Quelle  serializer.rs

/* This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

use crate::match_byte;
use dtoa_short::Notation;
use std::fmt::{self, Write};
use std::str;

use super::Token;

/// Trait for things the can serialize themselves in CSS syntax.
pub trait ToCss {
    /// Serialize `self` in CSS syntax, writing to `dest`.
    fn to_css<W>(&self, dest: &mut W) -> fmt::Result
    where
        W: fmt::Write;

    /// Serialize `self` in CSS syntax and return a string.
    ///
    /// (This is a convenience wrapper for `to_css` and probably should not be overridden.)
    #[inline]
    fn to_css_string(&self) -> String {
        let mut s = String::new();
        self.to_css(&mut s).unwrap();
        s
    }
}

#[inline]
fn write_numeric<W>(value: f32, int_value: Option<i32>, has_sign: bool, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    // `value.value >= 0` is true for negative 0.
    if has_sign && value.is_sign_positive() {
        dest.write_str("+")?;
    }

    let notation = if value == 0.0 && value.is_sign_negative() {
        // Negative zero. Work around #20596.
        dest.write_str("-0")?;
        Notation {
            decimal_point: false,
            scientific: false,
        }
    } else {
        dtoa_short::write(dest, value)?
    };

    if int_value.is_none() && value.fract() == 0. && !notation.decimal_point && !notation.scientific
    {
        dest.write_str(".0")?;
    }
    Ok(())
}

impl<'a> ToCss for Token<'a> {
    fn to_css<W>(&self, dest: &mut W) -> fmt::Result
    where
        W: fmt::Write,
    {
        match *self {
            Token::Ident(ref value) => serialize_identifier(value, dest)?,
            Token::AtKeyword(ref value) => {
                dest.write_str("@")?;
                serialize_identifier(value, dest)?;
            }
            Token::Hash(ref value) => {
                dest.write_str("#")?;
                serialize_name(value, dest)?;
            }
            Token::IDHash(ref value) => {
                dest.write_str("#")?;
                serialize_identifier(value, dest)?;
            }
            Token::QuotedString(ref value) => serialize_string(value, dest)?,
            Token::UnquotedUrl(ref value) => {
                dest.write_str("url(")?;
                serialize_unquoted_url(value, dest)?;
                dest.write_str(")")?;
            }
            Token::Delim(value) => dest.write_char(value)?,

            Token::Number {
                value,
                int_value,
                has_sign,
            } => write_numeric(value, int_value, has_sign, dest)?,
            Token::Percentage {
                unit_value,
                int_value,
                has_sign,
            } => {
                write_numeric(unit_value * 100., int_value, has_sign, dest)?;
                dest.write_str("%")?;
            }
            Token::Dimension {
                value,
                int_value,
                has_sign,
                ref unit,
            } => {
                write_numeric(value, int_value, has_sign, dest)?;
                // Disambiguate with scientific notation.
                let unit = &**unit;
                // TODO(emilio): This doesn't handle e.g. 100E1m, which gets us
                // an unit of "E1m"...
                if unit == "e" || unit == "E" || unit.starts_with("e-") || unit.starts_with("E-") {
                    dest.write_str("\\65 ")?;
                    serialize_name(&unit[1..], dest)?;
                } else {
                    serialize_identifier(unit, dest)?;
                }
            }

            Token::WhiteSpace(content) => dest.write_str(content)?,
            Token::Comment(content) => {
                dest.write_str("/*")?;
                dest.write_str(content)?;
                dest.write_str("*/")?
            }
            Token::Colon => dest.write_str(":")?,
            Token::Semicolon => dest.write_str(";")?,
            Token::Comma => dest.write_str(",")?,
            Token::IncludeMatch => dest.write_str("~=")?,
            Token::DashMatch => dest.write_str("|=")?,
            Token::PrefixMatch => dest.write_str("^=")?,
            Token::SuffixMatch => dest.write_str("$=")?,
            Token::SubstringMatch => dest.write_str("*=")?,
            Token::CDO => dest.write_str("<!--")?,
            Token::CDC => dest.write_str("-->")?,

