Issue #22341: Merge CRC doc from 3.5

Doc/library/binascii.rst

@@ -115,15 +115,16 @@ The :mod:`binascii` module defines the following functions:
    possibly the last fragment).
 
 
-.. function:: crc_hqx(data, crc)
+.. function:: crc_hqx(data, value)
 
-   Compute the binhex4 crc value of *data*, starting with an initial *crc* and
-   returning the result.
+   Compute the binhex4 crc value of *data*, starting with *value* as the
+   initial crc, and return the result.
 
 
-.. function:: crc32(data[, crc])
+.. function:: crc32(data[, value])
 
-   Compute CRC-32, the 32-bit checksum of data, starting with an initial crc.  This
+   Compute CRC-32, the 32-bit checksum of *data*, starting with an
+   initial CRC of *value*.  The default initial CRC is zero.  The algorithm
    is consistent with the ZIP file checksum.  Since the algorithm is designed for
    use as a checksum algorithm, it is not suitable for use as a general hash
    algorithm.  Use as follows::
@@ -131,15 +132,13 @@ The :mod:`binascii` module defines the following functions:
       print(binascii.crc32(b"hello world"))
       # Or, in two pieces:
       crc = binascii.crc32(b"hello")
-      crc = binascii.crc32(b" world", crc) & 0xffffffff
+      crc = binascii.crc32(b" world", crc)
       print('crc32 = {:#010x}'.format(crc))
 
-.. note::
-   To generate the same numeric value across all Python versions and
-   platforms use crc32(data) & 0xffffffff.  If you are only using
-   the checksum in packed binary format this is not necessary as the
-   return value is the correct 32bit binary representation
-   regardless of sign.
+   .. versionchanged:: 3.0
+      The result is always unsigned.
+      To generate the same numeric value across all Python versions and
+      platforms, use ``crc32(data) & 0xffffffff``.
 
 
 .. function:: b2a_hex(data)

Doc/library/zlib.rst

@@ -31,22 +31,19 @@ The available exception and functions in this module are:
 .. function:: adler32(data[, value])
 
    Computes an Adler-32 checksum of *data*.  (An Adler-32 checksum is almost as
-   reliable as a CRC32 but can be computed much more quickly.)  If *value* is
-   present, it is used as the starting value of the checksum; otherwise, a fixed
-   default value is used.  This allows computing a running checksum over the
+   reliable as a CRC32 but can be computed much more quickly.)  The result
+   is an unsigned 32-bit integer.  If *value* is present, it is used as
+   the starting value of the checksum; otherwise, a default value of 1
+   is used.  Passing in *value* allows computing a running checksum over the
    concatenation of several inputs.  The algorithm is not cryptographically
    strong, and should not be used for authentication or digital signatures.  Since
    the algorithm is designed for use as a checksum algorithm, it is not suitable
    for use as a general hash algorithm.
 
-   Always returns an unsigned 32-bit integer.
-
-.. note::
-   To generate the same numeric value across all Python versions and
-   platforms use adler32(data) & 0xffffffff.  If you are only using
-   the checksum in packed binary format this is not necessary as the
-   return value is the correct 32bit binary representation
-   regardless of sign.
+   .. versionchanged:: 3.0
+      Always returns an unsigned value.
+      To generate the same numeric value across all Python versions and
+      platforms, use ``adler32(data) & 0xffffffff``.
 
 
 .. function:: compress(data[, level])
@@ -97,23 +94,19 @@ The available exception and functions in this module are:
       single: Cyclic Redundancy Check
       single: checksum; Cyclic Redundancy Check
 
-   Computes a CRC (Cyclic Redundancy Check)  checksum of *data*. If *value* is
-   present, it is used as the starting value of the checksum; otherwise, a fixed
-   default value is used.  This allows computing a running checksum over the
+   Computes a CRC (Cyclic Redundancy Check) checksum of *data*. The
+   result is an unsigned 32-bit integer. If *value* is present, it is used
+   as the starting value of the checksum; otherwise, a default value of 0
+   is used.  Passing in *value* allows computing a running checksum over the
    concatenation of several inputs.  The algorithm is not cryptographically
    strong, and should not be used for authentication or digital signatures.  Since
    the algorithm is designed for use as a checksum algorithm, it is not suitable
    for use as a general hash algorithm.
 
-   Always returns an unsigned 32-bit integer.
-
-   .. note::
-
+   .. versionchanged:: 3.0
+      Always returns an unsigned value.
       To generate the same numeric value across all Python versions and
-      platforms, use ``crc32(data) & 0xffffffff``.  If you are only using
-      the checksum in packed binary format this is not necessary as the
-      return value is the correct 32-bit binary representation
-      regardless of sign.
+      platforms, use ``crc32(data) & 0xffffffff``.
 
