From 5074aaea69f7cec9673aa57ba277686c153d7862 Mon Sep 17 00:00:00 2001
From: Andreas Kling <kling@serenityos.org>
Date: Tue, 27 Apr 2021 13:05:50 +0200
Subject: [PATCH] LibWeb: Refactor Layout::TextNode splitting into a chunk
 iterator

Creating a ChunkIterator allows you to iterate over the text in a
Layout::TextNode at your leisure by calling next() when you want
another chunk.

This is one of many steps towards improving inline layout.
---
 Userland/Libraries/LibWeb/Layout/TextNode.cpp | 162 ++++++++++--------
 Userland/Libraries/LibWeb/Layout/TextNode.h   |  30 +++-
 2 files changed, 115 insertions(+), 77 deletions(-)

diff --git a/Userland/Libraries/LibWeb/Layout/TextNode.cpp b/Userland/Libraries/LibWeb/Layout/TextNode.cpp
index df362475e22..e10a57bd8ae 100644
--- a/Userland/Libraries/LibWeb/Layout/TextNode.cpp
+++ b/Userland/Libraries/LibWeb/Layout/TextNode.cpp
@@ -1,11 +1,11 @@
 /*
- * Copyright (c) 2018-2020, Andreas Kling <kling@serenityos.org>
+ * Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
  *
  * SPDX-License-Identifier: BSD-2-Clause
  */
 
+#include <AK/ScopeGuard.h>
 #include <AK/StringBuilder.h>
-#include <AK/Utf8View.h>
 #include <LibGfx/Painter.h>
 #include <LibWeb/DOM/Document.h>
 #include <LibWeb/Layout/BlockBox.h>
@@ -102,59 +102,6 @@ void TextNode::paint_cursor_if_needed(PaintContext& context, const LineBoxFragme
     context.painter().draw_rect(cursor_rect, computed_values().color());
 }
 
-template<typename Callback>
-void TextNode::for_each_chunk(Callback callback, LayoutMode layout_mode, bool do_wrap_lines, bool do_wrap_breaks) const
-{
-    Utf8View view(m_text_for_rendering);
-    if (view.is_empty())
-        return;
-
-    auto start_of_chunk = view.begin();
-
-    auto commit_chunk = [&](auto it, bool has_breaking_newline, bool must_commit = false) {
-        if (layout_mode == LayoutMode::OnlyRequiredLineBreaks && !must_commit)
-            return;
-
-        int start = view.byte_offset_of(start_of_chunk);
-        int length = view.byte_offset_of(it) - view.byte_offset_of(start_of_chunk);
-
-        if (has_breaking_newline || length > 0) {
-            auto chunk_view = view.substring_view(start, length);
-            callback(chunk_view, start, length, has_breaking_newline, is_all_whitespace(chunk_view.as_string()));
-        }
-
-        start_of_chunk = it;
-    };
-
-    bool last_was_space = isspace(*view.begin());
-    bool last_was_newline = false;
-    for (auto it = view.begin(); it != view.end();) {
-        if (layout_mode == LayoutMode::AllPossibleLineBreaks) {
-            commit_chunk(it, false);
-        }
-        if (last_was_newline) {
-            last_was_newline = false;
-            commit_chunk(it, true);
-        }
-        if (do_wrap_breaks && *it == '\n') {
-            last_was_newline = true;
-            commit_chunk(it, false);
-        }
-        if (do_wrap_lines) {
-            bool is_space = isspace(*it);
-            if (is_space != last_was_space) {
-                last_was_space = is_space;
-                commit_chunk(it, false);
-            }
-        }
-        ++it;
-    }
-    if (last_was_newline)
-        commit_chunk(view.end(), true);
-    if (start_of_chunk != view.end())
-        commit_chunk(view.end(), false, true);
-}
-
 void TextNode::split_into_lines_by_rules(InlineFormattingContext& context, LayoutMode layout_mode, bool do_collapse, bool do_wrap_lines, bool do_wrap_breaks)
 {
     auto& containing_block = context.containing_block();
@@ -193,22 +140,15 @@ void TextNode::split_into_lines_by_rules(InlineFormattingContext& context, Layou
         m_text_for_rendering = dom_node().data();
     }
 
-    // do_wrap_lines  => chunks_are_words
-    // !do_wrap_lines => chunks_are_lines
-    struct Chunk {
-        Utf8View view;
-        int start { 0 };
-        int length { 0 };
-        bool is_break { false };
-        bool is_all_whitespace { false };
-    };
     Vector<Chunk, 128> chunks;
+    ChunkIterator iterator(m_text_for_rendering, layout_mode, do_wrap_lines, do_wrap_breaks);
 
-    for_each_chunk(
-        [&](const Utf8View& view, int start, int length, bool is_break, bool is_all_whitespace) {
-            chunks.append({ Utf8View(view), start, length, is_break, is_all_whitespace });
-        },
-        layout_mode, do_wrap_lines, do_wrap_breaks);
+    for (;;) {
+        auto chunk = iterator.next();
+        if (!chunk.has_value())
+            break;
+        chunks.append(chunk.release_value());
+    }
 
     for (size_t i = 0; i < chunks.size(); ++i) {
         auto& chunk = chunks[i];
@@ -251,11 +191,9 @@ void TextNode::split_into_lines_by_rules(InlineFormattingContext& context, Layou
             }
         }
 
