Add sourcepos for the task item symbol

Author: kivikakk

src/cm.rs

@@ -6,7 +6,7 @@ use std::str;
 use crate::ctype::{isalpha, isdigit, ispunct, ispunct_char, isspace, isspace_char};
 use crate::nodes::{
     ListDelimType, ListType, Node, NodeAlert, NodeCodeBlock, NodeHeading, NodeHtmlBlock, NodeLink,
-    NodeList, NodeMath, NodeValue, NodeWikiLink, TableAlignment,
+    NodeList, NodeMath, NodeTaskItem, NodeValue, NodeWikiLink, TableAlignment,
 };
 use crate::parser::options::{Options, Plugins, WikiLinksMode};
 #[cfg(feature = "phoenix_heex")]
@@ -471,7 +471,7 @@ impl<'a, 'o, 'c, 'w> CommonMarkFormatter<'a, 'o, 'c, 'w> {
                 }
             }
             NodeValue::Emph => self.format_emph(node)?,
-            NodeValue::TaskItem(symbol) => self.format_task_item(symbol, node, entering)?,
+            NodeValue::TaskItem(ref nti) => self.format_task_item(nti, node, entering)?,
             NodeValue::Strikethrough => self.format_strikethrough()?,
             NodeValue::Highlight => self.format_highlight()?,
             NodeValue::Superscript => self.format_superscript()?,
@@ -855,7 +855,7 @@ impl<'a, 'o, 'c, 'w> CommonMarkFormatter<'a, 'o, 'c, 'w> {
 
     fn format_task_item(
         &mut self,
-        symbol: Option<char>,
+        nti: &NodeTaskItem,
         node: Node<'a>,
         entering: bool,
     ) -> fmt::Result {
@@ -867,7 +867,7 @@ impl<'a, 'o, 'c, 'w> CommonMarkFormatter<'a, 'o, 'c, 'w> {
             self.format_item(node, entering)?;
         }
         if entering {
-            write!(self, "[{}] ", symbol.unwrap_or(' '))?;
+            write!(self, "[{}] ", nti.symbol.unwrap_or(' '))?;
         }
         Ok(())
     }

src/html.rs

@@ -20,8 +20,8 @@ use crate::ctype::isspace;
 use crate::nodes::NodeShortCode;
 use crate::nodes::{
     ListType, Node, NodeAlert, NodeCode, NodeCodeBlock, NodeFootnoteDefinition,
-    NodeFootnoteReference, NodeHeading, NodeHtmlBlock, NodeLink, NodeList, NodeMath, NodeValue,
-    NodeWikiLink, TableAlignment,
+    NodeFootnoteReference, NodeHeading, NodeHtmlBlock, NodeLink, NodeList, NodeMath, NodeTaskItem,
+    NodeValue, NodeWikiLink, TableAlignment,
 };
 use crate::parser::options::{Options, Plugins};
 use crate::{node_matches, scanners};
@@ -444,7 +444,7 @@ pub fn format_node_default<T>(
         NodeValue::Table(_) => render_table(context, node, entering),
         NodeValue::TableCell => render_table_cell(context, node, entering),
         NodeValue::TableRow(thead) => render_table_row(context, node, entering, thead),
-        NodeValue::TaskItem(symbol) => render_task_item(context, node, entering, symbol),
+        NodeValue::TaskItem(ref nti) => render_task_item(context, node, entering, nti),
 
         // Extensions
         NodeValue::Alert(ref alert) => render_alert(context, node, entering, alert),
@@ -1204,7 +1204,7 @@ fn render_task_item<T>(
     context: &mut Context<T>,
     node: Node<'_>,
     entering: bool,
-    symbol: Option<char>,
+    nti: &NodeTaskItem,
 ) -> Result<ChildRendering, fmt::Error> {
     let write_li = node
         .parent()
@@ -1228,7 +1228,7 @@ fn render_task_item<T>(
         if context.options.render.tasklist_classes {
             context.write_str(" class=\"task-list-item-checkbox\"")?;
         }
-        if symbol.is_some() {
+        if nti.symbol.is_some() {
             context.write_str(" checked=\"\"")?;
         }
         context.write_str(" disabled=\"\" /> ")?;

src/nodes.rs

@@ -135,7 +135,7 @@ pub enum NodeValue {
     /// **Block**. [Task list item](https://github.github.com/gfm/#task-list-items-extension-).
     /// The value is the symbol that was used in the brackets to mark a task item as checked, or
     /// `None` if the item is unchecked.
-    TaskItem(Option<char>),
+    TaskItem(NodeTaskItem),
 
     /// **Inline**.  A [soft line break](https://github.github.com/gfm/#soft-line-breaks).  If
     /// the `hardbreaks` option is set in `Options` during formatting, it will be formatted
@@ -294,6 +294,17 @@ pub struct NodeTable {
     pub num_nonempty_cells: usize,
 }
 