 
 .. function:: decompress(data[, wbits[, bufsize]])

Lib/gzip.py

@@ -210,7 +210,7 @@ class GzipFile(_compression.BaseStream):
 
     def _init_write(self, filename):
         self.name = filename
-        self.crc = zlib.crc32(b"") & 0xffffffff
+        self.crc = zlib.crc32(b"")
         self.size = 0
         self.writebuf = []
         self.bufsize = 0
@@ -261,7 +261,7 @@ class GzipFile(_compression.BaseStream):
         if length > 0:
             self.fileobj.write(self.compress.compress(data))
             self.size += length
-            self.crc = zlib.crc32(data, self.crc) & 0xffffffff
+            self.crc = zlib.crc32(data, self.crc)
             self.offset += length
 
         return length
@@ -381,7 +381,7 @@ class _GzipReader(_compression.DecompressReader):
         self._last_mtime = None
 
     def _init_read(self):
-        self._crc = zlib.crc32(b"") & 0xffffffff
+        self._crc = zlib.crc32(b"")
         self._stream_size = 0  # Decompressed size of unconcatenated stream
 
     def _read_exact(self, n):
@@ -485,7 +485,7 @@ class _GzipReader(_compression.DecompressReader):
         return uncompress
 
     def _add_read_data(self, data):
-        self._crc = zlib.crc32(data, self._crc) & 0xffffffff
+        self._crc = zlib.crc32(data, self._crc)
         self._stream_size = self._stream_size + len(data)
 
     def _read_eof(self):

Lib/tarfile.py

@@ -459,13 +459,7 @@ class _Stream:
                 self.fileobj.write(self.buf)
                 self.buf = b""
                 if self.comptype == "gz":
-                    # The native zlib crc is an unsigned 32-bit integer, but
-                    # the Python wrapper implicitly casts that to a signed C
-                    # long.  So, on a 32-bit box self.crc may "look negative",
-                    # while the same crc on a 64-bit box may "look positive".
-                    # To avoid irksome warnings from the `struct` module, force
-                    # it to look positive on all boxes.
-                    self.fileobj.write(struct.pack("<L", self.crc & 0xffffffff))
+                    self.fileobj.write(struct.pack("<L", self.crc))
                     self.fileobj.write(struct.pack("<L", self.pos & 0xffffFFFF))
         finally:
             if not self._extfileobj:

Lib/test/test_zlib.py

@@ -47,16 +47,11 @@ class ChecksumTestCase(unittest.TestCase):
         self.assertEqual(zlib.adler32(b"", 1), 1)
         self.assertEqual(zlib.adler32(b"", 432), 432)
 
-    def assertEqual32(self, seen, expected):
-        # 32-bit values masked -- checksums on 32- vs 64- bit machines
-        # This is important if bit 31 (0x08000000L) is set.
-        self.assertEqual(seen & 0x0FFFFFFFF, expected & 0x0FFFFFFFF)
-
     def test_penguins(self):
-        self.assertEqual32(zlib.crc32(b"penguin", 0), 0x0e5c1a120)
-        self.assertEqual32(zlib.crc32(b"penguin", 1), 0x43b6aa94)
-        self.assertEqual32(zlib.adler32(b"penguin", 0), 0x0bcf02f6)
-        self.assertEqual32(zlib.adler32(b"penguin", 1), 0x0bd602f7)
+        self.assertEqual(zlib.crc32(b"penguin", 0), 0x0e5c1a120)
+        self.assertEqual(zlib.crc32(b"penguin", 1), 0x43b6aa94)
+        self.assertEqual(zlib.adler32(b"penguin", 0), 0x0bcf02f6)
+        self.assertEqual(zlib.adler32(b"penguin", 1), 0x0bd602f7)
 
         self.assertEqual(zlib.crc32(b"penguin"), zlib.crc32(b"penguin", 0))
         self.assertEqual(zlib.adler32(b"penguin"),zlib.adler32(b"penguin",1))

Lib/zipfile.py

@@ -734,7 +734,7 @@ class ZipExtFile(io.BufferedIOBase):
 
         if hasattr(zipinfo, 'CRC'):
             self._expected_crc = zipinfo.CRC
-            self._running_crc = crc32(b'') & 0xffffffff
+            self._running_crc = crc32(b'')
         else:
             self._expected_crc = None
 
@@ -856,7 +856,7 @@ class ZipExtFile(io.BufferedIOBase):
         if self._expected_crc is None:
             # No need to compute the CRC if we don't have a reference value
             return
-        self._running_crc = crc32(newdata, self._running_crc) & 0xffffffff
+        self._running_crc = crc32(newdata, self._running_crc)
         # Check the CRC if we're at the end of the file
         if self._eof and self._running_crc != self._expected_crc:
             raise BadZipFile("Bad CRC-32 for file %r" % self.name)
@@ -1492,7 +1492,7 @@ class ZipFile:
                     if not buf:
                         break
                     file_size = file_size + len(buf)
-                    CRC = crc32(buf, CRC) & 0xffffffff
+                    CRC = crc32(buf, CRC)
                     if cmpr:
                         buf = cmpr.compress(buf)
                         compress_size = compress_size + len(buf)
@@ -1567,7 +1567,7 @@ class ZipFile:
 
             self._writecheck(zinfo)
             self._didModify = True
-            zinfo.CRC = crc32(data) & 0xffffffff       # CRC-32 checksum
+            zinfo.CRC = crc32(data)       # CRC-32 checksum
             co = _get_compressor(zinfo.compress_type)
             if co:
                 data = co.compress(data) + co.flush()