Add support for serializing math nodes as markdown

Closes GH-148.

Reviewed-by: Titus Wormer <[email protected]>

Author: bnchi

mdast_util_to_markdown/src/configure.rs

@@ -65,6 +65,11 @@ pub struct Options {
     /// Setext headings cannot be used for empty headings or headings with a
     /// rank of three or more.
     pub setext: bool,
+    /// Whether to support math (text) with a single dollar (`bool`, default: `true`).
+    /// Single dollars work in Pandoc and many other places, but often interfere with “normal”
+    /// dollars in text.
+    /// If you turn this off, you can still use two or more dollars for text math.
+    pub single_dollar_text_math: bool,
     /// Marker to use for strong (`'*'` or `'_'`, default: `'*'`).
     pub strong: char,
     /// Whether to join definitions without a blank line (`bool`, default:
@@ -90,6 +95,7 @@ impl Default for Options {
             rule_repetition: 3,
             rule_spaces: false,
             setext: false,
+            single_dollar_text_math: true,
             strong: '*',
             tight_definitions: false,
         }

mdast_util_to_markdown/src/construct_name.rs

@@ -20,13 +20,6 @@ pub enum ConstructName {
     ///     ^
     /// ```
     Blockquote,
-    /// Whole code (indented).
-    ///
-    /// ```markdown
-    /// ␠␠␠␠console.log(1)
-    /// ^^^^^^^^^^^^^^^^^^
-    /// ```
-    CodeIndented,
     /// Whole code (fenced).
     ///
     /// ````markdown
@@ -74,6 +67,13 @@ pub enum ConstructName {
     ///   | ~~~
     /// ````
     CodeFencedMetaTilde,
+    /// Whole code (indented).
+    ///
+    /// ```markdown
+    /// ␠␠␠␠console.log(1)
+    /// ^^^^^^^^^^^^^^^^^^
+    /// ```
+    CodeIndented,
     /// Whole definition.
     ///
     /// ```markdown
@@ -186,6 +186,26 @@ pub enum ConstructName {
     ///     ^^^^
     /// ```
     ListItem,
+    /// Math (flow).
+    ///
+    /// ```markdown
+    /// > | $$
+    ///     ^^
+    /// > | a
+    ///     ^
+    /// > | $$
+    ///     ^^
+    /// ```
+    MathFlow,
+    /// Math (flow) meta flag.
+    ///
+    /// ```markdown
+    /// > | $$a
+    ///       ^
+    ///   | b
+    ///   | $$
+    /// ```
+    MathFlowMeta,
     /// Paragraph.
     ///
     /// ```markdown

