z!\e]`.
//!
//! - The `h`, `v` and `R` shorthands are also wrapped in a negative character
//!   class.
//!
//! - The `w`, `d` and `s` shorthands are negated by making them uppercase
//!   (`![w]` = `\W`), except when there is more than one item in the class
//!   (`![w '-']` = `[^\w\-]`)
//!
//! - `w`, `s`, `d` and Unicode categories/scripts/blocks can be negated
//!   individually _within a character class_, e.g. `[s !s]` = `[\s\S]`,
//!   `![!Latin 'a']` = `[^\P{Latin}a]`.
//!
//!   When a negated character class only contains 1 item, which is also
//!   negated, the class is   removed and the negations cancel each other out:
//!   `![!w]` = `\w`, `![!L]` = `\p{L}`.

use std::fmt;

use crate::{
    compile::{CompileResult, CompileState},
    diagnose::{CompileError, CompileErrorKind, Feature},
    exprs::literal,
    options::{CompileOptions, RegexFlavor},
    regex::{Regex, RegexProperty, RegexShorthand},
    unicode_set::UnicodeSet,
};

use pomsky_syntax::{
    exprs::{Category, CharClass, CodeBlock, GroupItem, GroupName, OtherProperties, Script},
    Span,
};

use super::Compile;

impl Compile for CharClass {
    fn compile(&self, options: CompileOptions, _state: &mut CompileState<'_>) -> CompileResult {
        // when single, a `[!w]` can be turned into `![w]`
        let is_single = self.inner.len() == 1;
        let mut group_negative = false;

        let mut set = UnicodeSet::new();
        for item in &self.inner {
            match *item {
                GroupItem::Char(c) => {
                    if !is_single {
                        validate_char_in_class(c, options.flavor, self.span)?;
                    }
                    set.add_char(c)
                }
                GroupItem::Range { first, last } => {
                    validate_char_in_class(first, options.flavor, self.span)?;
                    validate_char_in_class(last, options.flavor, self.span)?;
                    set.add_range(first..=last);
                }
                GroupItem::Named { name, negative, span } => {
                    if self.unicode_aware {
                        named_class_to_regex_unicode(
                            name,
                            negative,
                            &mut group_negative,
                            is_single,
                            options.flavor,
                            span,
                            &mut set,
                        )?;
                    } else {
                        named_class_to_regex_ascii(name, negative, options.flavor, span, &mut set)?;
                    }
                }
            }
        }

        // this makes it possible to use code points outside the BMP in .NET,
        // as long as there is only one in the character set
        if let Some(only_char) = set.try_into_char() {
            return Ok(Regex::Literal(only_char.to_string()));
        }

        Ok(Regex::CharSet(RegexCharSet { negative: group_negative, set }))
    }
}

fn validate_char_in_class(char: char, flavor: RegexFlavor, span: Span) -> Result<(), CompileError> {
    if flavor == RegexFlavor::DotNet && char > '\u{FFFF}' {
        Err(CompileErrorKind::Unsupported(Feature::LargeCodePointInCharClass(char), flavor)
            .at(span))
    } else {
        Ok(())
    }
}

pub(crate) fn check_char_class_empty(
    char_set: &RegexCharSet,
    span: Span,
) -> Result<(), CompileError> {
    if char_set.negative {
        if let Some((group1, group2)) = char_set.set.full_props() {
            return Err(CompileErrorKind::EmptyClassNegated { group1, group2 }.at(span));
        }
    }
    Ok(())
}

pub fn is_ascii_only_in_flavor(group: GroupName, flavor: RegexFlavor) -> bool {
    match flavor {
        RegexFlavor::JavaScript => matches!(group, GroupName::Word | GroupName::Digit),
        RegexFlavor::RE2 => matches!(group, GroupName::Word | GroupName::Digit | GroupName::Space),
        _ => false,
    }
}

fn named_class_to_regex_ascii(
    group: GroupName,
    negative: bool,
    flavor: RegexFlavor,
    span: Span,
    set: &mut UnicodeSet,
) -> Result<(), CompileError> {
    // In JS, \W and \D can be used for negation because they're ascii-only
    // Same goes for \W, \D and \S in RE2
    if negative && !is_ascii_only_in_flavor(group, flavor) {
        return Err(CompileErrorKind::NegativeShorthandInAsciiMode.at(span));
    }

