Merge pull request #751 from phip1611/str

cstr[ing]16: convenience functions

CHANGELOG.md

@@ -6,6 +6,19 @@
 
 - There is a new `fs` module that provides a high-level API for file-system
   access. The API is close to the `std::fs` module.
+- Multiple convenience methods for `CString16` and `CStr16`, including:
+  - `CStr16::as_slice()`
+  - `CStr16::num_chars()`
+  - `CStr16::is_empty()`
+  - `CString16::new()`
+  - `CString16::is_empty()`
+  - `CString16::num_chars()`
+  - `CString16::replace_char()`
+  - `CString16::push()`
+  - `CString16::push_str()`
+  - `From<&CStr16>` for `CString16`
+  - `From<&CStr16>` for `String`
+  - `From<&CString16>` for `String`
 
 ### Changed
 
@@ -16,6 +29,7 @@
 - `Error::new` and `Error::from` now panic if the status is `SUCCESS`.
 - `Image::get_image_file_system` now returns a `fs::FileSystem` instead of the
   protocol.
+- `CString16::default` now always contains a null character.
 
 ## uefi-macros - [Unreleased]
 

uefi/src/data_types/chars.rs

@@ -65,6 +65,17 @@ pub const NUL_8: Char8 = Char8(0);
 #[repr(transparent)]
 pub struct Char16(u16);
 
+impl Char16 {
+    /// Creates a UCS-2 character from a Rust character without checks.
+    ///
+    /// # Safety
+    /// The caller must be sure that the character is valid.
+    #[must_use]
+    pub const unsafe fn from_u16_unchecked(val: u16) -> Self {
+        Self(val)
+    }
+}
+
 impl TryFrom<char> for Char16 {
     type Error = CharConversionError;
 
@@ -125,4 +136,4 @@ impl fmt::Display for Char16 {
 }
 
 /// UCS-2 version of the NUL character
-pub const NUL_16: Char16 = Char16(0);
+pub const NUL_16: Char16 = unsafe { Char16::from_u16_unchecked(0) };

uefi/src/data_types/owned_strs.rs

@@ -4,6 +4,8 @@ use crate::data_types::strs::EqStrUntilNul;
 use crate::data_types::UnalignedSlice;
 use crate::polyfill::vec_into_raw_parts;
 use alloc::borrow::{Borrow, ToOwned};
+use alloc::string::String;
+use alloc::vec;
 use alloc::vec::Vec;
 use core::{fmt, ops};
 
@@ -47,15 +49,79 @@ impl core::error::Error for FromStrError {}
 /// let s = CString16::try_from("abc").unwrap();
 /// assert_eq!(s.to_string(), "abc");
 /// ```
-#[derive(Clone, Debug, Default, Eq, PartialEq, Ord, PartialOrd)]
+#[derive(Clone, Debug, Eq, PartialEq, Ord, PartialOrd)]
 pub struct CString16(Vec<Char16>);
 
+impl CString16 {
+    /// Creates a new empty string with a terminating null character.
+    #[must_use]
+    pub fn new() -> Self {
+        Self(vec![NUL_16])
+    }
+
+    /// Inserts a character at the end of the string, right before the null
+    /// character.
+    ///
+    /// # Panics
+    /// Panics if the char is a null character.
+    pub fn push(&mut self, char: Char16) {
+        assert_ne!(char, NUL_16, "Pushing a null-character is illegal");
+        let last_elem = self
+            .0
+            .last_mut()
+            .expect("There should be at least a null character");
+        *last_elem = char;
+        self.0.push(NUL_16);
+    }
+
+    /// Extends the string with the given [`CStr16`]. The null character is
+    /// automatically kept at the end.
+    pub fn push_str(&mut self, str: &CStr16) {
+        str.as_slice()
+            .iter()
+            .copied()
+            .for_each(|char| self.push(char));
+    }
+
+    /// Replaces all chars in the string with the replace value in-place.
+    pub fn replace_char(&mut self, search: Char16, replace: Char16) {
+        assert_ne!(search, NUL_16, "Replacing a null character is illegal");
+        assert_ne!(
+            replace, NUL_16,
+            "Replacing with a null character is illegal"
+        );
+        self.0
+            .as_mut_slice()
+            .iter_mut()
+            .filter(|char| **char == search)
+            .for_each(|char| *char = replace);
+    }
+
+    /// Returns the number of characters without the trailing null character.
+    #[must_use]
+    pub fn num_chars(&self) -> usize {
+        self.0.len() - 1
+    }
+
+    /// Returns if the string is empty. This ignores the null character.
+    #[must_use]
+    pub fn is_empty(&self) -> bool {
+        self.num_chars() == 0
+    }
+}
+
+impl Default for CString16 {
+    fn default() -> Self {
+        CString16::new()
+    }
+}
+
 impl TryFrom<&str> for CString16 {
     type Error = FromStrError;
 