            Token::Function(ref name) => {
                serialize_identifier(name, dest)?;
                dest.write_str("(")?;
            }
            Token::ParenthesisBlock => dest.write_str("(")?,
            Token::SquareBracketBlock => dest.write_str("[")?,
            Token::CurlyBracketBlock => dest.write_str("{")?,

            Token::BadUrl(ref contents) => {
                dest.write_str("url(")?;
                dest.write_str(contents)?;
                dest.write_char(')')?;
            }
            Token::BadString(ref value) => {
                // During tokenization, an unescaped newline after a quote causes
                // the token to be a BadString instead of a QuotedString.
                // The BadString token ends just before the newline
                // (which is in a separate WhiteSpace token),
                // and therefore does not have a closing quote.
                dest.write_char('"')?;
                CssStringWriter::new(dest).write_str(value)?;
            }
            Token::CloseParenthesis => dest.write_str(")")?,
            Token::CloseSquareBracket => dest.write_str("]")?,
            Token::CloseCurlyBracket => dest.write_str("}")?,
        }
        Ok(())
    }
}

fn hex_escape<W>(ascii_byte: u8, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    static HEX_DIGITS: &[u8; 16] = b"0123456789abcdef";
    let b3;
    let b4;
    let bytes = if ascii_byte > 0x0F {
        let high = (ascii_byte >> 4) as usize;
        let low = (ascii_byte & 0x0F) as usize;
        b4 = [b'\\', HEX_DIGITS[high], HEX_DIGITS[low], b' '];
        &b4[..]
    } else {
        b3 = [b'\\', HEX_DIGITS[ascii_byte as usize], b' '];
        &b3[..]
    };
    dest.write_str(unsafe { str::from_utf8_unchecked(bytes) })
}

fn char_escape<W>(ascii_byte: u8, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    let bytes = [b'\\', ascii_byte];
    dest.write_str(unsafe { str::from_utf8_unchecked(&bytes) })
}

/// Write a CSS identifier, escaping characters as necessary.
pub fn serialize_identifier<W>(mut value: &str, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    if value.is_empty() {
        return Ok(());
    }

    if let Some(value) = value.strip_prefix("--") {
        dest.write_str("--")?;
        serialize_name(value, dest)
    } else if value == "-" {
        dest.write_str("\\-")
    } else {
        if value.as_bytes()[0] == b'-' {
            dest.write_str("-")?;
            value = &value[1..];
        }
        if let digit @ b'0'..=b'9' = value.as_bytes()[0] {
            hex_escape(digit, dest)?;
            value = &value[1..];
        }
        serialize_name(value, dest)
    }
}

/// Write a CSS name, like a custom property name.
///
/// You should only use this when you know what you're doing, when in doubt,
/// consider using `serialize_identifier`.
pub fn serialize_name<W>(value: &str, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    let mut chunk_start = 0;
    for (i, b) in value.bytes().enumerate() {
        let escaped = match_byte! { b,
            b'0'..=b'9' | b'A'..=b'Z' | b'a'..=b'z' | b'_' | b'-' => continue,
            b'\0' => Some("\u{FFFD}"),
            b => {
                if !b.is_ascii() {
                    continue;
                }
                None
            },
        };
        dest.write_str(&value[chunk_start..i])?;
        if let Some(escaped) = escaped {
            dest.write_str(escaped)?;
        } else if (b'\x01'..=b'\x1F').contains(&b) || b == b'\x7F' {
            hex_escape(b, dest)?;
        } else {
            char_escape(b, dest)?;
        }
        chunk_start = i + 1;
    }
    dest.write_str(&value[chunk_start..])
}

fn serialize_unquoted_url<W>(value: &str, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    let mut chunk_start = 0;
    for (i, b) in value.bytes().enumerate() {
        let hex = match_byte! { b,
            b'\0'..=b' ' | b'\x7F' => true,
            b'(' | b')' | b'"' | b'\'' | b'\\' => false,
            _ => continue,
        };
        dest.write_str(&value[chunk_start..i])?;
        if hex {
            hex_escape(b, dest)?;
        } else {
            char_escape(b, dest)?;
        }
        chunk_start = i + 1;
    }
    dest.write_str(&value[chunk_start..])
}

