Merge branch 'notriddle/main'

Byron · Sep 6, 2024 · ed4ff6f · ed4ff6f
2 parents ab22a57 + c3a0f32
commit ed4ff6f
Show file tree

Hide file tree

Showing 6 changed files with 72 additions and 14 deletions.
diff --git a/src/lib.rs b/src/lib.rs
@@ -63,6 +63,8 @@ pub struct State<'a> {
     pub is_in_code_block: bool,
     /// True if the last event was text and the text does not have trailing newline. Used to inject additional newlines before code block end fence.
     pub last_was_text_without_trailing_newline: bool,
+    /// True if the last event was a paragraph start. Used to escape spaces at start of line (prevent spurrious indented code).
+    pub last_was_paragraph_start: bool,
     /// Currently open links
     pub link_stack: Vec<LinkCategory<'a>>,
     /// Currently open images
@@ -230,6 +232,8 @@ where
 
     let last_was_text_without_trailing_newline = state.last_was_text_without_trailing_newline;
     state.last_was_text_without_trailing_newline = false;
+    let last_was_paragraph_start = state.last_was_paragraph_start;
+    state.last_was_paragraph_start = false;
     match event.borrow() {
         Rule => {
             consume_newlines(formatter, state)?;
@@ -271,21 +275,25 @@ where
             let consumed_newlines = state.newlines_before_start != 0;
             consume_newlines(formatter, state)?;
             match tag {
-                Item => match state.list_stack.last_mut() {
-                    Some(inner) => {
-                        state.padding.push(padding_of(*inner));
-                        match inner {
-                            Some(n) => {
-                                let bullet_number = *n;
-                                if options.increment_ordered_list_bullets {
-                                    *n += 1;
+                Item => {
+                    // lazy lists act like paragraphs with no event
+                    state.last_was_paragraph_start = true;
+                    match state.list_stack.last_mut() {
+                        Some(inner) => {
+                            state.padding.push(padding_of(*inner));
+                            match inner {
+                                Some(n) => {
+                                    let bullet_number = *n;
+                                    if options.increment_ordered_list_bullets {
+                                        *n += 1;
+                                    }
+                                    write!(formatter, "{}{} ", bullet_number, options.ordered_list_token)
                                 }
-                                write!(formatter, "{}{} ", bullet_number, options.ordered_list_token)
+                                None => write!(formatter, "{} ", options.list_token),
                             }
-                            None => write!(formatter, "{} ", options.list_token),
                         }
+                        None => Ok(()),
                     }
-                    None => Ok(()),
                 },
                 Table(alignments) => {
                     state.table_alignments = alignments.iter().map(From::from).collect();
@@ -344,7 +352,10 @@ where
                     state.padding.push("    ".into());
                     write!(formatter, "[^{name}]: ")
                 }
-                Paragraph => Ok(()),
+                Paragraph => {
+                    state.last_was_paragraph_start = true;
+                    Ok(())
+                },
                 Heading {
                     level,
                     id,
@@ -614,13 +625,23 @@ where
         HardBreak => formatter.write_str("  \n").and(padding(formatter, &state.padding)),
         SoftBreak => formatter.write_char('\n').and(padding(formatter, &state.padding)),
         Text(text) => {
+            let mut text = &text[..];
             if let Some(shortcut_text) = state.current_shortcut_text.as_mut() {
                 shortcut_text.push_str(text);
             }
             if let Some(text_for_header) = state.text_for_header.as_mut() {
                 text_for_header.push_str(text);
             }
             consume_newlines(formatter, state)?;
+            if last_was_paragraph_start {
+                if text.starts_with('\t') {
+                    formatter.write_str("&#9;")?;
+                    text = &text[1..];
+                } else if text.starts_with(' ') {
+                    formatter.write_str("&#32;")?;
+                    text = &text[1..];
+                }
+            }
             state.last_was_text_without_trailing_newline = !text.ends_with('\n');
             print_text_without_trailing_newline(
                 &escape_leading_special_characters(text, state.is_in_code_block, options),

diff --git a/src/text_modifications.rs b/src/text_modifications.rs
@@ -23,10 +23,10 @@ where
 
 pub fn escape_leading_special_characters<'a>(
     t: &'a str,
-    is_in_block_quote: bool,
+    is_in_code_block: bool,
     options: &Options<'a>,
 ) -> Cow<'a, str> {
-    if is_in_block_quote || t.is_empty() {
+    if is_in_code_block || t.is_empty() {
         return Cow::Borrowed(t);
     }
 

diff --git a/tests/fixtures/common-mark.md b/tests/fixtures/common-mark.md
@@ -238,6 +238,16 @@ foo
 * | < > # 
 ````
 
+### Entity escapes
+
+&copy; 2280 one world government, inc
+
+&#9;paragraph starts with a literal tab.
+
+&#32;&#32;&#32;&#32;I'm using markdown like a typewriter,
+&#32;&#32;&#32;&#32;probably because this document started life in a WYSIWYG
+&#32;&#32;&#32;&#32;editor and got converted...
+
 [Links]: http://www.example.com/shortcut
 [`diam`]: http://www.example.com/shortcut_code_diam
 [`voluptua`]: http://www.example.com/shortcut_code_voluptua

diff --git a/tests/fixtures/snapshots/stupicat-event-by-event-output b/tests/fixtures/snapshots/stupicat-event-by-event-output
@@ -242,6 +242,16 @@ foo
 * | < > # 
 ````
 
+### Entity escapes
+
+© 2280 one world government, inc
+
+&#9;paragraph starts with a literal tab.
+
+&#32;   I'm using markdown like a typewriter,
+    probably because this document started life in a WYSIWYG
+    editor and got converted...
+
 [Links]: http://www.example.com/shortcut
 [`diam`]: http://www.example.com/shortcut_code_diam
 [`voluptua`]: http://www.example.com/shortcut_code_voluptua

diff --git a/tests/fixtures/snapshots/stupicat-output b/tests/fixtures/snapshots/stupicat-output
@@ -242,6 +242,16 @@ foo
 * | < > # 
 ````
 
+### Entity escapes
+
+© 2280 one world government, inc
+
+&#9;paragraph starts with a literal tab.
+
+&#32;   I'm using markdown like a typewriter,
+    probably because this document started life in a WYSIWYG
+    editor and got converted...
+
 [Links]: http://www.example.com/shortcut
 [`diam`]: http://www.example.com/shortcut_code_diam
 [`voluptua`]: http://www.example.com/shortcut_code_voluptua

diff --git a/tests/fmt.rs b/tests/fmt.rs
@@ -1131,6 +1131,13 @@ mod escapes {
             );
         });
     }
+
+    #[test]
+    fn entity_escape_is_not_code_block_indent() {
+        source_range_fmt::assert_events_eq("&#9;foo");
+        source_range_fmt::assert_events_eq("&#32;   foo");
+        source_range_fmt::assert_events_eq(" * &#32;   foo\n * &#9;foo");
+    }
 }
 
 mod list {