Avoid short reads in read_to_end

On Windows, the UTF-16 to UTF-8 translation is made simpler by ensuring we don't split code points.

Author: ChrisDenton

library/std/src/io/mod.rs

@@ -405,7 +405,7 @@ where
 // - pass large buffers to readers that do not initialize the spare capacity. this can amortize per-call overheads
 // - and finally pass not-too-small and not-too-large buffers to Windows read APIs because they manage to suffer from both problems
 //   at the same time, i.e. small reads suffer from syscall overhead, all reads incur costs proportional to buffer size (#110650)
-//
+// - also avoid <4 byte reads as this may split UTF-8 code points, which can be a problem for Windows console reads (#142847)
 pub(crate) fn default_read_to_end<R: Read + ?Sized>(
     r: &mut R,
     buf: &mut Vec<u8>,
@@ -452,7 +452,7 @@ pub(crate) fn default_read_to_end<R: Read + ?Sized>(
     let mut consecutive_short_reads = 0;
 
     loop {
-        if buf.len() == buf.capacity() && buf.capacity() == start_cap {
+        if buf.spare_capacity_mut().len() < PROBE_SIZE && buf.capacity() == start_cap {
             // The buffer might be an exact fit. Let's read into a probe buffer
             // and see if it returns `Ok(0)`. If so, we've avoided an
             // unnecessary doubling of the capacity. But if not, append the
@@ -464,7 +464,9 @@ pub(crate) fn default_read_to_end<R: Read + ?Sized>(
             }
         }
 
-        if buf.len() == buf.capacity() {
+        // Try to ensure we're not making unnecessarily short reads.
+        // This is skipped if the user's pre-allocated Vec may have been exactly sized.
+        if buf.capacity() != start_cap || buf.spare_capacity_mut().len() == 0 {
             // buf is full, need more space
             buf.try_reserve(PROBE_SIZE)?;
         }