avm2: Implement encodeURI and encodeURIComponent

core/src/avm2/globals.rs

@@ -468,6 +468,14 @@ pub fn load_player_globals<'gc>(
     function(activation, "", "parseInt", toplevel::parse_int, script)?;
     function(activation, "", "parseFloat", toplevel::parse_float, script)?;
     function(activation, "", "escape", toplevel::escape, script)?;
+    function(activation, "", "encodeURI", toplevel::encode_uri, script)?;
+    function(
+        activation,
+        "",
+        "encodeURIComponent",
+        toplevel::encode_uri_component,
+        script,
+    )?;
 
     avm2_system_class!(regexp, activation, regexp::create_class(activation), script);
     avm2_system_class!(vector, activation, vector::create_class(activation), script);

core/src/avm2/globals/toplevel.rs

@@ -1,5 +1,7 @@
 //! Global scope built-ins
 
+use ruffle_wstr::Units;
+
 use crate::avm2::activation::Activation;
 use crate::avm2::object::Object;
 use crate::avm2::value::Value;
@@ -7,6 +9,7 @@ use crate::avm2::Error;
 use crate::string::{AvmString, WStr, WString};
 use crate::stub::Stub;
 use std::borrow::Cow;
+use std::fmt::Write;
 
 pub fn trace<'gc>(
     activation: &mut Activation<'_, 'gc>,
@@ -263,3 +266,68 @@ pub fn escape<'gc>(
 
     Ok(AvmString::new(activation.context.gc_context, output).into())
 }
+
+pub fn encode_uri<'gc>(
+    activation: &mut Activation<'_, 'gc>,
+    _this: Option<Object<'gc>>,
+    args: &[Value<'gc>],
+) -> Result<Value<'gc>, Error<'gc>> {
+    encode_utf8_with_exclusions(
+        activation,
+        args,
+        // Characters that are not escaped, sourced from as3 docs
+        "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ@;/?:@&=+$,#-_.!~*'()",
+    )
+}
+
+pub fn encode_uri_component<'gc>(
+    activation: &mut Activation<'_, 'gc>,
+    _this: Option<Object<'gc>>,
+    args: &[Value<'gc>],
+) -> Result<Value<'gc>, Error<'gc>> {
+    encode_utf8_with_exclusions(
+        activation,
+        args,
+        // Characters that are not escaped, sourced from as3 docs
+        "0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-_.!~*'()",
+    )
+}
+
+fn encode_utf8_with_exclusions<'gc>(
+    activation: &mut Activation<'_, 'gc>,
+    args: &[Value<'gc>],
+    not_converted: &str,
+) -> Result<Value<'gc>, Error<'gc>> {
+    let value = match args.first() {
+        None => return Ok("undefined".into()),
+        Some(Value::Undefined) => return Ok("null".into()),
+        Some(value) => value,
+    };
+
+    let mut output = String::new();
+
+    let input = value.coerce_to_string(activation)?;
+    let input_string = match input.units() {
+        // Latin-1 values map directly to unicode codepoints,
+        // so we can directly convert to a `char`
+        Units::Bytes(bytes) => bytes.iter().map(|b| *b as char).collect(),
+        Units::Wide(wide) => String::from_utf16_lossy(wide),
+    };
+
+    for x in input_string.chars() {
+        if not_converted.contains(x) {
+            output.push(x);
+        } else {
+            let mut bytes = [0; 4];
+            let utf8_bytes = x.encode_utf8(&mut bytes);
+            let mut encoded = String::new();
+            // Each byte in the utf-8 encoding is encoded as a hex value
+            for byte in utf8_bytes.bytes() {
+                write!(encoded, "%{x:02X}", x = byte).unwrap();
+            }
+            output.push_str(&encoded);
+        }
+    }
+
+    Ok(AvmString::new_utf8(activation.context.gc_context, output).into())
+}

tests/tests/swfs/avm2/escape/Test.as

@@ -4,38 +4,44 @@
 	}
 }
 
-trace("// escape()");
-trace(escape());
-trace("");
+import flash.utils.getDefinitionByName;
 
