Improve language and add region controls

5 files changed, 68 insertions, 16 deletions

diff --git a/src/library/text/lang.rs b/src/library/text/lang.rs
index 343359d1..360827fa 100644
--- a/src/library/text/lang.rs
+++ b/src/library/text/lang.rs
@@ -1,24 +1,30 @@
 use crate::eval::Value;
 use crate::geom::Dir;
 
-/// A natural language.
+/// A code for a natural language.
 #[derive(Debug, Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
-pub struct Lang([u8; 2]);
+pub struct Lang([u8; 3], u8);
 
 impl Lang {
     /// The code for the english language.
-    pub const ENGLISH: Self = Self(*b"en");
+    pub const ENGLISH: Self = Self(*b"en ", 2);
 
-    /// Construct a language from a two-byte ISO 639-1 code.
+    /// Construct a language from a two- or three-byte ISO 639-1/2/3 code.
     pub fn from_str(iso: &str) -> Option<Self> {
-        let mut bytes: [u8; 2] = iso.as_bytes().try_into().ok()?;
-        bytes.make_ascii_lowercase();
-        Some(Self(bytes))
+        let len = iso.len();
+        if matches!(len, 2 ..= 3) && iso.is_ascii() {
+            let mut bytes = [b' '; 3];
+            bytes[.. len].copy_from_slice(iso.as_bytes());
+            bytes.make_ascii_lowercase();
+            Some(Self(bytes, len as u8))
+        } else {
+            None
+        }
     }
 
-    /// Return the language code as a string slice.
+    /// Return the language code as an all lowercase string slice.
     pub fn as_str(&self) -> &str {
-        std::str::from_utf8(&self.0).unwrap_or_default()
+        std::str::from_utf8(&self.0[.. usize::from(self.1)]).unwrap_or_default()
     }
 
     /// The default direction for the language.
@@ -35,5 +41,34 @@ castable! {
     Lang,
     Expected: "string",
     Value::Str(string) => Self::from_str(&string)
-        .ok_or("expected two letter language code")?,
+        .ok_or("expected two or three letter language code (ISO 639-1/2/3)")?,
+}
+
+/// A code for a region somewhere in the world.
+#[derive(Debug, Copy, Clone, Eq, PartialEq, Ord, PartialOrd, Hash)]
+pub struct Region([u8; 2]);
+
+impl Region {
+    /// Construct a region from its two-byte ISO 3166-1 alpha-2 code.
+    pub fn from_str(iso: &str) -> Option<Self> {
+        if iso.is_ascii() {
+            let mut bytes: [u8; 2] = iso.as_bytes().try_into().ok()?;
+            bytes.make_ascii_uppercase();
+            Some(Self(bytes))
+        } else {
+            None
+        }
+    }
+
+    /// Return the region code as an all uppercase string slice.
+    pub fn as_str(&self) -> &str {
+        std::str::from_utf8(&self.0).unwrap_or_default()
+    }
+}
+
+castable! {
+    Region,
+    Expected: "string",
+    Value::Str(string) => Self::from_str(&string)
+        .ok_or("expected two letter region code (ISO 3166-1 alpha-2)")?,
 }
diff --git a/src/library/text/mod.rs b/src/library/text/mod.rs
index 636b878c..0eb57339 100644
--- a/src/library/text/mod.rs
+++ b/src/library/text/mod.rs
@@ -65,8 +65,10 @@ impl TextNode {
     /// The bottom end of the text bounding box.
     pub const BOTTOM_EDGE: TextEdge = TextEdge::Metric(VerticalFontMetric::Baseline);
 
-    /// An ISO 639-1 language code.
+    /// An ISO 639-1/2/3 language code.
     pub const LANG: Lang = Lang::ENGLISH;
+    /// An ISO 3166-1 alpha-2 region code.
+    pub const REGION: Option<Region> = None;
     /// The direction for text and inline objects. When `auto`, the direction is
     /// automatically inferred from the language.
     #[property(resolve)]
diff --git a/src/library/text/par.rs b/src/library/text/par.rs
index 232a5d0f..fc978357 100644
--- a/src/library/text/par.rs
+++ b/src/library/text/par.rs
@@ -406,9 +406,9 @@ fn collect<'a>(
             ParChild::Quote(double) => {
                 let prev = full.len();
                 if styles.get(TextNode::SMART_QUOTES) {
-                    // TODO: Also get region.
                     let lang = styles.get(TextNode::LANG);
-                    let quotes = Quotes::from_lang(lang.as_str(), "");
+                    let region = styles.get(TextNode::REGION);
+                    let quotes = Quotes::from_lang(lang, region);
                     let peeked = iter.peek().and_then(|(child, _)| match child {
                         ParChild::Text(text) => text.chars().next(),
                         ParChild::Quote(_) => Some('"'),
diff --git a/src/library/text/quotes.rs b/src/library/text/quotes.rs
index 5f67bdb5..98402ca4 100644
--- a/src/library/text/quotes.rs
+++ b/src/library/text/quotes.rs
@@ -1,3 +1,4 @@
+use super::{Lang, Region};
 use crate::parse::is_newline;
 
 /// State machine for smart quote subtitution.
@@ -91,9 +92,10 @@ impl<'s> Quotes<'s> {
     /// 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") => ("‹", "›", "«", "»"),
+    pub fn from_lang(lang: Lang, region: Option<Region>) -> Self {
+        let region = region.as_ref().map(Region::as_str);
+        let (single_open, single_close, double_open, double_close) = match lang.as_str() {
+            "de" if matches!(region, Some("CH" | "LI")) => ("‹", "›", "«", "»"),
             "cs" | "da" | "de" | "et" | "is" | "lt" | "lv" | "sk" | "sl" => {
                 ("‚", "‘", "„", "“")
             }
diff --git a/src/library/text/shaping.rs b/src/library/text/shaping.rs
index 72f86a38..055761df 100644
--- a/src/library/text/shaping.rs
+++ b/src/library/text/shaping.rs
@@ -1,4 +1,5 @@
 use std::ops::Range;
+use std::str::FromStr;
 
 use rustybuzz::{Feature, UnicodeBuffer};
 
@@ -372,6 +373,7 @@ fn shape_segment<'a>(
     // Fill the buffer with our text.
     let mut buffer = UnicodeBuffer::new();
     buffer.push_str(text);
+    buffer.set_language(language(ctx.styles));
     buffer.set_direction(match ctx.dir {
         Dir::LTR => rustybuzz::Direction::LeftToRight,
         Dir::RTL => rustybuzz::Direction::RightToLeft,
@@ -613,3 +615,14 @@ fn tags(styles: StyleChain) -> Vec<Feature> {
 
     tags
 }
+
+/// Process the language and and region of a style chain into a
+/// rustybuzz-compatible BCP 47 language.
+fn language(styles: StyleChain) -> rustybuzz::Language {
+    let mut bcp: EcoString = styles.get(TextNode::LANG).as_str().into();
+    if let Some(region) = styles.get(TextNode::REGION) {
+        bcp.push('-');
+        bcp.push_str(region.as_str());
+    }
+    rustybuzz::Language::from_str(&bcp).unwrap()
+}