AK+Everywhere: Make UTF-16 to UTF-8 converter fallible

This could fail to allocate the underlying storage needed to store the
UTF-8 data. Propagate this error.

10 files changed, 25 insertions, 22 deletions

diff --git a/AK/Utf16View.cpp b/AK/Utf16View.cpp
index 4c23484bee..2c049f5e39 100644
--- a/AK/Utf16View.cpp
+++ b/AK/Utf16View.cpp
@@ -81,7 +81,7 @@ u32 Utf16View::decode_surrogate_pair(u16 high_surrogate, u16 low_surrogate)
     return ((high_surrogate - high_surrogate_min) << 10) + (low_surrogate - low_surrogate_min) + first_supplementary_plane_code_point;
 }
 
-DeprecatedString Utf16View::to_utf8(AllowInvalidCodeUnits allow_invalid_code_units) const
+ErrorOr<DeprecatedString> Utf16View::to_utf8(AllowInvalidCodeUnits allow_invalid_code_units) const
 {
     StringBuilder builder;
 
@@ -92,17 +92,17 @@ DeprecatedString Utf16View::to_utf8(AllowInvalidCodeUnits allow_invalid_code_uni
 
                 if ((next < end_ptr()) && is_low_surrogate(*next)) {
                     auto code_point = decode_surrogate_pair(*ptr, *next);
-                    builder.append_code_point(code_point);
+                    TRY(builder.try_append_code_point(code_point));
                     ++ptr;
                     continue;
                 }
             }
 
-            builder.append_code_point(static_cast<u32>(*ptr));
+            TRY(builder.try_append_code_point(static_cast<u32>(*ptr)));
         }
     } else {
         for (auto code_point : *this)
-            builder.append_code_point(code_point);
+            TRY(builder.try_append_code_point(code_point));
     }
 
     return builder.build();
diff --git a/AK/Utf16View.h b/AK/Utf16View.h
index 101f4e4892..1a46546da1 100644
--- a/AK/Utf16View.h
+++ b/AK/Utf16View.h
@@ -75,7 +75,7 @@ public:
         No,
     };
 
-    DeprecatedString to_utf8(AllowInvalidCodeUnits = AllowInvalidCodeUnits::No) const;
+    ErrorOr<DeprecatedString> to_utf8(AllowInvalidCodeUnits = AllowInvalidCodeUnits::No) const;
 
     bool is_null() const { return m_code_units.is_null(); }
     bool is_empty() const { return m_code_units.is_empty(); }
diff --git a/Tests/AK/TestUtf16.cpp b/Tests/AK/TestUtf16.cpp
index 56b2057e17..174c959e7a 100644
--- a/Tests/AK/TestUtf16.cpp
+++ b/Tests/AK/TestUtf16.cpp
@@ -56,14 +56,14 @@ TEST_CASE(encode_utf8)
         DeprecatedString utf8_string("Привет, мир! 😀 γειά σου κόσμος こんにちは世界");
         auto string = MUST(AK::utf8_to_utf16(utf8_string));
         Utf16View view { string };
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), utf8_string);
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), utf8_string);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes)), utf8_string);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No)), utf8_string);
     }
     {
         auto encoded = Array { (u16)0xd83d };
         Utf16View view { encoded };
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), "\xed\xa0\xbd"sv);
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), "\ufffd"sv);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes)), "\xed\xa0\xbd"sv);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No)), "\ufffd"sv);
     }
 }
 
@@ -269,14 +269,14 @@ TEST_CASE(substring_view)
         view = view.substring_view(7, 2);
 
         EXPECT(view.length_in_code_units() == 2);
-        EXPECT_EQ(view.to_utf8(), "😀"sv);
+        EXPECT_EQ(MUST(view.to_utf8()), "😀"sv);
     }
     {
         Utf16View view { string };
         view = view.substring_view(7, 1);
 
