editor: Fix DeleteToPreviousWordStart and DeleteToNextWordEnd interaction with newlines (#16848)

Closes #5285, #14389

Changes:
- `DeleteToPreviousWordStart` now deletes '\n' separately from preceding
words and whitespace.
- `DeleteToNextWordEnd` now deletes '\n' and any following whitespace
separately from subsequent words.
- Added an `ignore_newlines` flag to both actions to optionally retain
the old behavior.

These modifications align the behavior more closely with other popular
editors like VSCode and Sublime:
- `DeleteToPreviousWordStart` now matches the default <Ctrl+Backspace>
action in those editors.
- `DeleteToNextWordEnd` becomes more intuitive and closely resembles the
default <Ctrl+Delete> behavior in those editors.

Release Notes:

- Improved `DeleteToPreviousWordStart` and `DeleteToNextWordEnd`
interactions around newlines. You can opt-in into the previous behavior
by adding {"ignore_newlines": true} to either action's binds in your
keymap. ([#5285](https://github.com/zed-industries/zed/issues/5285),
[#14389](https://github.com/zed-industries/zed/issues/14389))

crates/editor/src/movement.rs

@@ -270,6 +270,19 @@ pub fn previous_word_start(map: &DisplaySnapshot, point: DisplayPoint) -> Displa
     })
 }
 
+/// Returns a position of the previous word boundary, where a word character is defined as either
+/// uppercase letter, lowercase letter, '_' character, language-specific word character (like '-' in CSS) or newline.
+pub fn previous_word_start_or_newline(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
+    let raw_point = point.to_point(map);
+    let classifier = map.buffer_snapshot.char_classifier_at(raw_point);
+
+    find_preceding_boundary_display_point(map, point, FindRange::MultiLine, |left, right| {
+        (classifier.kind(left) != classifier.kind(right) && !right.is_whitespace())
+            || left == '\n'
+            || right == '\n'
+    })
+}
+
 /// Returns a position of the previous subword boundary, where a subword is defined as a run of
 /// word characters of the same "subkind" - where subcharacter kinds are '_' character,
 /// lowerspace characters and uppercase characters.
@@ -299,6 +312,24 @@ pub fn next_word_end(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint
     })
 }
 
+/// Returns a position of the next word boundary, where a word character is defined as either
+/// uppercase letter, lowercase letter, '_' character, language-specific word character (like '-' in CSS) or newline.
+pub fn next_word_end_or_newline(map: &DisplaySnapshot, point: DisplayPoint) -> DisplayPoint {
+    let raw_point = point.to_point(map);
+    let classifier = map.buffer_snapshot.char_classifier_at(raw_point);
+
+    let mut on_starting_row = true;
+    find_boundary(map, point, FindRange::MultiLine, |left, right| {
+        if left == '\n' {
+            on_starting_row = false;
+        }
+        (classifier.kind(left) != classifier.kind(right)
+            && ((on_starting_row && !left.is_whitespace())
+                || (!on_starting_row && !right.is_whitespace())))
+            || right == '\n'
+    })
+}
+
 /// Returns a position of the next subword boundary, where a subword is defined as a run of
 /// word characters of the same "subkind" - where subcharacter kinds are '_' character,
 /// lowerspace characters and uppercase characters.