diff options
Diffstat (limited to 'Modules/cjkcodecs/multibytecodec.c')
| -rw-r--r-- | Modules/cjkcodecs/multibytecodec.c | 62 |
1 files changed, 44 insertions, 18 deletions
diff --git a/Modules/cjkcodecs/multibytecodec.c b/Modules/cjkcodecs/multibytecodec.c index 7b04f020c9..c032cdb0ed 100644 --- a/Modules/cjkcodecs/multibytecodec.c +++ b/Modules/cjkcodecs/multibytecodec.c @@ -316,15 +316,15 @@ multibytecodec_encerror(MultibyteCodec *codec, goto errorexit; if (!PyTuple_Check(retobj) || PyTuple_GET_SIZE(retobj) != 2 || - !PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) || + (!PyUnicode_Check((tobj = PyTuple_GET_ITEM(retobj, 0))) && !PyBytes_Check(tobj)) || !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { PyErr_SetString(PyExc_TypeError, "encoding error handler must return " - "(unicode, int) tuple"); + "(str, int) tuple"); goto errorexit; } - { + if (PyUnicode_Check(tobj)) { const Py_UNICODE *uraw = PyUnicode_AS_UNICODE(tobj); retstr = multibytecodec_encode(codec, state, &uraw, @@ -333,6 +333,10 @@ multibytecodec_encerror(MultibyteCodec *codec, if (retstr == NULL) goto errorexit; } + else { + Py_INCREF(tobj); + retstr = tobj; + } assert(PyBytes_Check(retstr)); retstrsize = PyBytes_GET_SIZE(retstr); @@ -439,14 +443,16 @@ multibytecodec_decerror(MultibyteCodec *codec, !PyLong_Check(PyTuple_GET_ITEM(retobj, 1))) { PyErr_SetString(PyExc_TypeError, "decoding error handler must return " - "(unicode, int) tuple"); + "(str, int) tuple"); goto errorexit; } + if (PyUnicode_AsUnicode(retuni) == NULL) + goto errorexit; retunisize = PyUnicode_GET_SIZE(retuni); if (retunisize > 0) { REQUIRE_DECODEBUFFER(buf, retunisize); - memcpy((char *)buf->outbuf, PyUnicode_AS_DATA(retuni), + memcpy((char *)buf->outbuf, PyUnicode_AS_UNICODE(retuni), retunisize * Py_UNICODE_SIZE); buf->outbuf += retunisize; } @@ -483,6 +489,7 @@ multibytecodec_encode(MultibyteCodec *codec, return PyBytes_FromStringAndSize(NULL, 0); buf.excobj = NULL; + buf.outobj = NULL; buf.inbuf = buf.inbuf_top = *data; buf.inbuf_end = buf.inbuf_top + datalen; @@ -573,8 +580,11 @@ MultibyteCodec_Encode(MultibyteCodecObject *self, } } - data = PyUnicode_AS_UNICODE(arg); - datalen = PyUnicode_GET_SIZE(arg); + data = PyUnicode_AsUnicodeAndSize(arg, &datalen); + if (data == NULL) { + Py_XDECREF(ucvt); + return NULL; + } errorcb = internal_error_callback(errors); if (errorcb == NULL) { @@ -627,7 +637,7 @@ MultibyteCodec_Decode(MultibyteCodecObject *self, if (datalen == 0) { PyBuffer_Release(&pdata); ERROR_DECREF(errorcb); - return make_tuple(PyUnicode_FromUnicode(NULL, 0), 0); + return make_tuple(PyUnicode_New(0, 0), 0); } buf.excobj = NULL; @@ -637,6 +647,8 @@ MultibyteCodec_Decode(MultibyteCodecObject *self, if (buf.outobj == NULL) goto errorexit; buf.outbuf = PyUnicode_AS_UNICODE(buf.outobj); + if (buf.outbuf == NULL) + goto errorexit; buf.outbuf_end = buf.outbuf + PyUnicode_GET_SIZE(buf.outobj); if (self->codec->decinit != NULL && @@ -742,6 +754,7 @@ encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, PyObject *ucvt, *r = NULL; Py_UNICODE *inbuf, *inbuf_end, *inbuf_tmp = NULL; Py_ssize_t datalen, origpending; + wchar_t *data; if (PyUnicode_Check(unistr)) ucvt = NULL; @@ -751,13 +764,15 @@ encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, return NULL; else if (!PyUnicode_Check(unistr)) { PyErr_SetString(PyExc_TypeError, - "couldn't convert the object to unicode."); + "couldn't convert the object to str."); Py_DECREF(ucvt); return NULL; } } - datalen = PyUnicode_GET_SIZE(unistr); + data = PyUnicode_AsUnicodeAndSize(unistr, &datalen); + if (data == NULL) + goto errorexit; origpending = ctx->pendingsize; if (origpending > 0) { @@ -848,7 +863,9 @@ decoder_prepare_buffer(MultibyteDecodeBuffer *buf, const char *data, buf->outobj = PyUnicode_FromUnicode(NULL, size); if (buf->outobj == NULL) return -1; - buf->outbuf = PyUnicode_AS_UNICODE(buf->outobj); + buf->outbuf = PyUnicode_AsUnicode(buf->outobj); + if (buf->outbuf == NULL) + return -1; buf->outbuf_end = buf->outbuf + PyUnicode_GET_SIZE(buf->outobj); } @@ -900,11 +917,17 @@ mbiencoder_encode(MultibyteIncrementalEncoderObject *self, static PyObject * mbiencoder_reset(MultibyteIncrementalEncoderObject *self) { - if (self->codec->decreset != NULL && - self->codec->decreset(&self->state, self->codec->config) != 0) - return NULL; + /* Longest output: 4 bytes (b'\x0F\x1F(B') with ISO 2022 */ + unsigned char buffer[4], *outbuf; + Py_ssize_t r; + if (self->codec->encreset != NULL) { + outbuf = buffer; + r = self->codec->encreset(&self->state, self->codec->config, + &outbuf, sizeof(buffer)); + if (r != 0) + return NULL; + } self->pendingsize = 0; - Py_RETURN_NONE; } @@ -1246,7 +1269,7 @@ mbstreamreader_iread(MultibyteStreamReaderObject *self, Py_ssize_t rsize, finalsize = 0; if (sizehint == 0) - return PyUnicode_FromUnicode(NULL, 0); + return PyUnicode_New(0, 0); buf.outobj = buf.excobj = NULL; cres = NULL; @@ -1572,12 +1595,13 @@ mbstreamwriter_iwrite(MultibyteStreamWriterObject *self, PyObject *unistr) { PyObject *str, *wr; + _Py_IDENTIFIER(write); str = encoder_encode_stateful(STATEFUL_ECTX(self), unistr, 0); if (str == NULL) return -1; - wr = PyObject_CallMethod(self->stream, "write", "O", str); + wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", str); Py_DECREF(str); if (wr == NULL) return -1; @@ -1643,7 +1667,9 @@ mbstreamwriter_reset(MultibyteStreamWriterObject *self) assert(PyBytes_Check(pwrt)); if (PyBytes_Size(pwrt) > 0) { PyObject *wr; - wr = PyObject_CallMethod(self->stream, "write", "O", pwrt); + _Py_IDENTIFIER(write); + + wr = _PyObject_CallMethodId(self->stream, &PyId_write, "O", pwrt); if (wr == NULL) { Py_DECREF(pwrt); return NULL; |