Fix bug. Read binary content that contains \x00 will break in read_cs…

…tring(). (#12)

* Fix Bug: If \x00 is the content of file, jfs.read() and FileIO.read() will not truncate it

* add test case for the scenario that content has \x00

* add test content in test_read_content_with_zero()

* shorten the CONTENT_WITH_ZERO constant

* if the jfs_ function return size, use buf.raw[:size] instead of read_cstring()

* add test_pread(), test case for pread.

* revert read_cstring()

juicefs/juicefs.py

@@ -18,7 +18,6 @@
  create_statvfs_result,
  create_summary,
  parse_xattrs,
- read_cstring,
 )
 from juicefs.utils import check_juicefs_error, create_os_error
 
@@ -234,8 +233,8 @@ def readlink(self, path: str) -> str:
  """
  bufsize = 4096
  buf = create_string_buffer(bufsize)
- self._lib.readlink[1](path, buf, bufsize)
- return read_cstring(BytesIO(buf.raw)).decode()
+ size = self._lib.readlink[1](path, buf, bufsize)
+ return buf.raw[:size].decode()
 
  def open(self, path: str, flags: int, mode: int = DEFAULT_FILE_MODE) -> int: # fd
  """Open the file *path* and set various flags according to *flags* and
@@ -306,17 +305,17 @@ def read(self, fd: int, size: int) -> bytes:
  If the end of the file referred to by fd has been reached, an empty bytes object is returned.
  """
  buf = create_string_buffer(size)
- self._lib[fd].read[0](buf, size)
- return bytes(read_cstring(BytesIO(buf.raw)))
+ size = self._lib[fd].read[0](buf, size)
+ return buf.raw[:size]
 
  def pread(self, fd: int, size: int, offset: int) -> bytes:
  """Read from a file descriptor, *fd*, at a position of *offset*.
 
  It will read up to *size* number of bytes. The file offset remains unchanged.
  """
  buf = create_string_buffer(size)
- self._lib[fd].pread[0](buf, size, offset)
- return bytes(read_cstring(BytesIO(buf.raw)))
+ size = self._lib[fd].pread[0](buf, size, offset)
+ return buf.raw[:size]
 
  def write(self, fd: int, content: bytes) -> int:
  """Write the *content* to file descriptor *fd*.

tests/test_juicefs.py

@@ -85,6 +85,9 @@ def os_dirname(jfs):
 
 CONTENT = b"text"
 CONTENT2 = b"text/text"
+CONTENT_WITH_ZERO = (
+ b"\x05\x00\x00\x00\x00\x00\x00\x00\x98\x00\x00\x00\x00\x00\x00\x00-\x01"
+)
 
 
 def test_rename(jfs, filename, filename2, dirname):
@@ -738,6 +741,39 @@ def test_summary(jfs, dirname):
  assert summary.dirs == 2
 
 
+def test_pread(jfs, filename):
+ jfs.create(filename)
+ fdw = jfs.open(filename, os.O_WRONLY)
+ jfs.write(fdw, CONTENT_WITH_ZERO)
+ jfs.close(fdw)
+
+ fdr = jfs.open(filename, os.O_RDONLY)
+ for offset in range(5):
+ for size in range(len(CONTENT_WITH_ZERO) + 5):
+ start_pos = min(offset, len(CONTENT_WITH_ZERO))
+ stop_pos = min(offset + size, len(CONTENT_WITH_ZERO))
+ assert jfs.pread(fdr, size, offset) == CONTENT_WITH_ZERO[start_pos:stop_pos]
+ jfs.close(fdr)
+
+
+def test_read_content_with_zero(jfs, filename):
+ jfs.create(filename)
+ fdw = jfs.open(filename, os.O_WRONLY)
+ content_len = jfs.write(fdw, CONTENT_WITH_ZERO)
+ assert content_len == len(CONTENT_WITH_ZERO)
+ jfs.close(fdw)
+
+ fdr = jfs.open(filename, os.O_RDONLY)
+ assert jfs.read(fdr, 7) == CONTENT_WITH_ZERO[:7]
+ assert jfs.read(fdr, len(CONTENT_WITH_ZERO) + 10) == CONTENT_WITH_ZERO[7:]
+ jfs.close(fdr)
+
+ from juicefs.io import open as _jfs_open
+
+ with _jfs_open(jfs, filename, "rb") as f:
+ f.read() == CONTENT_WITH_ZERO
+
+
 # TODO: read-write mode not supported in this version
 # def test_read_write_wbp(jfs, filename):
 # TEXT = b"hello world"