mdast_util_to_markdown/src/handle/inline_math.rs

@@ -0,0 +1,82 @@
+//! JS equivalent: https://github.com/syntax-tree/mdast-util-math/blob/main/lib/index.js#L241
+
+use super::Handle;
+use crate::state::{Info, State};
+use alloc::format;
+use markdown::{
+    mdast::{InlineMath, Node},
+    message::Message,
+};
+use regex::Regex;
+
+impl Handle for InlineMath {
+    fn handle(
+        &self,
+        state: &mut State,
+        _info: &Info,
+        _parent: Option<&Node>,
+        _node: &Node,
+    ) -> Result<alloc::string::String, Message> {
+        let mut size: usize = if !state.options.single_dollar_text_math {
+            2
+        } else {
+            1
+        };
+
+        let pattern = format!("(^|[^$]){}([^$]|$)", "\\$".repeat(size));
+        let mut dollar_sign_match = Regex::new(&pattern).unwrap();
+        while dollar_sign_match.is_match(&self.value) {
+            size += 1;
+            let pattern = format!("(^|[^$]){}([^$]|$)", "\\$".repeat(size));
+            dollar_sign_match = Regex::new(&pattern).unwrap();
+        }
+
+        let sequence = "$".repeat(size);
+
+        let no_whitespaces = !self.value.chars().all(char::is_whitespace);
+        let starts_with_whitespace = self.value.starts_with(char::is_whitespace);
+        let ends_with_whitespace = self.value.ends_with(char::is_whitespace);
+        let starts_with_dollar = self.value.starts_with('$');
+        let ends_with_dollar = self.value.ends_with('$');
+
+        let mut value = self.value.clone();
+        if no_whitespaces
+            && ((starts_with_whitespace && ends_with_whitespace)
+                || starts_with_dollar
+                || ends_with_dollar)
+        {
+            value = format!(" {} ", value);
+        }
+
+        for pattern in &mut state.r#unsafe {
+            if !pattern.at_break {
+                continue;
+            }
+
+            State::compile_pattern(pattern);
+
+            if let Some(regex) = &pattern.compiled {
+                while let Some(m) = regex.find(&value) {
+                    let position = m.start();
+
+                    let position = if position > 0
+                        && &value[position..m.len()] == "\n"
+                        && &value[position - 1..position] == "\r"
+                    {
+                        position - 1
+                    } else {
+                        position
+                    };
+
+                    value.replace_range(position..m.start() + 1, " ");
+                }
+            }
+        }
+
+        Ok(format!("{}{}{}", sequence, value, sequence))
+    }
+}
+
+pub fn peek_inline_math() -> char {
+    '$'
+}

mdast_util_to_markdown/src/handle/math.rs

@@ -0,0 +1,46 @@
+//! JS equivalent: https://github.com/syntax-tree/mdast-util-math/blob/main/lib/index.js#L204
+
+use super::Handle;
+use crate::{
+    construct_name::ConstructName,
+    state::{Info, State},
+    util::{longest_char_streak::longest_char_streak, safe::SafeConfig},
+};
+use alloc::string::String;
+use markdown::{
+    mdast::{Math, Node},
+    message::Message,
+};
+
+impl Handle for Math {
+    fn handle(
+        &self,
+        state: &mut State,
+        _info: &Info,
+        _parent: Option<&Node>,
+        _node: &Node,
+    ) -> Result<alloc::string::String, Message> {
+        let sequence = "$".repeat((longest_char_streak(&self.value, '$') + 1).max(2));
+        state.enter(ConstructName::MathFlow);
+
+        let mut value = String::new();
+        value.push_str(&sequence);
+
+        if let Some(meta) = &self.meta {
+            state.enter(ConstructName::MathFlowMeta);
+            value.push_str(&state.safe(meta, &SafeConfig::new(&value, "\n", Some('$'))));
+            state.exit();
+        }
+
+        value.push('\n');
+
+        if !self.value.is_empty() {
+            value.push_str(&self.value);
+            value.push('\n');
+        }
+
+        value.push_str(&sequence);
+        state.exit();
+        Ok(value)
+    }
+}

mdast_util_to_markdown/src/handle/mod.rs

@@ -12,10 +12,12 @@ pub mod html;
 pub mod image;
 pub mod image_reference;
 pub mod inline_code;
+pub mod inline_math;
 pub mod link;
 pub mod link_reference;
 mod list;
 mod list_item;
+mod math;
 mod paragraph;
 mod root;
 pub mod strong;

mdast_util_to_markdown/src/handle/root.rs

@@ -33,11 +33,12 @@ fn phrasing(child: &Node) -> bool {
         *child,
         Node::Break(_)
             | Node::Emphasis(_)
-            | Node::ImageReference(_)
             | Node::Image(_)
+            | Node::ImageReference(_)
             | Node::InlineCode(_)
-            | Node::LinkReference(_)
+            | Node::InlineMath(_)
             | Node::Link(_)
+            | Node::LinkReference(_)
             | Node::Strong(_)
             | Node::Text(_)
     )

