Merge branch 'wscanf' into 'master'

wscanf and vwscanf implementation proposal

See merge request redox-os/relibc!613

src/header/wchar/lookaheadreader.rs

@@ -1,13 +1,24 @@
 use super::{fseek_locked, ftell_locked, FILE, SEEK_SET};
 use crate::{
+    header::{
+        errno::EILSEQ,
+        wchar::{fgetwc, get_char_encoded_length, mbrtowc, MB_CUR_MAX},
+        wctype::WEOF,
+    },
     io::Read,
-    platform::types::{off_t, wint_t},
+    platform::{
+        types::{c_char, off_t, wchar_t, wint_t},
+        ERRNO,
+    },
 };
+use core::ptr;
+
 struct LookAheadBuffer {
     buf: *const wint_t,
     pos: isize,
     look_ahead: isize,
 }
+
 impl LookAheadBuffer {
     fn look_ahead(&mut self) -> Result<Option<wint_t>, i32> {
         let wchar = unsafe { *self.buf.offset(self.look_ahead) };
@@ -34,20 +45,104 @@ impl From<*const wint_t> for LookAheadBuffer {
     }
 }
 
-pub struct LookAheadReader(LookAheadBuffer);
+struct LookAheadFile<'a> {
+    f: &'a mut FILE,
+    look_ahead: i64,
+}
 
-impl LookAheadReader {
+impl<'a> LookAheadFile<'a> {
+    fn look_ahead(&mut self) -> Result<Option<wint_t>, i32> {
+        let buf = &mut [0; MB_CUR_MAX as usize];
+        let seek = unsafe { ftell_locked(self.f) };
+        unsafe { fseek_locked(self.f, self.look_ahead as off_t, SEEK_SET) };
+
+        let mut encoded_length = 0;
+        let mut bytes_read = 0;
+        loop {
+            match self.f.read(&mut buf[bytes_read..bytes_read + 1]) {
+                Ok(0) => {
+                    ERRNO.set(EILSEQ);
+                    return Ok(Some(WEOF));
+                }
+                Ok(_) => {}
+                Err(_) => return Err(-1),
+            }
+
+            bytes_read += 1;
+
+            if bytes_read == 1 {
+                encoded_length = if let Some(el) = get_char_encoded_length(buf[0]) {
+                    el
+                } else {
+                    ERRNO.set(EILSEQ);
+                    return Ok(Some(WEOF));
+                };
+            }
+
+            if bytes_read >= encoded_length {
+                break;
+            }
+        }
+
+        let mut wc: wchar_t = 0;
+        unsafe {
+            mbrtowc(
+                &mut wc,
+                buf.as_ptr() as *const c_char,
+                encoded_length,
+                ptr::null_mut(),
+            );
+
+            fseek_locked(self.f, seek, SEEK_SET);
+        }
+
+        self.look_ahead += encoded_length as i64;
+
+        Ok(Some(wc as wint_t))
+    }
+
+    fn commit(&mut self) {
+        unsafe { fseek_locked(self.f, self.look_ahead as off_t, SEEK_SET) };
+    }
+}
+
+impl<'a> From<&'a mut FILE> for LookAheadFile<'a> {
+    fn from(f: &'a mut FILE) -> LookAheadFile<'a> {
+        let look_ahead = unsafe { ftell_locked(f) } as i64;
+        LookAheadFile { f, look_ahead }
+    }
+}
+
+enum LookAheadReaderEnum<'a> {
+    FILE(LookAheadFile<'a>),
+    BUFFER(LookAheadBuffer),
+}
+
+pub struct LookAheadReader<'a>(LookAheadReaderEnum<'a>);
+
+impl LookAheadReader<'_> {
     pub fn lookahead1(&mut self) -> Result<Option<wint_t>, i32> {
-        self.0.look_ahead()
+        match &mut self.0 {
+            LookAheadReaderEnum::FILE(f) => f.look_ahead(),
+            LookAheadReaderEnum::BUFFER(b) => b.look_ahead(),
+        }
     }
-
     pub fn commit(&mut self) {
-        self.0.commit()
+        match &mut self.0 {
+            LookAheadReaderEnum::FILE(f) => f.commit(),
+            LookAheadReaderEnum::BUFFER(b) => b.commit(),
+        }
     }
 }
 
