Better smartquotes (#4849)

crates/typst/src/layout/inline/collect.rs

@@ -16,8 +16,6 @@ use crate::utils::Numeric;
 // paragraph's full text.
 const SPACING_REPLACE: &str = " "; // Space
 const OBJ_REPLACE: &str = "\u{FFFC}"; // Object Replacement Character
-const SPACING_REPLACE_CHAR: char = ' ';
-const OBJ_REPLACE_CHAR: char = '\u{FFFC}';
 
 // Unicode BiDi control characters.
 const LTR_EMBEDDING: &str = "\u{202A}";
@@ -125,8 +123,8 @@ pub fn collect<'a>(
     consecutive: bool,
 ) -> SourceResult<(String, Vec<Segment<'a>>, SpanMapper)> {
     let mut collector = Collector::new(2 + children.len());
-    let mut iter = children.iter(styles).peekable();
     let mut locator = locator.split();
+    let mut quoter = SmartQuoter::new();
 
     let outer_dir = TextElem::dir_in(*styles);
     let first_line_indent = ParElem::first_line_indent_in(*styles);
@@ -144,7 +142,7 @@ pub fn collect<'a>(
         collector.spans.push(1, Span::detached());
     }
 
-    while let Some((child, styles)) = iter.next() {
+    for (child, styles) in children.iter(styles) {
         let prev_len = collector.full.len();
 
         if child.is::<SpaceElem>() {
@@ -191,32 +189,16 @@ pub fn collect<'a>(
         } else if let Some(elem) = child.to_packed::<SmartQuoteElem>() {
             let double = elem.double(styles);
             if elem.enabled(styles) {
-                let quotes = SmartQuotes::new(
+                let quotes = SmartQuotes::get(
                     elem.quotes(styles),
                     TextElem::lang_in(styles),
                     TextElem::region_in(styles),
                     elem.alternative(styles),
                 );
-                let peeked = iter.peek().and_then(|(child, _)| {
-                    if let Some(elem) = child.to_packed::<TextElem>() {
-                        elem.text().chars().find(|c| !is_default_ignorable(*c))
-                    } else if child.is::<SmartQuoteElem>() {
-                        Some('"')
-                    } else if child.is::<SpaceElem>()
-                        || child.is::<HElem>()
-                        || child.is::<LinebreakElem>()
-                        // This is a temporary hack. We should rather skip these
-                        // and peek at the next child.
-                        || child.is::<TagElem>()
-                    {
-                        Some(SPACING_REPLACE_CHAR)
-                    } else {
-                        Some(OBJ_REPLACE_CHAR)
-                    }
-                });
-
-                let quote = collector.quoter.quote(&quotes, double, peeked);
-                collector.push_quote(quote, styles);
+                let before =
+                    collector.full.chars().rev().find(|&c| !is_default_ignorable(c));
+                let quote = quoter.quote(before, &quotes, double);
+                collector.push_text(quote, styles);
             } else {
                 collector.push_text(if double { "\"" } else { "'" }, styles);
             }
@@ -261,7 +243,6 @@ struct Collector<'a> {
     full: String,
     segments: Vec<Segment<'a>>,
     spans: SpanMapper,
-    quoter: SmartQuoter,
 }
 
 impl<'a> Collector<'a> {
@@ -270,13 +251,12 @@ impl<'a> Collector<'a> {
             full: String::new(),
             segments: Vec::with_capacity(capacity),
             spans: SpanMapper::new(),
-            quoter: SmartQuoter::new(),
         }
     }
 
     fn push_text(&mut self, text: &str, styles: StyleChain<'a>) {
         self.full.push_str(text);
-        self.push_segment(Segment::Text(text.len(), styles), false);
+        self.push_segment(Segment::Text(text.len(), styles));
     }
 
     fn build_text<F>(&mut self, styles: StyleChain<'a>, f: F)
@@ -286,24 +266,15 @@ impl<'a> Collector<'a> {
         let prev = self.full.len();
         f(&mut self.full);
         let len = self.full.len() - prev;
-        self.push_segment(Segment::Text(len, styles), false);
-    }
-
-    fn push_quote(&mut self, quote: &str, styles: StyleChain<'a>) {
-        self.full.push_str(quote);
-        self.push_segment(Segment::Text(quote.len(), styles), true);
+        self.push_segment(Segment::Text(len, styles));
     }
 
     fn push_item(&mut self, item: Item<'a>) {
         self.full.push_str(item.textual());
-        self.push_segment(Segment::Item(item), false);
-    }
-
-    fn push_segment(&mut self, segment: Segment<'a>, is_quote: bool) {
-        if let Some(last) = self.full.chars().rev().find(|c| !is_default_ignorable(*c)) {
-            self.quoter.last(last, is_quote);
+        self.push_segment(Segment::Item(item));
     }
 