-trace("// escape(undefined)");
-trace(escape(undefined));
-trace("");
+var fns = ["escape", "encodeURI", "encodeURIComponent"];
+for each (var fnName in fns) {
+	var fn = getDefinitionByName(fnName);
+	trace("// " + fnName + "()");
+	trace(fn());
+	trace("");
 
-trace("// typeof(escape(undefined))");
-trace(typeof(escape(undefined)));
-trace("");
+	trace("// " + fnName + "(undefined)");
+	trace(fn(undefined));
+	trace("");
 
-trace("// escape(null)");
-trace(escape(null));
-trace("");
+	trace("// typeof(" + fnName + "(undefined))");
+	trace(typeof(fn(undefined)));
+	trace("");
 
-var input = "test";
-trace("// escape(\"" + input + "\")");
-trace(escape(input));
-trace("");
+	trace("// " + fnName + "(null)");
+	trace(fn(null));
+	trace("");
 
-var input = "!\"£$%^&*()1234567890qwertyuiop[]asdfghjkl;'#\zxcvbnm,./QWERTYUIOP{}ASDFGHJKL:@~|ZXCVBNM<>?\u0010";
-trace("// escape(\"" + input + "\")");
-trace(escape(input));
-trace("");
+	var input = "test";
+	trace("// " + fnName + "(\"" + input + "\")");
+	trace(fn(input));
+	trace("");
 
-var input = "\x05";
-trace("// escape(\"\\x05\")");
-trace(escape(input));
-trace("");
+	var input = "!\"£$%^&*()1234567890qwertyuiop[]asdfghjkl;'#\zxcvbnm,./QWERTYUIOP{}ASDFGHJKL:@~|ZXCVBNM<>?\u0010";
+	trace("// " + fnName + "(\"" + input + "\")");
+	trace(fn(input));
+	trace("");
 
-var input = "😭";
-trace("// escape(\"" + input + "\")");
-trace(escape(input));
-trace("");
+	var input = "\x05";
+	trace("// " + fnName + "(\"\\x05\")");
+	trace(fn(input));
+	trace("");
+
+	var input = "😭";
+	trace("// " + fnName + "(\"" + input + "\")");
+	trace(fn(input));
+	trace("");	
+}

tests/tests/swfs/avm2/escape/output.txt

@@ -22,3 +22,51 @@ test
 // escape("😭")
 %uD83D%uDE2D
 
+// encodeURI()
+undefined
+
+// encodeURI(undefined)
+null
+
+// typeof(encodeURI(undefined))
+string
+
+// encodeURI(null)
+null
+
+// encodeURI("test")
+test
+
+// encodeURI("!"£$%^&*()1234567890qwertyuiop[]asdfghjkl;'#zxcvbnm,./QWERTYUIOP{}ASDFGHJKL:@~|ZXCVBNM<>?")
+!%22%C2%A3$%25%5E&*()1234567890qwertyuiop%5B%5Dasdfghjkl;'#zxcvbnm,./QWERTYUIOP%7B%7DASDFGHJKL:@~%7CZXCVBNM%3C%3E?%10
+
+// encodeURI("\x05")
+%05
+
+// encodeURI("😭")
+%F0%9F%98%AD
+
+// encodeURIComponent()
+undefined
+
+// encodeURIComponent(undefined)
+null
+
+// typeof(encodeURIComponent(undefined))
+string
+
+// encodeURIComponent(null)
+null
+
+// encodeURIComponent("test")
+test
+
+// encodeURIComponent("!"£$%^&*()1234567890qwertyuiop[]asdfghjkl;'#zxcvbnm,./QWERTYUIOP{}ASDFGHJKL:@~|ZXCVBNM<>?")
+!%22%C2%A3%24%25%5E%26*()1234567890qwertyuiop%5B%5Dasdfghjkl%3B'%23zxcvbnm%2C.%2FQWERTYUIOP%7B%7DASDFGHJKL%3A%40~%7CZXCVBNM%3C%3E%3F%10
+
+// encodeURIComponent("\x05")
+%05
+
+// encodeURIComponent("😭")
+%F0%9F%98%AD
+