LibJS: Implement and use the GetSubstitution abstract operation

Used by String.prototype.replace, String.prototype.replaceAll, and
RegExp.prototype [ @@replace ].

5 files changed, 129 insertions, 6 deletions

diff --git a/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp b/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
index 0f73cef6a7..cfa0d90eaf 100644
--- a/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
+++ b/Userland/Libraries/LibJS/Runtime/AbstractOperations.cpp
@@ -5,6 +5,7 @@
  * SPDX-License-Identifier: BSD-2-Clause
  */
 
+#include <AK/CharacterTypes.h>
 #include <AK/Function.h>
 #include <AK/Optional.h>
 #include <AK/Result.h>
@@ -574,4 +575,91 @@ Value canonical_numeric_index_string(GlobalObject& global_object, PropertyName c
     return n;
 }
 
+// 22.1.3.17.1 GetSubstitution ( matched, str, position, captures, namedCaptures, replacement ), https://tc39.es/ecma262/#sec-getsubstitution
+String get_substitution(GlobalObject& global_object, String const& matched, String const& str, size_t position, Vector<Value> const& captures, Value named_captures, Value replacement)
+{
+    auto& vm = global_object.vm();
+
+    auto replace_string = replacement.to_string(global_object);
+    if (vm.exception())
+        return {};
+
+    StringBuilder result;
+
+    for (size_t i = 0; i < replace_string.length(); ++i) {
+        char curr = replace_string[i];
+
+        if ((curr != '$') || (i + 1 >= replace_string.length())) {
+            result.append(curr);
+            continue;
+        }
+
+        char next = replace_string[i + 1];
+
+        if (next == '$') {
+            result.append(next);
+            ++i;
+        } else if (next == '&') {
+            result.append(matched);
+            ++i;
+        } else if (next == '`') {
+            result.append(str.substring_view(0, position));
+            ++i;
+        } else if (next == '\'') {
+            auto tail_pos = position + matched.length();
+            if (tail_pos < str.length())
+                result.append(str.substring_view(tail_pos));
+            ++i;
+        } else if (is_ascii_digit(next)) {
+            bool is_two_digits = (i + 2 < replace_string.length()) && is_ascii_digit(replace_string[i + 2]);
+
+            auto capture_postition_string = replace_string.substring_view(i + 1, is_two_digits ? 2 : 1);
+            auto capture_position = capture_postition_string.to_uint();
+
+            if (capture_position.has_value() && (*capture_position > 0) && (*capture_position <= captures.size())) {
+                auto& value = captures[*capture_position - 1];
+
+                if (!value.is_undefined()) {
+                    auto value_string = value.to_string(global_object);
+                    if (vm.exception())
+                        return {};
+
+                    result.append(value_string);
+                }
+
+                i += is_two_digits ? 2 : 1;
+            } else {
+                result.append(curr);
+            }
+        } else if (next == '<') {
+            auto start_position = i + 2;
+            auto end_position = replace_string.find('>', start_position);
+
+            if (named_captures.is_undefined() || !end_position.has_value()) {
+                result.append(curr);
+            } else {
+                auto group_name = replace_string.substring(start_position, *end_position - start_position);
+
+                auto capture = named_captures.as_object().get(group_name);
+                if (vm.exception())
+                    return {};
+
+                if (!capture.is_undefined()) {
+                    auto capture_string = capture.to_string(global_object);
+                    if (vm.exception())
+                        return {};
+
+                    result.append(capture_string);
+                }
+
+                i = *end_position;
+            }
+        } else {
+            result.append(curr);
+        }
+    }
+
+    return result.build();
+}
+
 }
diff --git a/Userland/Libraries/LibJS/Runtime/AbstractOperations.h b/Userland/Libraries/LibJS/Runtime/AbstractOperations.h
index 3c7ecdbb68..f70f51654b 100644
--- a/Userland/Libraries/LibJS/Runtime/AbstractOperations.h
+++ b/Userland/Libraries/LibJS/Runtime/AbstractOperations.h
@@ -30,6 +30,7 @@ Object* get_prototype_from_constructor(GlobalObject&, FunctionObject const& cons
 Object* create_unmapped_arguments_object(GlobalObject&, Vector<Value> const& arguments);
 Object* create_mapped_arguments_object(GlobalObject&, FunctionObject&, Vector<FunctionNode::Parameter> const&, Vector<Value> const& arguments, Environment&);
 Value canonical_numeric_index_string(GlobalObject&, PropertyName const&);
+String get_substitution(GlobalObject&, String const& matched, String const& str, size_t position, Vector<Value> const& captures, Value named_captures, Value replacement);
 