-impl From<*const wint_t> for LookAheadReader {
-    fn from(buff: *const wint_t) -> LookAheadReader {
-        LookAheadReader(buff.into())
+impl<'a> From<&'a mut FILE> for LookAheadReader<'a> {
+    fn from(f: &'a mut FILE) -> LookAheadReader<'a> {
+        LookAheadReader(LookAheadReaderEnum::FILE(f.into()))
+    }
+}
+
+impl<'a> From<*const wint_t> for LookAheadReader<'a> {
+    fn from(buff: *const wint_t) -> LookAheadReader<'a> {
+        LookAheadReader(LookAheadReaderEnum::BUFFER(buff.into()))
     }
 }

src/header/wchar/mod.rs

@@ -10,6 +10,7 @@ use crate::{
         stdlib::{malloc, MB_CUR_MAX, MB_LEN_MAX},
         string,
         time::*,
+        wchar::{lookaheadreader::LookAheadReader, utf8::get_char_encoded_length},
         wctype::*,
     },
     iter::{NulTerminated, NulTerminatedInclusive},
@@ -69,14 +70,8 @@ pub unsafe extern "C" fn fgetwc(stream: *mut FILE) -> wint_t {
         bytes_read += 1;
 
         if bytes_read == 1 {
-            encoded_length = if buf[0] >> 7 == 0 {
-                1
-            } else if buf[0] >> 5 == 6 {
-                2
-            } else if buf[0] >> 4 == 0xe {
-                3
-            } else if buf[0] >> 3 == 0x1e {
-                4
+            encoded_length = if let Some(el) = get_char_encoded_length(buf[0]) {
+                el
             } else {
                 ERRNO.set(EILSEQ);
                 return WEOF;
@@ -1006,9 +1001,21 @@ pub unsafe extern "C" fn wmemset(ws: *mut wchar_t, wc: wchar_t, n: size_t) -> *m
     ws
 }
 
-// #[no_mangle]
-pub extern "C" fn wscanf(format: *const wchar_t, ap: va_list) -> c_int {
-    unimplemented!();
+#[no_mangle]
+pub unsafe extern "C" fn vwscanf(format: *const wchar_t, __valist: va_list) -> c_int {
+    let mut file = (*stdin).lock();
+    if let Err(_) = file.try_set_byte_orientation_unlocked() {
+        return -1;
+    }
+
+    let f: &mut FILE = &mut *file;
+    let reader: LookAheadReader = f.into();
+    wscanf::scanf(reader, format, __valist)
+}
+
+#[no_mangle]
+pub unsafe extern "C" fn wscanf(format: *const wchar_t, mut __valist: ...) -> c_int {
+    vwscanf(format, __valist.as_va_list())
 }
 
 #[no_mangle]

src/header/wchar/utf8.rs

@@ -91,3 +91,18 @@ pub unsafe fn wcrtomb(s: *mut c_char, wc: wchar_t, ps: *mut mbstate_t) -> usize
 
     size
 }
+
+/// Gets the encoded length of a character. It is used to recognize wide characters
+pub fn get_char_encoded_length(first_byte: u8) -> Option<usize> {
+    if first_byte >> 7 == 0 {
+        Some(1)
+    } else if first_byte >> 5 == 6 {
+        Some(2)
+    } else if first_byte >> 4 == 0xe {
+        Some(3)
+    } else if first_byte >> 3 == 0x1e {
+        Some(4)
+    } else {
+        None
+    }
+}