gpui: Fix text wrapping for URLs (#35724)

Jason Lee and Mikayla Maki created 2 weeks ago

Close #35715

Release Notes:

- Fixed to wrap long URLs in editor.

<img width="836" height="740" alt="image"
src="https://github.com/user-attachments/assets/635ce792-5f19-4c76-b131-0d270d09b103"
/>

I remember when I was working on CJK line wrapping support in the early
days, I considered making `\` a line wrapping character, but for some
reason it was on the list of characters that were not allowed to wrap.

In reference to VS Code, it looks like `&`, `/`, `?` should wrap, so I
removed all of them.

---------

Co-authored-by: Mikayla Maki <[e-mail]>

Change summary

crates/gpui/examples/text_wrapper.rs        |  6 +++++-
crates/gpui/src/text_system/line_wrapper.rs | 17 +++++++++++------
2 files changed, 16 insertions(+), 7 deletions(-)

Detailed changes

crates/gpui/examples/text_wrapper.rs

@@ -7,7 +7,11 @@ struct HelloWorld {}
 
 impl Render for HelloWorld {
     fn render(&mut self, _window: &mut Window, _cx: &mut Context<Self>) -> impl IntoElement {
-        let text = "The longest word 你好世界这段是中文，こんにちはこの段落は日本語です in any of the major English language dictionaries is pneumonoultramicroscopicsilicovolcanoconiosis, a word that refers to a lung disease contracted from the inhalation of very fine silica particles, specifically from a volcano; medically, it is the same as silicosis.";
+        let text = "The longest word 你好世界这段是中文，こんにちはこの段落は日本語です in any of the major \
+            English language dictionaries is pneumonoultramicroscopicsilicovolcanoconiosis, a word that \
+            refers to a lung disease contracted from the inhalation of very fine silica particles, \
+            a url https://github.com/zed-industries/zed/pull/35724?query=foo&bar=2, \
+            specifically from a volcano; medically, it is the same as silicosis.";
         div()
             .id("page")
             .size_full()

crates/gpui/src/text_system/line_wrapper.rs

@@ -163,6 +163,8 @@ impl LineWrapper {
         line
     }
 
+    /// Any character in this list should be treated as a word character,
+    /// meaning it can be part of a word that should not be wrapped.
     pub(crate) fn is_word_char(c: char) -> bool {
         // ASCII alphanumeric characters, for English, numbers: `Hello123`, etc.
         c.is_ascii_alphanumeric() ||
@@ -180,10 +182,9 @@ impl LineWrapper {
         // https://en.wikipedia.org/wiki/Cyrillic_script_in_Unicode
         matches!(c, '\u{0400}'..='\u{04FF}') ||
         // Some other known special characters that should be treated as word characters,
-        // e.g. `a-b`, `var_name`, `I'm`, '@mention`, `#hashtag`, `100%`, `3.1415`, `2^3`, `a~b`, etc.
-        matches!(c, '-' | '_' | '.' | '\'' | '$' | '%' | '@' | '#' | '^' | '~' | ',' | '!' | ';' | '*') ||
-        // Characters that used in URL, e.g. `https://github.com/zed-industries/zed?a=1&b=2` for better wrapping a long URL.
-        matches!(c,  '/' | ':' | '?' | '&' | '=') ||
+        // e.g. `a-b`, `var_name`, `I'm`, '@mention`, `#hashtag`, `100%`, `3.1415`,
+        // `2^3`, `a~b`, `a=1`, `Self::new`, etc.
+        matches!(c, '-' | '_' | '.' | '\'' | '$' | '%' | '@' | '#' | '^' | '~' | ',' | '=' | ':') ||
         // `⋯` character is special used in Zed, to keep this at the end of the line.
         matches!(c, '⋯')
     }
@@ -644,15 +645,19 @@ mod tests {
         assert_word("@mention");
         assert_word("#hashtag");
         assert_word("$variable");
+        assert_word("a=1");
+        assert_word("Self::is_word_char");
         assert_word("more⋯");
 
         // Space
         assert_not_word("foo bar");
 
         // URL case
-        assert_word("https://github.com/zed-industries/zed/");
         assert_word("github.com");
-        assert_word("a=1&b=2");
+        assert_not_word("zed-industries/zed");
+        assert_not_word("zed-industries\\zed");
+        assert_not_word("a=1&b=2");
+        assert_not_word("foo?b=2");
 
         // Latin-1 Supplement
         assert_word("ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏ");