Issue #8397: Raise an error when attempting to mix iteration and regular

reads on a BZ2File object, rather than returning incorrect results.

3 files changed, 49 insertions, 0 deletions

diff --git a/Lib/test/test_bz2.py b/Lib/test/test_bz2.py
index 63b3009..0826458 100644
--- a/Lib/test/test_bz2.py
+++ b/Lib/test/test_bz2.py
@@ -304,6 +304,24 @@ class BZ2FileTest(BaseTest):
         finally:
             f.close()
 
+    def testMixedIterationReads(self):
+        # Issue #8397: mixed iteration and reads should be forbidden.
+        f = bz2.BZ2File(self.filename, 'wb')
+        try:
+            # The internal buffer size is hard-wired to 8192 bytes, we must
+            # write out more than that for the test to stop half through
+            # the buffer.
+            f.write(self.TEXT * 100)
+        finally:
+            f.close()
+        f = bz2.BZ2File(self.filename, 'rb')
+        try:
+            next(f)
+            self.assertRaises(ValueError, f.read)
+            self.assertRaises(ValueError, f.readline)
+            self.assertRaises(ValueError, f.readlines)
+        finally:
+            f.close()
 
 class BZ2CompressorTest(BaseTest):
     def testCompress(self):
diff --git a/Misc/NEWS b/Misc/NEWS
index 7d1317d..1126d4a 100644
--- a/Misc/NEWS
+++ b/Misc/NEWS
@@ -29,6 +29,9 @@ Extensions
 Library
 -------
 
+- Issue #8397: Raise an error when attempting to mix iteration and regular
+  reads on a BZ2File object, rather than returning incorrect results.
+
 - Issue #9448: Fix a leak of OS resources (mutexes or semaphores) when
   re-initializing a buffered IO object by calling its ``__init__`` method.
 
diff --git a/Modules/bz2module.c b/Modules/bz2module.c
index 29f8592..c44d3b9 100644
--- a/Modules/bz2module.c
+++ b/Modules/bz2module.c
@@ -138,6 +138,22 @@ typedef struct {
 /* ===================================================================== */
 /* Utility functions. */
 
+/* Refuse regular I/O if there's data in the iteration-buffer.
+ * Mixing them would cause data to arrive out of order, as the read*
+ * methods don't use the iteration buffer. */
+static int
+check_iterbuffered(BZ2FileObject *f)
+{
+    if (f->f_buf != NULL &&
+        (f->f_bufend - f->f_bufptr) > 0 &&
+        f->f_buf[0] != '\0') {
+        PyErr_SetString(PyExc_ValueError,
+            "Mixing iteration and read methods would lose data");
+        return -1;
+    }
+    return 0;
+}
+
 static int
 Util_CatchBZ2Error(int bzerror)
 {
@@ -427,6 +443,10 @@ BZ2File_read(BZ2FileObject *self, PyObject *args)
             goto cleanup;
     }
 
+    /* refuse to mix with f.next() */
+    if (check_iterbuffered(self))
+        goto cleanup;
+
     if (bytesrequested < 0)
         buffersize = Util_NewBufferSize((size_t)0);
     else
@@ -516,6 +536,10 @@ BZ2File_readline(BZ2FileObject *self, PyObject *args)
             goto cleanup;
     }
 
+    /* refuse to mix with f.next() */
+    if (check_iterbuffered(self))
+        goto cleanup;
+
     if (sizehint == 0)
         ret = PyBytes_FromStringAndSize("", 0);
     else
@@ -573,6 +597,10 @@ BZ2File_readlines(BZ2FileObject *self, PyObject *args)
             goto cleanup;
     }
 
+    /* refuse to mix with f.next() */
+    if (check_iterbuffered(self))
+        goto cleanup;
+
     if ((list = PyList_New(0)) == NULL)
         goto cleanup;