         EXPECT(view.length_in_code_units() == 1);
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes), "\xed\xa0\xbd"sv);
-        EXPECT_EQ(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No), "\ufffd"sv);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes)), "\xed\xa0\xbd"sv);
+        EXPECT_EQ(MUST(view.to_utf8(Utf16View::AllowInvalidCodeUnits::No)), "\ufffd"sv);
     }
 }
diff --git a/Userland/Applications/HexEditor/HexEditorWidget.cpp b/Userland/Applications/HexEditor/HexEditorWidget.cpp
index 0e363be188..122d377020 100644
--- a/Userland/Applications/HexEditor/HexEditorWidget.cpp
+++ b/Userland/Applications/HexEditor/HexEditorWidget.cpp
@@ -372,7 +372,7 @@ void HexEditorWidget::update_inspector_values(size_t position)
         if (valid_code_units == 0)
             value_inspector_model->set_parsed_value(ValueInspectorModel::ValueType::UTF16, "");
         else
-            value_inspector_model->set_parsed_value(ValueInspectorModel::ValueType::UTF16, utf16_view.unicode_substring_view(0, 1).to_utf8());
+            value_inspector_model->set_parsed_value(ValueInspectorModel::ValueType::UTF16, utf16_view.unicode_substring_view(0, 1).to_utf8().release_value_but_fixme_should_propagate_errors());
     } else {
         value_inspector_model->set_parsed_value(ValueInspectorModel::ValueType::UTF16, "");
     }
diff --git a/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp b/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
index ddf89aee18..cef3472432 100644
--- a/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
+++ b/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
@@ -1265,7 +1265,7 @@ ThrowCompletionOr<DeprecatedString> get_substitution(VM& vm, Utf16View const& ma
         } else if (is_ascii_digit(next)) {
             bool is_two_digits = (i + 2 < replace_view.length_in_code_units()) && is_ascii_digit(replace_view.code_unit_at(i + 2));
 
-            auto capture_position_string = replace_view.substring_view(i + 1, is_two_digits ? 2 : 1).to_utf8();
+            auto capture_position_string = TRY_OR_THROW_OOM(vm, replace_view.substring_view(i + 1, is_two_digits ? 2 : 1).to_utf8());
             auto capture_position = capture_position_string.to_uint();
 
             if (capture_position.has_value() && (*capture_position > 0) && (*capture_position <= captures.size())) {
@@ -1295,7 +1295,7 @@ ThrowCompletionOr<DeprecatedString> get_substitution(VM& vm, Utf16View const& ma
                 result.append(curr);
             } else {
                 auto group_name_view = replace_view.substring_view(start_position, *end_position - start_position);
-                auto group_name = group_name_view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes);
+                auto group_name = TRY_OR_THROW_OOM(vm, group_name_view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes));
 
                 auto capture = TRY(named_captures.as_object().get(group_name));
 
@@ -1311,7 +1311,7 @@ ThrowCompletionOr<DeprecatedString> get_substitution(VM& vm, Utf16View const& ma
         }
     }
 
-    return Utf16String(move(result)).to_utf8();
+    return TRY_OR_THROW_OOM(vm, Utf16View { result }.to_utf8());
 }
 
 }
diff --git a/Userland/Libraries/LibJS/Runtime/Intl/DateTimeFormat.cpp b/Userland/Libraries/LibJS/Runtime/Intl/DateTimeFormat.cpp
index 57973e4711..8f1f527b77 100644
--- a/Userland/Libraries/LibJS/Runtime/Intl/DateTimeFormat.cpp
+++ b/Userland/Libraries/LibJS/Runtime/Intl/DateTimeFormat.cpp
@@ -718,7 +718,7 @@ ThrowCompletionOr<Vector<PatternPartition>> format_date_time_pattern(VM& vm, Dat
                 if (formatted_value.length() > 2) {
                     Utf16String utf16_formatted_value { formatted_value };
                     if (utf16_formatted_value.length_in_code_units() > 2)
-                        formatted_value = utf16_formatted_value.substring_view(utf16_formatted_value.length_in_code_units() - 2).to_utf8();
+                        formatted_value = TRY_OR_THROW_OOM(vm, utf16_formatted_value.substring_view(utf16_formatted_value.length_in_code_units() - 2).to_utf8());
                 }
 