mdast_util_to_markdown/src/state.rs

@@ -7,8 +7,9 @@ use crate::{
     construct_name::ConstructName,
     handle::{
         emphasis::peek_emphasis, html::peek_html, image::peek_image,
-        image_reference::peek_image_reference, inline_code::peek_inline_code, link::peek_link,
-        link_reference::peek_link_reference, strong::peek_strong, Handle,
+        image_reference::peek_image_reference, inline_code::peek_inline_code,
+        inline_math::peek_inline_math, link::peek_link, link_reference::peek_link_reference,
+        strong::peek_strong, Handle,
     },
     r#unsafe::Unsafe,
     util::{
@@ -322,6 +323,8 @@ impl<'a> State<'a> {
             Node::Strong(strong) => strong.handle(self, info, parent, node),
             Node::Text(text) => text.handle(self, info, parent, node),
             Node::ThematicBreak(thematic_break) => thematic_break.handle(self, info, parent, node),
+            Node::Math(math) => math.handle(self, info, parent, node),
+            Node::InlineMath(inline_math) => inline_math.handle(self, info, parent, node),
             _ => Err(Message {
                 place: None,
                 reason: format!("Unexpected node type `{:?}`", node),
@@ -409,7 +412,7 @@ impl<'a> State<'a> {
             index_stack: Vec::new(),
             options,
             stack: Vec::new(),
-            r#unsafe: Unsafe::get_default_unsafe(),
+            r#unsafe: Unsafe::get_default_unsafe(options),
         }
     }
 
@@ -424,6 +427,7 @@ impl<'a> State<'a> {
             Node::LinkReference(_) => Some(peek_link_reference()),
             Node::Link(link) => Some(peek_link(link, node, self)),
             Node::Strong(_) => Some(peek_strong(self)),
+            Node::InlineMath(_) => Some(peek_inline_math()),
             _ => None,
         }
     }

mdast_util_to_markdown/src/unsafe.rs

@@ -3,7 +3,7 @@
 //! JS equivalent: <https://github.com/syntax-tree/mdast-util-to-markdown/blob/main/lib/unsafe.js>.
 //! Also: <https://github.com/syntax-tree/mdast-util-to-markdown/blob/fd6a508/lib/types.js#L287-L305>.
 
-use crate::construct_name::ConstructName;
+use crate::{construct_name::ConstructName, Options};
 use alloc::{vec, vec::Vec};
 use regex::Regex;
 
@@ -38,7 +38,7 @@ impl<'a> Unsafe<'a> {
         }
     }
 
-    pub fn get_default_unsafe() -> Vec<Self> {
+    pub fn get_default_unsafe(options: &Options) -> Vec<Self> {
         let full_phrasing_spans = vec![
             ConstructName::Autolink,
             ConstructName::DestinationLiteral,
@@ -87,6 +87,7 @@ impl<'a> Unsafe<'a> {
                     ConstructName::CodeFencedMetaTilde,
                     ConstructName::DestinationLiteral,
                     ConstructName::HeadingAtx,
+                    ConstructName::MathFlowMeta,
                 ],
                 vec![],
                 false,
@@ -102,6 +103,7 @@ impl<'a> Unsafe<'a> {
                     ConstructName::CodeFencedMetaTilde,
                     ConstructName::DestinationLiteral,
                     ConstructName::HeadingAtx,
+                    ConstructName::MathFlowMeta,
                 ],
                 vec![],
                 false,
@@ -308,6 +310,27 @@ impl<'a> Unsafe<'a> {
                 false,
             ),
             Self::new('~', None, None, vec![], vec![], true),
+            Self::new(
+                '$',
+                None,
+                if options.single_dollar_text_math {
+                    None
+                } else {
+                    "\\$".into()
+                },
+                vec![ConstructName::Phrasing],
+                vec![],
+                false,
+            ),
+            Self::new(
+                '$',
+                None,
+                None,
+                vec![ConstructName::MathFlowMeta],
+                vec![],
+                false,
+            ),
+            Self::new('$', None, "\\$".into(), vec![], vec![], true),
         ]
     }