-        if (do_wrap_breaks) {
-            if (chunk.is_break) {
-                containing_block.add_line_box();
-                available_width = context.available_width_at_line(line_boxes.size() - 1);
-            }
+        if (do_wrap_breaks && chunk.has_breaking_newline) {
+            containing_block.add_line_box();
+            available_width = context.available_width_at_line(line_boxes.size() - 1);
         }
     }
 }
@@ -319,4 +257,80 @@ void TextNode::handle_mousemove(Badge<EventHandler>, const Gfx::IntPoint& positi
     downcast<Label>(*parent()).handle_mousemove_on_label({}, position, button);
 }
 
+TextNode::ChunkIterator::ChunkIterator(StringView const& text, LayoutMode layout_mode, bool wrap_lines, bool wrap_breaks)
+    : m_layout_mode(layout_mode)
+    , m_wrap_lines(wrap_lines)
+    , m_wrap_breaks(wrap_breaks)
+    , m_utf8_view(text)
+    , m_start_of_chunk(m_utf8_view.begin())
+    , m_iterator(m_utf8_view.begin())
+{
+    m_last_was_space = !text.is_empty() && isspace(*m_utf8_view.begin());
+}
+
+Optional<TextNode::Chunk> TextNode::ChunkIterator::next()
+{
+    while (m_iterator != m_utf8_view.end()) {
+        auto guard = ScopeGuard([&] { ++m_iterator; });
+        if (m_layout_mode == LayoutMode::AllPossibleLineBreaks) {
+            if (auto result = try_commit_chunk(m_iterator, false); result.has_value())
+                return result.release_value();
+        }
+        if (m_last_was_newline) {
+            m_last_was_newline = false;
+            if (auto result = try_commit_chunk(m_iterator, true); result.has_value())
+                return result.release_value();
+        }
+        if (m_wrap_breaks && *m_iterator == '\n') {
+            m_last_was_newline = true;
+            if (auto result = try_commit_chunk(m_iterator, false); result.has_value())
+                return result.release_value();
+        }
+        if (m_wrap_lines) {
+            bool is_space = isspace(*m_iterator);
+            if (is_space != m_last_was_space) {
+                m_last_was_space = is_space;
+                if (auto result = try_commit_chunk(m_iterator, false); result.has_value())
+                    return result.release_value();
+            }
+        }
+    }
+
+    if (m_last_was_newline) {
+        m_last_was_newline = false;
+        if (auto result = try_commit_chunk(m_utf8_view.end(), true); result.has_value())
+            return result.release_value();
+    }
+    if (m_start_of_chunk != m_utf8_view.end()) {
+        if (auto result = try_commit_chunk(m_utf8_view.end(), false, true); result.has_value())
+            return result.release_value();
+    }
+
+    return {};
+}
+
+Optional<TextNode::Chunk> TextNode::ChunkIterator::try_commit_chunk(Utf8View::Iterator const& it, bool has_breaking_newline, bool must_commit)
+{
+    if (m_layout_mode == LayoutMode::OnlyRequiredLineBreaks && !must_commit)
+        return {};
+
+    auto start = m_utf8_view.byte_offset_of(m_start_of_chunk);
+    auto length = m_utf8_view.byte_offset_of(it) - m_utf8_view.byte_offset_of(m_start_of_chunk);
+
+    if (has_breaking_newline || length > 0) {
+        auto chunk_view = m_utf8_view.substring_view(start, length);
+        m_start_of_chunk = it;
+        return Chunk {
+            .view = chunk_view,
+            .start = start,
+            .length = length,
+            .has_breaking_newline = has_breaking_newline,
+            .is_all_whitespace = is_all_whitespace(chunk_view.as_string()),
+        };
+    }
+
+    m_start_of_chunk = it;
+    return {};
+}
+
 }
diff --git a/Userland/Libraries/LibWeb/Layout/TextNode.h b/Userland/Libraries/LibWeb/Layout/TextNode.h
index ded15114d32..5597f42df02 100644
--- a/Userland/Libraries/LibWeb/Layout/TextNode.h
+++ b/Userland/Libraries/LibWeb/Layout/TextNode.h
@@ -6,6 +6,7 @@
 
 #pragma once
 
+#include <AK/Utf8View.h>
 #include <LibWeb/DOM/Text.h>
 #include <LibWeb/Layout/Node.h>
 
@@ -26,6 +27,32 @@ public:
 
     virtual void split_into_lines(InlineFormattingContext&, LayoutMode) override;
 
+    struct Chunk {
+        Utf8View view;
+        size_t start { 0 };
+        size_t length { 0 };
+        bool has_breaking_newline { false };
+        bool is_all_whitespace { false };
+    };
+
+    class ChunkIterator {
+    public:
+        ChunkIterator(StringView const& text, LayoutMode, bool wrap_lines, bool wrap_breaks);
+        Optional<Chunk> next();
+
+    private:
+        Optional<Chunk> try_commit_chunk(Utf8View::Iterator const&, bool has_breaking_newline, bool must_commit = false);
+
+        const LayoutMode m_layout_mode;
+        const bool m_wrap_lines;
+        const bool m_wrap_breaks;
+        bool m_last_was_space { false };
+        bool m_last_was_newline { false };
+        Utf8View m_utf8_view;
+        Utf8View::Iterator m_start_of_chunk;
+        Utf8View::Iterator m_iterator;
+    };
+
 private:
     virtual bool is_text_node() const final { return true; }
     virtual bool wants_mouse_events() const override;
@@ -35,9 +62,6 @@ private:
     void split_into_lines_by_rules(InlineFormattingContext&, LayoutMode, bool do_collapse, bool do_wrap_lines, bool do_wrap_breaks);
     void paint_cursor_if_needed(PaintContext&, const LineBoxFragment&) const;
 
-    template<typename Callback>
-    void for_each_chunk(Callback, LayoutMode, bool do_wrap_lines, bool do_wrap_breaks) const;
-
     String m_text_for_rendering;
 };