https://github.com/python/cpython/commit/df0ee5aa1a27872f1d8b6c5265abb36a024c39ea commit: df0ee5aa1a27872f1d8b6c5265abb36a024c39ea branch: 3.14 author: Miss Islington (bot) <[email protected]> committer: gpshead <[email protected]> date: 2026-11-08T19:22:34-08:00 summary:
[3.14] gh-160324: Keep `io.StringIO` state after `MemoryError` (GH-160325) (#161042) * gh-160324: Keep `io.StringIO` state after `MemoryError` (GH-160325) Allocate before consuming the accumulated writes in realize() and make_intermediate(), and reset the size when the contents are lost, so that later calls do not index outside the internal buffer. Shrink the buffer of the accumulating writer before finishing it, so that getvalue(), read() and the other methods leave the contents unchanged when they fail with MemoryError instead of dropping them. ANT-2026-MPHGX2FL (cherry picked from commit 83b047ef80836c24672cf3e2e807e68efed7d495) Co-authored-by: Gregory P. Smith <[email protected]> * Avoid _PyUnicodeWriter_SetBuffer() and _PyUnicode_ResizeCompact() They do not exist in 3.14. Use PyUnicode_Resize(), which leaves the buffer unchanged on failure, and update the writer fields directly. * Do not copy the intermediate string in make_intermediate() In 3.14, writing a string to a new PyUnicodeWriter copies it, so a MemoryError there lost the contents of the StringIO. Disable overallocation for this write so the writer shares the string (copy-on-write) instead, as it does in main. --------- Co-authored-by: Gregory P. Smith <[email protected]> Co-authored-by: Gregory P. Smith <[email protected]> files: A Misc/NEWS.d/next/Library/2026-10-22-08-15-56.gh-issue-160324.sIoRlz.rst M Lib/test/test_memoryio.py M Modules/_io/stringio.c diff --git a/Lib/test/test_memoryio.py b/Lib/test/test_memoryio.py index d134df18d9e8479..7fd03b202aea58b 100644 --- a/Lib/test/test_memoryio.py +++ b/Lib/test/test_memoryio.py @@ -1067,6 +1067,39 @@ def __str__(self): memio2.write(MyStr("world")) self.assertEqual(memio2.getvalue(), "hello world") + @support.nomemtest + def test_memory_error_consistent_state(self): + # A MemoryError while copying the accumulated writes into the + # internal buffer must leave the size of the stream consistent with + # the buffer, so that using the object again does not read or write + # outside it. + piece = "0123456789" * 10 + text = piece * 50 + calls = [("readline",), ("read", 5), ("read",), ("getvalue",), + ("truncate", 3), ("write", "ab"), ("__next__",)] + for name, *args in calls: + for n in range(3): + with self.subTest(method=name, n=n): + memio = self.ioclass() + for _ in range(50): + memio.write(piece) + memio.seek(0) + method = getattr(memio, name) + try: + with support.inject_memory_error_cm(n, n + 1): + method(*args) + except MemoryError: + pass + else: + continue + # The contents are unchanged. + self.assertEqual(memio.getvalue(), text) + self.assertEqual(memio.seek(0, 2), len(text)) + memio.seek(0) + self.assertEqual(memio.readline(), text) + memio.write("!") + self.assertEqual(memio.getvalue(), text + "!") + class CStringIOPickleTest(PyStringIOPickleTest): UnsupportedOperation = io.UnsupportedOperation diff --git a/Misc/NEWS.d/next/Library/2026-10-22-08-15-56.gh-issue-160324.sIoRlz.rst b/Misc/NEWS.d/next/Library/2026-10-22-08-15-56.gh-issue-160324.sIoRlz.rst new file mode 100644 index 000000000000000..d54069128e2c0eb --- /dev/null +++ b/Misc/NEWS.d/next/Library/2026-10-22-08-15-56.gh-issue-160324.sIoRlz.rst @@ -0,0 +1,3 @@ +Fix a crash or garbage data when an :class:`io.StringIO` object is used again +after one of its methods failed with :exc:`MemoryError`. The object now keeps +its contents. diff --git a/Modules/_io/stringio.c b/Modules/_io/stringio.c index 8f7801dab96dd1d..2d4e1a5c70d83b1 100644 --- a/Modules/_io/stringio.c +++ b/Modules/_io/stringio.c @@ -129,25 +129,60 @@ resize_buffer(stringio *self, size_t size) return -1; } +/* Shrink the buffer of the writer to its contents, so that the following + PyUnicodeWriter_Finish() has nothing to allocate and cannot fail. + If this fails, the writer is unchanged. */ +static int +shrink_writer(stringio *self) +{ + _PyUnicodeWriter *writer = (_PyUnicodeWriter *)self->writer; + if (writer->readonly || writer->buffer == NULL + || writer->pos == 0 || writer->pos == writer->size) + { + return 0; + } + /* On failure, writer->buffer is left unchanged. */ + if (PyUnicode_Resize(&writer->buffer, writer->pos) < 0) { + return -1; + } + PyObject *buffer = writer->buffer; + writer->data = PyUnicode_DATA(buffer); + writer->kind = PyUnicode_KIND(buffer); + writer->maxchar = PyUnicode_MAX_CHAR_VALUE(buffer); + writer->size = PyUnicode_GET_LENGTH(buffer); + return 0; +} + static PyObject * make_intermediate(stringio *self) { - PyObject *intermediate = PyUnicodeWriter_Finish(self->writer); - self->writer = NULL; - self->state = STATE_REALIZED; - if (intermediate == NULL) + /* Create the new writer and shrink the old one first: if this fails, + the object keeps its contents. */ + PyUnicodeWriter *writer = PyUnicodeWriter_Create(0); + if (writer == NULL) return NULL; + if (shrink_writer(self) < 0) { + PyUnicodeWriter_Discard(writer); + return NULL; + } - self->writer = PyUnicodeWriter_Create(0); - if (self->writer == NULL) { - Py_DECREF(intermediate); + PyObject *intermediate = PyUnicodeWriter_Finish(self->writer); + self->writer = writer; + if (intermediate == NULL) { + /* The contents are lost, the new writer is empty. */ + self->string_size = 0; return NULL; } - if (PyUnicodeWriter_WriteStr(self->writer, intermediate)) { + /* Without overallocation, writing to an empty writer shares the string + (copy-on-write) instead of copying it, so it does not allocate. */ + ((_PyUnicodeWriter *)writer)->overallocate = 0; + int res = PyUnicodeWriter_WriteStr(writer, intermediate); + ((_PyUnicodeWriter *)writer)->overallocate = 1; + if (res < 0) { Py_DECREF(intermediate); + self->string_size = 0; return NULL; } - self->state = STATE_ACCUMULATING; return intermediate; } @@ -160,23 +195,30 @@ realize(stringio *self) if (self->state == STATE_REALIZED) return 0; assert(self->state == STATE_ACCUMULATING); + + /* Allocate the internal buffer and shrink the writer first: if this + fails, the object stays in the accumulating state and keeps its + contents. */ + if (resize_buffer(self, self->string_size) < 0) + return -1; + if (shrink_writer(self) < 0) + return -1; self->state = STATE_REALIZED; intermediate = PyUnicodeWriter_Finish(self->writer); self->writer = NULL; - if (intermediate == NULL) + if (intermediate == NULL) { + /* The contents are lost, the internal buffer contains nothing. */ + self->string_size = 0; return -1; + } - /* Append the intermediate string to the internal buffer. - The length should be equal to the current cursor position. - */ + /* Copy the intermediate string to the internal buffer. */ len = PyUnicode_GET_LENGTH(intermediate); - if (resize_buffer(self, len) < 0) { - Py_DECREF(intermediate); - return -1; - } + assert(len == self->string_size); if (!PyUnicode_AsUCS4(intermediate, self->buf, len, 0)) { Py_DECREF(intermediate); + self->string_size = 0; return -1; } @@ -344,7 +386,9 @@ _io_StringIO_read_impl(stringio *self, Py_ssize_t size) /* Optimization for seek(0); read() */ if (self->state == STATE_ACCUMULATING && self->pos == 0 && size == n) { PyObject *result = make_intermediate(self); - self->pos = self->string_size; + if (result != NULL) { + self->pos = self->string_size; + } return result; } _______________________________________________ Python-checkins mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3//lists/python-checkins.python.org Member address: [email protected]
