Andrew Geissler | 5f35090 | 2021-07-23 13:09:54 -0400 | [diff] [blame] | 1 | # |
| 2 | # SPDX-License-Identifier: GPL-2.0-only |
| 3 | # |
| 4 | # Helper library to implement streaming compression and decompression using an |
| 5 | # external process |
| 6 | # |
| 7 | # This library should be used directly by end users; a wrapper library for the |
| 8 | # specific compression tool should be created |
| 9 | |
| 10 | import builtins |
| 11 | import io |
| 12 | import os |
| 13 | import subprocess |
| 14 | |
| 15 | |
| 16 | def open_wrap( |
| 17 | cls, filename, mode="rb", *, encoding=None, errors=None, newline=None, **kwargs |
| 18 | ): |
| 19 | """ |
| 20 | Open a compressed file in binary or text mode. |
| 21 | |
| 22 | Users should not call this directly. A specific compression library can use |
| 23 | this helper to provide it's own "open" command |
| 24 | |
| 25 | The filename argument can be an actual filename (a str or bytes object), or |
| 26 | an existing file object to read from or write to. |
| 27 | |
| 28 | The mode argument can be "r", "rb", "w", "wb", "x", "xb", "a" or "ab" for |
| 29 | binary mode, or "rt", "wt", "xt" or "at" for text mode. The default mode is |
| 30 | "rb". |
| 31 | |
| 32 | For binary mode, this function is equivalent to the cls constructor: |
| 33 | cls(filename, mode). In this case, the encoding, errors and newline |
| 34 | arguments must not be provided. |
| 35 | |
| 36 | For text mode, a cls object is created, and wrapped in an |
| 37 | io.TextIOWrapper instance with the specified encoding, error handling |
| 38 | behavior, and line ending(s). |
| 39 | """ |
| 40 | if "t" in mode: |
| 41 | if "b" in mode: |
| 42 | raise ValueError("Invalid mode: %r" % (mode,)) |
| 43 | else: |
| 44 | if encoding is not None: |
| 45 | raise ValueError("Argument 'encoding' not supported in binary mode") |
| 46 | if errors is not None: |
| 47 | raise ValueError("Argument 'errors' not supported in binary mode") |
| 48 | if newline is not None: |
| 49 | raise ValueError("Argument 'newline' not supported in binary mode") |
| 50 | |
| 51 | file_mode = mode.replace("t", "") |
| 52 | if isinstance(filename, (str, bytes, os.PathLike)): |
| 53 | binary_file = cls(filename, file_mode, **kwargs) |
| 54 | elif hasattr(filename, "read") or hasattr(filename, "write"): |
| 55 | binary_file = cls(None, file_mode, fileobj=filename, **kwargs) |
| 56 | else: |
| 57 | raise TypeError("filename must be a str or bytes object, or a file") |
| 58 | |
| 59 | if "t" in mode: |
| 60 | return io.TextIOWrapper( |
| 61 | binary_file, encoding, errors, newline, write_through=True |
| 62 | ) |
| 63 | else: |
| 64 | return binary_file |
| 65 | |
| 66 | |
| 67 | class CompressionError(OSError): |
| 68 | pass |
| 69 | |
| 70 | |
| 71 | class PipeFile(io.RawIOBase): |
| 72 | """ |
| 73 | Class that implements generically piping to/from a compression program |
| 74 | |
| 75 | Derived classes should add the function get_compress() and get_decompress() |
| 76 | that return the required commands. Input will be piped into stdin and the |
| 77 | (de)compressed output should be written to stdout, e.g.: |
| 78 | |
| 79 | class FooFile(PipeCompressionFile): |
| 80 | def get_decompress(self): |
| 81 | return ["fooc", "--decompress", "--stdout"] |
| 82 | |
| 83 | def get_compress(self): |
| 84 | return ["fooc", "--compress", "--stdout"] |
| 85 | |
| 86 | """ |
| 87 | |
| 88 | READ = 0 |
| 89 | WRITE = 1 |
| 90 | |
| 91 | def __init__(self, filename=None, mode="rb", *, stderr=None, fileobj=None): |
| 92 | if "t" in mode or "U" in mode: |
| 93 | raise ValueError("Invalid mode: {!r}".format(mode)) |
| 94 | |
| 95 | if not "b" in mode: |
| 96 | mode += "b" |
| 97 | |
| 98 | if mode.startswith("r"): |
| 99 | self.mode = self.READ |
| 100 | elif mode.startswith("w"): |
| 101 | self.mode = self.WRITE |
| 102 | else: |
| 103 | raise ValueError("Invalid mode %r" % mode) |
| 104 | |
| 105 | if fileobj is not None: |
| 106 | self.fileobj = fileobj |
| 107 | else: |
| 108 | self.fileobj = builtins.open(filename, mode or "rb") |
| 109 | |
| 110 | if self.mode == self.READ: |
| 111 | self.p = subprocess.Popen( |
| 112 | self.get_decompress(), |
| 113 | stdin=self.fileobj, |
| 114 | stdout=subprocess.PIPE, |
| 115 | stderr=stderr, |
| 116 | close_fds=True, |
| 117 | ) |
| 118 | self.pipe = self.p.stdout |
| 119 | else: |
| 120 | self.p = subprocess.Popen( |
| 121 | self.get_compress(), |
| 122 | stdin=subprocess.PIPE, |
| 123 | stdout=self.fileobj, |
| 124 | stderr=stderr, |
| 125 | close_fds=True, |
| 126 | ) |
| 127 | self.pipe = self.p.stdin |
| 128 | |
| 129 | self.__closed = False |
| 130 | |
| 131 | def _check_process(self): |
| 132 | if self.p is None: |
| 133 | return |
| 134 | |
| 135 | returncode = self.p.wait() |
| 136 | if returncode: |
| 137 | raise CompressionError("Process died with %d" % returncode) |
| 138 | self.p = None |
| 139 | |
| 140 | def close(self): |
| 141 | if self.closed: |
| 142 | return |
| 143 | |
| 144 | self.pipe.close() |
| 145 | if self.p is not None: |
| 146 | self._check_process() |
| 147 | self.fileobj.close() |
| 148 | |
| 149 | self.__closed = True |
| 150 | |
| 151 | @property |
| 152 | def closed(self): |
| 153 | return self.__closed |
| 154 | |
| 155 | def fileno(self): |
| 156 | return self.pipe.fileno() |
| 157 | |
| 158 | def flush(self): |
| 159 | self.pipe.flush() |
| 160 | |
| 161 | def isatty(self): |
| 162 | return self.pipe.isatty() |
| 163 | |
| 164 | def readable(self): |
| 165 | return self.mode == self.READ |
| 166 | |
| 167 | def writable(self): |
| 168 | return self.mode == self.WRITE |
| 169 | |
| 170 | def readinto(self, b): |
| 171 | if self.mode != self.READ: |
| 172 | import errno |
| 173 | |
| 174 | raise OSError( |
| 175 | errno.EBADF, "read() on write-only %s object" % self.__class__.__name__ |
| 176 | ) |
| 177 | size = self.pipe.readinto(b) |
| 178 | if size == 0: |
| 179 | self._check_process() |
| 180 | return size |
| 181 | |
| 182 | def write(self, data): |
| 183 | if self.mode != self.WRITE: |
| 184 | import errno |
| 185 | |
| 186 | raise OSError( |
| 187 | errno.EBADF, "write() on read-only %s object" % self.__class__.__name__ |
| 188 | ) |
| 189 | data = self.pipe.write(data) |
| 190 | |
| 191 | if not data: |
| 192 | self._check_process() |
| 193 | |
| 194 | return data |