    match group {
        GroupName::Word => {
            if let RegexFlavor::JavaScript | RegexFlavor::RE2 = flavor {
                let s = if negative { RegexShorthand::NotWord } else { RegexShorthand::Word };
                set.add_prop(RegexCharSetItem::Shorthand(s));
            } else {
                // we already checked above if negative
                set.add_range('a'..='z');
                set.add_range('A'..='Z');
                set.add_range('0'..='9');
                set.add_char('_');
            }
        }
        GroupName::Digit => {
            if let RegexFlavor::JavaScript | RegexFlavor::RE2 = flavor {
                let s = if negative { RegexShorthand::NotDigit } else { RegexShorthand::Digit };
                set.add_prop(RegexCharSetItem::Shorthand(s));
            } else {
                // we already checked above if negative
                set.add_range('0'..='9');
            }
        }
        GroupName::Space => {
            if let RegexFlavor::RE2 = flavor {
                let s = if negative { RegexShorthand::NotSpace } else { RegexShorthand::Space };
                set.add_prop(RegexCharSetItem::Shorthand(s));
            } else {
                set.add_char(' ');
                set.add_range('\x09'..='\x0D'); // \t\n\v\f\r
            }
        }
        GroupName::HorizSpace => set.add_char('\t'),
        GroupName::VertSpace => set.add_range('\x0A'..='\x0D'),
        _ => return Err(CompileErrorKind::UnicodeInAsciiMode.at(span)),
    }
    Ok(())
}

fn named_class_to_regex_unicode(
    group: GroupName,
    negative: bool,
    group_negative: &mut bool,
    is_single: bool,
    flavor: RegexFlavor,
    span: Span,
    set: &mut UnicodeSet,
) -> Result<(), CompileError> {
    match group {
        GroupName::Word => {
            if flavor == RegexFlavor::RE2 {
                return Err(CompileErrorKind::Unsupported(Feature::ShorthandW, flavor).at(span));
            } else if flavor == RegexFlavor::JavaScript {
                if negative {
                    if is_single {
                        *group_negative ^= true;
                    } else {
                        return Err(CompileErrorKind::Unsupported(
                            Feature::NegativeShorthandW,
                            flavor,
                        )
                        .at(span));
                    }
                }
                set.add_prop(
                    RegexProperty::Other(OtherProperties::Alphabetic).negative_item(false),
                );
                set.add_prop(RegexProperty::Category(Category::Mark).negative_item(false));
                set.add_prop(
                    RegexProperty::Category(Category::Decimal_Number).negative_item(false),
                );
                set.add_prop(
                    RegexProperty::Category(Category::Connector_Punctuation).negative_item(false),
                );
            } else {
                let s = if negative { RegexShorthand::NotWord } else { RegexShorthand::Word };
                set.add_prop(RegexCharSetItem::Shorthand(s));
            }
        }
        GroupName::Digit => {
            if matches!(flavor, RegexFlavor::JavaScript | RegexFlavor::RE2) {
                set.add_prop(
                    RegexProperty::Category(Category::Decimal_Number).negative_item(negative),
                );
            } else {
                let s = if negative { RegexShorthand::NotDigit } else { RegexShorthand::Digit };
                set.add_prop(RegexCharSetItem::Shorthand(s));
            }
        }

        GroupName::Space => {
            if flavor == RegexFlavor::RE2 {
                if negative {
                    if is_single {
                        *group_negative ^= true;
                    } else {
                        return Err(CompileErrorKind::Unsupported(
                            Feature::NegativeShorthandS,
                            flavor,
                        )
                        .at(span));
                    }
                }

                // [ \f\n\r\t\u000b\u00a0\u1680\u2000-\u200a\u2028\u2029\u202f\u205f\u3000\ufeff]
                set.add_prop(RegexCharSetItem::Shorthand(RegexShorthand::Space));
                set.add_char('\x0b');
                set.add_char('\u{a0}');
                set.add_char('\u{1680}');
                set.add_range('\u{2000}'..='\u{200a}');
                set.add_range('\u{2028}'..='\u{2029}');
                set.add_char('\u{202f}');
                set.add_char('\u{205f}');
                set.add_char('\u{3000}');
                set.add_char('\u{feff}');
            } else {
                set.add_prop(RegexCharSetItem::Shorthand(if negative {
                    RegexShorthand::NotSpace
                } else {
                    RegexShorthand::Space
                }))
            }
        }

        GroupName::HorizSpace | GroupName::VertSpace if negative => {
            return Err(CompileErrorKind::NegatedHorizVertSpace.at(span));
        }

        GroupName::HorizSpace | GroupName::VertSpace
            if matches!(flavor, RegexFlavor::Pcre | RegexFlavor::Java) =>
        {
            set.add_prop(RegexCharSetItem::Shorthand(if group == GroupName::HorizSpace {
                RegexShorthand::HorizSpace
            } else {
                RegexShorthand::VertSpace
            }));
        }
        GroupName::HorizSpace => {
            set.add_char('\t');
            if flavor == RegexFlavor::Python {
                return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));
            } else {
                set.add_prop(
                    RegexProperty::Category(Category::Space_Separator).negative_item(false),
                );
            }
        }
        GroupName::VertSpace => {
            set.add_range('\x0A'..='\x0D');
            set.add_char('\u{85}');
            set.add_char('\u{2028}');
            set.add_char('\u{2029}');
        }

