block: parse captions

src/block.rs

@@ -65,6 +65,9 @@ pub enum Leaf {
     /// Has zero or one inline for the cell contents.
     TableCell(Alignment),
 
+    /// Span is '^' character.
+    Caption,
+
     /// Span is the link tag.
     /// Inlines are lines of the URL.
     LinkDefinition,
@@ -268,7 +271,25 @@ impl<'s> TreeParser<'s> {
                         self.alignments.clear();
                         self.tree.enter(Node::Container(Table), span);
                         let mut last_row_node = None;
-                        for row in lines {
+                        let caption_line = lines
+                            .iter()
+                            .position(|sp| sp.of(self.src).trim_start().starts_with('^'))
+                            .map_or(lines.len(), |caption_line| {
+                                self.tree.enter(Node::Leaf(Caption), span);
+                                lines[caption_line] =
+                                    lines[caption_line].trim_start(self.src).skip("^ ".len());
+                                lines[lines.len() - 1] = lines[lines.len() - 1].trim_end(self.src);
+                                for line in &lines[caption_line..] {
+                                    self.tree.inline(*line);
+                                }
+                                self.tree.exit(); // caption, will insert inlines later if any
+                                caption_line
+                            });
+                        for row in &lines[..caption_line] {
+                            let row = row.trim(self.src);
+                            if row.is_empty() {
+                                break;
+                            }
                             let row_node = self
                                 .tree
                                 .enter(Node::Container(TableRow { head: false }), row.with_len(1));
@@ -468,6 +489,7 @@ struct BlockParser {
     kind: Block,
     span: Span,
     fence: Option<(char, usize)>,
+    caption: bool,
 }
 
 impl BlockParser {
@@ -595,6 +617,7 @@ impl BlockParser {
             kind,
             span,
             fence,
+            caption: false,
         }
     }
 
@@ -638,14 +661,26 @@ impl BlockParser {
                 !((&mut c).take(fence_length).all(|c| c == fence)
                     && c.next().map_or(true, char::is_whitespace))
             }
-            Block::Container(List { .. } | DescriptionList | TableRow { .. })
+            Block::Container(Table) if self.caption => !line.trim().is_empty(),
-            | Block::Leaf(TableCell(..)) => {
-                panic!()
-            }
             Block::Container(Table) => {
                 let line = line.trim();
                 let l = line.len();
-                line.as_bytes()[l - 1] == b'|' && line.as_bytes()[l - 2] != b'\\'
+                match l {
+                    0 => true,
+                    1..=2 => false,
+                    _ => {
+                        if line.starts_with("^ ") {
+                            self.caption = true;
+                            true
+                        } else {
+                            line.as_bytes()[l - 1] == b'|' && line.as_bytes()[l - 2] != b'\\'
+                        }
+                    }
+                }
+            }
+            Block::Container(List { .. } | DescriptionList | TableRow { .. })
+            | Block::Leaf(TableCell(..) | Caption) => {
+                panic!()
             }
         }
     }
@@ -1470,6 +1505,69 @@ mod test {
         );
     }
 
