|
| 1 | +# Copyright 2019 The Emscripten Authors. All rights reserved. |
| 2 | +# Emscripten is available under two separate licenses, the MIT license and the |
| 3 | +# University of Illinois/NCSA Open Source License. Both these licenses can be |
| 4 | +# found in the LICENSE file. |
| 5 | + |
| 6 | +"""Utility functions for parsing 'ar' files. |
| 7 | +
|
| 8 | +This is needed in emsripten because command line tools such as llvm-ar are not |
| 9 | +able to deal with archives containing many files with the same name. Despite |
| 10 | +this linker are expected to handle this case and emscripten needs to emulate |
| 11 | +linker behaviour when using the fastcomp backend. |
| 12 | +
|
| 13 | +See https://en.wikipedia.org/wiki/Ar_(Unix) |
| 14 | +
|
| 15 | +""" |
| 16 | + |
| 17 | +from __future__ import print_function |
| 18 | + |
| 19 | +from __builtin__ import open as builtin_open |
| 20 | +import struct |
| 21 | +import os |
| 22 | +import sys |
| 23 | + |
| 24 | +MAGIC = '!<arch>\n' |
| 25 | + |
| 26 | + |
| 27 | +class ArError(Exception): |
| 28 | + """Base exception.""" |
| 29 | + pass |
| 30 | + |
| 31 | + |
| 32 | +class ArInfo(object): |
| 33 | + def __init__(self, name, offset, timestamp, owner, group, mode, size, data): |
| 34 | + self.name = name |
| 35 | + self.offset = offset |
| 36 | + self.timestamp = timestamp |
| 37 | + self.owner = owner |
| 38 | + self.group = group |
| 39 | + self.mode = mode |
| 40 | + self.size = size |
| 41 | + self.data = data |
| 42 | + |
| 43 | + |
| 44 | +class ArFile(object): |
| 45 | + def __init__(self, filename): |
| 46 | + self.filename = filename |
| 47 | + self._file = builtin_open(filename, 'r') |
| 48 | + magic = self._file.read(len(MAGIC)) |
| 49 | + if MAGIC != magic: |
| 50 | + raise ArError('not an ar file: ' + filename) |
| 51 | + self.members = [] |
| 52 | + self.members_map = {} |
| 53 | + self.offset_to_info = {} |
| 54 | + |
| 55 | + def _read_member(self): |
| 56 | + offset = self._file.tell() |
| 57 | + name = self._file.read(16) |
| 58 | + if len(name) == 0: |
| 59 | + return None |
| 60 | + name = name.strip() |
| 61 | + timestamp = self._file.read(12).strip() |
| 62 | + owner = self._file.read(6).strip() |
| 63 | + group = self._file.read(6).strip() |
| 64 | + mode = self._file.read(8).strip() |
| 65 | + size = int(self._file.read(10)) |
| 66 | + ending = self._file.read(2) |
| 67 | + if ending != '\x60\n': |
| 68 | + raise ArError('invalid ar header') |
| 69 | + data = self._file.read(size) |
| 70 | + if mode.strip(): |
| 71 | + mode = int(mode) |
| 72 | + if owner.strip(): |
| 73 | + owner = int(owner) |
| 74 | + if group.strip(): |
| 75 | + group = int(group) |
| 76 | + if size % 2: |
| 77 | + if self._file.read(1) != '\n': |
| 78 | + raise ArError('invalid ar header') |
| 79 | + |
| 80 | + return ArInfo(name, offset, timestamp, owner, group, mode, size, data) |
| 81 | + |
| 82 | + def next(self): |
| 83 | + while True: |
| 84 | + # Keep reading entries until we find a non-special one |
| 85 | + info = self._read_member() |
| 86 | + if not info: |
| 87 | + return None |
| 88 | + if info.name == '//': |
| 89 | + # Special file containing long filenames |
| 90 | + self.name_data = info.data |
| 91 | + elif info.name == '/': |