        _ if flavor == RegexFlavor::Python => {
            return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));
        }
        GroupName::Category(c) => {
            if let (RegexFlavor::Rust, Category::Surrogate)
            | (RegexFlavor::DotNet | RegexFlavor::RE2, Category::Cased_Letter) = (flavor, c)
            {
                return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));
            }
            set.add_prop(RegexProperty::Category(c).negative_item(negative));
        }
        GroupName::Script(s) => {
            if flavor == RegexFlavor::DotNet {
                return Err(CompileErrorKind::Unsupported(Feature::UnicodeScript, flavor).at(span));
            }
            if let (RegexFlavor::Ruby, Script::Kawi | Script::Nag_Mundari)
            | (RegexFlavor::Rust, Script::Unknown) = (flavor, s)
            {
                return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));
            }
            set.add_prop(RegexProperty::Script(s).negative_item(negative));
        }
        GroupName::CodeBlock(b) => match flavor {
            RegexFlavor::DotNet | RegexFlavor::Java | RegexFlavor::Ruby => {
                match (flavor, b) {
                    (RegexFlavor::Java, CodeBlock::No_Block)
                    | (
                        // These should work since Oniguruma updated to Unicode 15.1
                        // ... but our C bindings for Oniguruma are unmaintained!
                        RegexFlavor::Ruby,
                        CodeBlock::Arabic_Extended_C
                        | CodeBlock::CJK_Unified_Ideographs_Extension_H
                        | CodeBlock::Cyrillic_Extended_D
                        | CodeBlock::Devanagari_Extended_A
                        | CodeBlock::Kaktovik_Numerals,
                    ) => {
                        return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));
                    }
                    (RegexFlavor::DotNet, _) => {
                        let dotnet_name = b.as_str().replace("_And_", "_and_").replace('_', "");
                        if pomsky_syntax::blocks_supported_in_dotnet()
                            .binary_search(&dotnet_name.as_str())
                            .is_err()
                        {
                            return Err(
                                CompileErrorKind::unsupported_specific_prop_in(flavor).at(span)
                            );
                        }
                    }
                    _ => {}
                }

                set.add_prop(RegexProperty::Block(b).negative_item(negative));
            }
            _ => return Err(CompileErrorKind::Unsupported(Feature::UnicodeBlock, flavor).at(span)),
        },
        GroupName::OtherProperties(o) => {
            use OtherProperties as OP;
            use RegexFlavor as RF;

            if let RF::JavaScript | RF::Rust | RF::Pcre | RF::Ruby = flavor {
                match (flavor, o) {
                    (RF::JavaScript, _) => {}
                    (_, OP::Changes_When_NFKC_Casefolded)
                    | (RF::Pcre, OP::Assigned)
                    | (RF::Ruby, OP::Bidi_Mirrored) => {
                        return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));
                    }
                    _ => {}
                }
                set.add_prop(RegexProperty::Other(o).negative_item(negative));
            } else {
                return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));
            }
        }
    }
    Ok(())
}

#[cfg_attr(feature = "dbg", derive(Debug))]
#[derive(Default)]
pub(crate) struct RegexCharSet {
    pub(crate) negative: bool,
    pub(crate) set: UnicodeSet,
}

impl RegexCharSet {
    pub(crate) fn new(items: UnicodeSet) -> Self {
        Self { negative: false, set: items }
    }

    pub(crate) fn negate(mut self) -> Self {
        self.negative = !self.negative;
        self
    }

    pub(crate) fn codegen(&self, buf: &mut String, flavor: RegexFlavor) {
        if self.set.len() == 1 {
            if let Some(range) = self.set.ranges().next() {
                let (first, last) = range.as_chars();
                if first == last && !self.negative {
                    return literal::codegen_char_esc(first, buf, flavor);
                }
            } else if let Some(prop) = self.set.props().next() {
                match prop {
                    RegexCharSetItem::Shorthand(s) => {
                        let shorthand = if self.negative { s.negate() } else { Some(s) };
                        if let Some(shorthand) = shorthand {
                            return shorthand.codegen(buf);
                        }
                    }
                    RegexCharSetItem::Property { negative, value } => {
                        return value.codegen(buf, negative ^ self.negative, flavor);
                    }
                }
            }
        }

        if self.negative {
            buf.push_str("[^");
        } else {
            buf.push('[');
        }

        let mut is_first = true;
        for prop in self.set.props() {
            match prop {
                RegexCharSetItem::Shorthand(s) => s.codegen(buf),
                RegexCharSetItem::Property { negative, value } => {
                    value.codegen(buf, negative, flavor);
                }
            }
            is_first = false;
        }
        for range in self.set.ranges() {
            let (first, last) = range.as_chars();
            if first == last {
                literal::compile_char_esc_in_class(first, buf, is_first, flavor);
            } else {
                literal::compile_char_esc_in_class(first, buf, is_first, flavor);
                if range.first + 1 < range.last {
                    buf.push('-');
                }
                literal::compile_char_esc_in_class(last, buf, false, flavor);
            }
            is_first = false;
        }

