From a710229d9d3e31430ccffeaa68a7316367382e28 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?K=C3=A9vin=20Commaille?= <zecakeh@tedomum.fr>
Date: Wed, 19 Oct 2022 17:21:07 +0200
Subject: [PATCH] events: Fix markdown content detection

Encoded HTML entities would be detected as a change.
Use the parsed events instead to check if markdown syntax was detected.
---
 crates/ruma-common/src/events/message.rs      |  7 +--
 crates/ruma-common/src/events/room/message.rs | 47 ++++++++++++++++---
 .../ruma-common/tests/events/room_message.rs  | 22 +++++++++
 3 files changed, 65 insertions(+), 11 deletions(-)

diff --git a/crates/ruma-common/src/events/message.rs b/crates/ruma-common/src/events/message.rs
index 7c129790..9351d3ad 100644
--- a/crates/ruma-common/src/events/message.rs
+++ b/crates/ruma-common/src/events/message.rs
@@ -238,12 +238,9 @@ impl Text {
     /// Returns `None` if no Markdown formatting was found.
     #[cfg(feature = "markdown")]
     pub fn markdown(body: impl AsRef<str>) -> Option<Self> {
-        let body = body.as_ref();
-        let mut html_body = String::new();
+        use super::room::message::parse_markdown;
 
-        pulldown_cmark::html::push_html(&mut html_body, pulldown_cmark::Parser::new(body));
-
-        (html_body != format!("<p>{}</p>\n", body)).then(|| Self::html(html_body))
+        parse_markdown(body.as_ref()).map(Self::html)
     }
 
     fn default_mimetype() -> String {
diff --git a/crates/ruma-common/src/events/room/message.rs b/crates/ruma-common/src/events/room/message.rs
index 21816aeb..299a422c 100644
--- a/crates/ruma-common/src/events/room/message.rs
+++ b/crates/ruma-common/src/events/room/message.rs
@@ -592,12 +592,7 @@ impl FormattedBody {
     /// Returns `None` if no Markdown formatting was found.
     #[cfg(feature = "markdown")]
     pub fn markdown(body: impl AsRef<str>) -> Option<Self> {
-        let body = body.as_ref();
-        let mut html_body = String::new();
-
-        pulldown_cmark::html::push_html(&mut html_body, pulldown_cmark::Parser::new(body));
-
-        (html_body != format!("<p>{}</p>\n", body)).then(|| Self::html(html_body))
+        parse_markdown(body.as_ref()).map(Self::html)
     }
 
     /// Sanitize this `FormattedBody` if its format is `MessageFormat::Html`.
@@ -636,3 +631,43 @@ pub struct CustomEventContent {
     #[serde(flatten)]
     data: JsonObject,
 }
+
+#[cfg(feature = "markdown")]
+pub(crate) fn parse_markdown(text: &str) -> Option<String> {
+    use pulldown_cmark::{Event, Parser, Tag};
+
+    let mut found_first_paragraph = false;
+
+    let has_markdown = Parser::new(text).any(|ref event| {
+        let is_text = matches!(event, Event::Text(_));
+        let is_break = matches!(event, Event::SoftBreak | Event::HardBreak);
+        let is_first_paragraph_start = if matches!(event,
+            Event::Start(tag)
+                if matches!(tag, Tag::Paragraph)
+        ) {
+            if found_first_paragraph {
+                false
+            } else {
+                found_first_paragraph = true;
+                true
+            }
+        } else {
+            false
+        };
+        let is_paragraph_end = matches!(event,
+                Event::End(tag)
+                if matches!(tag, Tag::Paragraph)
+        );
+
+        !is_text && !is_break && !is_first_paragraph_start && !is_paragraph_end
+    });
+
+    if !has_markdown {
+        return None;
+    }
+
+    let mut html_body = String::new();
+    pulldown_cmark::html::push_html(&mut html_body, Parser::new(text));
+
+    Some(html_body)
+}
diff --git a/crates/ruma-common/tests/events/room_message.rs b/crates/ruma-common/tests/events/room_message.rs
index 1cbe66d8..255a7777 100644
--- a/crates/ruma-common/tests/events/room_message.rs
+++ b/crates/ruma-common/tests/events/room_message.rs
@@ -302,6 +302,28 @@ fn markdown_content_serialization() {
     );
 }
 
+#[test]
+#[cfg(feature = "markdown")]
+fn markdown_detection() {
+    use ruma_common::events::room::message::FormattedBody;
+
+    // No markdown
+    let formatted_body = FormattedBody::markdown("A simple message.");
+    assert_matches!(formatted_body, None);
+
+    // Multiple paragraphs trigger markdown
+    let formatted_body = FormattedBody::markdown("A message\nwith\n\nmultiple\n\nparagraphs");
+    formatted_body.unwrap();
+
+    // HTML entities don't trigger markdown.
+    let formatted_body = FormattedBody::markdown("A message with & HTML < entities");
+    assert_matches!(formatted_body, None);
+
+    // HTML triggers markdown.
+    let formatted_body = FormattedBody::markdown("<span>An HTML message</span>");
+    formatted_body.unwrap();
+}
+
 #[test]
 fn verification_request_deserialization() {
     let user_id = user_id!("@example2:localhost");