Merge pull request #90 from chriskrycho/escape-code-in-tables

Escape code in tables

Author: Byron

src/lib.rs

@@ -51,6 +51,7 @@ pub enum CodeBlockKind {
 /// This does not only allow introspection, but enables the user
 /// to halt the serialization at any time, and resume it later.
 #[derive(Clone, Default, Debug, PartialEq, Eq, PartialOrd, Ord, Hash)]
+#[non_exhaustive]
 pub struct State<'a> {
     /// The amount of newlines to insert after `Event::Start(...)`
     pub newlines_before_start: usize,
@@ -79,6 +80,8 @@ pub struct State<'a> {
     pub image_stack: Vec<ImageLink<'a>>,
     /// Keeps track of the last seen heading's id, classes, and attributes
     pub current_heading: Option<Heading<'a>>,
+    /// True whenever between `Start(TableCell)` and `End(TableCell)`
+    pub in_table_cell: bool,
 
     /// Keeps track of the last seen shortcut/link
     pub current_shortcut_text: Option<String>,
@@ -305,10 +308,25 @@ where
                 text_for_header.push_str(text);
                 text_for_header.push('`');
             }
+
+            // (re)-escape `|` when it appears as part of inline code in the
+            // body of a table.
+            //
+            // NOTE: This does not do *general* escaped-character handling
+            // because the only character which *requires* this handling in this
+            // spot in earlier versions of `pulldown-cmark` is a pipe character
+            // in inline code in a table. Other escaping is handled when `Text`
+            // events are emitted.
+            let text = if state.in_table_cell {
+                Cow::Owned(text.replace('|', "\\|"))
+            } else {
+                Cow::Borrowed(text.as_ref())
+            };
+
             if text.chars().all(|ch| ch == ' ') {
                 write!(formatter, "`{text}`")
             } else {
-                let backticks = "`".repeat(count_consecutive(text, '`') + 1);
+                let backticks = "`".repeat(count_consecutive(&text, '`') + 1);
                 let space = match text.as_bytes() {
                     &[b'`', ..] | &[.., b'`'] => " ", // Space needed to separate backtick.
                     &[b' ', .., b' '] => " ",         // Space needed to escape inner space.
@@ -355,6 +373,7 @@ where
                 TableRow => Ok(()),
                 TableCell => {
                     state.text_for_header = Some(String::new());
+                    state.in_table_cell = true;
                     formatter.write_char('|')
                 }
                 Link {
@@ -689,6 +708,7 @@ where
                 state
                     .table_headers
                     .push(state.text_for_header.take().unwrap_or_default());
+                state.in_table_cell = false;
                 Ok(())
             }
             t @ (TagEnd::TableRow | TagEnd::TableHead) => {

tests/cat.sh

@@ -99,3 +99,9 @@ title "stupicat"
     WITH_SNAPSHOT="$snapshot/stupicat-toml-frontmatter-output" \
     expect_run_sh $SUCCESSFULLY "${exe[*]} $fixture/toml-frontmatter.md 2>/dev/null"
 )
+
+(with "table with escaped characters in cells"
+  it "succeeds in reproducing the escapes" && \
+    WITH_SNAPSHOT="$snapshot/stupicat-table-with-escapes-output" \
+    expect_run_sh $SUCCESSFULLY "${exe[*]} $fixture/table-with-escapes.md 2>/dev/null"
+)

tests/fixtures/snapshots/stupicat-table-with-escapes-output

@@ -0,0 +1,6 @@
+|First|Second|Third|Fourth|
+|-----|------|-----|------|
+|`\|`|`\\`|`a\[b\]`|`>>`|
+
+Normal inline code with pipes `|` should not be re-escaped, because it does not
+cause a problem!

tests/fixtures/table-with-escapes.md

@@ -0,0 +1,6 @@
+| First | Second | Third    | Fourth |
+| ----- | ------ | -------- | ------ |
+| `\|`  | `\\`   | `a\[b\]` | `>>`   |
+
+Normal inline code with pipes `|` should not be re-escaped, because it does not
+cause a problem!