        buf.push(']');
    }
}

#[derive(Clone, Copy, PartialEq, Eq)]
pub(crate) enum RegexCharSetItem {
    Shorthand(RegexShorthand),
    Property { negative: bool, value: RegexProperty },
}

impl RegexCharSetItem {
    pub(crate) fn negate(self) -> Option<Self> {
        match self {
            RegexCharSetItem::Shorthand(s) => s.negate().map(RegexCharSetItem::Shorthand),
            RegexCharSetItem::Property { negative, value } => {
                Some(RegexCharSetItem::Property { negative: !negative, value })
            }
        }
    }
}

impl fmt::Debug for RegexCharSetItem {
    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
        match self {
            Self::Shorthand(s) => f.write_str(s.as_str()),
            &Self::Property { value, negative } => {
                if negative {
                    f.write_str("!")?;
                }
                f.write_str(value.as_str())
            }
        }
    }
}

1	//! Implements _character classes_. The analogue in the regex world are
2	//! [character classes](https://www.regular-expressions.info/charclass.html),
3	//! [shorthand character classes](https://www.regular-expressions.info/shorthand.html),
4	//! [non-printable characters](https://www.regular-expressions.info/nonprint.html),
5	//! [Unicode categories/scripts/blocks](https://www.regular-expressions.info/unicode.html#category),
6	//! [POSIX classes](https://www.regular-expressions.info/posixbrackets.html#class) and the
7	//! [dot](https://www.regular-expressions.info/dot.html).
8	//!
9	//! All kinds of character classes mentioned above require `[` square brackets
10	//! `]` in Pomsky. A character class can be negated by putting the keyword `not`
11	//! after the opening bracket. For example, `![.]` compiles to `\n`.
12	//!
13	//! ## Items
14	//!
15	//! A character class can contain multiple _items_, which can be
16	//!
17	//! - A __code point__, e.g. `['a']` or `[U+107]`
18	//!
19	//! - This includes [non-printable characters](https://www.regular-expressions.info/nonprint.html).\
20	//! Supported are `[n]`, `[r]`, `[t]`, `[a]`, `[e]` and `[f]`.
21	//!
22	//! - A __range of code points__. For example, `[U+10 - U+200]` matches any code
23	//! point P where `U+10 ≤ P ≤ U+200`
24	//!
25	//! - A __named character class__, which can be one of
26	//!
27	//! - a [shorthand character class](https://www.regular-expressions.info/shorthand.html).\
28	//! Supported are `[w]`, `[d]`, `[s]`, `[h]`, `[v]` and `[R]`.
29	//!
30	//! - a [POSIX class](https://www.regular-expressions.info/posixbrackets.html#class).\
31	//! Supported are `[ascii_alnum]`, `[ascii_alpha]`, `[ascii]`,
32	//! `[ascii_blank]`, `[ascii_cntrl]`, `[ascii_digit]`, `[ascii_graph]`,
33	//! `[ascii_lower]`, `[ascii_print]`, `[ascii_punct]`, ´ `[ascii_space]`,
34	//! `[ascii_upper]`, `[ascii_word]` and `[ascii_xdigit]`.\ _Note_: POSIX
35	//! classes are not Unicode aware!\ _Note_: They're converted to ranges,
36	//! e.g. `[ascii_alpha]` = `[a-zA-Z]`.
37	//!
38	//! - a [Unicode category, script or block](https://www.regular-expressions.info/unicode.html#category).\
39	//! For example: `[Letter]` compiles to `\p{Letter}`. Pomsky currently
40	//! treats any uppercase identifier except `R` as Unicode class.
41	//!
42	//! ## Compilation
43	//!
44	//! When a character class contains only a single item (e.g. `[w]`), the
45	//! character class is "flattened":
46	//!
47	//! - `['a']` = `a`
48	//! - `[w]` = `\w`
49	//! - `[Letter]` = `\p{Letter}`
50	//!
51	//! When there is more than one item or a range (e.g. `['a'-'z' '!']`), a regex
52	//! character class is created:
53	//!
54	//! - `['a'-'z' '!']` = `[a-z!]`
55	//! - `[w e Punctuation]` = `[\w\e\p{Punctuation}]`
56	//!
57	//! ### Negation
58	//!
59	//! Negation is implemented as follows:
60	//!
