LibWeb: Add Document helpers to move its cursor to word boundaries

This implementation is based on the same feature I added to Serenity's
TextEditor:

https://github.com/SerenityOS/serenity/pull/17477

Userland/Libraries/LibWeb/DOM/CharacterData.cpp

@@ -130,6 +130,8 @@ WebIDL::ExceptionOr<void> CharacterData::replace_data(size_t offset, size_t coun
 
     if (m_grapheme_segmenter)
         m_grapheme_segmenter->set_segmented_text(m_data);
+    if (m_word_segmenter)
+        m_word_segmenter->set_segmented_text(m_data);
 
     return {};
 }
@@ -165,4 +167,14 @@ Unicode::Segmenter& CharacterData::grapheme_segmenter()
     return *m_grapheme_segmenter;
 }
 
+Unicode::Segmenter& CharacterData::word_segmenter()
+{
+    if (!m_word_segmenter) {
+        m_word_segmenter = Unicode::Segmenter::create(Unicode::SegmenterGranularity::Word);
+        m_word_segmenter->set_segmented_text(m_data);
+    }
+
+    return *m_word_segmenter;
+}
+
 }

Userland/Libraries/LibWeb/DOM/CharacterData.h

@@ -41,6 +41,7 @@ public:
     WebIDL::ExceptionOr<void> replace_data(size_t offset_in_utf16_code_units, size_t count_in_utf16_code_units, String const&);
 
     Unicode::Segmenter& grapheme_segmenter();
+    Unicode::Segmenter& word_segmenter();
 
 protected:
     CharacterData(Document&, NodeType, String const&);
@@ -51,6 +52,7 @@ private:
     String m_data;
 
     OwnPtr<Unicode::Segmenter> m_grapheme_segmenter;
+    OwnPtr<Unicode::Segmenter> m_word_segmenter;
 };
 
 }

Userland/Libraries/LibWeb/DOM/Document.cpp

@@ -5423,6 +5423,24 @@ bool Document::decrement_cursor_position_offset()
     return true;
 }
 
+bool Document::increment_cursor_position_to_next_word()
+{
+    if (!m_cursor_position->increment_offset_to_next_word())
+        return false;
+
+    reset_cursor_blink_cycle();
+    return true;
+}
+
+bool Document::decrement_cursor_position_to_previous_word()
+{
+    if (!m_cursor_position->decrement_offset_to_previous_word())
+        return false;
+
+    reset_cursor_blink_cycle();
+    return true;
+}
+
 void Document::user_did_edit_document_text(Badge<EditEventHandler>)
 {
     reset_cursor_blink_cycle();

Userland/Libraries/LibWeb/DOM/Document.h

@@ -693,6 +693,8 @@ public:
     void set_cursor_position(JS::NonnullGCPtr<DOM::Position>);
     bool increment_cursor_position_offset();
     bool decrement_cursor_position_offset();
+    bool increment_cursor_position_to_next_word();
+    bool decrement_cursor_position_to_previous_word();
 
     bool cursor_blink_state() const { return m_cursor_blink_state; }
 

Userland/Libraries/LibWeb/DOM/Position.cpp

@@ -6,6 +6,7 @@
  */
 
 #include <AK/Utf8View.h>
+#include <LibUnicode/CharacterTypes.h>
 #include <LibUnicode/Segmenter.h>
 #include <LibWeb/DOM/Node.h>
 #include <LibWeb/DOM/Position.h>
@@ -66,6 +67,60 @@ bool Position::decrement_offset()
     return false;
 }
 
+static bool should_continue_beyond_word(Utf8View const& word)
+{
+    for (auto code_point : word) {
+        if (!Unicode::code_point_has_punctuation_general_category(code_point) && !Unicode::code_point_has_separator_general_category(code_point))
+            return false;
+    }
+
+    return true;
+}
+
+bool Position::increment_offset_to_next_word()
+{
+    if (!is<DOM::Text>(*m_node) || offset_is_at_end_of_node())
+        return false;
+
+    auto& node = static_cast<DOM::Text&>(*m_node);
+
+    while (true) {
+        if (auto offset = node.word_segmenter().next_boundary(m_offset); offset.has_value()) {
+            auto word = node.data().code_points().substring_view(m_offset, *offset - m_offset);
+            m_offset = *offset;
+
+            if (should_continue_beyond_word(word))
+                continue;
+        }
+
+        break;
+    }
+
+    return true;
+}
+
+bool Position::decrement_offset_to_previous_word()
+{
+    if (!is<DOM::Text>(*m_node) || m_offset == 0)
+        return false;
+
+    auto& node = static_cast<DOM::Text&>(*m_node);
+
+    while (true) {
+        if (auto offset = node.word_segmenter().previous_boundary(m_offset); offset.has_value()) {
+            auto word = node.data().code_points().substring_view(*offset, m_offset - *offset);
+            m_offset = *offset;
+
+            if (should_continue_beyond_word(word))
+                continue;
+        }
+
+        break;
+    }
+
+    return true;
+}
+
 bool Position::offset_is_at_end_of_node() const
 {
     if (!is<DOM::Text>(*m_node))

Userland/Libraries/LibWeb/DOM/Position.h

@@ -36,6 +36,9 @@ public:
     bool increment_offset();
     bool decrement_offset();
 
+    bool increment_offset_to_next_word();
+    bool decrement_offset_to_previous_word();
+
     bool equals(JS::NonnullGCPtr<Position> other) const
     {
         return m_node.ptr() == other->m_node.ptr() && m_offset == other->m_offset;