+/// A task list item's contents, and where it was found
+#[derive(Debug, Clone, PartialEq, Eq, Copy)]
+pub struct NodeTaskItem {
+    /// The symbol within the brackets used to mark the task item as checked,
+    /// or `None` if unchecked.
+    pub symbol: Option<char>,
+
+    /// The source position of the symbol itself.
+    pub symbol_sourcepos: Sourcepos,
+}
+
 /// An inline [code span](https://github.github.com/gfm/#code-spans).
 #[derive(Default, Debug, Clone, PartialEq, Eq)]
 pub struct NodeCode {

src/parser/mod.rs

@@ -20,7 +20,7 @@ use crate::node_matches;
 use crate::nodes::{
     self, AlertType, Ast, ListDelimType, ListType, Node, NodeAlert, NodeCodeBlock,
     NodeDescriptionItem, NodeFootnoteDefinition, NodeHeading, NodeHtmlBlock, NodeList,
-    NodeMultilineBlockQuote, NodeValue, Sourcepos,
+    NodeMultilineBlockQuote, NodeTaskItem, NodeValue, Sourcepos,
 };
 use crate::parser::inlines::RefMap;
 pub use crate::parser::options::Options;
@@ -2308,7 +2308,7 @@ where
         sourcepos: &mut Sourcepos,
         spx: &mut Spx,
     ) {
-        let (end, matched) = match scanners::tasklist(text) {
+        let (end, matched, symbol_range) = match scanners::tasklist(text) {
             Some(p) => p,
             None => return,
         };
@@ -2328,6 +2328,13 @@ where
             return;
         }
 
+        let symbol_sourcepos: Sourcepos = (
+            sourcepos.start.column_add(symbol_range.start as isize),
+            // `- 1`: symbol_range is end-exclusive, but sourcepos are inclusive.
+            sourcepos.start.column_add(symbol_range.end as isize - 1),
+        )
+            .into();
+
         let parent = node.parent().unwrap();
 
         if node_matches!(parent, NodeValue::TableCell) {
@@ -2349,7 +2356,10 @@ where
             parent.prepend(
                 self.arena.alloc(
                     Ast::new_with_sourcepos(
-                        NodeValue::TaskItem(if symbol == ' ' { None } else { Some(symbol) }),
+                        NodeValue::TaskItem(NodeTaskItem {
+                            symbol: if symbol == ' ' { None } else { Some(symbol) },
+                            symbol_sourcepos,
+                        }),
                         *sourcepos,
                     )
                     .into(),
@@ -2387,8 +2397,10 @@ where
                 parent.data_mut().sourcepos.start.column = adjust;
             }
 
-            grandparent.data_mut().value =
-                NodeValue::TaskItem(if symbol == ' ' { None } else { Some(symbol) });
+            grandparent.data_mut().value = NodeValue::TaskItem(NodeTaskItem {
+                symbol: if symbol == ' ' { None } else { Some(symbol) },
+                symbol_sourcepos,
+            });
 
             if let NodeValue::List(ref mut list) = &mut great_grandparent.data_mut().value {
                 list.is_task_list = true;

src/scanners.re

@@ -470,10 +470,11 @@ pub fn close_multiline_block_quote_fence(s: &str) -> Option<usize> {
 */
 }
 
-// Returns both the length of the match, and the tasklist item contents.
-// It is not guaranteed to be one byte, or one "character" long; the caller must ascertain
-// its fitness for purpose.
-pub fn tasklist(s: &str) -> Option<(usize, &str)> {
+// Returns the length of the match, the tasklist item contents, and the range of
+// the tasklist item match.
+// The match is not guaranteed to be one byte or one "character" long; the
+// caller must ascertain its fitness for purpose.
+pub fn tasklist(s: &str) -> Option<(usize, &str, std::ops::Range<usize>)> {
     let mut cursor = 0;
     let mut marker = 0;
     let len = s.len();
@@ -491,7 +492,7 @@ pub fn tasklist(s: &str) -> Option<(usize, &str)> {
         if cursor == len + 1 {
             cursor -= 1;
         }
-        return Some((cursor, &s[t1..t2]));
+        return Some((cursor, &s[t1..t2], t1..t2));
     }
     * { return None; }
 */

src/scanners.rs

@@ -1,4 +1,5 @@
 use super::*;
+use pretty_assertions::assert_eq;
 
 #[test]
 fn tasklist() {
@@ -516,3 +517,38 @@ fn tasklist_relaxed_unicode() {
         ])
     );
 }
+
+#[test]
+fn tasklist_symbol_sourcepos() {
+    let mut options = Options::default();
+    options.extension.tasklist = true;
+
+    let arena = Arena::new();
+    let md = r#"Henlo!
+
+> - [x] well
+>
+>   1. * > 3. [ ] true
+
++ [x] ok
+"#;
+
+    let root = parse_document(&arena, md, &options);
+    let sourceposes = root
+        .descendants()
+        .filter_map(|n| {
+            let NodeValue::TaskItem(ref nti) = n.data().value else {
+                return None;
+            };
+            Some(nti.symbol_sourcepos)
+        })
+        .collect::<Vec<_>>();
+    assert_eq!(
+        sourceposes,
+        vec![
+            (3, 6, 3, 6).into(),
+            (5, 16, 5, 16).into(),
+            (7, 4, 7, 4).into()
+        ]
+    );
+}

src/tests/tasklist.rs

@@ -271,11 +271,12 @@ impl<'o, 'c> XmlFormatter<'o, 'c> {
                     self.escape(&nfr.name)?;
                     self.output.write_str("\"")?;
                 }
-                NodeValue::TaskItem(Some(_)) => {
-                    self.output.write_str(" completed=\"true\"")?;
-                }
-                NodeValue::TaskItem(None) => {
-                    self.output.write_str(" completed=\"false\"")?;
+                NodeValue::TaskItem(ref nti) => {
+                    if nti.symbol.is_some() {
+                        self.output.write_str(" completed=\"true\"")?;
+                    } else {
+                        self.output.write_str(" completed=\"false\"")?;
+                    }
                 }
                 #[cfg(feature = "shortcodes")]
                 NodeValue::ShortCode(ref nsc) => {