61	//! - Ranges and chars such as `!['a'-'z' '!' e]` are wrapped in a negative
62	//! character class, e.g. `[^a-z!\e]`.
63	//!
64	//! - The `h`, `v` and `R` shorthands are also wrapped in a negative character
65	//! class.
66	//!
67	//! - The `w`, `d` and `s` shorthands are negated by making them uppercase
68	//! (`![w]` = `\W`), except when there is more than one item in the class
69	//! (`![w '-']` = `[^\w\-]`)
70	//!
71	//! - `w`, `s`, `d` and Unicode categories/scripts/blocks can be negated
72	//! individually _within a character class_, e.g. `[s !s]` = `[\s\S]`,
73	//! `![!Latin 'a']` = `[^\P{Latin}a]`.
74	//!
75	//! When a negated character class only contains 1 item, which is also
76	//! negated, the class is removed and the negations cancel each other out:
77	//! `![!w]` = `\w`, `![!L]` = `\p{L}`.
78
79	use std::fmt;
80
81	use crate::{
82	compile::{CompileResult, CompileState},
83	diagnose::{CompileError, CompileErrorKind, Feature},
84	exprs::literal,
85	options::{CompileOptions, RegexFlavor},
86	regex::{Regex, RegexProperty, RegexShorthand},
87	unicode_set::UnicodeSet,
88	};
89
90	use pomsky_syntax::{
91	exprs::{Category, CharClass, CodeBlock, GroupItem, GroupName, OtherProperties, Script},
92	Span,
93	};
94
95	use super::Compile;
96
97	impl Compile for CharClass {
98	fn compile(&self, options: CompileOptions, _state: &mut CompileState<'_>) -> CompileResult {	234✔
99	// when single, a `[!w]` can be turned into `![w]`	234✔
100	let is_single = self.inner.len() == 1;	234✔
101	let mut group_negative = false;	234✔
102		234✔
103	let mut set = UnicodeSet::new();	234✔
104	for item in &self.inner {	600✔
105	match *item {	374✔
106	GroupItem::Char(c) => {	204✔
107	if !is_single {	204✔
108	validate_char_in_class(c, options.flavor, self.span)?;	140✔
109	}	64✔
110	set.add_char(c)	204✔
111	}
112	GroupItem::Range { first, last } => {	30✔
113	validate_char_in_class(first, options.flavor, self.span)?;	30✔
114	validate_char_in_class(last, options.flavor, self.span)?;	30✔
115	set.add_range(first..=last);	30✔
116	}
117	GroupItem::Named { name, negative, span } => {	140✔
118	if self.unicode_aware {	140✔
119	named_class_to_regex_unicode(	110✔
120	name,	110✔
121	negative,	110✔
122	&mut group_negative,	110✔
123	is_single,	110✔
124	options.flavor,	110✔
125	span,	110✔
126	&mut set,	110✔
127	)?;	110✔
128	} else {
129	named_class_to_regex_ascii(name, negative, options.flavor, span, &mut set)?;	30✔
130	}
131	}
132	}
133	}
134
135	// this makes it possible to use code points outside the BMP in .NET,
136	// as long as there is only one in the character set
137	if let Some(only_char) = set.try_into_char() {	226✔
138	return Ok(Regex::Literal(only_char.to_string()));	63✔
139	}	163✔
140		163✔
141	Ok(Regex::CharSet(RegexCharSet { negative: group_negative, set }))	163✔
142	}	234✔
143	}
144
145	fn validate_char_in_class(char: char, flavor: RegexFlavor, span: Span) -> Result<(), CompileError> {	200✔
146	if flavor == RegexFlavor::DotNet && char > '\u{FFFF}' {	200✔
147	Err(CompileErrorKind::Unsupported(Feature::LargeCodePointInCharClass(char), flavor)	×
148	.at(span))	×
149	} else {
150	Ok(())	200✔
151	}
152	}	200✔
153
154	pub(crate) fn check_char_class_empty(	49✔
155	char_set: &RegexCharSet,	49✔
156	span: Span,	49✔
157	) -> Result<(), CompileError> {	49✔
158	if char_set.negative {	49✔
159	if let Some((group1, group2)) = char_set.set.full_props() {	48✔
160	return Err(CompileErrorKind::EmptyClassNegated { group1, group2 }.at(span));	3✔
161	}	45✔
162	}	1✔
163	Ok(())	46✔
164	}	49✔
165
166	pub fn is_ascii_only_in_flavor(group: GroupName, flavor: RegexFlavor) -> bool {	1✔
167	match flavor {	1✔
NEW 168	RegexFlavor::JavaScript => matches!(group, GroupName::Word \| GroupName::Digit),	×
NEW 169	RegexFlavor::RE2 => matches!(group, GroupName::Word \| GroupName::Digit \| GroupName::Space),	×
170	_ => false,	1✔
171	}
