Refactor sub- and superscript substitution (#5120)

crates/typst/src/text/shift.rs

@@ -51,19 +51,18 @@ impl Show for Packed<SubElem> {
     #[typst_macros::time(name = "sub", span = self.span())]
     fn show(&self, engine: &mut Engine, styles: StyleChain) -> SourceResult<Content> {
         let body = self.body().clone();
-        let mut transformed = None;
+
         if self.typographic(styles) {
-            if let Some(text) = search_text(&body, true) {
+            if let Some(text) = convert_script(&body, true) {
                 if is_shapable(engine, &text, styles) {
-                    transformed = Some(TextElem::packed(text));
+                    return Ok(TextElem::packed(text));
                 }
             }
         };
 
-        Ok(transformed.unwrap_or_else(|| {
+        Ok(body
-            body.styled(TextElem::set_baseline(self.baseline(styles)))
+            .styled(TextElem::set_baseline(self.baseline(styles)))
-                .styled(TextElem::set_size(self.size(styles)))
+            .styled(TextElem::set_size(self.size(styles))))
-        }))
     }
 }
 
@@ -111,38 +110,38 @@ impl Show for Packed<SuperElem> {
     #[typst_macros::time(name = "super", span = self.span())]
     fn show(&self, engine: &mut Engine, styles: StyleChain) -> SourceResult<Content> {
         let body = self.body().clone();
-        let mut transformed = None;
+
         if self.typographic(styles) {
-            if let Some(text) = search_text(&body, false) {
+            if let Some(text) = convert_script(&body, false) {
                 if is_shapable(engine, &text, styles) {
-                    transformed = Some(TextElem::packed(text));
+                    return Ok(TextElem::packed(text));
                 }
             }
         };
 
-        Ok(transformed.unwrap_or_else(|| {
+        Ok(body
-            body.styled(TextElem::set_baseline(self.baseline(styles)))
+            .styled(TextElem::set_baseline(self.baseline(styles)))
-                .styled(TextElem::set_size(self.size(styles)))
+            .styled(TextElem::set_size(self.size(styles))))
-        }))
     }
 }
 
 /// Find and transform the text contained in `content` to the given script kind
 /// if and only if it only consists of `Text`, `Space`, and `Empty` leaves.
-fn search_text(content: &Content, sub: bool) -> Option<EcoString> {
+fn convert_script(content: &Content, sub: bool) -> Option<EcoString> {
     if content.is::<SpaceElem>() {
         Some(' '.into())
     } else if let Some(elem) = content.to_packed::<TextElem>() {
-        convert_script(elem.text(), sub)
+        if sub {
-    } else if let Some(sequence) = content.to_packed::<SequenceElem>() {
+            elem.text().chars().map(to_subscript_codepoint).collect()
-        let mut full = EcoString::new();
+        } else {
-        for item in &sequence.children {
+            elem.text().chars().map(to_superscript_codepoint).collect()
-            match search_text(item, sub) {
-                Some(text) => full.push_str(&text),
-                None => return None,
-            }
         }
-        Some(full)
+    } else if let Some(sequence) = content.to_packed::<SequenceElem>() {
+        sequence
+            .children
+            .iter()
+            .map(|item| convert_script(item, sub))
+            .collect()
     } else {
         None
     }
@@ -165,65 +164,47 @@ fn is_shapable(engine: &Engine, text: &str, styles: StyleChain) -> bool {
     false
 }
 
-/// Convert a string to sub- or superscript codepoints if all characters
-/// can be mapped to such a codepoint.
-fn convert_script(text: &str, sub: bool) -> Option<EcoString> {
-    let mut result = EcoString::with_capacity(text.len());
-    let converter = if sub { to_subscript_codepoint } else { to_superscript_codepoint };
-
-    for c in text.chars() {
-        match converter(c) {
-            Some(c) => result.push(c),
-            None => return None,
-        }
-    }
-
-    Some(result)
-}
-
 /// Convert a character to its corresponding Unicode superscript.
 fn to_superscript_codepoint(c: char) -> Option<char> {
-    char::from_u32(match c {
+    match c {
-        '0' => 0x2070,
+        '1' => Some('¹'),
-        '1' => 0x00B9,
+        '2' => Some('²'),
-        '2' => 0x00B2,
+        '3' => Some('³'),
-        '3' => 0x00B3,
+        '0' | '4'..='9' => char::from_u32(c as u32 - '0' as u32 + '⁰' as u32),
-        '4'..='9' => 0x2070 + (c as u32 + 4 - '4' as u32),
+        '+' => Some('⁺'),
-        '+' => 0x207A,
+        '−' => Some('⁻'),
-        '-' => 0x207B,
+        '=' => Some('⁼'),
-        '=' => 0x207C,
+        '(' => Some('⁽'),
-        '(' => 0x207D,
+        ')' => Some('⁾'),
-        ')' => 0x207E,
+        'n' => Some('ⁿ'),
-        'n' => 0x207F,
+        'i' => Some('ⁱ'),
-        'i' => 0x2071,
+        ' ' => Some(' '),
-        ' ' => 0x0020,
+        _ => None,
-        _ => return None,
+    }
-    })
 }
 
 /// Convert a character to its corresponding Unicode subscript.
 fn to_subscript_codepoint(c: char) -> Option<char> {
-    char::from_u32(match c {
+    match c {
-        '0' => 0x2080,
+        '0'..='9' => char::from_u32(c as u32 - '0' as u32 + '₀' as u32),
-        '1'..='9' => 0x2080 + (c as u32 - '0' as u32),
+        '+' => Some('₊'),
-        '+' => 0x208A,
+        '−' => Some('₋'),
-        '-' => 0x208B,
+        '=' => Some('₌'),
-        '=' => 0x208C,
+        '(' => Some('₍'),
-        '(' => 0x208D,
+        ')' => Some('₎'),
-        ')' => 0x208E,
+        'a' => Some('ₐ'),
-        'a' => 0x2090,
+        'e' => Some('ₑ'),
-        'e' => 0x2091,
+        'o' => Some('ₒ'),
-        'o' => 0x2092,
+        'x' => Some('ₓ'),
-        'x' => 0x2093,
+        'h' => Some('ₕ'),
-        'h' => 0x2095,
+        'k' => Some('ₖ'),
-        'k' => 0x2096,
+        'l' => Some('ₗ'),
-        'l' => 0x2097,
+        'm' => Some('ₘ'),
-        'm' => 0x2098,
+        'n' => Some('ₙ'),
-        'n' => 0x2099,
+        'p' => Some('ₚ'),
-        'p' => 0x209A,
+        's' => Some('ₛ'),
-        's' => 0x209B,
+        't' => Some('ₜ'),
-        't' => 0x209C,
+        ' ' => Some(' '),
-        ' ' => 0x0020,
+        _ => None,
-        _ => return None,
+    }
-    })
 }