From 7760c007140f8acbd6f214898a7c8e7a565f4875 Mon Sep 17 00:00:00 2001
From: Linus Groh <mail@linusgroh.de>
Date: Sat, 1 Oct 2022 18:39:40 +0100
Subject: [PATCH] LibWeb: Move strip_and_collapse_whitespace() to Infra/

...and make it spec compliant by considering all ASCII whitespace,
greatly simplifying it in the process :^)
---
 Userland/Libraries/LibWeb/CMakeLists.txt      |  1 +
 .../LibWeb/HTML/HTMLOptionElement.cpp         | 30 ++---------------
 Userland/Libraries/LibWeb/Infra/Strings.cpp   | 32 +++++++++++++++++++
 Userland/Libraries/LibWeb/Infra/Strings.h     | 15 +++++++++
 4 files changed, 50 insertions(+), 28 deletions(-)
 create mode 100644 Userland/Libraries/LibWeb/Infra/Strings.cpp
 create mode 100644 Userland/Libraries/LibWeb/Infra/Strings.h

diff --git a/Userland/Libraries/LibWeb/CMakeLists.txt b/Userland/Libraries/LibWeb/CMakeLists.txt
index 867988cc29..8d5ae47578 100644
--- a/Userland/Libraries/LibWeb/CMakeLists.txt
+++ b/Userland/Libraries/LibWeb/CMakeLists.txt
@@ -277,6 +277,7 @@ set(SOURCES
     HighResolutionTime/Performance.cpp
     Infra/ByteSequences.cpp
     Infra/JSON.cpp
+    Infra/Strings.cpp
     IntersectionObserver/IntersectionObserver.cpp
     Layout/AvailableSpace.cpp
     Layout/BlockContainer.cpp
diff --git a/Userland/Libraries/LibWeb/HTML/HTMLOptionElement.cpp b/Userland/Libraries/LibWeb/HTML/HTMLOptionElement.cpp
index 30022010d0..270337d22d 100644
--- a/Userland/Libraries/LibWeb/HTML/HTMLOptionElement.cpp
+++ b/Userland/Libraries/LibWeb/HTML/HTMLOptionElement.cpp
@@ -13,8 +13,7 @@
 #include <LibWeb/HTML/HTMLOptionElement.h>
 #include <LibWeb/HTML/HTMLScriptElement.h>
 #include <LibWeb/HTML/HTMLSelectElement.h>
-#include <LibWeb/Infra/CharacterTypes.h>
-#include <ctype.h>
+#include <LibWeb/Infra/Strings.h>
 
 namespace Web::HTML {
 
@@ -76,31 +75,6 @@ void HTMLOptionElement::set_value(String value)
     set_attribute(HTML::AttributeNames::value, value);
 }
 
-// https://infra.spec.whatwg.org/#strip-and-collapse-ascii-whitespace
-static String strip_and_collapse_whitespace(StringView string)
-{
-    // Replace any sequence of one or more consecutive code points that are ASCII whitespace in the string with a single U+0020 SPACE code point.
-    StringBuilder builder;
-    for (size_t i = 0; i < string.length(); ++i) {
-        if (isspace(string[i])) {
-            builder.append(' ');
-            while (i < string.length()) {
-                if (isspace(string[i])) {
-                    ++i;
-                    continue;
-                }
-                builder.append(string[i]);
-                break;
-            }
-            continue;
-        }
-        builder.append(string[i]);
-    }
-
-    // ...and then remove any leading and trailing ASCII whitespace from that string.
-    return builder.to_string().trim(Infra::ASCII_WHITESPACE);
-}
-
 static void concatenate_descendants_text_content(DOM::Node const* node, StringBuilder& builder)
 {
     // FIXME: SVGScriptElement should also be skipped, but it doesn't exist yet.
@@ -126,7 +100,7 @@ String HTMLOptionElement::text() const
     });
 
     // Return the result of stripping and collapsing ASCII whitespace from the above concatenation.
-    return strip_and_collapse_whitespace(builder.to_string());
+    return Infra::strip_and_collapse_whitespace(builder.string_view());
 }
 
 // https://html.spec.whatwg.org/multipage/form-elements.html#dom-option-text
diff --git a/Userland/Libraries/LibWeb/Infra/Strings.cpp b/Userland/Libraries/LibWeb/Infra/Strings.cpp
new file mode 100644
index 0000000000..3b415bd6d3
--- /dev/null
+++ b/Userland/Libraries/LibWeb/Infra/Strings.cpp
@@ -0,0 +1,32 @@
+/*
+ * Copyright (c) 2022, Linus Groh <linusg@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <AK/String.h>
+#include <AK/Utf8View.h>
+#include <LibWeb/Infra/CharacterTypes.h>
+#include <LibWeb/Infra/Strings.h>
+
+namespace Web::Infra {
+
+// https://infra.spec.whatwg.org/#strip-and-collapse-ascii-whitespace
+String strip_and_collapse_whitespace(StringView string)
+{
+    // Replace any sequence of one or more consecutive code points that are ASCII whitespace in the string with a single U+0020 SPACE code point.
+    StringBuilder builder;
+    for (auto code_point : Utf8View { string }) {
+        if (Infra::is_ascii_whitespace(code_point)) {
+            if (!builder.string_view().ends_with(' '))
+                builder.append(' ');
+            continue;
+        }
+        builder.append_code_point(code_point);
+    }
+
+    // ...and then remove any leading and trailing ASCII whitespace from that string.
+    return builder.string_view().trim(Infra::ASCII_WHITESPACE);
+}
+
+}
diff --git a/Userland/Libraries/LibWeb/Infra/Strings.h b/Userland/Libraries/LibWeb/Infra/Strings.h
new file mode 100644
index 0000000000..f73b435519
--- /dev/null
+++ b/Userland/Libraries/LibWeb/Infra/Strings.h
@@ -0,0 +1,15 @@
+/*
+ * Copyright (c) 2022, Linus Groh <linusg@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/Forward.h>
+
+namespace Web::Infra {
+
+String strip_and_collapse_whitespace(StringView string);
+
+}