Smart quotes

Co-Authored-By: Martin Haug <mhaug@live.de>

src/library/text/mod.rs

@@ -3,12 +3,14 @@
 mod deco;
 mod link;
 mod par;
+mod quotes;
 mod raw;
 mod shaping;
 
 pub use deco::*;
 pub use link::*;
 pub use par::*;
+pub use quotes::*;
 pub use raw::*;
 pub use shaping::*;
 
@@ -72,6 +74,8 @@ impl TextNode {
     /// will will be hyphenated if and only if justification is enabled.
     #[property(resolve)]
     pub const HYPHENATE: Smart<Hyphenate> = Smart::Auto;
+    /// Whether to apply smart quotes.
+    pub const SMART_QUOTES: bool = true;
 
     /// Whether to apply kerning ("kern").
     pub const KERNING: bool = true;

src/library/text/par.rs

@@ -4,7 +4,7 @@ use unicode_bidi::{BidiInfo, Level};
 use unicode_script::{Script, UnicodeScript};
 use xi_unicode::LineBreakIterator;
 
-use super::{shape, Lang, ShapedText, TextNode};
+use super::{shape, Lang, Quoter, Quotes, ShapedText, TextNode};
 use crate::font::FontStore;
 use crate::library::layout::Spacing;
 use crate::library::prelude::*;
@@ -386,9 +386,11 @@ fn collect<'a>(
     styles: &'a StyleChain<'a>,
 ) -> (String, Vec<(Segment<'a>, StyleChain<'a>)>) {
     let mut full = String::new();
+    let mut quoter = Quoter::new();
     let mut segments = vec![];
+    let mut iter = par.0.iter().peekable();
 
-    for (child, map) in par.0.iter() {
+    while let Some((child, map)) = iter.next() {
         let styles = map.chain(&styles);
         let segment = match child {
             ParChild::Text(text) => {
@@ -402,7 +404,25 @@ fn collect<'a>(
             }
             ParChild::Quote(double) => {
                 let prev = full.len();
-                full.push(if *double { '"' } else { '\'' });
+                if styles.get(TextNode::SMART_QUOTES) {
+                    // TODO: Also get region.
+                    let lang = styles.get(TextNode::LANG);
+                    let quotes = lang
+                        .as_ref()
+                        .map(|lang| Quotes::from_lang(lang.as_str(), ""))
+                        .unwrap_or_default();
+
+                    let peeked = iter.peek().and_then(|(child, _)| match child {
+                        ParChild::Text(text) => text.chars().next(),
+                        ParChild::Quote(_) => Some('"'),
+                        ParChild::Spacing(_) => Some(SPACING_REPLACE),
+                        ParChild::Node(_) => Some(NODE_REPLACE),
+                    });
+
+                    full.push_str(quoter.quote(&quotes, *double, peeked));
+                } else {
+                    full.push(if *double { '"' } else { '\'' });
+                }
                 Segment::Text(full.len() - prev)
             }
             ParChild::Spacing(spacing) => {
@@ -415,6 +435,10 @@ fn collect<'a>(
             }
         };
 
+        if let Some(last) = full.chars().last() {
+            quoter.last(last);
+        }
+
         if let (Some((Segment::Text(last_len), last_styles)), Segment::Text(len)) =
             (segments.last_mut(), segment)
         {