 enum class CallerMode {
     Strict,
diff --git a/Userland/Libraries/LibJS/Runtime/RegExpPrototype.cpp b/Userland/Libraries/LibJS/Runtime/RegExpPrototype.cpp
index 4c9759fd6e..29d20760d0 100644
--- a/Userland/Libraries/LibJS/Runtime/RegExpPrototype.cpp
+++ b/Userland/Libraries/LibJS/Runtime/RegExpPrototype.cpp
@@ -399,8 +399,7 @@ JS_DEFINE_NATIVE_FUNCTION(RegExpPrototype::symbol_replace)
             if (vm.exception())
                 return {};
         } else {
-            // FIXME: Implement the GetSubstituion algorithm for substituting placeholder '$' characters - https://tc39.es/ecma262/#sec-getsubstitution
-            replacement = replace_value.to_string(global_object);
+            replacement = get_substitution(global_object, matched, string, position, captures, named_captures, replace_value);
             if (vm.exception())
                 return {};
         }
diff --git a/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp b/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
index a7a3510400..2e1b423cae 100644
--- a/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
+++ b/Userland/Libraries/LibJS/Runtime/StringPrototype.cpp
@@ -822,8 +822,7 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::replace)
         if (vm.exception())
             return {};
     } else {
-        // FIXME: Implement the GetSubstituion algorithm for substituting placeholder '$' characters - https://tc39.es/ecma262/#sec-getsubstitution
-        replacement = replace_value.to_string(global_object);
+        replacement = get_substitution(global_object, search_string, string, *position, {}, js_undefined(), replace_value);
         if (vm.exception())
             return {};
     }
@@ -902,8 +901,7 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::replace_all)
             if (vm.exception())
                 return {};
         } else {
-            // FIXME: Implement the GetSubstituion algorithm for substituting placeholder '$' characters - https://tc39.es/ecma262/#sec-getsubstitution
-            replacement = replace_value.to_string(global_object);
+            replacement = get_substitution(global_object, search_string, string, position, {}, js_undefined(), replace_value);
             if (vm.exception())
                 return {};
         }
diff --git a/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replace.js b/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replace.js
index 5cce84c653..a570a68ad0 100644
--- a/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replace.js
+++ b/Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.replace.js
@@ -102,3 +102,40 @@ test("functional regex replacement", () => {
         })
     ).toBe("xd");
 });
+
+test("replacement with substitution", () => {
+    expect("abc".replace("b", "$")).toBe("a$c");
+    expect("abc".replace("b", "$.")).toBe("a$.c");
+
+    expect("abc".replace("b", "$$")).toBe("a$c");
+    expect("abc".replace("b", ">$$<")).toBe("a>$<c");
+    expect("abc".replace("b", "$$$$")).toBe("a$$c");
+
+    expect("abc".replace("b", "$&")).toBe("abc");
+    expect("a123c".replace(/\d+/, "$&")).toBe("a123c");
+
+    expect("abc".replace("b", "$`")).toBe("aac");
+    expect("aabc".replace("b", "$`")).toBe("aaaac");
+    expect("a123c".replace(/\d+/, "$`")).toBe("aac");
+
+    expect("abc".replace("b", "$'")).toBe("acc");
+    expect("abcc".replace("b", "$'")).toBe("acccc");
+    expect("a123c".replace(/\d+/, "$'")).toBe("acc");
+
+    expect("abc".replace("b", "$0")).toBe("a$0c");
+    expect("abc".replace("b", "$99")).toBe("a$99c");
+    expect("abc".replace("b", "$100")).toBe("a$100c");
+    expect("abc".replace(/(a)b(c)/, "$0")).toBe("$0");
+    expect("abc".replace(/(a)b(c)/, "$1")).toBe("a");
+    expect("abc".replace(/(a)b(c)/, "$2")).toBe("c");
+    expect("abc".replace(/(a)b(c)/, "$3")).toBe("$3");
+    expect("abc".replace(/(a)b(c)/, "$2b$1")).toBe("cba");
+
+    expect("abc".replace("b", "$<val>")).toBe("a$<val>c");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<")).toBe("$<");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<not_terminated")).toBe("$<not_terminated");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<not_found>")).toBe("");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<val1>")).toBe("a");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<val2>")).toBe("c");
+    expect("abc".replace(/(?<val1>a)b(?<val2>c)/, "$<val2>b$<val1>")).toBe("cba");
+});