preserves/implementations/rust/preserves/src/value/packed/reader.rs

477 lines
16 KiB
Rust

use num::bigint::BigInt;
use num::traits::cast::{FromPrimitive, ToPrimitive};
use std::borrow::Cow;
use std::convert::TryFrom;
use std::convert::TryInto;
use std::marker::PhantomData;
use super::constants::Tag;
use super::super::{
FALSE,
IOResult,
IOValue,
Map,
NestedValue,
Record,
Set,
TRUE,
Value,
reader::{
BinarySource,
BytesBinarySource,
ConfiguredReader,
IOBinarySource,
Reader,
ReaderResult,
},
repr::Annotations,
signed_integer::SignedInteger,
};
use crate::error::{self, ExpectedKind, Received, is_eof_io_error, io_syntax_error};
pub struct PackedReader<'de, S: BinarySource<'de>> {
pub source: S,
phantom: PhantomData<&'de ()>,
}
impl<'de, S: BinarySource<'de>> BinarySource<'de> for PackedReader<'de, S> {
fn skip(&mut self) -> IOResult<()> {
self.source.skip()
}
fn peek(&mut self) -> IOResult<u8> {
self.source.peek()
}
fn readbytes(&mut self, count: usize) -> IOResult<Cow<'de, [u8]>> {
self.source.readbytes(count)
}
fn readbytes_into(&mut self, bs: &mut [u8]) -> IOResult<()> {
self.source.readbytes_into(bs)
}
}
fn out_of_range<I: Into<BigInt>>(i: I) -> error::Error {
error::Error::NumberOutOfRange(i.into())
}
impl<'de> PackedReader<'de, BytesBinarySource<'de>> {
pub fn decode_bytes(bytes: &'de [u8]) -> Self {
PackedReader::new(BytesBinarySource::new(bytes))
}
}
impl<'de, 'a, IOR: std::io::Read> PackedReader<'de, IOBinarySource<'a, IOR>> {
pub fn decode_read(read: &'a mut IOR) -> Self {
PackedReader::new(IOBinarySource::new(read))
}
}
impl<'de, S: BinarySource<'de>> PackedReader<'de, S> {
pub fn new(source: S) -> Self {
PackedReader { source, phantom: PhantomData }
}
fn read(&mut self) -> IOResult<u8> {
let v = self.peek()?;
self.skip()?;
Ok(v)
}
fn expected(&mut self, k: ExpectedKind) -> error::Error {
match self.demand_next(true) {
Ok(v) => error::Error::Expected(k, Received::ReceivedOtherValue(v)),
Err(e) => e.into()
}
}
fn varint(&mut self) -> IOResult<usize> {
let mut shift = 0;
let mut acc: usize = 0;
loop {
let v = self.read()?;
acc |= ((v & 0x7f) as usize) << shift;
shift += 7;
if v & 0x80 == 0 { return Ok(acc) }
}
}
fn peekend(&mut self) -> IOResult<bool> {
if self.peek()? == Tag::End.into() {
self.skip()?;
Ok(true)
} else {
Ok(false)
}
}
fn peek_next_nonannotation_tag(&mut self) -> ReaderResult<Tag> {
loop {
match Tag::try_from(self.peek()?)? {
Tag::Annotation => {
self.skip()?;
self.skip_value()?;
},
other => return Ok(other),
}
}
}
fn next_atomic(&mut self, expected_tag: Tag, k: ExpectedKind) -> ReaderResult<Cow<'de, [u8]>> {
let actual_tag = self.peek_next_nonannotation_tag()?;
if actual_tag == expected_tag {
self.skip()?;
let count = self.varint()?;
Ok(self.readbytes(count)?)
