prepare for newio

Cargo.toml

@@ -15,3 +15,4 @@ rustc-serialize = "0.2.10"
 
 [dependencies.byteorder]
 git = "https://github.com/BurntSushi/byteorder.git"
+branch = "newio"

src/lib.rs

@@ -2,18 +2,21 @@
 #![crate_type = "rlib"]
 #![crate_type = "dylib"]
 
-#![allow(unstable)]
+#![feature(hash, core, io, unicode, collections)]
 
 extern crate "rustc-serialize" as rustc_serialize;
 extern crate byteorder;
+extern crate unicode;
 
-use std::old_io::{Buffer, MemWriter};
+//use std::old_io::{Buffer, MemWriter};
 use rustc_serialize::{Encodable, Decodable};
 
 pub use writer::{EncoderWriter, EncodingResult, EncodingError};
 pub use reader::{DecoderReader, DecodingResult, DecodingError};
 use writer::SizeChecker;
 
+use std::io::{Write, BufRead};
+
 mod writer;
 mod reader;
 #[cfg(test)] mod test;
@@ -43,30 +46,6 @@ mod reader;
 ///! }
 ///! ```
 ///!
-///! ### Using Into/From Functions
-///!
-///! ```rust
-///! #![allow(unstable)]
-///! extern crate bincode;
-///! use std::old_io::pipe::PipeStream;
-///! use std::old_io::BufferedReader;
-///! fn main() {
-///!     // The pipes that we will be using to send values across.
-///!     let streams = PipeStream::pair().unwrap();
-///!     let (mut reader, mut writer) = (BufferedReader::new(streams.reader),
-///!                                     streams.writer);
-///!     // The object that we will send across.
-///!     let target = Some(5u32);
-///!     // The max-size of the encoded bytes.
-///!     let limit = bincode::SizeLimit::Bounded(10);
-///!
-///!     // Do the actual encoding and decoding.
-///!     bincode::encode_into(&target, &mut writer, limit).ok();
-///!     let out: Option<u32> = bincode::decode_from(&mut reader, limit).unwrap();
-///!     assert_eq!(target, out);
-///! }
-///! ```
-///!
 
 /// A limit on the size of bytes to be read or written.
 ///
@@ -97,9 +76,19 @@ pub enum SizeLimit {
 /// If the encoding would take more bytes than allowed by `size_limit`,
 /// an error is returned.
 pub fn encode<T: Encodable>(t: &T, size_limit: SizeLimit) -> EncodingResult<Vec<u8>> {
-    let mut w = MemWriter::new();
+    // Since we are putting values directly into a vector, we can do size
-    match encode_into(t, &mut w, size_limit) {
+    // computation out here and pre-allocate a buffer of *exactly*
-        Ok(()) => Ok(w.into_inner()),
+    // the right size.
+    let mut w = if let SizeLimit::Bounded(l) = size_limit {
+        let actual_size = encoded_size_bounded(t, l);
+        let actual_size = try!(actual_size.ok_or(EncodingError::SizeLimit));
+        Vec::with_capacity(actual_size as usize)
+    } else {
+        vec![]
+    };
+
+    match encode_into(t, &mut w, SizeLimit::Infinite) {
+        Ok(()) => Ok(w),
         Err(e) => Err(e)
     }
 }
@@ -121,7 +110,7 @@ pub fn decode<T: Decodable>(b: &[u8]) -> DecodingResult<T> {
 /// If this returns an `EncodingError` (other than SizeLimit), assume that the
 /// writer is in an invalid state, as writing could bail out in the middle of
 /// encoding.
-pub fn encode_into<T: Encodable, W: Writer>(t: &T, w: &mut W, size_limit: SizeLimit) -> EncodingResult<()> {
+pub fn encode_into<T: Encodable, W: Write>(t: &T, w: &mut W, size_limit: SizeLimit) -> EncodingResult<()> {
     try!(match size_limit {
         SizeLimit::Infinite => Ok(()),
         SizeLimit::Bounded(x) => {
@@ -130,7 +119,7 @@ pub fn encode_into<T: Encodable, W: Writer>(t: &T, w: &mut W, size_limit: SizeLi
         }
     });
 
-    t.encode(&mut writer::EncoderWriter::new(w, size_limit))
+    t.encode(&mut writer::EncoderWriter::new(w))
 }
 
