test/test_extract.py

   1 """Extract tests.
   2 """
   3
   4 import io
   5 import os
   6 import sys
   7 from datetime import datetime
   8
   9 import pytest
  10
  11 import rarfile
  12
  13
  14 def get_props(rf, name):
  15     inf = rf.getinfo(name)
  16     return "".join([
  17         inf.is_file() and "F" or "-",
  18         inf.is_dir() and "D" or "-",
  19         inf.is_symlink() and "L" or "-",
  20     ])
  21
  22
  23 def san_unix(fn):
  24     return rarfile.sanitize_filename(fn, "/", False)
  25
  26
  27 def san_win32(fn):
  28     return rarfile.sanitize_filename(fn, "/", True)
  29
  30
  31 def test_sanitize_unix():
  32     assert san_unix("asd/asd") == "asd/asd"
  33     assert san_unix("asd/../asd") == "asd/asd"
  34     assert san_unix("c:/a/x") == r"c:/a/x"
  35     assert san_unix("z/./../a /b./c") == "z/a /b./c"
  36     assert san_unix("z<>*?:") == "z____:"
  37
  38
  39 def test_sanitize_win32():
  40     assert san_win32("asd/asd") == "asd/asd"
  41     assert san_win32("asd/../asd") == "asd/asd"
  42     assert san_win32("c:/a/x") == "a/x"
  43     assert san_win32("z/./../a /b./c") == "z/a_/b_/c"
  44     assert san_win32("z<>*?:\\^") == "z_______"
  45
  46
  47 def checktime(fn, exp_mtime):
  48     # cannot check subsecond precision as filesystem may not support it
  49     cut = len("0000-00-00 00:00:00")
  50     st = os.stat(fn)
  51     got_mtime = datetime.fromtimestamp(st.st_mtime, exp_mtime.tzinfo)
  52     exp_stamp = exp_mtime.isoformat(" ", "seconds")[:cut]
  53     got_stamp = got_mtime.isoformat(" ", "seconds")[:cut]
  54     assert exp_stamp == got_stamp
  55
  56
  57 def checkfile(fn, data, mtime):
  58     with open(fn, "r", encoding="utf8") as f:
  59         got = f.read()
  60         assert got.strip() == data
  61
  62     checktime(fn, mtime)
  63
  64
  65 def check_subdir(rf, tmp_path):
  66     # pre-mkdir
  67     ext1 = tmp_path / "ext1"
  68     inf = rf.getinfo("sub/dir1/file1.txt")
  69     os.mkdir(ext1)
  70     rf.extract(inf, ext1)
  71     assert sorted(os.listdir(tmp_path)) == ["ext1"]
  72     assert os.listdir(ext1 / "sub") == ["dir1"]
  73     checkfile(ext1 / "sub/dir1/file1.txt", "file1", inf.mtime)
  74
  75     # no mkdir
  76     ext2 = tmp_path / "ext2"
  77     inf = rf.getinfo("sub/dir2/file2.txt")
  78     rf.extract("sub/dir2/file2.txt", ext2)
  79     assert sorted(os.listdir(tmp_path)) == ["ext1", "ext2"]
  80     assert os.listdir(ext2 / "sub") == ["dir2"]
  81     checkfile(ext2 / "sub/dir2/file2.txt", "file2", inf.mtime)
  82
  83     # spaced
  84     ext3 = tmp_path / "ext3"
  85     inf = rf.getinfo("sub/with space/long fn.txt")
  86     rf.extract("sub/with space/long fn.txt", ext3)
  87     checkfile(ext3 / "sub/with space/long fn.txt", "long fn", inf.mtime)
  88
  89     # unicode
  90     ext4 = tmp_path / "ext4"
  91     inf = rf.getinfo("sub/üȵĩöḋè/file.txt")
  92     rf.extract("sub/üȵĩöḋè/file.txt", ext4)
  93     checkfile(ext4 / "sub/üȵĩöḋè/file.txt", "file", inf.mtime)
  94
  95     # dir only
  96     ext5 = tmp_path / "ext5"
  97     inf = rf.getinfo("sub/dir2")
  98     rf.extract("sub/dir2", ext5)
  99     assert os.listdir(ext5 / "sub") == ["dir2"]
 100     assert os.listdir(ext5 / "sub/dir2") == []
 101     checktime(ext5 / "sub/dir2", inf.mtime)
 102
 103     # cwd
 104     ext6 = tmp_path / "ext6"
 105     os.mkdir(ext6)
 106     old = os.getcwd()
 107     try:
 108         os.chdir(ext6)
 109         rf.extract("sub/dir1")
 110         assert os.listdir(".") == ["sub"]
 111         assert os.listdir("sub") == ["dir1"]
 112         assert os.listdir("sub/dir1") == []
 113     finally:
 114         os.chdir(old)
 115
 116     # errors
 117     with pytest.raises(io.UnsupportedOperation):
 118         rf.open("sub/dir1")
 119
 120
 121 @pytest.mark.parametrize("fn", [
 122     "test/files/rar3-subdirs.rar",
 123     "test/files/rar5-subdirs.rar",
 124 ])
 125 def test_subdirs(fn, tmp_path):
 126     with rarfile.RarFile(fn) as rf:
 127         check_subdir(rf, tmp_path)
