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::super::signed_integer::SignedInteger; use super::super::value::{Value, NestedValue, IOValue, FALSE, TRUE, Map, Set, Record, Annotations}; use super::constants::{Op, InvalidOp, AtomMinor, CompoundMinor}; use super::super::reader::{ BinarySource, BytesBinarySource, CompoundBody, CompoundLimit, ConfiguredReader, IOBinarySource, IOResult, Reader, ReaderResult, }; 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 { self.source.peek() } fn readbytes(&mut self, count: usize) -> IOResult> { self.source.readbytes(count) } fn readbytes_into(&mut self, bs: &mut [u8]) -> IOResult<()> { self.source.readbytes_into(bs) } } fn out_of_range>(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 { 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 { let mut shift = 0; let mut acc: usize = 0; loop { let v = self.read()?; acc = acc | (((v & 0x7f) as usize) << shift); shift = shift + 7; if v & 0x80 == 0 { return Ok(acc) } } } fn wirelength(&mut self, arg: u8) -> IOResult { if arg < 15 { Ok(usize::from(arg)) } else { self.varint() } } fn peekend(&mut self) -> IOResult { if self.peek()? == 4 { self.skip()?; Ok(true) } else { Ok(false) } } fn gather_chunks(&mut self) -> IOResult> { let mut bs = Vec::with_capacity(256); while !self.peekend()? { match decodeop(self.peek()?)? { (Op::Atom(AtomMinor::ByteString), arg) => { self.skip()?; let count = self.wirelength(arg)?; if count == 0 { return Err(io_syntax_error("Empty binary chunks are forbidden")); } bs.extend_from_slice(&self.readbytes(count)?) }, _ => return Err(io_syntax_error("Unexpected non-format-B-ByteString chunk")) } } Ok(bs) } fn peek_next_nonannotation_op(&mut self) -> ReaderResult<(Op, u8)> { loop { match decodeop(self.peek()?)? { (Op::Misc(0), 5) => { self.skip()?; self.skip_value()?; }, other => return Ok(other), } } } fn next_atomic(&mut self, minor: AtomMinor, k: ExpectedKind) -> ReaderResult> { match self.peek_next_nonannotation_op()? { (Op::Atom(actual_minor), arg) if actual_minor == minor => { self.skip()?; let count = self.wirelength(arg)?; Ok(self.readbytes(count)?) } (Op::Misc(2), arg) => match Op::try_from(arg)? { Op::Atom(actual_minor) if actual_minor == minor => { self.skip()?; Ok(Cow::Owned(self.gather_chunks()?)) } _ => Err(self.expected(k)), }, _ => Err(self.expected(k)), } } fn next_compound(&mut self, minor: CompoundMinor, k: ExpectedKind) -> ReaderResult> { match self.peek_next_nonannotation_op()? { (Op::Compound(actual_minor), arg) if actual_minor == minor => { self.skip()?; Ok(CompoundBody::counted(minor, self.wirelength(arg)?)) } (Op::Misc(2), arg) => match Op::try_from(arg)? { Op::Compound(actual_minor) if actual_minor == minor => { self.skip()?; Ok(CompoundBody::streaming(minor)) } _ => Err(self.expected(k)), }, _ => Err(self.expected(k)), } } fn read_number_format_b(&mut self, count: usize) -> IOResult { if count == 0 { return Ok(SignedInteger::from(0 as 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 = 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 = 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(&mut self, f: F) -> ReaderResult where F: FnOnce(u128) -> Option { match self.peek_next_nonannotation_op()? { (Op::Misc(3), arg) => { self.skip()?; if arg > 12 { Err(out_of_range((arg as i8) - 16)) } else { f(arg as u128).ok_or_else(|| out_of_range(arg)) } } (Op::Atom(AtomMinor::SignedInteger), arg) => { self.skip()?; let count = self.wirelength(arg)?; let n = &self.read_number_format_b(count)?; let i = n.try_into().or_else(|_| Err(out_of_range(n)))?; f(i).ok_or_else(|| out_of_range(i)) } _ => { let n_value = self.demand_next(false)?; let n = n_value.value().to_signedinteger()?; let i = n.try_into().or_else(|_| Err(out_of_range(n)))?; f(i).ok_or_else(|| out_of_range(i)) } } } fn next_signed(&mut self, f: F) -> ReaderResult where F: FnOnce(i128) -> Option { match self.peek_next_nonannotation_op()? { (Op::Misc(3), arg) => { self.skip()?; let n = arg as i128; let n = if n > 12 { n - 16 } else { n }; f(n).ok_or_else(|| out_of_range(n)) } (Op::Atom(AtomMinor::SignedInteger), arg) => { self.skip()?; let count = self.wirelength(arg)?; let n = &self.read_number_format_b(count)?; let i = n.try_into().or_else(|_| Err(out_of_range(n)))?; f(i).ok_or_else(|| out_of_range(i)) } _ => { let n_value = self.demand_next(false)?; let n = n_value.value().to_signedinteger()?; let i = n.try_into().or_else(|_| Err(out_of_range(n)))?; f(i).ok_or_else(|| out_of_range(i)) } } } } impl<'de, S: BinarySource<'de>> Reader<'de> for PackedReader<'de, S> { type CompoundInfo = CompoundMinor; fn next(&mut self, read_annotations: bool) -> IOResult> { match self.peek() { Err(e) if is_eof_io_error(&e) => return Ok(None), Err(e) => return Err(e), Ok(_) => (), } loop { return Ok(Some(match decodeop(self.read()?)? { (Op::Misc(0), 0) => FALSE.clone(), (Op::Misc(0), 1) => TRUE.clone(), (Op::Misc(0), 2) => { let mut bs = [0; 4]; self.readbytes_into(&mut bs)?; Value::from(f32::from_bits(u32::from_be_bytes(bs))).wrap() } (Op::Misc(0), 3) => { let mut bs = [0; 8]; self.readbytes_into(&mut bs)?; Value::from(f64::from_bits(u64::from_be_bytes(bs))).wrap() } (Op::Misc(0), 5) => { if read_annotations { let mut annotations = vec![self.demand_next(read_annotations)?]; while decodeop(self.peek()?)? == (Op::Misc(0), 5) { 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()?; continue; } } (Op::Misc(0), _) => Err(io_syntax_error("Invalid format A encoding"))?, (Op::Misc(1), _) => Err(io_syntax_error("Invalid format A encoding"))?, (Op::Misc(2), arg) => match Op::try_from(arg)? { Op::Atom(minor) => decodebinary(minor, Cow::Owned(self.gather_chunks()?))?, Op::Compound(minor) => decodecompound(minor, DelimitedStream { reader: self.configured(read_annotations), })?, _ => Err(io_syntax_error("Invalid format C start byte"))?, } (Op::Misc(3), arg) => { let n = if arg > 12 { i32::from(arg) - 16 } else { i32::from(arg) }; // TODO: prebuild these in value.rs Value::from(n).wrap() } (Op::Misc(_), _) => unreachable!(), (Op::Atom(AtomMinor::SignedInteger), arg) => { let count = self.wirelength(arg)?; let n = self.read_number_format_b(count)?; Value::SignedInteger(n).wrap() } (Op::Atom(minor), arg) => { let count = self.wirelength(arg)?; decodebinary(minor, self.readbytes(count)?)? } (Op::Compound(minor), arg) => { let count = self.wirelength(arg)?; decodecompound(minor, CountedStream { reader: self.configured(read_annotations), count, })? } (Op::Reserved(3), 15) => continue, (Op::Reserved(_), _) => return Err(InvalidOp.into()), })) } } fn open_record(&mut self, arity: Option) -> ReaderResult> { if let Some(expected_arity) = arity { let compound_format = self.next_compound(CompoundMinor::Record, ExpectedKind::Record(arity))?; if let CompoundLimit::Counted(count) = compound_format.limit { if count != expected_arity + 1 /* we add 1 for the label */ { return Err(error::Error::Expected(ExpectedKind::Record(arity), Received::ReceivedSomethingElse)); } } Ok(compound_format) } else { self.next_compound(CompoundMinor::Record, ExpectedKind::Record(None)) } } fn open_sequence_or_set(&mut self) -> ReaderResult> { match self.peek_next_nonannotation_op()? { (Op::Compound(minor), arg) if CompoundMinor::Sequence == minor || CompoundMinor::Set == minor => { self.skip()?; Ok(CompoundBody::counted(minor, self.wirelength(arg)?)) } (Op::Misc(2), arg) => match Op::try_from(arg)? { Op::Compound(minor) if CompoundMinor::Sequence == minor || CompoundMinor::Set == minor => { self.skip()?; Ok(CompoundBody::streaming(minor)) } _ => Err(self.expected(ExpectedKind::SequenceOrSet)), } _ => Err(self.expected(ExpectedKind::SequenceOrSet)), } } fn open_sequence(&mut self) -> ReaderResult> { self.next_compound(CompoundMinor::Sequence, ExpectedKind::Sequence) } fn open_set(&mut self) -> ReaderResult> { self.next_compound(CompoundMinor::Set, ExpectedKind::Set) } fn open_dictionary(&mut