 /// Decoes an object directly from a `Buffer`ed Reader.
@@ -142,7 +131,7 @@ pub fn encode_into<T: Encodable, W: Writer>(t: &T, w: &mut W, size_limit: SizeLi
 /// If this returns an `DecodingError`, assume that the buffer that you passed
 /// in is in an invalid state, as the error could be returned during any point
 /// in the reading.
-pub fn decode_from<R: Buffer, T: Decodable>(r: &mut R, size_limit: SizeLimit) ->
+pub fn decode_from<R: BufRead, T: Decodable>(r: &mut R, size_limit: SizeLimit) ->
 DecodingResult<T> {
     Decodable::decode(&mut reader::DecoderReader::new(r, size_limit))
 }
@@ -158,3 +147,13 @@ pub fn encoded_size<T: Encodable>(t: &T) -> u64 {
     t.encode(&mut size_checker).ok();
     size_checker.written
 }
+
+/// Given a maximum size limit, check how large an object would be if it
+/// were to be encoded.
+///
+/// If it can be encoded in `max` or fewer bytes, that number will be returned
+/// inside `Some`.  If it goes over bounds, then None is returned.
+pub fn encoded_size_bounded<T: Encodable>(t: &T, max: u64) -> Option<u64> {
+    let mut size_checker = SizeChecker::new(max);
+    t.encode(&mut size_checker).ok().map(|_| size_checker.written)
+}

src/reader.rs

@@ -1,11 +1,14 @@
-use std::old_io::{Buffer, Reader, IoError};
+use std::io::{BufRead, Read};
+use std::io::Error as IoError;
+use std::io::Result as IoResult;
 use std::num::{cast, NumCast};
 use std::error::{Error, FromError};
 use std::fmt;
 
 use rustc_serialize::Decoder;
 
-use byteorder::{BigEndian, ReaderBytesExt};
+use byteorder::{BigEndian, ReadBytesExt};
+use unicode;
 
 use super::SizeLimit;
 
@@ -98,7 +101,7 @@ pub struct DecoderReader<'a, R: 'a> {
     read: u64
 }
 
-impl<'a, R: Buffer> DecoderReader<'a, R> {
+impl<'a, R: BufRead> DecoderReader<'a, R> {
     pub fn new(r: &'a mut R, size_limit: SizeLimit) -> DecoderReader<'a, R> {
         DecoderReader {
             reader: r,
@@ -125,7 +128,7 @@ impl <'a, A> DecoderReader<'a, A> {
     }
 }
 
-impl<'a, R: Buffer> Decoder for DecoderReader<'a, R> {
+impl<'a, R: BufRead> Decoder for DecoderReader<'a, R> {
     type Error = DecodingError;
 
     fn read_nil(&mut self) -> DecodingResult<()> {
@@ -189,16 +192,46 @@ impl<'a, R: Buffer> Decoder for DecoderReader<'a, R> {
         self.reader.read_f32::<BigEndian>().map_err(wrap_io)
     }
     fn read_char(&mut self) -> DecodingResult<char> {
-        let c = try!(self.reader.read_char().map_err(wrap_io));
+        use std::str;
-        try!(self.read_bytes(c.len_utf8()));
-        Ok(c)
 