172	}	1✔
173
174	fn named_class_to_regex_ascii(	30✔
175	group: GroupName,	30✔
176	negative: bool,	30✔
177	flavor: RegexFlavor,	30✔
178	span: Span,	30✔
179	set: &mut UnicodeSet,	30✔
180	) -> Result<(), CompileError> {	30✔
181	// In JS, \W and \D can be used for negation because they're ascii-only	30✔
182	// Same goes for \W, \D and \S in RE2	30✔
183	if negative && !is_ascii_only_in_flavor(group, flavor) {	30✔
184	return Err(CompileErrorKind::NegativeShorthandInAsciiMode.at(span));	1✔
185	}	29✔
186		29✔
187	match group {	29✔
188	GroupName::Word => {
189	if let RegexFlavor::JavaScript \| RegexFlavor::RE2 = flavor {	8✔
190	let s = if negative { RegexShorthand::NotWord } else { RegexShorthand::Word };	2✔
191	set.add_prop(RegexCharSetItem::Shorthand(s));	2✔
192	} else {	6✔
193	// we already checked above if negative	6✔
194	set.add_range('a'..='z');	6✔
195	set.add_range('A'..='Z');	6✔
196	set.add_range('0'..='9');	6✔
197	set.add_char('_');	6✔
198	}	6✔
199	}
200	GroupName::Digit => {
201	if let RegexFlavor::JavaScript \| RegexFlavor::RE2 = flavor {	12✔
202	let s = if negative { RegexShorthand::NotDigit } else { RegexShorthand::Digit };	2✔
203	set.add_prop(RegexCharSetItem::Shorthand(s));	2✔
204	} else {	10✔
205	// we already checked above if negative	10✔
206	set.add_range('0'..='9');	10✔
207	}	10✔
208	}
209	GroupName::Space => {
210	if let RegexFlavor::RE2 = flavor {	8✔
211	let s = if negative { RegexShorthand::NotSpace } else { RegexShorthand::Space };	1✔
212	set.add_prop(RegexCharSetItem::Shorthand(s));	1✔
213	} else {	7✔
214	set.add_char(' ');	7✔
215	set.add_range('\x09'..='\x0D'); // \t\n\v\f\r	7✔
216	}	7✔
217	}
218	GroupName::HorizSpace => set.add_char('\t'),	×
219	GroupName::VertSpace => set.add_range('\x0A'..='\x0D'),	×
220	_ => return Err(CompileErrorKind::UnicodeInAsciiMode.at(span)),	1✔
221	}
222	Ok(())	28✔
223	}	30✔
224
225	fn named_class_to_regex_unicode(	110✔
226	group: GroupName,	110✔
227	negative: bool,	110✔
228	group_negative: &mut bool,	110✔
229	is_single: bool,	110✔
230	flavor: RegexFlavor,	110✔
231	span: Span,	110✔
232	set: &mut UnicodeSet,	110✔
233	) -> Result<(), CompileError> {	110✔
234	match group {	5✔
235	GroupName::Word => {
236	if flavor == RegexFlavor::RE2 {	22✔
NEW 237	return Err(CompileErrorKind::Unsupported(Feature::ShorthandW, flavor).at(span));	×
238	} else if flavor == RegexFlavor::JavaScript {	22✔
239	if negative {	5✔
240	if is_single {	2✔
241	*group_negative ^= true;	1✔
242	} else {	1✔
243	return Err(CompileErrorKind::Unsupported(	1✔
244	Feature::NegativeShorthandW,	1✔
245	flavor,	1✔
246	)	1✔
247	.at(span));	1✔
248	}
249	}	3✔
250	set.add_prop(	4✔
251	RegexProperty::Other(OtherProperties::Alphabetic).negative_item(false),	4✔
252	);	4✔
253	set.add_prop(RegexProperty::Category(Category::Mark).negative_item(false));	4✔
254	set.add_prop(	4✔
255	RegexProperty::Category(Category::Decimal_Number).negative_item(false),	4✔
256	);	4✔
257	set.add_prop(	4✔
258	RegexProperty::Category(Category::Connector_Punctuation).negative_item(false),	4✔
259	);	4✔
260	} else {
261	let s = if negative { RegexShorthand::NotWord } else { RegexShorthand::Word };	17✔
262	set.add_prop(RegexCharSetItem::Shorthand(s));	17✔
263	}
264	}
265	GroupName::Digit => {
266	if matches!(flavor, RegexFlavor::JavaScript \| RegexFlavor::RE2) {	20✔
267	set.add_prop(	8✔
268	RegexProperty::Category(Category::Decimal_Number).negative_item(negative),	8✔
269	);	8✔
270	} else {	8✔
271	let s = if negative { RegexShorthand::NotDigit } else { RegexShorthand::Digit };	12✔
272	set.add_prop(RegexCharSetItem::Shorthand(s));	12✔
273	}
274	}
275
276	GroupName::Space => {
277	if flavor == RegexFlavor::RE2 {	16✔
278	if negative {	4✔
279	if is_single {	1✔
280	*group_negative ^= true;	1✔
281	} else {	1✔
NEW 282	return Err(CompileErrorKind::Unsupported(	×
NEW 283	Feature::NegativeShorthandS,	×
