implement from_utf16_lossy

* add proptests
* add doc tests
* add to fuzzing harness

compact_str/src/lib.rs

@@ -332,6 +332,36 @@ impl CompactString {
         Ok(ret)
     }
 
+    /// Decode a UTF-16–encoded slice `v` into a `CompactString`, replacing invalid data with 
+    /// the replacement character (`U+FFFD`), �.
+    ///
+    /// # Examples
+    ///
+    /// Basic usage:
+    ///
+    /// ```
+    /// # use compact_str::CompactString;
+    /// // 𝄞mus<invalid>ic<invalid>
+    /// let v = &[0xD834, 0xDD1E, 0x006d, 0x0075,
+    ///           0x0073, 0xDD1E, 0x0069, 0x0063,
+    ///           0xD834];
+    ///
+    /// assert_eq!(CompactString::from("𝄞mus\u{FFFD}ic\u{FFFD}"),
+    ///            CompactString::from_utf16_lossy(v));
+    /// ```
+    #[inline]
+    pub fn from_utf16_lossy<B: AsRef<[u16]>>(buf: B) -> Self {
+        let buf = buf.as_ref();
+        let mut ret = CompactString::with_capacity(buf.len());
+        for c in std::char::decode_utf16(buf.iter().copied()) {
+            match c {
+                Ok(c) => ret.push(c),
+                Err(_) => ret.push_str("�"),
+            }
+        }
+        ret
+    }
+
     /// Returns the length of the [`CompactString`] in `bytes`, not [`char`]s or graphemes.
     ///
     /// When using `UTF-8` encoding (which all strings in Rust do) a single character will be 1 to 4

compact_str/src/tests.rs

@@ -259,6 +259,23 @@ fn proptest_from_utf16_random(#[strategy(rand_u16s())] buf: Vec<u16>) {
     }
 }
 
+#[proptest]
+#[cfg_attr(miri, ignore)]
+fn proptest_from_utf16_lossy_roundtrips(#[strategy(rand_unicode())] control: String) {
+    let utf16_buf: Vec<u16> = control.encode_utf16().collect();
+    let compact = CompactString::from_utf16_lossy(&utf16_buf);
+
+    assert_eq!(compact, control);
+}
+
+#[proptest]
+#[cfg_attr(miri, ignore)]
+fn proptest_from_utf16_lossy_random(#[strategy(rand_u16s())] buf: Vec<u16>) {
+    let control = String::from_utf16_lossy(&buf);
+    let compact = CompactString::from_utf16_lossy(&buf);
+    assert_eq!(compact, control);
+}
+
 #[proptest]
 #[cfg_attr(miri, ignore)]
 fn proptest_remove(#[strategy(rand_unicode_with_range(1..80))] mut control: String, val: u8) {

fuzz/src/creation.rs

@@ -26,6 +26,8 @@ pub enum Creation<'a> {
     BytesUnchecked(&'a [u8]),
     /// Create using [`CompactString::from_utf16`]
     BytesUtf16(Vec<u16>),
+    /// Create using [`CompactString::from_utf16_lossy`]
+    BytesUtf16Lossy(Vec<u16>),
     /// Create using [`CompactString::from_utf8_buf`]
     Buf(&'a [u8]),
     /// Create using [`CompactString::from_utf8_buf_unchecked`]
@@ -342,6 +344,13 @@ impl Creation<'_> {
                     _ => panic!("CompactString and String read UTF-16 differently?"),
                 }
             }
+            BytesUtf16Lossy(data) => {
+                let compact = CompactString::from_utf16_lossy(&data);
+                let std_str = String::from_utf16_lossy(&data);
+
+                assert_eq!(compact, std_str);
+                Some((compact, std_str))
+            }
             Buf(data) => {
                 let mut buffer = Cursor::new(data);