Merge pull request #488 from liamwhite/strikethrough-subscript-unify

Add support for subscript extension
kivikakk · Nov 20, 2024 · fc4292c · fc4292c
2 parents 0f0f659 + 3d772f9
commit fc4292c
Show file tree

Hide file tree

Showing 10 changed files with 100 additions and 8 deletions.
diff --git a/src/cm.rs b/src/cm.rs
@@ -397,6 +397,7 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
             NodeValue::Math(ref math) => self.format_math(math, allow_wrap, entering),
             NodeValue::WikiLink(ref nl) => return self.format_wikilink(nl, entering),
             NodeValue::Underline => self.format_underline(),
+            NodeValue::Subscript => self.format_subscript(),
             NodeValue::SpoileredText => self.format_spoiler(),
             NodeValue::EscapedTag(ref net) => self.format_escaped_tag(net),
         };
@@ -712,7 +713,7 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
     }
 
     fn format_strikethrough(&mut self) {
-        write!(self, "~").unwrap();
+        write!(self, "~~").unwrap();
     }
 
     fn format_superscript(&mut self) {
@@ -723,6 +724,10 @@ impl<'a, 'o> CommonMarkFormatter<'a, 'o> {
         write!(self, "__").unwrap();
     }
 
+    fn format_subscript(&mut self) {
+        write!(self, "~").unwrap();
+    }
+
     fn format_spoiler(&mut self) {
         write!(self, "||").unwrap();
     }

diff --git a/src/html.rs b/src/html.rs
@@ -1139,6 +1139,18 @@ impl<'o> HtmlFormatter<'o> {
                     self.output.write_all(b"</u>")?;
                 }
             }
+            NodeValue::Subscript => {
+                // Unreliable sourcepos.
+                if entering {
+                    self.output.write_all(b"<sub")?;
+                    if self.options.render.experimental_inline_sourcepos {
+                        self.render_sourcepos(node)?;
+                    }
+                    self.output.write_all(b">")?;
+                } else {
+                    self.output.write_all(b"</sub>")?;
+                }
+            }
             NodeValue::SpoileredText => {
                 // Unreliable sourcepos.
                 if entering {

diff --git a/src/main.rs b/src/main.rs
@@ -184,6 +184,7 @@ enum Extension {
     WikilinksTitleAfterPipe,
     WikilinksTitleBeforePipe,
     Underline,
+    Subscript,
     Spoiler,
     Greentext,
 }
@@ -267,6 +268,7 @@ fn main() -> Result<(), Box<dyn Error>> {
         .wikilinks_title_after_pipe(exts.contains(&Extension::WikilinksTitleAfterPipe))
         .wikilinks_title_before_pipe(exts.contains(&Extension::WikilinksTitleBeforePipe))
         .underline(exts.contains(&Extension::Underline))
+        .subscript(exts.contains(&Extension::Subscript))
         .spoiler(exts.contains(&Extension::Spoiler))
         .greentext(exts.contains(&Extension::Greentext))
         .maybe_front_matter_delimiter(cli.front_matter_delimiter);

diff --git a/src/nodes.rs b/src/nodes.rs
@@ -189,6 +189,9 @@ pub enum NodeValue {
     /// **Inline**.  Underline. Enabled with `underline` option.
     Underline,
 
+    /// **Inline**.  Subscript. Enabled with `subscript` options.
+    Subscript,
+
     /// **Inline**.  Spoilered text.  Enabled with `spoiler` option.
     SpoileredText,
 
@@ -514,6 +517,7 @@ impl NodeValue {
             NodeValue::Math(..) => "math",
             NodeValue::WikiLink(..) => "wikilink",
             NodeValue::Underline => "underline",
+            NodeValue::Subscript => "subscript",
             NodeValue::SpoileredText => "spoiler",
             NodeValue::EscapedTag(_) => "escaped_tag",
         }
@@ -764,6 +768,7 @@ pub fn can_contain_type<'a>(node: &'a AstNode<'a>, child: &NodeValue) -> bool {
         | NodeValue::Superscript
         | NodeValue::SpoileredText
         | NodeValue::Underline
+        | NodeValue::Subscript
         // XXX: this is quite a hack: the EscapedTag _contains_ whatever was
         // possibly going to fall into the spoiler. This should be fixed in
         // inlines.
@@ -791,6 +796,7 @@ pub fn can_contain_type<'a>(node: &'a AstNode<'a>, child: &NodeValue) -> bool {
                 | NodeValue::Superscript
                 | NodeValue::SpoileredText
                 | NodeValue::Underline
+                | NodeValue::Subscript
         ),
 
         #[cfg(feature = "shortcodes")]
@@ -810,6 +816,7 @@ pub fn can_contain_type<'a>(node: &'a AstNode<'a>, child: &NodeValue) -> bool {
             | NodeValue::Superscript
             | NodeValue::SpoileredText
             | NodeValue::Underline
+            | NodeValue::Subscript
             | NodeValue::ShortCode(..)
         ),
 

diff --git a/src/parser/inlines.rs b/src/parser/inlines.rs
@@ -149,7 +149,7 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
             s.special_chars[b':' as usize] = true;
             s.special_chars[b'w' as usize] = true;
         }
-        if options.extension.strikethrough {
+        if options.extension.strikethrough || options.extension.subscript {
             s.special_chars[b'~' as usize] = true;
             s.skip_chars[b'~' as usize] = true;
         }
@@ -281,7 +281,9 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
                     ))
                 }
             }
-            '~' if self.options.extension.strikethrough => Some(self.handle_delim(b'~')),
+            '~' if self.options.extension.strikethrough || self.options.extension.subscript => {
+                Some(self.handle_delim(b'~'))
+            }
             '^' if self.options.extension.superscript && !self.within_brackets => {
                 Some(self.handle_delim(b'^'))
             }
@@ -333,7 +335,7 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
 