self) -> ReaderResult> { self.next_compound(CompoundMinor::Dictionary, ExpectedKind::Dictionary) } fn close_compound_counted(&mut self, _minor: Self::CompoundInfo) -> ReaderResult<()> { // Nothing to do -- no close delimiter to consume Ok(()) } fn close_compound_stream(&mut self, _minor: Self::CompoundInfo) -> ReaderResult { Ok(self.peekend()?) } fn next_boolean(&mut self) -> ReaderResult { match self.peek_next_nonannotation_op()? { (Op::Misc(0), 0) => { self.skip()?; Ok(false) } (Op::Misc(0), 1) => { self.skip()?; Ok(true) } _ => Err(self.expected(ExpectedKind::Boolean)), } } fn next_i8(&mut self) -> ReaderResult { self.next_signed(|n| n.to_i8()) } fn next_i16(&mut self) -> ReaderResult { self.next_signed(|n| n.to_i16()) } fn next_i32(&mut self) -> ReaderResult { self.next_signed(|n| n.to_i32()) } fn next_i64(&mut self) -> ReaderResult { self.next_signed(|n| n.to_i64()) } fn next_i128(&mut self) -> ReaderResult { self.next_signed(|n| n.to_i128()) } fn next_u8(&mut self) -> ReaderResult { self.next_unsigned(|n| n.to_u8()) } fn next_u16(&mut self) -> ReaderResult { self.next_unsigned(|n| n.to_u16()) } fn next_u32(&mut self) -> ReaderResult { self.next_unsigned(|n| n.to_u32()) } fn next_u64(&mut self) -> ReaderResult { self.next_unsigned(|n| n.to_u64()) } fn next_u128(&mut self) -> ReaderResult { self.next_unsigned(|n| n.to_u128()) } fn next_float(&mut self) -> ReaderResult { match self.peek_next_nonannotation_op()? { (Op::Misc(0), 2) => { self.skip()?; let mut bs = [0; 4]; self.readbytes_into(&mut bs)?; Ok(f32::from_bits(u32::from_be_bytes(bs))) }, (Op::Misc(0), 3) => { 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_double(&mut self) -> ReaderResult { match self.peek_next_nonannotation_op()? { (Op::Misc(0), 2) => { self.skip()?; let mut bs = [0; 4]; self.readbytes_into(&mut bs)?; Ok(f32::from_bits(u32::from_be_bytes(bs)) as f64) }, (Op::Misc(0), 3) => { 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> { Ok(decodestr(self.next_atomic(AtomMinor::String, ExpectedKind::Symbol)?)?) } fn next_bytestring(&mut self) -> ReaderResult> { self.next_atomic(AtomMinor::ByteString, ExpectedKind::Symbol) } fn next_symbol(&mut self) -> ReaderResult> { Ok(decodestr(self.next_atomic(AtomMinor::Symbol, ExpectedKind::Symbol)?)?) } } struct CountedStream<'de, R: Reader<'de>> { reader: ConfiguredReader<'de, R>, count: usize, } impl<'de, R: Reader<'de>> Iterator for CountedStream<'de, R> { type Item = IOResult; fn next(&mut self) -> Option { if self.count == 0 { return None } self.count -= 1; Some(self.reader.reader.demand_next(self.reader.read_annotations)) } } 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; fn next(&mut self) -> Option { 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)), } } } pub fn decodeop(b: u8) -> IOResult<(Op, u8)> { Ok((Op::try_from(b >> 4)?, b & 15)) } pub fn decodestr<'de>(cow: Cow<'de, [u8]>) -> IOResult> { 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"))?.to_owned())), } } pub fn decodebinary<'de>(minor: AtomMinor, bs: Cow<'de, [u8]>) -> IOResult { Ok(match minor { AtomMinor::SignedInteger => Value::from(&BigInt::from_signed_bytes_be(&bs)).wrap(), AtomMinor::String => Value::String(decodestr(bs)?.into_owned()).wrap(), AtomMinor::ByteString => Value::ByteString(bs.into_owned()).wrap(), AtomMinor::Symbol => Value::Symbol(decodestr(bs)?.into_owned()).wrap(), }) } pub fn decodecompound<'de, I: Iterator>>( minor: CompoundMinor, mut iter: I ) -> IOResult { match minor { CompoundMinor::Record => { let vs = iter.collect::>>()?; if vs.len() < 1 { Err(io_syntax_error("Too few elements in encoded record")) } else { Ok(Value::Record(Record(vs)).wrap()) } } CompoundMinor::Sequence => { let vs = iter.collect::>>()?; Ok(Value::Sequence(vs).wrap()) } CompoundMinor::Set => { let mut s = Set::new(); for res in iter { s.insert(res?); } Ok(Value::Set(s).wrap()) } CompoundMinor::Dictionary => { 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")), } } Ok(Value::Dictionary(d).wrap()) } } }