use crate::{ValueClass, AtomClass, Atom}; use crate::error::{self, ExpectedKind}; use num_bigint::BigInt; use num_traits::cast::{FromPrimitive, ToPrimitive}; use std::borrow::Cow; use std::convert::TryFrom; use std::convert::TryInto; use std::io; use std::marker::PhantomData; use super::constants::Tag; use super::super::{ CompoundClass, boundary as B, reader::{ NextToken, Reader, ReaderResult, }, signed_integer::SignedInteger, source::BinarySource, }; pub struct PackedReader<'de, S: BinarySource<'de>> { pub source: S, phantom: PhantomData<&'de ()>, } impl<'de, S: BinarySource<'de>> BinarySource<'de> for PackedReader<'de, S> { #[inline(always)] fn mark(&mut self) -> io::Result { self.source.mark() } #[inline(always)] fn restore(&mut self, mark: usize) -> io::Result<()> { self.source.restore(mark) } #[inline(always)] fn skip(&mut self) -> io::Result<()> { self.source.skip() } #[inline(always)] fn peek(&mut self) -> io::Result> { self.source.peek() } #[inline(always)] fn readbytes(&mut self, count: u64) -> io::Result> { self.source.readbytes(count) } #[inline(always)] fn readbytes_into(&mut self, bs: &mut [u8]) -> io::Result<()> { self.source.readbytes_into(bs) } #[inline(always)] fn input_position(&mut self) -> io::Result> { self.source.input_position() } #[inline(always)] fn discard(&mut self, count: u64) -> io::Result<()> { self.source.discard(count) } #[inline(always)] fn read_to_end(&mut self) -> io::Result> { self.source.read_to_end() } } fn out_of_range>(i: I) -> error::Error { error::Error::NumberOutOfRange(i.into()) } impl<'de, S: BinarySource<'de>> PackedReader<'de, S> { #[inline(always)] pub fn new(source: S) -> Self { PackedReader { source, phantom: PhantomData } } #[inline(always)] fn varint(&mut self) -> io::Result { let mut shift = 0; let mut acc: u64 = 0; loop { let v = self.read()?; if shift == 63 && v > 1 { Err(error::Error::Message("PackedReader length too long".to_string()))? } acc |= ((v & 0x7f) as u64) << shift; shift += 7; if v & 0x80 == 0 { return Ok(acc) } if shift >= 70 { Err(error::Error::Message("PackedReader length too long".to_string()))? } } } #[inline(always)] fn peekend(&mut self) -> io::Result { if self.peek()? == Some(Tag::End.into()) { self.skip()?; Ok(true) } else { Ok(false) } } #[inline(always)] fn peek_next_nonannotation_tag(&mut self) -> ReaderResult { loop { match Tag::try_from(self.peek_noeof()?)? { Tag::Annotation => { self.skip()?; self.skip_value()?; }, other => return Ok(other), } } } fn next_atomic(&mut self, expected_tag: Tag, k: ExpectedKind) -> ReaderResult> { 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(error::Error::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(error::Error::Expected(k)) } } #[inline(always)] fn read_signed_integer(&mut self, count: u64) -> io::Result { let count_u64 = count; let count: usize = count.try_into().map_err( |_| error::Error::Message("Signed integer too long".to_string()))?; if count == 0 { return Ok(SignedInteger::from(0_i128)); } if count > 16 { let bs = self.readbytes(count_u64)?; 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))) } } #[inline(always)] fn next_unsigned(&mut self, f: F) -> ReaderResult where F: FnOnce(u128) -> Option { 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(error::Error::Expected(ExpectedKind::SignedInteger)) } } #[inline(always)] fn next_signed(&mut self, f: F) -> ReaderResult where F: FnOnce(i128) -> Option { 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(error::Error::Expected(ExpectedKind::SignedInteger)) } } #[inline(always)] fn decodestr<'a>(&mut self, cow: Cow<'a, [u8]>) -> io::Result> { match cow { Cow::Borrowed(bs) => Ok(Cow::Borrowed(std::str::from_utf8(bs).map_err(|_| self.syntax_error("Invalid UTF-8"))?)), Cow::Owned(bs) => Ok(Cow::Owned(String::from_utf8(bs).map_err(|_| self.syntax_error("Invalid UTF-8"))?)), } } } impl From for Option { #[inline(always)] fn from(tag: Tag) -> Self { match tag { Tag::False => Some(NextToken::Value(ValueClass::Atomic(AtomClass::Boolean))), Tag::True => Some(NextToken::Value(ValueClass::Atomic(AtomClass::Boolean))), Tag::Float => Some(NextToken::Value(ValueClass::Atomic(AtomClass::Float))), Tag::Double => Some(NextToken::Value(ValueClass::Atomic(AtomClass::Double))), Tag::Annotation => Some(NextToken::Annotation), Tag::Embedded => Some(NextToken::Value(ValueClass::Embedded)), Tag::SmallInteger(_) => Some(NextToken::Value(ValueClass::Atomic(AtomClass::SignedInteger))), Tag::MediumInteger(_) => Some(NextToken::Value(ValueClass::Atomic(AtomClass::SignedInteger))), Tag::SignedInteger => Some(NextToken::Value(ValueClass::Atomic(AtomClass::SignedInteger))), Tag::String => Some(NextToken::Value(ValueClass::Atomic(AtomClass::String))), Tag::ByteString => Some(NextToken::Value(ValueClass::Atomic(AtomClass::ByteString))), Tag::Symbol => Some(NextToken::Value(ValueClass::Atomic(AtomClass::Symbol))), Tag::Record => Some(NextToken::Value(ValueClass::Compound(CompoundClass::Record))), Tag::Sequence => Some(NextToken::Value(ValueClass::Compound(CompoundClass::Sequence))), Tag::Set => Some(NextToken::Value(ValueClass::Compound(CompoundClass::Set))), Tag::Dictionary => Some(NextToken::Value(ValueClass::Compound(CompoundClass::Dictionary))), Tag::End => None, } } } impl<'de, S: BinarySource<'de>> Reader<'de> for PackedReader<'de, S> { fn peek_class(&mut self) -> io::Result> { let tag = Tag::try_from(match self.peek()? { None => return Ok(None), Some(b) => b, })?; match tag.into() { Some(n) => Ok(Some(n)), None => Err(self.syntax_error("Unexpected tag")), } } fn next_atom(&mut self) -> ReaderResult> { match Tag::try_from(self.read()?)? { Tag::False => Ok(Atom::Boolean(false)), Tag::True => Ok(Atom::Boolean(true)), Tag::Float => { let mut bs = [0; 4]; self.readbytes_into(&mut bs)?; Ok(Atom::Float(f32::from_bits(u32::from_be_bytes(bs)))) } Tag::Double => { let mut bs = [0; 8]; self.readbytes_into(&mut bs)?; Ok(Atom::Double(f64::from_bits(u64::from_be_bytes(bs)))) } Tag::SmallInteger(v) => Ok(Atom::SignedInteger(Cow::Owned(v.into()))), Tag::MediumInteger(count) => Ok(Atom::SignedInteger(Cow::Owned(self.read_signed_integer(count.into())?))), Tag::SignedInteger => { let count = self.varint()?; Ok(Atom::SignedInteger(Cow::Owned(self.read_signed_integer(count)?))) } Tag::String => { let count = self.varint()?; let bs = self.readbytes(count)?; Ok(Atom::String(self.decodestr(bs)?)) } Tag::ByteString => { let count = self.varint()?; Ok(Atom::ByteString(self.readbytes(count)?)) } Tag::Symbol => { let count = self.varint()?; let bs = self.readbytes(count)?; Ok(Atom::Symbol(self.decodestr(bs)?)) } _ => Err(self.syntax_error("Expected atomic value"))?, } } #[inline(always)] fn boundary(&mut self, _b: &B::Type) -> ReaderResult<()> { Ok(()) } #[inline(always)] fn open_record(&mut self) -> ReaderResult<()> { self.next_compound(Tag::Record, ExpectedKind::Record) } #[inline(always)] fn open_sequence(&mut self) -> ReaderResult<()> { self.next_compound(Tag::Sequence, ExpectedKind::Sequence) } #[inline(always)] fn open_set(&mut self) -> ReaderResult<()> { self.next_compound(Tag::Set, ExpectedKind::Set) } #[inline(always)] fn open_dictionary(&mut self) -> ReaderResult<()> { self.next_compound(Tag::Dictionary, ExpectedKind::Dictionary) } #[inline(always)] fn close_compound(&mut self, _b: &mut B::Type, _i: &B::Item) -> ReaderResult { Ok(self.peekend()?) } #[inline(always)] fn open_embedded(&mut self) -> ReaderResult<()> { self.next_compound(Tag::Embedded, ExpectedKind::Embedded) } #[inline(always)] fn close_embedded(&mut self) -> ReaderResult<()> { Ok(()) } #[inline(always)] fn mark(&mut self) -> io::Result { self.source.mark() } #[inline(always)] fn restore(&mut self, mark: usize) -> io::Result<()> { self.source.restore(mark) } #[inline(always)] fn next_boolean(&mut self) -> ReaderResult { match self.peek_next_nonannotation_tag()? { Tag::False => { self.skip()?; Ok(false) } Tag::True => { self.skip()?; Ok(true) } _ => Err(error::Error::Expected(ExpectedKind::Boolean)), } } fn next_signedinteger(&mut self) -> ReaderResult { let tag = self.peek_next_nonannotation_tag()?; match tag { Tag::SmallInteger(v) => { self.skip()?; Ok(SignedInteger::from(v as i32)) } Tag::MediumInteger(count) => { self.skip()?; Ok(self.read_signed_integer(count.into())?) } Tag::SignedInteger => { self.skip()?; let count = self.varint()?; Ok(self.read_signed_integer(count)?) } _ => Err(error::Error::Expected(ExpectedKind::SignedInteger)) } } 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_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(error::Error::Expected(ExpectedKind::Float)), } } fn next_double(&mut self) -> ReaderResult { 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(error::Error::Expected(ExpectedKind::Double)), } } fn next_str(&mut self) -> ReaderResult> { let bs = self.next_atomic(Tag::String, ExpectedKind::Symbol)?; Ok(self.decodestr(bs)?) } fn next_bytestring(&mut self) -> ReaderResult> { self.next_atomic(Tag::ByteString, ExpectedKind::Symbol) } fn next_symbol(&mut self) -> ReaderResult> { let bs = self.next_atomic(Tag::Symbol, ExpectedKind::Symbol)?; Ok(self.decodestr(bs)?) } fn open_annotation(&mut self) -> ReaderResult<()> { if let Tag::Annotation = Tag::try_from(self.peek_noeof()?)? { Ok(self.skip()?) } else { Err(error::Error::Expected(ExpectedKind::Annotation)) } } fn close_annotation(&mut self) -> ReaderResult<()> { Ok(()) } fn specialized(&mut self) -> Option<(&str, &mut dyn BinarySource<'de>)> { Some(("packed", &mut self.source)) } }