                 break;
diff --git a/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp b/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
index c29c4b0cc8..3d4f87de34 100644
--- a/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
+++ b/Userland/Libraries/LibJS/Runtime/PrimitiveString.cpp
@@ -68,7 +68,8 @@ DeprecatedString const& PrimitiveString::deprecated_string() const
 {
     resolve_rope_if_needed();
     if (!m_has_utf8_string) {
-        m_utf8_string = m_utf16_string.to_utf8();
+        // FIXME: Propagate this error.
+        m_utf8_string = MUST(m_utf16_string.to_utf8(vm()));
         m_has_utf8_string = true;
     }
     return m_utf8_string;
diff --git a/Userland/Libraries/LibJS/Runtime/Utf16String.cpp b/Userland/Libraries/LibJS/Runtime/Utf16String.cpp
index e6a4b4caf2..098d761e67 100644
--- a/Userland/Libraries/LibJS/Runtime/Utf16String.cpp
+++ b/Userland/Libraries/LibJS/Runtime/Utf16String.cpp
@@ -6,6 +6,7 @@
 
 #include <AK/StringView.h>
 #include <LibJS/Runtime/Utf16String.h>
+#include <LibJS/Runtime/VM.h>
 
 namespace JS {
 namespace Detail {
@@ -96,9 +97,9 @@ Utf16View Utf16String::substring_view(size_t code_unit_offset) const
     return view().substring_view(code_unit_offset);
 }
 
-DeprecatedString Utf16String::to_utf8() const
+ThrowCompletionOr<DeprecatedString> Utf16String::to_utf8(VM& vm) const
 {
-    return view().to_utf8(Utf16View::AllowInvalidCodeUnits::Yes);
+    return TRY_OR_THROW_OOM(vm, view().to_utf8(Utf16View::AllowInvalidCodeUnits::Yes));
 }
 
 u16 Utf16String::code_unit_at(size_t index) const
diff --git a/Userland/Libraries/LibJS/Runtime/Utf16String.h b/Userland/Libraries/LibJS/Runtime/Utf16String.h
index fdcc014252..4b101a8261 100644
--- a/Userland/Libraries/LibJS/Runtime/Utf16String.h
+++ b/Userland/Libraries/LibJS/Runtime/Utf16String.h
@@ -12,6 +12,7 @@
 #include <AK/Types.h>
 #include <AK/Utf16View.h>
 #include <AK/Vector.h>
+#include <LibJS/Runtime/Completion.h>
 
 namespace JS {
 namespace Detail {
@@ -49,7 +50,7 @@ public:
     Utf16View substring_view(size_t code_unit_offset, size_t code_unit_length) const;
     Utf16View substring_view(size_t code_unit_offset) const;
 
-    DeprecatedString to_utf8() const;
+    ThrowCompletionOr<DeprecatedString> to_utf8(VM&) const;
     u16 code_unit_at(size_t index) const;
 
     size_t length_in_code_units() const;
diff --git a/Userland/Libraries/LibRegex/RegexMatch.h b/Userland/Libraries/LibRegex/RegexMatch.h
index 9943d55909..47bcf7f39d 100644
--- a/Userland/Libraries/LibRegex/RegexMatch.h
+++ b/Userland/Libraries/LibRegex/RegexMatch.h
@@ -385,7 +385,7 @@ public:
     {
         return m_view.visit(
             [](StringView view) { return view.to_deprecated_string(); },
-            [](Utf16View view) { return view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes); },
+            [](Utf16View view) { return view.to_utf8(Utf16View::AllowInvalidCodeUnits::Yes).release_value_but_fixme_should_propagate_errors(); },
             [](auto& view) {
                 StringBuilder builder;
                 for (auto it = view.begin(); it != view.end(); ++it)