/// Write a double-quoted CSS string token, escaping content as necessary.
pub fn serialize_string<W>(value: &str, dest: &mut W) -> fmt::Result
where
    W: fmt::Write,
{
    dest.write_str("\"")?;
    CssStringWriter::new(dest).write_str(value)?;
    dest.write_str("\"")?;
    Ok(())
}

/// A `fmt::Write` adapter that escapes text for writing as a double-quoted CSS string.
/// Quotes are not included.
///
/// Typical usage:
///
/// ```{rust,ignore}
/// fn write_foo<W>(foo: &Foo, dest: &mut W) -> fmt::Result where W: fmt::Write {
///     dest.write_str("\"")?;
///     {
///         let mut string_dest = CssStringWriter::new(dest);
///         // Write into string_dest...
///     }
///     dest.write_str("\"")?;
///     Ok(())
/// }
/// ```
pub struct CssStringWriter<'a, W> {
    inner: &'a mut W,
}

impl<'a, W> CssStringWriter<'a, W>
where
    W: fmt::Write,
{
    /// Wrap a text writer to create a `CssStringWriter`.
    pub fn new(inner: &'a mut W) -> CssStringWriter<'a, W> {
        CssStringWriter { inner }
    }
}

impl<'a, W> fmt::Write for CssStringWriter<'a, W>
where
    W: fmt::Write,
{
    fn write_str(&mut self, s: &str) -> fmt::Result {
        let mut chunk_start = 0;
        for (i, b) in s.bytes().enumerate() {
            let escaped = match_byte! { b,
                b'"' => Some("\\\""),
                b'\\' => Some("\\\\"),
                b'\0' => Some("\u{FFFD}"),
                b'\x01'..=b'\x1F' | b'\x7F' => None,
                _ => continue,
            };
            self.inner.write_str(&s[chunk_start..i])?;
            match escaped {
                Some(x) => self.inner.write_str(x)?,
                None => hex_escape(b, self.inner)?,
            };
            chunk_start = i + 1;
        }
        self.inner.write_str(&s[chunk_start..])
    }
}

macro_rules! impl_tocss_for_int {
    ($T: ty) => {
        impl ToCss for $T {
            fn to_css<W>(&self, dest: &mut W) -> fmt::Result
            where
                W: fmt::Write,
            {
                let mut buf = itoa::Buffer::new();
                dest.write_str(buf.format(*self))
            }
        }
    };
}

impl_tocss_for_int!(i8);
impl_tocss_for_int!(u8);
impl_tocss_for_int!(i16);
impl_tocss_for_int!(u16);
impl_tocss_for_int!(i32);
impl_tocss_for_int!(u32);
impl_tocss_for_int!(i64);
impl_tocss_for_int!(u64);

macro_rules! impl_tocss_for_float {
    ($T: ty) => {
        impl ToCss for $T {
            fn to_css<W>(&self, dest: &mut W) -> fmt::Result
            where
                W: fmt::Write,
            {
                dtoa_short::write(dest, *self).map(|_| ())
            }
        }
    };
}

impl_tocss_for_float!(f32);
impl_tocss_for_float!(f64);

/// A category of token. See the `needs_separator_when_before` method.
#[derive(Copy, Clone, Eq, PartialEq, Debug, Default)]
pub enum TokenSerializationType {
    /// No token serialization type.
    #[default]
    Nothing,

