|
| 1 | +pub(crate) use _bz2::make_module; |
| 2 | + |
| 3 | +#[pymodule] |
| 4 | +mod _bz2 { |
| 5 | + use crate::common::lock::PyMutex; |
| 6 | + use crate::vm::{ |
| 7 | + builtins::{PyBytesRef, PyTypeRef}, |
| 8 | + function::{ArgBytesLike, OptionalArg}, |
| 9 | + object::{PyPayload, PyResult}, |
| 10 | + types::Constructor, |
| 11 | + VirtualMachine, |
| 12 | + }; |
| 13 | + use bzip2::{write::BzEncoder, Decompress, Status}; |
| 14 | + use std::{fmt, io::Write}; |
| 15 | + |
| 16 | + // const BUFSIZ: i32 = 8192; |
| 17 | + |
| 18 | + struct DecompressorState { |
| 19 | + decoder: Decompress, |
| 20 | + eof: bool, |
| 21 | + needs_input: bool, |
| 22 | + // input_buffer: Vec<u8>, |
| 23 | + // output_buffer: Vec<u8>, |
| 24 | + } |
| 25 | + |
| 26 | + #[pyattr] |
| 27 | + #[pyclass(name = "BZ2Decompressor")] |
| 28 | + #[derive(PyPayload)] |
| 29 | + struct BZ2Decompressor { |
| 30 | + state: PyMutex<DecompressorState>, |
| 31 | + } |
| 32 | + |
| 33 | + impl fmt::Debug for BZ2Decompressor { |
| 34 | + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 35 | + write!(f, "_bz2.BZ2Decompressor") |
| 36 | + } |
| 37 | + } |
| 38 | + |
| 39 | + impl Constructor for BZ2Decompressor { |
| 40 | + type Args = (); |
| 41 | + |
| 42 | + fn py_new(cls: PyTypeRef, _: Self::Args, vm: &VirtualMachine) -> PyResult { |
| 43 | + Self { |
| 44 | + state: PyMutex::new(DecompressorState { |
| 45 | + decoder: Decompress::new(false), |
| 46 | + eof: false, |
| 47 | + needs_input: true, |
| 48 | + // input_buffer: Vec::new(), |
| 49 | + // output_buffer: Vec::new(), |
| 50 | + }), |
| 51 | + } |
| 52 | + .into_ref_with_type(vm, cls) |
| 53 | + .map(Into::into) |
| 54 | + } |
| 55 | + } |
| 56 | + |
| 57 | + #[pyclass(with(Constructor))] |
| 58 | + impl BZ2Decompressor { |
| 59 | + #[pymethod] |
| 60 | + fn decompress( |
| 61 | + &self, |
| 62 | + data: ArgBytesLike, |
| 63 | + // TODO: PyIntRef |
| 64 | + max_length: OptionalArg<i32>, |
| 65 | + vm: &VirtualMachine, |
| 66 | + ) -> PyResult<PyBytesRef> { |
| 67 | + let max_length = max_length.unwrap_or(-1); |
| 68 | + if max_length >= 0 { |
| 69 | + return Err(vm.new_not_implemented_error( |
| 70 | + "the max_value argument is not implemented yet".to_owned(), |
| 71 | + )); |
| 72 | + } |
| 73 | + // let max_length = if max_length < 0 || max_length >= BUFSIZ { |
| 74 | + // BUFSIZ |
| 75 | + // } else { |
| 76 | + // max_length |
| 77 | + // }; |
| 78 | + |
| 79 | + let mut state = self.state.lock(); |
| 80 | + let DecompressorState { |
| 81 | + decoder, |
| 82 | + eof, |
| 83 | + .. |
| 84 | + // needs_input, |
| 85 | + // input_buffer, |
| 86 | + // output_buffer, |
| 87 | + } = &mut *state; |
| 88 | + |
| 89 | + if *eof { |
| 90 | + return Err(vm.new_exception_msg( |
| 91 | + vm.ctx.exceptions.eof_error.to_owned(), |
| 92 | + "End of stream already reached".to_owned(), |
| 93 | + )); |
| 94 | + } |
| 95 | + |
| 96 | + // data.with_ref(|data| input_buffer.extend(data)); |
| 97 | + |
| 98 | + // If max_length is negative: |
| 99 | + // read the input X bytes at a time, compress it and append it to output. |
| 100 | + // Once you're out of input, setting needs_input to true and return the |
| 101 | + // output as bytes. |
| 102 | + // |
| 103 | + // TODO: |
| 104 | + // If max_length is non-negative: |
| 105 | + // Read the input X bytes at a time, compress it and append it to |
| 106 | + // the output. If output reaches `max_length` in size, return |
| 107 | + // it (up to max_length), and store the rest of the output |
| 108 | + // for later. |
| 109 | + |
| 110 | + // TODO: arbitrary choice, not the right way to do it. |
| 111 | + let mut buf = Vec::with_capacity(data.len() * 32); |
| 112 | + |
| 113 | + let before = decoder.total_in(); |
| 114 | + let res = data.with_ref(|data| decoder.decompress_vec(data, &mut buf)); |
| 115 | + let _written = (decoder.total_in() - before) as usize; |
| 116 | + |
| 117 | + let res = match res { |
| 118 | + Ok(x) => x, |
| 119 | + // TODO: error message |
| 120 | + _ => return Err(vm.new_os_error("Invalid data stream".to_owned())), |
| 121 | + }; |
| 122 | + |
| 123 | + if res == Status::StreamEnd { |
| 124 | + *eof = true; |
| 125 | + } |
| 126 | + Ok(vm.ctx.new_bytes(buf.to_vec())) |
| 127 | + } |
| 128 | + |
| 129 | + #[pyproperty] |
| 130 | + fn eof(&self) -> bool { |