     // After parsing a block (and sometimes during), this function traverses the
     // stack of `Delimiters`, tokens ("*", "_", etc.) that may delimit regions
-    // of text for special rendering: emphasis, strong, superscript,
+    // of text for special rendering: emphasis, strong, superscript, subscript,
     // spoilertext; looking for pairs of opening and closing delimiters,
     // with the goal of placing the intervening nodes into new emphasis,
     // etc AST nodes.
@@ -461,7 +463,8 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
                 // both get passed.
                 if c.delim_char == b'*'
                     || c.delim_char == b'_'
-                    || (self.options.extension.strikethrough && c.delim_char == b'~')
+                    || ((self.options.extension.strikethrough || self.options.extension.subscript)
+                        && c.delim_char == b'~')
                     || (self.options.extension.superscript && c.delim_char == b'^')
                     || (self.options.extension.spoiler && c.delim_char == b'|')
                 {
@@ -1068,7 +1071,7 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
         opener_num_chars -= use_delims;
         closer_num_chars -= use_delims;
 
-        if self.options.extension.strikethrough
+        if (self.options.extension.strikethrough || self.options.extension.subscript)
             && opener_char == b'~'
             && (opener_num_chars != closer_num_chars || opener_num_chars > 0)
         {
@@ -1101,8 +1104,19 @@ impl<'a, 'r, 'o, 'd, 'i> Subject<'a, 'r, 'o, 'd, 'i> {
         }
 
         let emph = self.make_inline(
-            if self.options.extension.strikethrough && opener_char == b'~' {
-                NodeValue::Strikethrough
+            if self.options.extension.subscript && opener_char == b'~' && use_delims == 1 {
+                NodeValue::Subscript
+            } else if opener_char == b'~' {
+                // Not emphasis
+                // Unlike for |, these cases have to be handled because they will match
+                // in the event subscript but not strikethrough is enabled
+                if self.options.extension.strikethrough {
+                    NodeValue::Strikethrough
+                } else if use_delims == 1 {
+                    NodeValue::EscapedTag("~".to_owned())
+                } else {
+                    NodeValue::EscapedTag("~~".to_owned())
+                }
             } else if self.options.extension.superscript && opener_char == b'^' {
                 NodeValue::Superscript
             } else if self.options.extension.spoiler && opener_char == b'|' {

diff --git a/src/parser/mod.rs b/src/parser/mod.rs
@@ -515,6 +515,26 @@ pub struct ExtensionOptions {
     #[builder(default)]
     pub underline: bool,
 
+    /// Enables subscript text using single tildes.
+    ///
+    /// If the strikethrough option is also enabled, this overrides the single
+    /// tilde case to output subscript text.
+    ///
+    /// ```md
+    /// H~2~O
+    /// ```
+    ///
+    /// ```
+    /// # use comrak::{markdown_to_html, Options};
+    /// let mut options = Options::default();
+    /// options.extension.subscript = true;
+    ///
+    /// assert_eq!(markdown_to_html("H~2~O", &options),
+    ///            "<p>H<sub>2</sub>O</p>\n");
+    /// ```
+    #[builder(default)]
+    pub subscript: bool,
+
     /// Enables spoilers using double vertical bars
     ///
     /// ```md

diff --git a/src/tests.rs b/src/tests.rs
@@ -25,6 +25,7 @@ mod rewriter;
 mod shortcodes;
 mod spoiler;
 mod strikethrough;
+mod subscript;
 mod superscript;
 mod table;
 mod tagfilter;

diff --git a/src/tests/api.rs b/src/tests/api.rs
@@ -72,6 +72,7 @@ fn exercise_full_api() {
         .wikilinks_title_after_pipe(true)
         .wikilinks_title_before_pipe(true)
         .underline(true)
+        .subscript(true)
         .spoiler(true)
         .greentext(true);
 
@@ -271,6 +272,7 @@ fn exercise_full_api() {
             let _: String = nl.url;
         }
         nodes::NodeValue::Underline => {}
+        nodes::NodeValue::Subscript => {}
         nodes::NodeValue::SpoileredText => {}
         nodes::NodeValue::EscapedTag(data) => {
             let _: &String = data;

diff --git a/src/tests/subscript.rs b/src/tests/subscript.rs
@@ -0,0 +1,28 @@
+use super::*;
+
+#[test]
+fn subscript() {
+    html_opts!(
+        [extension.subscript],
+        concat!("H~2~O\n"),
+        concat!("<p>H<sub>2</sub>O</p>\n"),
+    );
+}
+
+#[test]
+fn strikethrough_and_subscript() {
+    html_opts!(
+        [extension.subscript, extension.strikethrough],
+        concat!("~~H~2~O~~\n"),
+        concat!("<p><del>H<sub>2</sub>O</del></p>\n"),
+    );
+}
+
+#[test]
+fn no_strikethrough_when_only_subscript() {
+    html_opts!(
+        [extension.subscript],
+        concat!("~~H~2~O~~\n"),
+        concat!("<p>~~H<sub>2</sub>O~~</p>\n"),
+    );
+}
diff --git a/src/xml.rs b/src/xml.rs
@@ -285,6 +285,7 @@ impl<'o> XmlFormatter<'o> {
                     self.output.write_all(b"\"")?;
                 }
                 NodeValue::Underline => {}
+                NodeValue::Subscript => {}
                 NodeValue::SpoileredText => {}
                 NodeValue::EscapedTag(ref data) => {
                     self.output.write_all(data.as_bytes())?;