    /// The [`<whitespace-token>`](https://drafts.csswg.org/css-syntax/#whitespace-token-diagram)
    /// type.
    WhiteSpace,

    /// The [`<at-keyword-token>`](https://drafts.csswg.org/css-syntax/#at-keyword-token-diagram)
    /// type, the "[`<hash-token>`](https://drafts.csswg.org/css-syntax/#hash-token-diagram) with
    /// the type flag set to 'unrestricted'" type, or the
    /// "[`<hash-token>`](https://drafts.csswg.org/css-syntax/#hash-token-diagram) with the type
    /// flag set to 'id'" type.
    AtKeywordOrHash,

    /// The [`<number-token>`](https://drafts.csswg.org/css-syntax/#number-token-diagram) type.
    Number,

    /// The [`<dimension-token>`](https://drafts.csswg.org/css-syntax/#dimension-token-diagram)
    /// type.
    Dimension,

    /// The [`<percentage-token>`](https://drafts.csswg.org/css-syntax/#percentage-token-diagram)
    /// type.
    Percentage,

    /// The [`<url-token>`](https://drafts.csswg.org/css-syntax/#url-token-diagram) or
    /// `<bad-url-token>` type.
    UrlOrBadUrl,

    /// The [`<function-token>`](https://drafts.csswg.org/css-syntax/#function-token-diagram) type.
    Function,

    /// The [`<ident-token>`](https://drafts.csswg.org/css-syntax/#ident-token-diagram) type.
    Ident,

    /// The `-->` [`<CDC-token>`](https://drafts.csswg.org/css-syntax/#CDC-token-diagram) type.
    CDC,

    /// The `|=`
    /// [`<dash-match-token>`](https://drafts.csswg.org/css-syntax/#dash-match-token-diagram) type.
    DashMatch,

    /// The `*=`
    /// [`<substring-match-token>`](https://drafts.csswg.org/css-syntax/#substring-match-token-diagram)
    /// type.
    SubstringMatch,

    /// The `<(-token>` type.
    OpenParen,

    /// The `#` `<delim-token>` type.
    DelimHash,

    /// The `@` `<delim-token>` type.
    DelimAt,

    /// The `.` or `+` `<delim-token>` type.
    DelimDotOrPlus,

    /// The `-` `<delim-token>` type.
    DelimMinus,

    /// The `?` `<delim-token>` type.
    DelimQuestion,

    /// The `$`, `^`, or `~` `<delim-token>` type.
    DelimAssorted,

    /// The `=` `<delim-token>` type.
    DelimEquals,

    /// The `|` `<delim-token>` type.
    DelimBar,

    /// The `/` `<delim-token>` type.
    DelimSlash,

    /// The `*` `<delim-token>` type.
    DelimAsterisk,

    /// The `%` `<delim-token>` type.
    DelimPercent,

    /// A type indicating any other token.
    Other,
}

impl TokenSerializationType {
    /// Return a value that represents the absence of a token, e.g. before the start of the input.
    #[deprecated(
        since = "0.32.1",
        note = "use TokenSerializationType::Nothing or TokenSerializationType::default() instead"
    )]
    pub fn nothing() -> TokenSerializationType {
        Default::default()
    }

    /// If this value is `TokenSerializationType::Nothing`, set it to the given value instead.
    pub fn set_if_nothing(&mut self, new_value: TokenSerializationType) {
        if matches!(self, TokenSerializationType::Nothing) {
            *self = new_value
        }
    }