| 131 | + let state = self.state.lock(); |
| 132 | + state.eof |
| 133 | + } |
| 134 | + |
| 135 | + #[pyproperty] |
| 136 | + fn unused_data(&self, vm: &VirtualMachine) -> PyBytesRef { |
| 137 | + // Data found after the end of the compressed stream. |
| 138 | + // If this attribute is accessed before the end of the stream |
| 139 | + // has been reached, its value will be b''. |
| 140 | + vm.ctx.new_bytes(b"".to_vec()) |
| 141 | + // alternatively, be more honest: |
| 142 | + // Err(vm.new_not_implemented_error( |
| 143 | + // "unused_data isn't implemented yet".to_owned(), |
| 144 | + // )) |
| 145 | + // |
| 146 | + // TODO |
| 147 | + // let state = self.state.lock(); |
| 148 | + // if state.eof { |
| 149 | + // vm.ctx.new_bytes(state.input_buffer.to_vec()) |
| 150 | + // else { |
| 151 | + // vm.ctx.new_bytes(b"".to_vec()) |
| 152 | + // } |
| 153 | + } |
| 154 | + |
| 155 | + #[pyproperty] |
| 156 | + fn needs_input(&self) -> bool { |
| 157 | + // False if the decompress() method can provide more |
| 158 | + // decompressed data before requiring new uncompressed input. |
| 159 | + let state = self.state.lock(); |
| 160 | + state.needs_input |
| 161 | + } |
| 162 | + |
| 163 | + // TODO: mro()? |
| 164 | + } |
| 165 | + |
| 166 | + struct CompressorState { |
| 167 | + flushed: bool, |
| 168 | + encoder: Option<BzEncoder<Vec<u8>>>, |
| 169 | + } |
| 170 | + |
| 171 | + #[pyattr] |
| 172 | + #[pyclass(name = "BZ2Compressor")] |
| 173 | + #[derive(PyPayload)] |
| 174 | + struct BZ2Compressor { |
| 175 | + state: PyMutex<CompressorState>, |
| 176 | + } |
| 177 | + |
| 178 | + impl fmt::Debug for BZ2Compressor { |
| 179 | + fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
| 180 | + write!(f, "_bz2.BZ2Compressor") |
| 181 | + } |
| 182 | + } |
| 183 | + |
| 184 | + impl Constructor for BZ2Compressor { |
| 185 | + type Args = (OptionalArg<i32>,); |
| 186 | + |
| 187 | + fn py_new(cls: PyTypeRef, args: Self::Args, vm: &VirtualMachine) -> PyResult { |
| 188 | + let (compresslevel,) = args; |
| 189 | + // TODO: seriously? |
| 190 | + // compresslevel.unwrap_or(bzip2::Compression::best().level().try_into().unwrap()); |
| 191 | + let compresslevel = compresslevel.unwrap_or(9); |
| 192 | + let level = match compresslevel { |
| 193 | + valid_level @ 1..=9 => bzip2::Compression::new(valid_level as u32), |
| 194 | + _ => { |
| 195 | + return Err( |
| 196 | + vm.new_value_error("compresslevel must be between 1 and 9".to_owned()) |
| 197 | + ) |
| 198 | + } |
| 199 | + }; |
| 200 | + |
| 201 | + Self { |
| 202 | + state: PyMutex::new(CompressorState { |
| 203 | + flushed: false, |
| 204 | + encoder: Some(BzEncoder::new(Vec::new(), level)), |
| 205 | + }), |
| 206 | + } |
| 207 | + .into_ref_with_type(vm, cls) |
| 208 | + .map(Into::into) |
| 209 | + } |
| 210 | + } |
| 211 | + |
| 212 | + // TODO: return partial results from compress() instead of returning everything in flush() |
| 213 | + #[pyclass(with(Constructor))] |
| 214 | + impl BZ2Compressor { |
| 215 | + #[pymethod] |
| 216 | + fn compress(&self, data: ArgBytesLike, vm: &VirtualMachine) -> PyResult<PyBytesRef> { |
| 217 | + let mut state = self.state.lock(); |
| 218 | + if state.flushed { |
| 219 | + return Err(vm.new_value_error("Compressor has been flushed".to_owned())); |
| 220 | + } |
| 221 | + |
| 222 | + // let CompressorState { flushed, encoder } = &mut *state; |
| 223 | + let CompressorState { encoder, .. } = &mut *state; |
| 224 | + |
| 225 | + // TODO: handle Err |
| 226 | + data.with_ref(|input_bytes| encoder.as_mut().unwrap().write_all(input_bytes).unwrap()); |
| 227 | + Ok(vm.ctx.new_bytes(Vec::new())) |
| 228 | + } |
| 229 | + |
| 230 | + #[pymethod] |
| 231 | + fn flush(&self, vm: &VirtualMachine) -> PyResult<PyBytesRef> { |
| 232 | + let mut state = self.state.lock(); |
| 233 | + if state.flushed { |
| 234 | + return Err(vm.new_value_error("Repeated call to flush()".to_owned())); |
| 235 | + } |
| 236 | + |
| 237 | + // let CompressorState { flushed, encoder } = &mut *state; |
| 238 | + let CompressorState { encoder, .. } = &mut *state; |
| 239 | + |
| 240 | + // TODO: handle Err |
| 241 | + let out = encoder.take().unwrap().finish().unwrap(); |
| 242 | + state.flushed = true; |
| 243 | + Ok(vm.ctx.new_bytes(out.to_vec())) |
| 244 | + } |
| 245 | + } |
| 246 | +} |
0 commit comments