src/library/text/quotes.rs

@@ -0,0 +1,146 @@
+use crate::parse::is_newline;
+
+/// State machine for smart quote subtitution.
+#[derive(Debug, Clone)]
+pub struct Quoter {
+    /// 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,
+}
+
+impl Quoter {
+    /// Start quoting.
+    pub fn new() -> Self {
+        Self {
+            quote_depth: 0,
+            expect_opening: true,
+            last_num: false,
+        }
+    }
+
+    /// Process the last seen character.
+    pub fn last(&mut self, c: char) {
+        self.expect_opening = is_ignorable(c) || is_opening_bracket(c);
+        self.last_num = c.is_numeric();
+    }
+
+    /// Process and substitute a quote.
+    pub fn quote<'a>(
+        &mut self,
+        quotes: &Quotes<'a>,
+        double: bool,
+        peeked: Option<char>,
+    ) -> &'a str {
+        let peeked = peeked.unwrap_or(' ');
+        if self.expect_opening {
+            self.quote_depth += 1;
+            quotes.open(double)
+        } else if self.quote_depth > 0
+            && (peeked.is_ascii_punctuation() || is_ignorable(peeked))
+        {
+            self.quote_depth -= 1;
+            quotes.close(double)
+        } else if self.last_num {
+            quotes.prime(double)
+        } else {
+            quotes.fallback(double)
+        }
+    }
+}
+
+impl Default for Quoter {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+fn is_ignorable(c: char) -> bool {
+    c.is_whitespace() || is_newline(c)
+}
+
+fn is_opening_bracket(c: char) -> bool {
+    matches!(c, '(' | '{' | '[')
+}
+
+/// Decides which quotes to subtitute smart quotes with.
+pub struct Quotes<'s> {
+    /// The opening single quote.
+    pub single_open: &'s str,
+    /// The closing single quote.
+    pub single_close: &'s str,
+    /// The opening double quote.
+    pub double_open: &'s str,
+    /// The closing double quote.
+    pub double_close: &'s str,
+}
+
+impl<'s> Quotes<'s> {
+    /// Create a new `Quotes` struct with the defaults for a language and
+    /// region.
+    ///
+    /// The language should be specified as an all-lowercase ISO 639-1 code, the
+    /// region as an all-uppercase ISO 3166-alpha2 code.
+    ///
+    /// Currently, the supported languages are: English, Czech, Danish, German,
+    /// Swiss / Liechtensteinian German, Estonian, Icelandic, Lithuanian,
+    /// Latvian, Slovak, Slovenian, Bosnian, Finnish, Swedish, French,
+    /// Hungarian, Polish, Romanian, Japanese, Traditional Chinese, Russian, and
+    /// Norwegian.
+    ///
+    /// For unknown languages, the English quotes are used.
+    pub fn from_lang(language: &str, region: &str) -> Self {
+        let (single_open, single_close, double_open, double_close) = match language {
+            "de" if matches!(region, "CH" | "LI") => ("‹", "›", "«", "»"),
+            "cs" | "da" | "de" | "et" | "is" | "lt" | "lv" | "sk" | "sl" => {
+                ("‚", "‘", "„", "“")
+            }
+            "fr" => ("‹\u{00A0}", "\u{00A0}›", "«\u{00A0}", "\u{00A0}»"),
+            "bs" | "fi" | "sv" => ("’", "’", "”", "”"),
+            "hu" | "pl" | "ro" => ("’", "’", "„", "”"),
+            "ru" | "no" | "nn" => ("’", "’", "«", "»"),
+            _ => return Self::default(),
+        };
+
+        Self {
+            single_open,
+            single_close,
+            double_open,
+            double_close,
+        }
+    }
+
+    /// The opening quote.
+    fn open(&self, double: bool) -> &'s str {
+        if double { self.double_open } else { self.single_open }
+    }
+
+    /// The closing quote.
+    fn close(&self, double: bool) -> &'s str {
+        if double { self.double_close } else { self.single_close }
+    }
+
+    /// Which character should be used as a prime.
+    fn prime(&self, double: bool) -> &'static str {
+        if double { "″" } else { "′" }
+    }
+
+    /// Which character should be used as a fallback quote.
+    fn fallback(&self, double: bool) -> &'static str {
+        if double { "\"" } else { "’" }
+    }
+}
+
+impl Default for Quotes<'_> {
+    /// Returns the english quotes as default.
+    fn default() -> Self {
+        Self {
+            single_open: "‘",
+            single_close: "’",
+            double_open: "“",
+            double_close: "”",
+        }
+    }
+}

src/library/text/raw.rs

@@ -100,6 +100,7 @@ impl Show for RawNode {
         let mut map = StyleMap::new();
         map.set(TextNode::OVERHANG, false);
         map.set(TextNode::HYPHENATE, Smart::Custom(Hyphenate(false)));
+        map.set(TextNode::SMART_QUOTES, false);
 
         if let Smart::Custom(family) = styles.get(Self::FAMILY) {
             map.set_family(family.clone(), styles);

tests/typ/code/closure.typ

@@ -5,11 +5,10 @@
 // Don't parse closure directly in content.
 // Ref: true
 
-#let x = "\"hi\""
+#let x = "x"
 
-// Should output `"hi" => "bye"`.
-#set text(overhang: false)
-#x => "bye"
+// Should output `x => y`.
+#x => y
 
 ---
 // Basic closure without captures.

tests/typ/text/escape.typ

@@ -2,7 +2,8 @@
 
 ---
 // Escapable symbols.
-\\ \/ \[ \] \{ \} \# \* \_ \= \~ \` \$
+\\ \/ \[ \] \{ \} \# \* \_ \
+\= \~ \` \$ \" \'
 
 // No need to escape.
 ( ) ; < >

tests/typ/text/quotes.typ

@@ -0,0 +1,54 @@
+// Test smart quotes.
+
+---
+#set page(width: 200pt)
+
+// Test simple quotations in various languages.
+#set text(lang: "en")
+"The horse eats no cucumber salad" was the first sentence ever uttered on the 'telephone.'
+
+#set text(lang: "de")
+"Das Pferd frisst keinen Gurkensalat" war der erste jemals am 'Fernsprecher' gesagte Satz.
+
+#set text(lang: "fr")
+"Le cheval ne mange pas de salade de concombres" est la première phrase jamais prononcée au 'téléphone'.
+
+#set text(lang: "fi")
+"Hevonen ei syö kurkkusalaattia" oli ensimmäinen koskaan 'puhelimessa' lausuttu lause.
+
+#set text(lang: "ro")
+"Calul nu mănâncă salată de castraveți" a fost prima propoziție rostită vreodată la 'telefon'.
+
+#set text(lang: "ru")
+"Лошадь не ест салат из огурцов" - это была первая фраза, сказанная по 'телефону'.
+
+---
+// Test single pair of quotes.
+#set text(lang: "en")
+""
+
+---
+// Test sentences with numbers and apostrophes.
+#set text(lang: "en")
+The 5'11" 'quick' brown fox jumps over the "lazy" dog's ear.
+
+He said "I'm a big fella."
+
+---
+// Test escape sequences.
+The 5\'11\" 'quick\' brown fox jumps over the \"lazy" dog\'s ear.
+
+---
+// Test turning smart quotes off.
+#set text(lang: "en")
+He's told some books contain questionable "example text".
+
+#set text(smart-quotes: false)
+He's told some books contain questionable "example text".
+
+---
+// Test changing properties within text.
+#set text(lang: "en")
+"She suddenly started speaking french: #text(lang: "fr")['Je suis une banane.']" Roman told me.
+
+Some people's thought on this would be #text(smart-quotes: false)["strange."]