#19395: Raise exception when pickling a (BZ2|LZMA)(Compressor|Decompressor).

The underlying C libraries provide no mechanism for serializing compressor and
decompressor objects, so actually pickling these classes is impractical.
Previously, these objects would be pickled without error, but attempting to use
a deserialized instance would segfault the interpreter.

Lib/test/test_bz2.py

@@ -5,6 +5,7 @@ from test.support import TESTFN, bigmemtest, _4G
 import unittest
 from io import BytesIO
 import os
+import pickle
 import random
 import subprocess
 import sys
@@ -621,6 +622,11 @@ class BZ2CompressorTest(BaseTest):
         finally:
             data = None
 
+    def testPickle(self):
+        with self.assertRaises(TypeError):
+            pickle.dumps(BZ2Compressor())
+
+
 class BZ2DecompressorTest(BaseTest):
     def test_Constructor(self):
         self.assertRaises(TypeError, BZ2Decompressor, 42)
@@ -672,6 +678,10 @@ class BZ2DecompressorTest(BaseTest):
             compressed = None
             decompressed = None
 
+    def testPickle(self):
+        with self.assertRaises(TypeError):
+            pickle.dumps(BZ2Decompressor())
+
 
 class CompressDecompressTest(BaseTest):
     def testCompress(self):

Lib/test/test_lzma.py

@@ -1,5 +1,6 @@
 from io import BytesIO, UnsupportedOperation
 import os
+import pickle
 import random
 import unittest
 
@@ -216,6 +217,14 @@ class CompressorDecompressorTestCase(unittest.TestCase):
         finally:
             input = cdata = ddata = None
 
+    # Pickling raises an exception; there's no way to serialize an lzma_stream.
+
+    def test_pickle(self):
+        with self.assertRaises(TypeError):
+            pickle.dumps(LZMACompressor())
+        with self.assertRaises(TypeError):
+            pickle.dumps(LZMADecompressor())
+
 
 class CompressDecompressFunctionTestCase(unittest.TestCase):
 

Misc/NEWS

@@ -92,6 +92,10 @@ Core and Builtins
 Library
 -------
 
+- Issue #19395: Raise an exception when attempting to pickle a bz2 or lzma
+  compressor/decompressor object, rather than creating a pickle that would
+  cause a segfault when loaded and used.
+
 - Issue #19227: Try to fix deadlocks caused by re-seeding then OpenSSL
   pseudo-random number generator on fork().
 

Modules/_bz2module.c

@@ -250,6 +250,14 @@ BZ2Compressor_flush(BZ2Compressor *self, PyObject *noargs)
     return result;
 }
 
+static PyObject *
+BZ2Compressor_getstate(BZ2Compressor *self, PyObject *noargs)
+{
+    PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object",
+                 Py_TYPE(self)->tp_name);
+    return NULL;
+}
+
 static int
 BZ2Compressor_init(BZ2Compressor *self, PyObject *args, PyObject *kwargs)
 {
@@ -298,10 +306,11 @@ BZ2Compressor_dealloc(BZ2Compressor *self)
 }
 
 static PyMethodDef BZ2Compressor_methods[] = {
-    {"compress", (PyCFunction)BZ2Compressor_compress, METH_VARARGS,
+    {"compress",     (PyCFunction)BZ2Compressor_compress, METH_VARARGS,
      BZ2Compressor_compress__doc__},
-    {"flush",    (PyCFunction)BZ2Compressor_flush,    METH_NOARGS,
+    {"flush",        (PyCFunction)BZ2Compressor_flush,    METH_NOARGS,
      BZ2Compressor_flush__doc__},
+    {"__getstate__", (PyCFunction)BZ2Compressor_getstate, METH_NOARGS},
     {NULL}
 };
 
@@ -452,6 +461,14 @@ BZ2Decompressor_decompress(BZ2Decompressor *self, PyObject *args)
     return result;
 }
 
+static PyObject *
+BZ2Decompressor_getstate(BZ2Decompressor *self, PyObject *noargs)
+{
+    PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object",
+                 Py_TYPE(self)->tp_name);
+    return NULL;
+}
+
 static int
 BZ2Decompressor_init(BZ2Decompressor *self, PyObject *args, PyObject *kwargs)
 {
@@ -502,6 +519,7 @@ BZ2Decompressor_dealloc(BZ2Decompressor *self)
 static PyMethodDef BZ2Decompressor_methods[] = {
     {"decompress", (PyCFunction)BZ2Decompressor_decompress, METH_VARARGS,
      BZ2Decompressor_decompress__doc__},
+    {"__getstate__", (PyCFunction)BZ2Decompressor_getstate, METH_NOARGS},
     {NULL}
 };
 

Modules/_lzmamodule.c

@@ -546,6 +546,14 @@ Compressor_flush(Compressor *self, PyObject *noargs)
     return result;
 }
 
+static PyObject *
+Compressor_getstate(Compressor *self, PyObject *noargs)
+{
+    PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object",
+                 Py_TYPE(self)->tp_name);
+    return NULL;
+}
+
 static int
 Compressor_init_xz(lzma_stream *lzs, int check, uint32_t preset,
                    PyObject *filterspecs)
@@ -712,6 +720,7 @@ static PyMethodDef Compressor_methods[] = {
      Compressor_compress_doc},
     {"flush", (PyCFunction)Compressor_flush, METH_NOARGS,
      Compressor_flush_doc},
+    {"__getstate__", (PyCFunction)Compressor_getstate, METH_NOARGS},
     {NULL}
 };
 
@@ -869,6 +878,14 @@ Decompressor_decompress(Decompressor *self, PyObject *args)
     return result;
 }
 
+static PyObject *
+Decompressor_getstate(Decompressor *self, PyObject *noargs)
+{
+    PyErr_Format(PyExc_TypeError, "cannot serialize '%s' object",
+                 Py_TYPE(self)->tp_name);
+    return NULL;
+}
+
 static int
 Decompressor_init_raw(lzma_stream *lzs, PyObject *filterspecs)
 {
@@ -991,6 +1008,7 @@ Decompressor_dealloc(Decompressor *self)
 static PyMethodDef Decompressor_methods[] = {
     {"decompress", (PyCFunction)Decompressor_decompress, METH_VARARGS,
      Decompressor_decompress_doc},
+    {"__getstate__", (PyCFunction)Decompressor_getstate, METH_NOARGS},
     {NULL}
 };