| 92 | + # Special file containing symbol table |
| 93 | + num_entries = struct.unpack('>I', info.data[:4])[0] |
| 94 | + self.sym_offsets = struct.unpack('>%dI' % num_entries, info.data[4:4 + 4 * num_entries]) |
| 95 | + symbol_data = info.data[4 + 4 * num_entries:-1] |
| 96 | + self.symbols = symbol_data.rstrip('\0').split('\0') |
| 97 | + if len(self.symbols) != num_entries: |
| 98 | + raise ArError('invalid symbol table') |
| 99 | + else: |
| 100 | + break |
| 101 | + |
| 102 | + # This entry has a name from the "//" name section. |
| 103 | + if info.name[0] == '/': |
| 104 | + name_offset = int(info.name[1:]) |
| 105 | + if name_offset < 0 or name_offset >= len(self.name_data): |
| 106 | + raise ArError('invalid extended filename section') |
| 107 | + name_end = self.name_data.find('\n', name_offset) |
| 108 | + info.name = self.name_data[name_offset:name_end] |
| 109 | + info.name = info.name.rstrip('/') |
| 110 | + self.members.append(info) |
| 111 | + self.members_map[info.name] = info |
| 112 | + self.offset_to_info[info.offset] = info |
| 113 | + return info |
| 114 | + |
| 115 | + def getsymbols(self): |
| 116 | + return zip(self.symbols, self.sym_offsets) |
| 117 | + |
| 118 | + def getmember(self, id): |
| 119 | + """Polymophic member accessor that takes either and index or a name.""" |
| 120 | + if isinstance(id, int): |
| 121 | + return self.getmember_by_index(id) |
| 122 | + return self.getmember_by_name(id) |
| 123 | + |
| 124 | + def getmember_by_name(self, name): |
| 125 | + self.getmembers() |
| 126 | + return self.members_map[name] |
| 127 | + |
| 128 | + def getmember_by_index(self, index): |
| 129 | + self.getmembers() |
| 130 | + return self.members[index] |
| 131 | + |
| 132 | + def getmembers(self): |
| 133 | + while self.next(): |
| 134 | + pass |
| 135 | + return self.members |
| 136 | + |
| 137 | + def list(self): |
| 138 | + for m in self.getmembers(): |
| 139 | + sys.stdout.write(m.name + '\n') |
| 140 | + |
| 141 | + def extractall(self, path="."): |
| 142 | + names_written = set() |
| 143 | + for m in self.getmembers(): |
| 144 | + filename = m.name |
| 145 | + if filename in names_written: |
| 146 | + basename = filename |
| 147 | + count = 1 |
| 148 | + while filename in names_written: |
| 149 | + filename = basename + '.' + str(count) |
| 150 | + count += 1 |
| 151 | + |
| 152 | + names_written.add(filename) |
| 153 | + full_name = os.path.join(path, filename) |
| 154 | + with builtin_open(full_name, 'w') as f: |
| 155 | + f.write(m.data) |
| 156 | + |
| 157 | + return list(names_written) |
| 158 | + |
| 159 | + def close(self): |
| 160 | + self._file.close() |
| 161 | + |
| 162 | + def __enter__(self): |
| 163 | + return self |
| 164 | + |
| 165 | + def __exit__(self, type, value, traceback): |
| 166 | + self.close() |
| 167 | + |
| 168 | + |
| 169 | +def open(filename): |
| 170 | + return ArFile(filename) |
| 171 | + |
| 172 | + |
| 173 | +def is_arfile(filename): |
| 174 | + """Return True if name points to a ar archive that we |
| 175 | + are able to handle, else return False. |
| 176 | + """ |
| 177 | + try: |
| 178 | + t = open(filename) |
| 179 | + t.close() |
| 180 | + return True |
| 181 | + except ArError: |
| 182 | + return False |
0 commit comments