diff options
| author | John Sloboda <sloboda@gmail.com> | 2024-03-17 04:58:42 (GMT) |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-03-17 04:58:42 (GMT) |
| commit | 649857a1574a02235ccfac9e2ac1c12914cf8fe0 (patch) | |
| tree | cf08eb67a84db277e1a687eb7177c709d16616b6 /Modules/cjkcodecs | |
| parent | c514a975abe35fa4604cd3541e2286168ef67d10 (diff) | |
| download | cpython-649857a1574a02235ccfac9e2ac1c12914cf8fe0.zip cpython-649857a1574a02235ccfac9e2ac1c12914cf8fe0.tar.gz cpython-649857a1574a02235ccfac9e2ac1c12914cf8fe0.tar.bz2 | |
gh-85287: Change codecs to raise precise UnicodeEncodeError and UnicodeDecodeError (#113674)
Co-authored-by: Inada Naoki <songofacandy@gmail.com>
Diffstat (limited to 'Modules/cjkcodecs')
| -rw-r--r-- | Modules/cjkcodecs/multibytecodec.c | 42 |
1 files changed, 37 insertions, 5 deletions
diff --git a/Modules/cjkcodecs/multibytecodec.c b/Modules/cjkcodecs/multibytecodec.c index 2125da4..e5433d7 100644 --- a/Modules/cjkcodecs/multibytecodec.c +++ b/Modules/cjkcodecs/multibytecodec.c @@ -825,8 +825,15 @@ encoder_encode_stateful(MultibyteStatefulEncoderContext *ctx, if (inpos < datalen) { if (datalen - inpos > MAXENCPENDING) { /* normal codecs can't reach here */ - PyErr_SetString(PyExc_UnicodeError, - "pending buffer overflow"); + PyObject *excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError, + "sOnns", + ctx->codec->encoding, + inbuf, + inpos, datalen, + "pending buffer overflow"); + if (excobj == NULL) goto errorexit; + PyErr_SetObject(PyExc_UnicodeEncodeError, excobj); + Py_DECREF(excobj); goto errorexit; } ctx->pending = PyUnicode_Substring(inbuf, inpos, datalen); @@ -857,7 +864,16 @@ decoder_append_pending(MultibyteStatefulDecoderContext *ctx, npendings = (Py_ssize_t)(buf->inbuf_end - buf->inbuf); if (npendings + ctx->pendingsize > MAXDECPENDING || npendings > PY_SSIZE_T_MAX - ctx->pendingsize) { - PyErr_SetString(PyExc_UnicodeError, "pending buffer overflow"); + Py_ssize_t bufsize = (Py_ssize_t)(buf->inbuf_end - buf->inbuf_top); + PyObject *excobj = PyUnicodeDecodeError_Create(ctx->codec->encoding, + (const char *)buf->inbuf_top, + bufsize, + 0, + bufsize, + "pending buffer overflow"); + if (excobj == NULL) return -1; + PyErr_SetObject(PyExc_UnicodeDecodeError, excobj); + Py_DECREF(excobj); return -1; } memcpy(ctx->pending + ctx->pendingsize, buf->inbuf, npendings); @@ -938,7 +954,17 @@ _multibytecodec_MultibyteIncrementalEncoder_getstate_impl(MultibyteIncrementalEn return NULL; } if (pendingsize > MAXENCPENDING*4) { - PyErr_SetString(PyExc_UnicodeError, "pending buffer too large"); + PyObject *excobj = PyObject_CallFunction(PyExc_UnicodeEncodeError, + "sOnns", + self->codec->encoding, + self->pending, + 0, PyUnicode_GET_LENGTH(self->pending), + "pending buffer too large"); + if (excobj == NULL) { + return NULL; + } + PyErr_SetObject(PyExc_UnicodeEncodeError, excobj); + Py_DECREF(excobj); return NULL; } statebytes[0] = (unsigned char)pendingsize; @@ -1267,7 +1293,13 @@ _multibytecodec_MultibyteIncrementalDecoder_setstate_impl(MultibyteIncrementalDe } if (buffersize > MAXDECPENDING) { - PyErr_SetString(PyExc_UnicodeError, "pending buffer too large"); + PyObject *excobj = PyUnicodeDecodeError_Create(self->codec->encoding, + PyBytes_AS_STRING(buffer), buffersize, + 0, buffersize, + "pending buffer too large"); + if (excobj == NULL) return NULL; + PyErr_SetObject(PyExc_UnicodeDecodeError, excobj); + Py_DECREF(excobj); return NULL; } |