+    #[test]
+    fn parse_table_caption() {
+        test_parse!(
+            "|a|\n^ caption",
+            (Enter(Container(Table)), ""),
+            (Enter(Leaf(Caption)), ""),
+            (Inline, "caption"),
+            (Exit(Leaf(Caption)), ""),
+            (Enter(Container(TableRow { head: false })), "|"),
+            (Enter(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Inline, "a"),
+            (Exit(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Exit(Container(TableRow { head: false })), "|"),
+            (Exit(Container(Table)), ""),
+        );
+    }
+
+    #[test]
+    fn parse_table_caption_multiline() {
+        test_parse!(
+            concat!(
+                "|a|\n",       //
+                "\n",          //
+                "^ caption\n", //
+                "continued\n", //
+                "\n",          //
+                "para\n",      //
+            ),
+            (Enter(Container(Table)), ""),
+            (Enter(Leaf(Caption)), ""),
+            (Inline, "caption\n"),
+            (Inline, "continued"),
+            (Exit(Leaf(Caption)), ""),
+            (Enter(Container(TableRow { head: false })), "|"),
+            (Enter(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Inline, "a"),
+            (Exit(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Exit(Container(TableRow { head: false })), "|"),
+            (Exit(Container(Table)), ""),
+            (Atom(Blankline), "\n"),
+            (Enter(Leaf(Paragraph)), ""),
+            (Inline, "para"),
+            (Exit(Leaf(Paragraph)), ""),
+        );
+    }
+
+    #[test]
+    fn parse_table_caption_empty() {
+        test_parse!(
+            "|a|\n^ ",
+            (Enter(Container(Table)), ""),
+            (Enter(Container(TableRow { head: false })), "|"),
+            (Enter(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Inline, "a"),
+            (Exit(Leaf(TableCell(Alignment::Unspecified))), "|"),
+            (Exit(Container(TableRow { head: false })), "|"),
+            (Exit(Container(Table)), ""),
+            (Enter(Leaf(Paragraph)), ""),
+            (Inline, "^"),
+            (Exit(Leaf(Paragraph)), ""),
+        );
+    }
+
     #[test]
     fn parse_table_sep_row_only() {
         test_parse!(

src/html.rs

@@ -157,6 +157,7 @@ impl<'s, I: Iterator<Item = Event<'s>>, W: std::fmt::Write> Writer<'s, I, W> {
                         Container::Heading { level } => write!(self.out, "<h{}", level)?,
                         Container::TableCell { head: false, .. } => self.out.write_str("<td")?,
                         Container::TableCell { head: true, .. } => self.out.write_str("<th")?,
+                        Container::Caption => self.out.write_str("<caption")?,
                         Container::DescriptionTerm => self.out.write_str("<dt")?,
                         Container::CodeBlock { .. } => self.out.write_str("<pre")?,
                         Container::Span | Container::Math { .. } => self.out.write_str("<span")?,
@@ -338,6 +339,7 @@ impl<'s, I: Iterator<Item = Event<'s>>, W: std::fmt::Write> Writer<'s, I, W> {
                         Container::Heading { level } => write!(self.out, "</h{}>", level)?,
                         Container::TableCell { head: false, .. } => self.out.write_str("</td>")?,
                         Container::TableCell { head: true, .. } => self.out.write_str("</th>")?,
+                        Container::Caption => self.out.write_str("</caption>")?,
                         Container::DescriptionTerm => self.out.write_str("</dt>")?,
                         Container::CodeBlock { .. } => self.out.write_str("</code></pre>")?,
                         Container::Span => self.out.write_str("</span>")?,

src/lib.rs

@@ -56,6 +56,8 @@ pub enum Container<'s> {
     Heading { level: usize },
     /// A cell element of row within a table.
     TableCell { alignment: Alignment, head: bool },
+    /// A caption within a table.
+    Caption,
     /// A term within a description list.
     DescriptionTerm,
     /// A block with raw markup for a specific output format.
@@ -111,6 +113,7 @@ impl<'s> Container<'s> {
             | Self::Paragraph
             | Self::Heading { .. }
             | Self::TableCell { .. }
+            | Self::Caption
             | Self::DescriptionTerm
             | Self::RawBlock { .. }
             | Self::CodeBlock { .. } => true,
@@ -148,6 +151,7 @@ impl<'s> Container<'s> {
             Self::Paragraph
             | Self::Heading { .. }
             | Self::TableCell { .. }
+            | Self::Caption
             | Self::DescriptionTerm
             | Self::RawBlock { .. }
             | Self::CodeBlock { .. }
@@ -541,6 +545,7 @@ impl<'s> Parser<'s> {
                                     alignment,
                                     head: self.table_head_row,
                                 },
+                                block::Leaf::Caption => Container::Caption,
                                 block::Leaf::LinkDefinition => unreachable!(),
                             }
                         }