     fn try_from(input: &str) -> Result<Self, Self::Error> {
         // Initially allocate one Char16 for each byte of the input, plus
-        // one for the null byte. This should be a good guess for ASCII-ish
+        // one for the null character. This should be a good guess for ASCII-ish
         // input.
         let mut output = Vec::with_capacity(input.len() + 1);
 
@@ -112,6 +178,20 @@ impl<'a> TryFrom<&UnalignedSlice<'a, u16>> for CString16 {
     }
 }
 
+impl From<&CStr16> for CString16 {
+    fn from(value: &CStr16) -> Self {
+        let vec = value.as_slice_with_nul().to_vec();
+        Self(vec)
+    }
+}
+
+impl From<&CString16> for String {
+    fn from(value: &CString16) -> Self {
+        let slice: &CStr16 = value.as_ref();
+        String::from(slice)
+    }
+}
+
 impl<'a> UnalignedSlice<'a, u16> {
     /// Copies `self` to a new [`CString16`].
     pub fn to_cstring16(&self) -> Result<CString16, FromSliceWithNulError> {
@@ -256,4 +336,45 @@ mod tests {
             ]
         );
     }
+
+    /// This tests the following UCS-2 string functions:
+    /// - runtime constructor
+    /// - len()
+    /// - push() / push_str()
+    /// - to rust string
+    #[test]
+    fn test_push_str() {
+        let mut str1 = CString16::new();
+        assert_eq!(str1.num_bytes(), 2, "Should have null character");
+        assert_eq!(str1.num_chars(), 0);
+        str1.push(Char16::try_from('h').unwrap());
+        str1.push(Char16::try_from('i').unwrap());
+        assert_eq!(str1.num_chars(), 2);
+
+        let mut str2 = CString16::new();
+        str2.push(Char16::try_from('!').unwrap());
+
+        str2.push_str(str1.as_ref());
+        assert_eq!(str2.num_chars(), 3);
+
+        let rust_str = String::from(&str2);
+        assert_eq!(rust_str, "!hi");
+    }
+
+    #[test]
+    #[should_panic]
+    fn test_push_str_panic() {
+        CString16::new().push(NUL_16);
+    }
+
+    #[test]
+    fn test_char_replace_all_in_place() {
+        let mut input = CString16::try_from("foo/bar/foobar//").unwrap();
+        let search = Char16::try_from('/').unwrap();
+        let replace = Char16::try_from('\\').unwrap();
+        input.replace_char(search, replace);
+
+        let input = String::from(&input);
+        assert_eq!(input, "foo\\bar\\foobar\\\\")
+    }
 }

uefi/src/data_types/strs.rs

@@ -192,7 +192,7 @@ impl CStr16 {
     /// # Safety
     ///
     /// The function will start accessing memory from `ptr` until the first
-    /// null byte. It's the callers responsibility to ensure `ptr` points to
+    /// null character. It's the callers responsibility to ensure `ptr` points to
     /// a valid string, in accessible memory.
     #[must_use]
     pub unsafe fn from_ptr<'ptr>(ptr: *const Char16) -> &'ptr Self {
@@ -232,7 +232,7 @@ impl CStr16 {
     /// # Safety
     ///
     /// It's the callers responsibility to ensure chars is a valid UCS-2
-    /// null-terminated string, with no interior null bytes.
+    /// null-terminated string, with no interior null characters.
     #[must_use]
     pub const unsafe fn from_u16_with_nul_unchecked(codes: &[u16]) -> &Self {
         &*(codes as *const [u16] as *const Self)
@@ -309,26 +309,32 @@ impl CStr16 {
         })
     }
 
-    /// Returns the inner pointer to this C string
+    /// Returns the inner pointer to this C16 string.
     #[must_use]
     pub const fn as_ptr(&self) -> *const Char16 {
         self.0.as_ptr()
     }
 