} else {
Err(self.expected(k))
}
}
fn next_compound(&mut self, expected_tag: Tag, k: ExpectedKind) -> ReaderResult<()>
{
let actual_tag = self.peek_next_nonannotation_tag()?;
if actual_tag == expected_tag {
self.skip()?;
Ok(())
} else {
Err(self.expected(k))
}
}
fn read_signed_integer(&mut self, count: usize) -> IOResult<SignedInteger> {
if count == 0 {
return Ok(SignedInteger::from(0_i128));
}
if count > 16 {
let bs = self.readbytes(count)?;
if (bs[0] & 0x80) == 0 {
// Positive or zero.
let mut i = 0;
while i < count && bs[i] == 0 { i += 1; }
if count - i <= 16 {
Ok(SignedInteger::from(u128::from_be_bytes(bs[bs.len() - 16..].try_into().unwrap())))
} else {
Ok(SignedInteger::from(Cow::Owned(BigInt::from_bytes_be(num::bigint::Sign::Plus, &bs[i..]))))
}
} else {
// Negative.
let mut i = 0;
while i < count && bs[i] == 0xff { i += 1; }
if count - i <= 16 {
Ok(SignedInteger::from(i128::from_be_bytes(bs[bs.len() - 16..].try_into().unwrap())))
} else {
Ok(SignedInteger::from(Cow::Owned(BigInt::from_signed_bytes_be(&bs))))
}
}
} else {
let first_byte = self.read()?;
let prefix_byte = if (first_byte & 0x80) == 0 { 0x00 } else { 0xff };
let mut bs = [prefix_byte; 16];
bs[16 - count] = first_byte;
self.readbytes_into(&mut bs[16 - (count - 1)..])?;
Ok(SignedInteger::from(i128::from_be_bytes(bs)))
}
}
fn next_unsigned<T: FromPrimitive, F>(&mut self, f: F) -> ReaderResult<T>
where
F: FnOnce(u128) -> Option<T>
{
let tag = self.peek_next_nonannotation_tag()?;
match tag {
Tag::SmallInteger(v) => {
self.skip()?;
if v < 0 {
Err(out_of_range(v))
} else {
f(v as u128).ok_or_else(|| out_of_range(v))
}
}
Tag::MediumInteger(count) => {
self.skip()?;
let n = &self.read_signed_integer(count.into())?;
let i = n.try_into().map_err(|_| out_of_range(n))?;
f(i).ok_or_else(|| out_of_range(i))
}
Tag::SignedInteger => {
self.skip()?;
let count = self.varint()?;
let n = &self.read_signed_integer(count)?;
let i = n.try_into().map_err(|_| out_of_range(n))?;
f(i).ok_or_else(|| out_of_range(i))
}
_ => Err(self.expected(ExpectedKind::SignedInteger))
}
}
fn next_signed<T: FromPrimitive, F>(&mut self, f: F) -> ReaderResult<T>
where
F: FnOnce(i128) -> Option<T>
{
let tag = self.peek_next_nonannotation_tag()?;
match tag {
Tag::SmallInteger(v) => {
self.skip()?;
f(v.into()).ok_or_else(|| out_of_range(v))
}
Tag::MediumInteger(count) => {
self.skip()?;
let n = &self.read_signed_integer(count.into())?;
let i = n.try_into().map_err(|_| out_of_range(n))?;
f(i).ok_or_else(|| out_of_range(i))
}
Tag::SignedInteger => {
self.skip()?;
let count = self.varint()?;
let n = &self.read_signed_integer(count)?;
let i = n.try_into().map_err(|_| out_of_range(n))?;
f(i).ok_or_else(|| out_of_range(i))
}
_ => Err(self.expected(ExpectedKind::SignedInteger))
}
}
}
impl<'de, S: BinarySource<'de>> Reader<'de> for PackedReader<'de, S> {
fn next(&mut self, read_annotations: bool) -> IOResult<Option<IOValue>> {
match self.peek() {
Err(e) if is_eof_io_error(&e) => return Ok(None),
Err(e) => return Err(e),
Ok(_) => (),
}
Ok(Some(match Tag::try_from(self.read()?)? {
Tag::False => FALSE.clone(),
Tag::True => TRUE.clone(),
Tag::Float => {
let mut bs = [0; 4];
self.readbytes_into(&mut bs)?;
Value::from(f32::from_bits(u32::from_be_bytes(bs))).wrap()
}
Tag::Double => {
let mut bs = [0; 8];
self.readbytes_into(&mut bs)?;
Value::from(f64::from_bits(u64::from_be_bytes(bs))).wrap()
}
Tag::Annotation => {
if read_annotations {
let mut annotations = vec![self.demand_next(read_annotations)?];
while Tag::try_from(self.peek()?)? == Tag::Annotation {
self.skip()?;
annotations.push(self.demand_next(read_annotations)?);
}
let (existing_annotations, v) = self.demand_next(read_annotations)?.pieces();
annotations.extend_from_slice(existing_annotations.slice());
IOValue::wrap(Annotations::new(Some(annotations)), v)
} else {
self.skip_value()?;
while Tag::try_from(self.peek()?)? == Tag::Annotation {
self.skip()?;
self.skip_value()?;
}
self.demand_next(read_annotations)?