+    fn push_segment(&mut self, segment: Segment<'a>) {
         if let (Some(Segment::Text(last_len, last_styles)), Segment::Text(len, styles)) =
             (self.segments.last_mut(), &segment)
         {

crates/typst/src/model/quote.rs

@@ -159,7 +159,7 @@ impl Show for Packed<QuoteElem> {
         let block = self.block(styles);
 
         if self.quotes(styles) == Smart::Custom(true) || !block {
-            let quotes = SmartQuotes::new(
+            let quotes = SmartQuotes::get(
                 SmartQuoteElem::quotes_in(styles),
                 TextElem::lang_in(styles),
                 TextElem::region_in(styles),

crates/typst/src/text/smartquote.rs

@@ -97,68 +97,80 @@ impl PlainText for Packed<SmartQuoteElem> {
     }
 }
 
-/// State machine for smart quote substitution.
+/// A smart quote substitutor with zero lookahead.
 #[derive(Debug, Clone)]
 pub struct SmartQuoter {
-    /// How many quotes have been opened.
-    quote_depth: usize,
-    /// Whether an opening quote might follow.
-    expect_opening: bool,
-    /// Whether the last character was numeric.
-    last_num: bool,
-    /// The previous type of quote character, if it was an opening quote.
-    prev_quote_type: Option<bool>,
+    /// The amount of quotes that have been opened.
+    depth: u8,
+    /// Each bit indicates whether the quote at this nesting depth is a double.
+    /// Maximum supported depth is thus 32.
+    kinds: u32,
 }
 
 impl SmartQuoter {
     /// Start quoting.
     pub fn new() -> Self {
-        Self {
-            quote_depth: 0,
-            expect_opening: true,
-            last_num: false,
-            prev_quote_type: None,
-        }
+        Self { depth: 0, kinds: 0 }
     }
 
-    /// Process the last seen character.
-    pub fn last(&mut self, c: char, is_quote: bool) {
-        self.expect_opening = is_exterior_to_quote(c) || is_opening_bracket(c);
-        self.last_num = c.is_numeric();
-        if !is_quote {
-            self.prev_quote_type = None;
-        }
-    }
-
-    /// Process and substitute a quote.
+    /// Determine which smart quote to substitute given this quoter's nesting
+    /// state and the character immediately preceding the quote.
     pub fn quote<'a>(
         &mut self,
+        before: Option<char>,
         quotes: &SmartQuotes<'a>,
         double: bool,
-        peeked: Option<char>,
     ) -> &'a str {
-        let peeked = peeked.unwrap_or(' ');
-        let mut expect_opening = self.expect_opening;
-        if let Some(prev_double) = self.prev_quote_type.take() {
-            if double != prev_double {
-                expect_opening = true;
+        let opened = self.top();
+        let before = before.unwrap_or(' ');
+
+        // If we are after a number and haven't most recently opened a quote of
+        // this kind, produce a prime. Otherwise, we prefer a closing quote.
+        if before.is_numeric() && opened != Some(double) {
+            return if double { "″" } else { "′" };
+        }
+
+        // If we have a single smart quote, didn't recently open a single
+        // quotation, and are after an alphabetic char, interpret this as an
+        // apostrophe.
+        if !double && opened != Some(false) && before.is_alphabetic() {
+            return "’";
+        }
+
+        // If the most recently opened quotation is of this kind and the
+        // previous char does not indicate a nested quotation, close it.
+        if opened == Some(double)
+            && !before.is_whitespace()
+            && !is_newline(before)
+            && !is_opening_bracket(before)
+        {
+            self.pop();
+            return quotes.close(double);
+        }
+
+        // Otherwise, open a new the quotation.
+        self.push(double);
+        quotes.open(double)
+    }
+
+    /// The top of our quotation stack. Returns `Some(double)` for the most
+    /// recently opened quote or `None` if we didn't open one.
+    fn top(&self) -> Option<bool> {
+        self.depth.checked_sub(1).map(|i| (self.kinds >> i) & 1 == 1)
+    }
+
+    /// Push onto the quotation stack.
+    fn push(&mut self, double: bool) {
+        if self.depth < 32 {
+            self.kinds |= (double as u32) << self.depth;
+            self.depth += 1;
         }
     }
 