NEW 284	flavor,	×
NEW 285	)	×
NEW 286	.at(span));	×
287	}
288	}	3✔
289
290	// [ \f\n\r\t\u000b\u00a0\u1680\u2000-\u200a\u2028\u2029\u202f\u205f\u3000\ufeff]
291	set.add_prop(RegexCharSetItem::Shorthand(RegexShorthand::Space));	4✔
292	set.add_char('\x0b');	4✔
293	set.add_char('\u{a0}');	4✔
294	set.add_char('\u{1680}');	4✔
295	set.add_range('\u{2000}'..='\u{200a}');	4✔
296	set.add_range('\u{2028}'..='\u{2029}');	4✔
297	set.add_char('\u{202f}');	4✔
298	set.add_char('\u{205f}');	4✔
299	set.add_char('\u{3000}');	4✔
300	set.add_char('\u{feff}');	4✔
301	} else {
302	set.add_prop(RegexCharSetItem::Shorthand(if negative {	12✔
303	RegexShorthand::NotSpace	3✔
304	} else {
305	RegexShorthand::Space	9✔
306	}))
307	}
308	}
309
310	GroupName::HorizSpace \| GroupName::VertSpace if negative => {	×
311	return Err(CompileErrorKind::NegatedHorizVertSpace.at(span));	×
312	}
313
314	GroupName::HorizSpace \| GroupName::VertSpace
315	if matches!(flavor, RegexFlavor::Pcre \| RegexFlavor::Java) =>	5✔
316	{
317	set.add_prop(RegexCharSetItem::Shorthand(if group == GroupName::HorizSpace {	6✔
318	RegexShorthand::HorizSpace	3✔
319	} else {
320	RegexShorthand::VertSpace	3✔
321	}));
322	}
323	GroupName::HorizSpace => {
324	set.add_char('\t');	2✔
325	if flavor == RegexFlavor::Python {	2✔
326	return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));	×
327	} else {	2✔
328	set.add_prop(	2✔
329	RegexProperty::Category(Category::Space_Separator).negative_item(false),	2✔
330	);	2✔
331	}	2✔
332	}
333	GroupName::VertSpace => {	2✔
334	set.add_range('\x0A'..='\x0D');	2✔
335	set.add_char('\u{85}');	2✔
336	set.add_char('\u{2028}');	2✔
337	set.add_char('\u{2029}');	2✔
338	}	2✔
339
340	_ if flavor == RegexFlavor::Python => {	42✔
341	return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));	2✔
342	}
343	GroupName::Category(c) => {	5✔
344	if let (RegexFlavor::Rust, Category::Surrogate)	5✔
345	\| (RegexFlavor::DotNet \| RegexFlavor::RE2, Category::Cased_Letter) = (flavor, c)	5✔
346	{
347	return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));	×
348	}	5✔
349	set.add_prop(RegexProperty::Category(c).negative_item(negative));	5✔
350	}
351	GroupName::Script(s) => {	21✔
352	if flavor == RegexFlavor::DotNet {	21✔
353	return Err(CompileErrorKind::Unsupported(Feature::UnicodeScript, flavor).at(span));	1✔
354	}	20✔
355	if let (RegexFlavor::Ruby, Script::Kawi \| Script::Nag_Mundari)	20✔
356	\| (RegexFlavor::Rust, Script::Unknown) = (flavor, s)	20✔
357	{
358	return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));	×
359	}	20✔
360	set.add_prop(RegexProperty::Script(s).negative_item(negative));	20✔
361	}
362	GroupName::CodeBlock(b) => match flavor {	7✔
363	RegexFlavor::DotNet \| RegexFlavor::Java \| RegexFlavor::Ruby => {
364	match (flavor, b) {	6✔
365	(RegexFlavor::Java, CodeBlock::No_Block)
366	\| (
367	// These should work since Oniguruma updated to Unicode 15.1
368	// ... but our C bindings for Oniguruma are unmaintained!
369	RegexFlavor::Ruby,
370	CodeBlock::Arabic_Extended_C
371	\| CodeBlock::CJK_Unified_Ideographs_Extension_H
372	\| CodeBlock::Cyrillic_Extended_D
373	\| CodeBlock::Devanagari_Extended_A
374	\| CodeBlock::Kaktovik_Numerals,
375	) => {
376	return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));	×
377	}
378	(RegexFlavor::DotNet, _) => {
379	let dotnet_name = b.as_str().replace("_And_", "_and_").replace('_', "");	2✔
380	if pomsky_syntax::blocks_supported_in_dotnet()	2✔
381	.binary_search(&dotnet_name.as_str())	2✔
382	.is_err()	2✔
383	{
384	return Err(	×
385	CompileErrorKind::unsupported_specific_prop_in(flavor).at(span)	×
386	);	×
387	}	2✔
388	}
389	_ => {}	4✔
390	}
391
392	set.add_prop(RegexProperty::Block(b).negative_item(negative));	6✔
393	}
394	_ => return Err(CompileErrorKind::Unsupported(Feature::UnicodeBlock, flavor).at(span)),	1✔
395	},