}
}
Tag::Embedded => {
let v = self.demand_next(read_annotations)?;
Value::Embedded(v).wrap()
}
Tag::SmallInteger(v) => {
// TODO: prebuild these in value.rs
Value::from(v).wrap()
}
Tag::MediumInteger(count) => {
let n = self.read_signed_integer(count.into())?;
Value::SignedInteger(n).wrap()
}
Tag::SignedInteger => {
let count = self.varint()?;
let n = self.read_signed_integer(count)?;
Value::SignedInteger(n).wrap()
}
Tag::String => {
let count = self.varint()?;
Value::String(decodestr(self.readbytes(count)?)?.into_owned()).wrap()
}
Tag::ByteString => {
let count = self.varint()?;
Value::ByteString(self.readbytes(count)?.into_owned()).wrap()
}
Tag::Symbol => {
let count = self.varint()?;
Value::Symbol(decodestr(self.readbytes(count)?)?.into_owned()).wrap()
}
Tag::Record => {
let iter = DelimitedStream { reader: self.configured(read_annotations) };
let vs = iter.collect::<IOResult<Vec<IOValue>>>()?;
if vs.is_empty() {
return Err(io_syntax_error("Too few elements in encoded record"))
}
Value::Record(Record(vs)).wrap()
}
Tag::Sequence => {
let iter = DelimitedStream { reader: self.configured(read_annotations) };
let vs = iter.collect::<IOResult<Vec<IOValue>>>()?;
Value::Sequence(vs).wrap()
}
Tag::Set => {
let iter = DelimitedStream { reader: self.configured(read_annotations) };
let mut s = Set::new();
for res in iter { s.insert(res?); }
Value::Set(s).wrap()
}
Tag::Dictionary => {
let mut iter = DelimitedStream { reader: self.configured(read_annotations) };
let mut d = Map::new();
while let Some(kres) = iter.next() {
let k = kres?;
match iter.next() {
Some(vres) => {
let v = vres?;
d.insert(k, v);
}
None => return Err(io_syntax_error("Missing dictionary value")),
}
}
Value::Dictionary(d).wrap()
}
tag @ Tag::End => {
return Err(io_syntax_error(&format!("Invalid tag: {:?}", tag)));
}
}))
}
fn open_record(&mut self, arity: Option<usize>) -> ReaderResult<()> {
self.next_compound(Tag::Record, ExpectedKind::Record(arity))?;
self.ensure_more_expected()
}
fn open_sequence_or_set(&mut self) -> ReaderResult<()> {
match self.peek_next_nonannotation_tag()? {
Tag::Sequence | Tag::Set => {
self.skip()?;
Ok(())
}
_ => Err(self.expected(ExpectedKind::SequenceOrSet)),
}
}
fn open_sequence(&mut self) -> ReaderResult<()> {
self.next_compound(Tag::Sequence, ExpectedKind::Sequence)
}
fn open_set(&mut self) -> ReaderResult<()> {
self.next_compound(Tag::Set, ExpectedKind::Set)
}
fn open_dictionary(&mut self) -> ReaderResult<()> {
self.next_compound(Tag::Dictionary, ExpectedKind::Dictionary)
}
fn close_compound(&mut self) -> ReaderResult<bool> {
Ok(self.peekend()?)