-        if expect_opening {
-            self.quote_depth += 1;
-            self.prev_quote_type = Some(double);
-            quotes.open(double)
-        } else if self.quote_depth > 0
-            && (peeked.is_ascii_punctuation() || is_exterior_to_quote(peeked))
-        {
-            self.quote_depth -= 1;
-            quotes.close(double)
-        } else if self.last_num {
-            quotes.prime(double)
-        } else {
-            quotes.fallback(double)
-        }
+    /// Pop from the quotation stack.
+    fn pop(&mut self) {
+        self.depth -= 1;
+        self.kinds &= (1 << self.depth) - 1;
     }
 }
 
@@ -168,10 +180,7 @@ impl Default for SmartQuoter {
     }
 }
 
-fn is_exterior_to_quote(c: char) -> bool {
-    c.is_whitespace() || is_newline(c)
-}
-
+/// Whether the character is an opening bracket, parenthesis, or brace.
 fn is_opening_bracket(c: char) -> bool {
     matches!(c, '(' | '{' | '[')
 }
@@ -196,13 +205,13 @@ impl<'s> SmartQuotes<'s> {
     /// region as an all-uppercase ISO 3166-alpha2 code.
     ///
     /// Currently, the supported languages are: English, Czech, Danish, German,
-    /// Swiss / Liechtensteinian German, Estonian, Icelandic, Italian, Latin, Lithuanian,
-    /// Latvian, Slovak, Slovenian, Spanish, Bosnian, Finnish, Swedish, French,
-    /// Hungarian, Polish, Romanian, Japanese, Traditional Chinese, Russian, and
-    /// Norwegian.
+    /// Swiss / Liechtensteinian German, Estonian, Icelandic, Italian, Latin,
+    /// Lithuanian, Latvian, Slovak, Slovenian, Spanish, Bosnian, Finnish,
+    /// Swedish, French, Hungarian, Polish, Romanian, Japanese, Traditional
+    /// Chinese, Russian, and Norwegian.
     ///
     /// For unknown languages, the English quotes are used as fallback.
-    pub fn new(
+    pub fn get(
         quotes: &'s Smart<SmartQuoteDict>,
         lang: Lang,
         region: Option<Region>,
@@ -281,24 +290,6 @@ impl<'s> SmartQuotes<'s> {
             self.single_close
         }
     }
-
-    /// Which character should be used as a prime.
-    pub fn prime(&self, double: bool) -> &'static str {
-        if double {
-            "″"
-        } else {
-            "′"
-        }
-    }
-
-    /// Which character should be used as a fallback quote.
-    pub fn fallback(&self, double: bool) -> &'static str {
-        if double {
-            "\""
-        } else {
-            "’"
-        }
-    }
 }
 
 /// An opening and closing quote.

tests/suite/text/smartquote.typ

@@ -56,7 +56,28 @@ He said "I'm a big fella."
 
 --- smartquote-escape ---
 // Test escape sequences.
-The 5\'11\" 'quick\' brown fox jumps over the \"lazy" dog\'s ear.
+The 5\'11\" 'quick\' brown fox jumps over the \"lazy' dog\'s ear.
+
+--- smartquote-slash ---
+// Test that smartquotes can open before non-whitespace if not nested.
+"Hello"/"World" \
+'"Hello"/"World"' \
+""Hello"/"World""
+
+--- smartquote-close-before-letter ---
+// Test that smartquotes can close before alphabetic letters.
+Straight "A"s and "B"s
+
+--- smartquote-prime ---
+// Test that primes result after numbers when possible.
+A 2" nail. \
+'A 2" nail.' \
+"A 2" nail."
+
+--- smartquote-bracket ---
+// Test that brackets indicate an opening quote.
+"a ["b"] c" \
+"a b"c"d e"
 
 --- smartquote-disable ---
 // Test turning smart quotes off.