    /// Return true if, when a token of category `self` is serialized just before
    /// a token of category `other` with no whitespace in between,
    /// an empty comment `/**/` needs to be inserted between them
    /// so that they are not re-parsed as a single token.
    ///
    /// See https://drafts.csswg.org/css-syntax/#serialization
    ///
    /// See https://github.com/w3c/csswg-drafts/issues/4088 for the
    /// `DelimPercent` bits.
    pub fn needs_separator_when_before(self, other: TokenSerializationType) -> bool {
        use self::TokenSerializationType::*;
        match self {
            Ident => matches!(
                other,
                Ident
                    | Function
                    | UrlOrBadUrl
                    | DelimMinus
                    | Number
                    | Percentage
                    | Dimension
                    | CDC
                    | OpenParen
            ),
            AtKeywordOrHash | Dimension => matches!(
                other,
                Ident | Function | UrlOrBadUrl | DelimMinus | Number | Percentage | Dimension | CDC
            ),
            DelimHash | DelimMinus => matches!(
                other,
                Ident | Function | UrlOrBadUrl | DelimMinus | Number | Percentage | Dimension
            ),
            Number => matches!(
                other,
                Ident
                    | Function
                    | UrlOrBadUrl
                    | DelimMinus
                    | Number
                    | Percentage
                    | DelimPercent
                    | Dimension
            ),
            DelimAt => matches!(other, Ident | Function | UrlOrBadUrl | DelimMinus),
            DelimDotOrPlus => matches!(other, Number | Percentage | Dimension),
            DelimAssorted | DelimAsterisk => matches!(other, DelimEquals),
            DelimBar => matches!(other, DelimEquals | DelimBar | DashMatch),
            DelimSlash => matches!(other, DelimAsterisk | SubstringMatch),
            Nothing | WhiteSpace | Percentage | UrlOrBadUrl | Function | CDC | OpenParen
            | DashMatch | SubstringMatch | DelimQuestion | DelimEquals | DelimPercent | Other => {
                false
            }
        }
    }
}

impl<'a> Token<'a> {
    /// Categorize a token into a type that determines when `/**/` needs to be inserted
    /// between two tokens when serialized next to each other without whitespace in between.
    ///
    /// See the `TokenSerializationType::needs_separator_when_before` method.
    pub fn serialization_type(&self) -> TokenSerializationType {
        use self::TokenSerializationType::*;
        match self {
            Token::Ident(_) => Ident,
            Token::AtKeyword(_) | Token::Hash(_) | Token::IDHash(_) => AtKeywordOrHash,
            Token::UnquotedUrl(_) | Token::BadUrl(_) => UrlOrBadUrl,
            Token::Delim('#') => DelimHash,
            Token::Delim('@') => DelimAt,
            Token::Delim('.') | Token::Delim('+') => DelimDotOrPlus,
            Token::Delim('-') => DelimMinus,
            Token::Delim('?') => DelimQuestion,
            Token::Delim('$') | Token::Delim('^') | Token::Delim('~') => DelimAssorted,
            Token::Delim('%') => DelimPercent,
            Token::Delim('=') => DelimEquals,
            Token::Delim('|') => DelimBar,
            Token::Delim('/') => DelimSlash,
            Token::Delim('*') => DelimAsterisk,
            Token::Number { .. } => Number,
            Token::Percentage { .. } => Percentage,
            Token::Dimension { .. } => Dimension,
            Token::WhiteSpace(_) => WhiteSpace,
            Token::Comment(_) => DelimSlash,
            Token::DashMatch => DashMatch,
            Token::SubstringMatch => SubstringMatch,
            Token::CDC => CDC,
            Token::Function(_) => Function,
            Token::ParenthesisBlock => OpenParen,
            Token::SquareBracketBlock
            | Token::CurlyBracketBlock
            | Token::CloseParenthesis
            | Token::CloseSquareBracket
            | Token::CloseCurlyBracket
            | Token::QuotedString(_)
            | Token::BadString(_)
            | Token::Delim(_)
            | Token::Colon
            | Token::Semicolon
            | Token::Comma
            | Token::CDO
            | Token::IncludeMatch
            | Token::PrefixMatch
            | Token::SuffixMatch => Other,
        }
    }
}