Merge pull request #10 from vidstige/better-tests

BSD support

.gitignore

@@ -3,6 +3,9 @@ __pycache__/
 
 /*.egg-info
 
-/test_data/
+/test_data/file0.txt
+/test_data/file1.bin
+/test_data/test.a
+
 /dist/
 /build/

ar/archive.py

@@ -74,19 +74,22 @@ def lookup(data, offset):
     return data[start:end - 1].decode()
 
 
-def load(stream):
-    actual = stream.read(len(MAGIC))
-    if actual != MAGIC:
-        raise ArchiveError(f"Unexpected magic: {actual}")
+ENTRY_FORMAT = '16s12s6s6s8s10sbb'
+
 
-    fmt = '16s12s6s6s8s10sbb'
+def load(stream):
+    magic = stream.read(len(MAGIC))
+    if not isinstance(magic, bytes):
+        raise ArchiveError("Stream must be binary")
+    if magic != MAGIC:
+        raise ArchiveError(f"Unexpected magic: {magic!r}")
 
     lookup_data = None
     while True:
-        buffer = stream.read(struct.calcsize(fmt))
-        if len(buffer) < struct.calcsize(fmt):
+        buffer = stream.read(struct.calcsize(ENTRY_FORMAT))
+        if len(buffer) < struct.calcsize(ENTRY_FORMAT):
             break
-        name, timestamp, owner, group, mode, size, _, _ = struct.unpack(fmt, buffer)
+        name, timestamp, owner, group, mode, size, _, _ = struct.unpack(ENTRY_FORMAT, buffer)
         del timestamp, owner, group, mode
         name = name.decode().rstrip()
         size = int(size.decode().rstrip())
@@ -103,6 +106,13 @@ def load(stream):
             offset = stream.tell()
             stream.seek(pad(size, 2), 1)
             yield ArPath(expanded_name, offset, size)
+        elif name.startswith('#1/'):
+            # BSD long filenames
+            name_length = int(name[len('#1/'):])
+            long_name = stream.read(name_length).rstrip(b'\00').decode()
+            offset = stream.tell()
+            stream.seek(pad(size - name_length, 2), 1)
+            yield ArPath(long_name, offset, size - name_length)
         else:
             offset = stream.tell()
             stream.seek(pad(size, 2), 1)

ar/tests/test_bad_archive.py

@@ -0,0 +1,15 @@
+from pathlib import Path
+
+import pytest
+
+from ar import Archive, ArchiveError
+
+
+BAD_ARCHIVE = Path('test_data/bad.a')
+
+
+def test_bad_file():
+    with BAD_ARCHIVE.open('rb') as f:
+        with pytest.raises(ArchiveError) as exception_info:
+            Archive(f)
+        assert str(exception_info.value) == "Unexpected magic: b'nope, no'"

ar/tests/test_bsd.py

@@ -0,0 +1,46 @@
+# pylint: disable=redefined-outer-name
+import subprocess
+from pathlib import Path
+
+from ar import Archive, ArchiveError
+
+
+ARCHIVE = Path('test_data/bsd.a')
+
+
+def test_list():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        actual = [entry.name for entry in archive]
+        expected = ['file0.txt', 'file1.bin', 'long-filename.txt']
+        assert expected == actual
+
+
+def test_read_content():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file0.txt')
+        assert file0.read(1) == 'H'
+        assert file0.read() == 'ello'
+
+
+def test_read_content_long_filename():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('long-filename.txt')
+        assert file0.read(1) == 'l'
+        assert file0.read() == 'ong filename\n'
+
+def test_read_binary():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file1.bin', 'rb')
+        assert file0.read() == b'\xc3\x28'
+
+
+def test_seek_basic():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file0.txt')
+        file0.seek(1)
+        assert file0.read(3) == 'ell'

ar/tests/test_linux.py

@@ -0,0 +1,37 @@
+# pylint: disable=redefined-outer-name
+import subprocess
+from pathlib import Path
+
+from ar import Archive, ArchiveError
+
+
+ARCHIVE = Path('test_data/linux.a')
+
+
+def test_list():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        assert ['file0.txt', 'file1.bin'] == [entry.name for entry in archive]
+
+
+def test_read_content():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file0.txt')
+        assert file0.read(1) == 'H'
+        assert file0.read() == 'ello'
+
+
+def test_read_binary():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file1.bin', 'rb')
+        assert file0.read() == b'\xc3\x28'
+
+
+def test_seek_basic():
+    with ARCHIVE.open('rb') as f:
+        archive = Archive(f)
+        file0 = archive.open('file0.txt')
+        file0.seek(1)
+        assert file0.read(3) == 'ell'

ar/tests/test_ar.py → ar/tests/test_roundtrip.py

@@ -21,13 +21,6 @@ def simple_archive():
     return TEST_DATA / 'test.a'
 
 
-@pytest.fixture
-def bad_archive():
-    path = TEST_DATA / 'bad.a'
-    path.write_bytes(b'nope, not an ar file')
-    return path
-
-
 def test_list(simple_archive):
     with simple_archive.open('rb') as f:
         archive = Archive(f)
@@ -55,10 +48,3 @@ def test_seek_basic(simple_archive):
         file0 = archive.open('file0.txt')
         file0.seek(1)
         assert file0.read(3) == 'ell'
-
-
-def test_bad_file(bad_archive):
-    with bad_archive.open('rb') as f:
-        with pytest.raises(ArchiveError) as exception_info:
-            Archive(f)
-        assert str(exception_info.value) == "Unexpected magic: b'nope, no'"

ar/tests/test_wrong_mody.py

@@ -0,0 +1,15 @@
+from pathlib import Path
+
+import pytest
+
+from ar import Archive, ArchiveError
+
+
+BAD_ARCHIVE = Path('test_data/bad.a')
+
+
+def test_bad_file():
+    with BAD_ARCHIVE.open('rt') as f:
+        with pytest.raises(ArchiveError) as exception_info:
+            Archive(f)
+        assert str(exception_info.value) == "Stream must be binary"

test_data/bad.a

@@ -0,0 +1 @@
+nope, not an ar file

test_data/linux.a

@@ -0,0 +1,5 @@
+!<arch>
+file0.txt/      0           0     0     644     5         `
+Hello
+file1.bin/      0           0     0     644     2         `
+�(