 128
 129
 130 @pytest.mark.parametrize("fn", [
 131     "test/files/rar3-readonly-unix.rar",
 132     "test/files/rar3-readonly-win.rar",
 133     "test/files/rar5-readonly-unix.rar",
 134     "test/files/rar5-readonly-win.rar",
 135 ])
 136 def test_readonly(fn, tmp_path):
 137     with rarfile.RarFile(fn) as rf:
 138         assert get_props(rf, "ro_dir") == "-D-"
 139         assert get_props(rf, "ro_dir/ro_file.txt") == "F--"
 140
 141         rf.extractall(tmp_path)
 142
 143     assert os.access(tmp_path / "ro_dir/ro_file.txt", os.R_OK)
 144     assert not os.access(tmp_path / "ro_dir/ro_file.txt", os.W_OK)
 145
 146     if sys.platform != "win32":
 147         assert os.access(tmp_path / "ro_dir", os.R_OK)
 148         assert not os.access(tmp_path / "ro_dir", os.W_OK)
 149
 150
 151 @pytest.mark.parametrize("fn", [
 152     "test/files/rar3-symlink-unix.rar",
 153     "test/files/rar5-symlink-unix.rar",
 154 ])
 155 def test_symlink(fn, tmp_path):
 156     with rarfile.RarFile(fn) as rf:
 157         assert get_props(rf, "data.txt") == "F--"
 158         assert get_props(rf, "data_link") == "--L"
 159         assert get_props(rf, "random_link") == "--L"
 160
 161         rf.extractall(tmp_path)
 162
 163         assert sorted(os.listdir(tmp_path)) == ["data.txt", "data_link", "random_link"]
 164
 165         data = rf.getinfo("data.txt")
 166         data_link = rf.getinfo("data_link")
 167         random_link = rf.getinfo("random_link")
 168
 169         assert not data.is_symlink()
 170         assert data_link.is_symlink()
 171         assert random_link.is_symlink()
 172
 173         assert rf.read(data) == b"data\n"
 174         assert rf.read(data_link) == b"data.txt"
 175         assert rf.read(random_link) == b"../random123"
 176
 177         assert os.path.isfile(tmp_path / "data.txt")
 178         assert os.path.islink(tmp_path / "data_link")
 179         assert os.path.islink(tmp_path / "random_link")
 180
 181         # str - work around pypy3 bug
 182         assert os.readlink(str(tmp_path / "data_link")) == "data.txt"
 183         assert os.readlink(str(tmp_path / "random_link")) == "../random123"
 184
 185
 186 def test_symlink_win(tmp_path):
 187     fn = "test/files/rar5-symlink-win.rar"
 188     with rarfile.RarFile(fn) as rf:
 189         assert get_props(rf, "content/dir1") == "-D-"
 190         assert get_props(rf, "content/dir2") == "-D-"
 191         assert get_props(rf, "content/file.txt") == "F--"
 192         assert get_props(rf, "links/bad_link") == "--L"
 193         assert get_props(rf, "links/dir_junction") == "--L"
 194         assert get_props(rf, "links/dir_link") == "--L"
 195         assert get_props(rf, "links/file_link") == "--L"
 196
 197         with pytest.warns(rarfile.UnsupportedWarning):
 198             rf.extractall(tmp_path)
 199
 200         assert sorted(os.listdir(tmp_path)) == ["content", "links"]
 201         assert sorted(os.listdir(tmp_path / "content")) == ["dir1", "dir2", "file.txt"]
 202         assert sorted(os.listdir(tmp_path / "links")) == ["bad_link", "dir_link", "file_link"]
 203
 204         assert os.path.islink(tmp_path / "links/bad_link")
 205         assert os.path.islink(tmp_path / "links/dir_link")
 206         assert os.path.islink(tmp_path / "links/file_link")
 207
 208 @pytest.mark.parametrize("fn", [
 209     "test/files/rar3-old.rar",
 210     "test/files/rar3-vols.part1.rar",
 211     "test/files/rar5-vols.part1.rar",
 212 ])
 213 def test_vols(fn, tmp_path):
 214     with rarfile.RarFile(fn) as rf:
 215         rarfile.FORCE_TOOL = True
 216         try:
 217             rf.extractall(str(tmp_path))
 218         finally:
 219             rarfile.FORCE_TOOL = False
 220
 221
 222         assert sorted(os.listdir(tmp_path)) == ["vols"]
 223         assert sorted(os.listdir(tmp_path / "vols")) == ["bigfile.txt", "smallfile.txt"]
 224
 225         assert os.path.isfile(tmp_path / "vols" / "bigfile.txt")
 226         assert os.path.isfile(tmp_path / "vols" / "smallfile.txt")
 227