}
fn open_embedded(&mut self) -> ReaderResult<()> {
self.next_compound(Tag::Embedded, ExpectedKind::Embedded)
}
fn close_embedded(&mut self) -> ReaderResult<()> {
Ok(())
}
fn next_boolean(&mut self) -> ReaderResult<bool> {
match self.peek_next_nonannotation_tag()? {
Tag::False => { self.skip()?; Ok(false) }
Tag::True => { self.skip()?; Ok(true) }
_ => Err(self.expected(ExpectedKind::Boolean)),
}
}
fn next_i8(&mut self) -> ReaderResult<i8> { self.next_signed(|n| n.to_i8()) }
fn next_i16(&mut self) -> ReaderResult<i16> { self.next_signed(|n| n.to_i16()) }
fn next_i32(&mut self) -> ReaderResult<i32> { self.next_signed(|n| n.to_i32()) }
fn next_i64(&mut self) -> ReaderResult<i64> { self.next_signed(|n| n.to_i64()) }
fn next_i128(&mut self) -> ReaderResult<i128> { self.next_signed(|n| n.to_i128()) }
fn next_u8(&mut self) -> ReaderResult<u8> { self.next_unsigned(|n| n.to_u8()) }
fn next_u16(&mut self) -> ReaderResult<u16> { self.next_unsigned(|n| n.to_u16()) }
fn next_u32(&mut self) -> ReaderResult<u32> { self.next_unsigned(|n| n.to_u32()) }
fn next_u64(&mut self) -> ReaderResult<u64> { self.next_unsigned(|n| n.to_u64()) }
fn next_u128(&mut self) -> ReaderResult<u128> { self.next_unsigned(|n| n.to_u128()) }
fn next_f32(&mut self) -> ReaderResult<f32> {
match self.peek_next_nonannotation_tag()? {
Tag::Float => {
self.skip()?;
let mut bs = [0; 4];
self.readbytes_into(&mut bs)?;
Ok(f32::from_bits(u32::from_be_bytes(bs)))
},
Tag::Double => {
self.skip()?;
let mut bs = [0; 8];
self.readbytes_into(&mut bs)?;
Ok(f64::from_bits(u64::from_be_bytes(bs)) as f32)
},
_ => Err(self.expected(ExpectedKind::Float)),
}
}
fn next_f64(&mut self) -> ReaderResult<f64> {
match self.peek_next_nonannotation_tag()? {
Tag::Float => {
self.skip()?;
let mut bs = [0; 4];
self.readbytes_into(&mut bs)?;
Ok(f32::from_bits(u32::from_be_bytes(bs)) as f64)
},
Tag::Double => {
self.skip()?;
let mut bs = [0; 8];
self.readbytes_into(&mut bs)?;
Ok(f64::from_bits(u64::from_be_bytes(bs)))
},
_ => Err(self.expected(ExpectedKind::Double)),
}
}
fn next_str(&mut self) -> ReaderResult<Cow<'de, str>> {
Ok(decodestr(self.next_atomic(Tag::String, ExpectedKind::Symbol)?)?)
}
fn next_bytestring(&mut self) -> ReaderResult<Cow<'de, [u8]>> {
self.next_atomic(Tag::ByteString, ExpectedKind::Symbol)
}
fn next_symbol(&mut self) -> ReaderResult<Cow<'de, str>> {
Ok(decodestr(self.next_atomic(Tag::Symbol, ExpectedKind::Symbol)?)?)
}
}
struct DelimitedStream<'a, 'de, S: BinarySource<'de>> {
reader: ConfiguredReader<'de, &'a mut PackedReader<'de, S>>,
}
impl<'a, 'de, S: BinarySource<'de>> Iterator for DelimitedStream<'a, 'de, S>
{
type Item = IOResult<IOValue>;
fn next(&mut self) -> Option<Self::Item> {
match self.reader.reader.peekend() {
Err(e) => Some(Err(e)),
Ok(true) => None,
Ok(false) => Some(self.reader.reader.demand_next(self.reader.read_annotations)),
}
}
}
fn decodestr(cow: Cow<'_, [u8]>) -> IOResult<Cow<'_, str>> {
match cow {
Cow::Borrowed(bs) =>
Ok(Cow::Borrowed(std::str::from_utf8(bs).map_err(|_| io_syntax_error("Invalid UTF-8"))?)),
Cow::Owned(bs) =>
Ok(Cow::Owned(String::from_utf8(bs).map_err(|_| io_syntax_error("Invalid UTF-8"))?)),
}
}