396	GroupName::OtherProperties(o) => {	7✔
397	use OtherProperties as OP;
398	use RegexFlavor as RF;
399
400	if let RF::JavaScript \| RF::Rust \| RF::Pcre \| RF::Ruby = flavor {	7✔
401	match (flavor, o) {	7✔
402	(RF::JavaScript, _) => {}	4✔
403	(_, OP::Changes_When_NFKC_Casefolded)
404	\| (RF::Pcre, OP::Assigned)
405	\| (RF::Ruby, OP::Bidi_Mirrored) => {
406	return Err(CompileErrorKind::unsupported_specific_prop_in(flavor).at(span));	1✔
407	}
408	_ => {}	2✔
409	}
410	set.add_prop(RegexProperty::Other(o).negative_item(negative));	6✔
411	} else {
412	return Err(CompileErrorKind::Unsupported(Feature::UnicodeProp, flavor).at(span));	×
413	}
414	}
415	}
416	Ok(())	104✔
417	}	110✔
418
419	#[cfg_attr(feature = "dbg", derive(Debug))]
420	#[derive(Default)]
421	pub(crate) struct RegexCharSet {
422	pub(crate) negative: bool,
423	pub(crate) set: UnicodeSet,
424	}
425
426	impl RegexCharSet {
427	pub(crate) fn new(items: UnicodeSet) -> Self {	244✔
428	Self { negative: false, set: items }	244✔
429	}	244✔
430
431	pub(crate) fn negate(mut self) -> Self {	49✔
432	self.negative = !self.negative;	49✔
433	self	49✔
434	}	49✔
435
436	pub(crate) fn codegen(&self, buf: &mut String, flavor: RegexFlavor) {	401✔
437	if self.set.len() == 1 {	401✔
438	if let Some(range) = self.set.ranges().next() {	308✔
439	let (first, last) = range.as_chars();	242✔
440	if first == last && !self.negative {	242✔
441	return literal::codegen_char_esc(first, buf, flavor);	75✔
442	}	167✔
443	} else if let Some(prop) = self.set.props().next() {	66✔
444	match prop {	66✔
445	RegexCharSetItem::Shorthand(s) => {	27✔
446	let shorthand = if self.negative { s.negate() } else { Some(s) };	27✔
447	if let Some(shorthand) = shorthand {	27✔
448	return shorthand.codegen(buf);	25✔
449	}	2✔
450	}
451	RegexCharSetItem::Property { negative, value } => {	39✔
452	return value.codegen(buf, negative ^ self.negative, flavor);	39✔
453	}
454	}
455	}	×
456	}	93✔
457
458	if self.negative {	262✔
459	buf.push_str("[^");	30✔
460	} else {	232✔
461	buf.push('[');	232✔
462	}	232✔
463
464	let mut is_first = true;	262✔
465	for prop in self.set.props() {	262✔
466	match prop {	76✔
467	RegexCharSetItem::Shorthand(s) => s.codegen(buf),	55✔
468	RegexCharSetItem::Property { negative, value } => {	21✔
469	value.codegen(buf, negative, flavor);	21✔
470	}	21✔
471	}
472	is_first = false;	76✔
473	}
474	for range in self.set.ranges() {	378✔
475	let (first, last) = range.as_chars();	378✔
476	if first == last {	378✔
477	literal::compile_char_esc_in_class(first, buf, is_first, flavor);	119✔
478	} else {	119✔
479	literal::compile_char_esc_in_class(first, buf, is_first, flavor);	259✔
480	if range.first + 1 < range.last {	259✔
481	buf.push('-');	207✔
482	}	207✔
483	literal::compile_char_esc_in_class(last, buf, false, flavor);	259✔
484	}
485	is_first = false;	378✔
486	}
487
488	buf.push(']');	262✔
489	}	401✔
490	}
491
492	#[derive(Clone, Copy, PartialEq, Eq)]
493	pub(crate) enum RegexCharSetItem {
494	Shorthand(RegexShorthand),
495	Property { negative: bool, value: RegexProperty },
496	}
497
498	impl RegexCharSetItem {
499	pub(crate) fn negate(self) -> Option<Self> {	47✔
500	match self {	47✔
501	RegexCharSetItem::Shorthand(s) => s.negate().map(RegexCharSetItem::Shorthand),	21✔
502	RegexCharSetItem::Property { negative, value } => {	26✔
503	Some(RegexCharSetItem::Property { negative: !negative, value })	26✔
504	}
505	}
506	}	47✔
507	}
508
509	impl fmt::Debug for RegexCharSetItem {
510	fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {	6✔
511	match self {	6✔
512	Self::Shorthand(s) => f.write_str(s.as_str()),	4✔
513	&Self::Property { value, negative } => {	2✔
514	if negative {	2✔
515	f.write_str("!")?;	1✔
516	}	1✔
517	f.write_str(value.as_str())	2✔
518	}
519	}
520	}	6✔
521	}

pomsky-lang / pomsky / 12075733362

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous