Pandoc fuzzer

pulldown-cmark · Mar 5, 2024 · 99ca650 · 99ca650
1 parent 206ffcf
commit 99ca650
Show file tree

Hide file tree

Showing 2 changed files with 19 additions and 38 deletions.
diff --git a/fuzz/fuzz_targets/pandoc.rs b/fuzz/fuzz_targets/pandoc.rs
@@ -24,6 +24,9 @@ fuzz_target!(|text: String| {
         return;
     }
 
+    if text.contains("<") {
+        return;
+    }
     if text.contains("[^ ") {
         return;
     }
@@ -111,9 +114,24 @@ fuzz_target!(|text: String| {
     }
     let pulldown_cmark_events = normalize_pandoc(pulldown_cmark_events);
     let mut footstack = vec![];
+    let mut liststack = 0;
     for event in &pulldown_cmark_events {
         use pulldown_cmark::{Event, Tag, TagEnd};
         match event {
+            Event::Start(Tag::CodeBlock(..)) => {
+                // differences in list tightness
+                if liststack != 0 {
+                    return;
+                }
+            }
+            Event::Start(Tag::Item) => {
+                // differences in list tightness
+                liststack += 1;
+            }
+            Event::End(TagEnd::Item) => {
+                // differences in list tightness
+                liststack -= 1;
+            }
             Event::Start(Tag::FootnoteDefinition(id)) => {
                 footstack.push(id.to_string());
             }

diff --git a/fuzz/src/lib.rs b/fuzz/src/lib.rs
@@ -573,7 +573,7 @@ pub fn normalize_pandoc(events: Vec<Event<'_>>) -> Vec<Event<'_>> {
     for event in normalized.iter_mut() {
         match event {
             // Pandoc collapses spaces
-            Event::Text(p) if p.contains("  ") || p.contains("\n") => {
+            Event::Text(p) | Event::InlineMath(p) | Event::DisplayMath(p) if p.contains("  ") || p.contains("\n") => {
                 let mut st = String::with_capacity(p.len());
                 let mut prev_char_is_space = false;
                 for c in p.chars() {
@@ -599,29 +599,6 @@ pub fn normalize_pandoc(events: Vec<Event<'_>>) -> Vec<Event<'_>> {
             Event::Start(Tag::CodeBlock(fence)) => {
                 *fence = CodeBlockKind::Fenced("".into());
             }
-            Event::InlineMath(p) if p.contains("  ") || p.contains("\n") => {
-                let mut st = String::with_capacity(p.len());
-                let mut prev_char_is_space = false;
-                for c in p.chars() {
-                    if c == ' ' {
-                        if prev_char_is_space {
-                            continue;
-                        }
-                        prev_char_is_space = true;
-                        st.push(' ');
-                    } else if c == '\n' {
-                        if prev_char_is_space {
-                            continue;
-                        }
-                        prev_char_is_space = true;
-                        st.push('\n');
-                    } else {
-                        prev_char_is_space = false;
-                        st.push(c);
-                    }
-                }
-                *p = st.into();
-            }
             _ => {},
         }
     }
@@ -891,20 +868,6 @@ pub fn normalize(events: Vec<Event<'_>>) -> Vec<Event<'_>> {
             (Some(Event::Text(prev)), Event::Text(next)) => *prev = format!("{prev}{next}").into(),
             (Some(Event::Html(prev)), Event::Html(next)) => *prev = format!("{prev}{next}").into(),
 
-            // commonmark.js wraps non-empty list items in a paragraph.
-            (Some(Event::Start(Tag::Item)), next)
-                if next != &Event::Start(Tag::Paragraph) && next != &Event::End(TagEnd::Item) =>
-            {
-                normalized.push(Event::Start(Tag::Paragraph));
-                normalized.push(event);
-            }
-            (Some(prev), Event::End(TagEnd::Item))
-                if prev != &Event::End(TagEnd::Paragraph) && prev != &Event::Start(Tag::Item) =>
-            {
-                normalized.push(Event::End(TagEnd::Paragraph));
-                normalized.push(event);
-            }
-
             // commonmark.js always adds a final newline to code blocks.
             (Some(Event::Text(prev)), Event::End(TagEnd::CodeBlock)) => {
                 *prev = prev.trim_end().to_string().into();