+        let error = DecodingError::InvalidEncoding(InvalidEncoding {
+            desc: "Invalid char encoding",
+            detail: None
+        });
+
+        let mut buf = [0];
+
+        let _ = try!(self.reader.read(&mut buf[]));
+        let first_byte = buf[0];
+        let width = unicode::str::utf8_char_width(first_byte);
+        if width == 1 { return Ok(first_byte as char) }
+        if width == 0 { return Err(error)}
+        let mut buf = [first_byte, 0, 0, 0];
+        {
+            let mut start = 1;
+            while start < width {
+                match try!(self.reader.read(&mut buf[start .. width])) {
+                    n if n == width - start => break,
+                    n if n < width - start => { start += n; }
+                    _ => return Err(error)
+                }
+            }
+        }
+
+        let res = try!(match str::from_utf8(&buf[..width]).ok() {
+            Some(s) => Ok(s.char_at(0)),
+            None => Err(error)
+        });
+
+        try!(self.read_bytes(res.len_utf8()));
+        Ok(res)
     }
+
     fn read_str(&mut self) -> DecodingResult<String> {
         let len = try!(self.read_usize());
-
         try!(self.read_bytes(len));
-        let vector = try!(self.reader.read_exact(len));
+
+        let vector = try!(read_exact(&mut self.reader, len));
         match String::from_utf8(vector) {
             Ok(s) => Ok(s),
             Err(err) => Err(DecodingError::InvalidEncoding(InvalidEncoding {
@@ -309,3 +342,77 @@ impl<'a, R: Buffer> Decoder for DecoderReader<'a, R> {
         })
     }
 }
+
+fn read_at_least<R: Read>(reader: &mut R, min: usize, buf: &mut [u8]) -> IoResult<usize> {
+    use std::io::ErrorKind;
+    if min > buf.len() {
+        return Err(IoError::new(
+            ErrorKind::InvalidInput, "the buffer is too short", None));
+    }
+
+    let mut read = 0;
+    while read < min {
+        let mut zeroes = 0;
+        loop {
+            match reader.read(&mut buf[read..]) {
+                Ok(0) => {
+                    zeroes += 1;
+                    if zeroes >= 1000 {
+                        return Err(IoError::new(ErrorKind::Other,
+                                                "no progress was made",
+                                                None ));
+                    }
+                }
+                Ok(n) => {
+                    read += n;
+                    break;
+                }
+                err@Err(_) => return err
+            }
+        }
+    }
+    Ok(read)
+}
+
+unsafe fn slice_vec_capacity<'a, T>(v: &'a mut Vec<T>, start: usize, end: usize) -> &'a mut [T] {
+    use std::raw::Slice;
+    use std::ptr::PtrExt;
+    use std::mem::transmute;
+
+    assert!(start <= end);
+    assert!(end <= v.capacity());
+    transmute(Slice {
+        data: v.as_ptr().offset(start as isize),
+        len: end - start
+    })
+}
+
+
+fn push_at_least<R: Read>(reader: &mut R, min: usize, len: usize, buf: &mut Vec<u8>) -> IoResult<usize> {
+    use std::io::ErrorKind;
+    if min > len {
+        return Err(IoError::new(ErrorKind::InvalidInput, "the buffer is too short", None));
+    }
+
+    let start_len = buf.len();
+    buf.reserve(len);
+
+
+    let mut read = 0;
+    while read < min {
+        read += {
+            let s = unsafe { slice_vec_capacity(buf, start_len + read, start_len + len) };
+            try!(read_at_least(reader, 1, s))
+        };
+        unsafe { buf.set_len(start_len + read) };
+    }
+    Ok(read)
+}
+
+fn read_exact<R: Read>(reader: &mut R, len: usize) -> IoResult<Vec<u8>> {
+    let mut buf = Vec::with_capacity(len);
+    match push_at_least(reader, len, len, &mut buf) {
+        Ok(_) => Ok(buf),
+        Err(e) => Err(e),
+    }
+}

src/writer.rs

@@ -1,12 +1,11 @@
-use std::old_io::{Writer, IoError};
+use std::io::Write;
+use std::io::Error as IoError;
 use std::error::Error;
 use std::num::Int;
 use std::fmt;
 
 use rustc_serialize::Encoder;
-use byteorder::{BigEndian, WriterBytesExt};
+use byteorder::{BigEndian, WriteBytesExt};
-
-use super::SizeLimit;
 
 pub type EncodingResult<T> = Result<T, EncodingError>;
 
@@ -29,7 +28,6 @@ pub enum EncodingError {
 /// For most cases, prefer the `encode_into` function.
 pub struct EncoderWriter<'a, W: 'a> {
     writer: &'a mut W,
-    _size_limit: SizeLimit
 }
 
 pub struct SizeChecker {
@@ -66,11 +64,10 @@ impl Error for EncodingError {
     }
 }
 
-impl <'a, W: Writer> EncoderWriter<'a, W> {
+impl <'a, W: Write> EncoderWriter<'a, W> {
-    pub fn new(w: &'a mut W, size_limit: SizeLimit) -> EncoderWriter<'a, W> {
+    pub fn new(w: &'a mut W) -> EncoderWriter<'a, W> {
         EncoderWriter {
             writer: w,
-            _size_limit: size_limit
         }
     }
 }
@@ -98,7 +95,7 @@ impl SizeChecker {
     }
 }
 
-impl<'a, W: Writer> Encoder for EncoderWriter<'a, W> {
+impl<'a, W: Write> Encoder for EncoderWriter<'a, W> {
     type Error = EncodingError;
 
     fn emit_nil(&mut self) -> EncodingResult<()> { Ok(()) }