-    /// Get the underlying [`Char16`] slice, including the trailing null.
+    /// Get the underlying [`Char16`]s as slice without the trailing null.
+    #[must_use]
+    pub fn as_slice(&self) -> &[Char16] {
+        &self.0[..self.num_chars()]
+    }
+
+    /// Get the underlying [`Char16`]s as slice including the trailing null.
     #[must_use]
     pub const fn as_slice_with_nul(&self) -> &[Char16] {
         &self.0
     }
 
-    /// Converts this C string to a u16 slice
+    /// Converts this C string to a u16 slice without the trailing null.
     #[must_use]
     pub fn to_u16_slice(&self) -> &[u16] {
         let chars = self.to_u16_slice_with_nul();
         &chars[..chars.len() - 1]
     }
 
-    /// Converts this C string to a u16 slice containing the trailing 0 char
+    /// Converts this C string to a u16 slice containing the trailing null.
     #[must_use]
     pub const fn to_u16_slice_with_nul(&self) -> &[u16] {
         unsafe { &*(&self.0 as *const [Char16] as *const [u16]) }
@@ -343,7 +349,19 @@ impl CStr16 {
         }
     }
 
-    /// Get the number of bytes in the string (including the trailing null character).
+    /// Returns the number of characters without the trailing null. character
+    #[must_use]
+    pub const fn num_chars(&self) -> usize {
+        self.0.len() - 1
+    }
+
+    /// Returns if the string is empty. This ignores the null character.
+    #[must_use]
+    pub fn is_empty(&self) -> bool {
+        self.num_chars() == 0
+    }
+
+    /// Get the number of bytes in the string (including the trailing null).
     #[must_use]
     pub const fn num_bytes(&self) -> usize {
         self.0.len() * 2
@@ -373,6 +391,20 @@ impl CStr16 {
     }
 }
 
+#[cfg(feature = "alloc")]
+impl From<&CStr16> for alloc::string::String {
+    fn from(value: &CStr16) -> Self {
+        value
+            .as_slice()
+            .iter()
+            .copied()
+            .map(u16::from)
+            .map(|int| int as u32)
+            .map(|int| char::from_u32(int).expect("Should be encodable as UTF-8"))
+            .collect::<alloc::string::String>()
+    }
+}
+
 impl<StrType: AsRef<str> + ?Sized> EqStrUntilNul<StrType> for CStr16 {
     fn eq_str_until_nul(&self, other: &StrType) -> bool {
         let other = other.as_ref();
@@ -391,7 +423,7 @@ impl<StrType: AsRef<str> + ?Sized> EqStrUntilNul<StrType> for CStr16 {
     }
 }
 
-/// An iterator over `CStr16`.
+/// An iterator over the [`Char16`]s in a [`CStr16`].
 #[derive(Debug)]
 pub struct CStr16Iter<'a> {
     inner: &'a CStr16,
@@ -451,11 +483,11 @@ impl<'a> UnalignedSlice<'a, u16> {
 /// get the other direction (`right.eq_str_until_nul(&left)`) for free. Hence, the relation is
 /// reflexive.
 pub trait EqStrUntilNul<StrType: ?Sized> {
-    /// Checks if the provided Rust string `StrType` is equal to [Self] until the first null-byte
-    /// is found. An exception is the terminating null-byte of [Self] which is ignored.
+    /// Checks if the provided Rust string `StrType` is equal to [Self] until the first null character
+    /// is found. An exception is the terminating null character of [Self] which is ignored.
     ///
-    /// As soon as the first null byte in either `&self` or `other` is found, this method returns.
-    /// Note that Rust strings are allowed to contain null-bytes that do not terminate the string.
+    /// As soon as the first null character in either `&self` or `other` is found, this method returns.
+    /// Note that Rust strings are allowed to contain null bytes that do not terminate the string.
     /// Although this is rather unusual, you can compare `"foo\0bar"` with an instance of [Self].
     /// In that case, only `foo"` is compared against [Self] (if [Self] is long enough).
     fn eq_str_until_nul(&self, other: &StrType) -> bool;
@@ -575,6 +607,16 @@ mod tests {
         );
     }
 
+    #[test]
+    fn test_cstr16_as_slice() {
+        let string: &CStr16 = cstr16!("a");
+        assert_eq!(string.as_slice(), &[Char16::try_from('a').unwrap()]);
+        assert_eq!(
+            string.as_slice_with_nul(),
+            &[Char16::try_from('a').unwrap(), NUL_16]
+        );
+    }
+
     // Code generation helper for the compare tests of our CStrX types against "str" and "String"
     // from the standard library.
     #[allow(non_snake_case)]