| 1 | n/a | /* Accumulator struct implementation */ |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | #include "Python.h" |
|---|
| 4 | n/a | #include "accu.h" |
|---|
| 5 | n/a | |
|---|
| 6 | n/a | static PyObject * |
|---|
| 7 | n/a | join_list_unicode(PyObject *lst) |
|---|
| 8 | n/a | { |
|---|
| 9 | n/a | /* return ''.join(lst) */ |
|---|
| 10 | n/a | PyObject *sep, *ret; |
|---|
| 11 | n/a | sep = PyUnicode_FromStringAndSize("", 0); |
|---|
| 12 | n/a | ret = PyUnicode_Join(sep, lst); |
|---|
| 13 | n/a | Py_DECREF(sep); |
|---|
| 14 | n/a | return ret; |
|---|
| 15 | n/a | } |
|---|
| 16 | n/a | |
|---|
| 17 | n/a | int |
|---|
| 18 | n/a | _PyAccu_Init(_PyAccu *acc) |
|---|
| 19 | n/a | { |
|---|
| 20 | n/a | /* Lazily allocated */ |
|---|
| 21 | n/a | acc->large = NULL; |
|---|
| 22 | n/a | acc->small = PyList_New(0); |
|---|
| 23 | n/a | if (acc->small == NULL) |
|---|
| 24 | n/a | return -1; |
|---|
| 25 | n/a | return 0; |
|---|
| 26 | n/a | } |
|---|
| 27 | n/a | |
|---|
| 28 | n/a | static int |
|---|
| 29 | n/a | flush_accumulator(_PyAccu *acc) |
|---|
| 30 | n/a | { |
|---|
| 31 | n/a | Py_ssize_t nsmall = PyList_GET_SIZE(acc->small); |
|---|
| 32 | n/a | if (nsmall) { |
|---|
| 33 | n/a | int ret; |
|---|
| 34 | n/a | PyObject *joined; |
|---|
| 35 | n/a | if (acc->large == NULL) { |
|---|
| 36 | n/a | acc->large = PyList_New(0); |
|---|
| 37 | n/a | if (acc->large == NULL) |
|---|
| 38 | n/a | return -1; |
|---|
| 39 | n/a | } |
|---|
| 40 | n/a | joined = join_list_unicode(acc->small); |
|---|
| 41 | n/a | if (joined == NULL) |
|---|
| 42 | n/a | return -1; |
|---|
| 43 | n/a | if (PyList_SetSlice(acc->small, 0, nsmall, NULL)) { |
|---|
| 44 | n/a | Py_DECREF(joined); |
|---|
| 45 | n/a | return -1; |
|---|
| 46 | n/a | } |
|---|
| 47 | n/a | ret = PyList_Append(acc->large, joined); |
|---|
| 48 | n/a | Py_DECREF(joined); |
|---|
| 49 | n/a | return ret; |
|---|
| 50 | n/a | } |
|---|
| 51 | n/a | return 0; |
|---|
| 52 | n/a | } |
|---|
| 53 | n/a | |
|---|
| 54 | n/a | int |
|---|
| 55 | n/a | _PyAccu_Accumulate(_PyAccu *acc, PyObject *unicode) |
|---|
| 56 | n/a | { |
|---|
| 57 | n/a | Py_ssize_t nsmall; |
|---|
| 58 | n/a | assert(PyUnicode_Check(unicode)); |
|---|
| 59 | n/a | |
|---|
| 60 | n/a | if (PyList_Append(acc->small, unicode)) |
|---|
| 61 | n/a | return -1; |
|---|
| 62 | n/a | nsmall = PyList_GET_SIZE(acc->small); |
|---|
| 63 | n/a | /* Each item in a list of unicode objects has an overhead (in 64-bit |
|---|
| 64 | n/a | * builds) of: |
|---|
| 65 | n/a | * - 8 bytes for the list slot |
|---|
| 66 | n/a | * - 56 bytes for the header of the unicode object |
|---|
| 67 | n/a | * that is, 64 bytes. 100000 such objects waste more than 6MB |
|---|
| 68 | n/a | * compared to a single concatenated string. |
|---|
| 69 | n/a | */ |
|---|
| 70 | n/a | if (nsmall < 100000) |
|---|
| 71 | n/a | return 0; |
|---|
| 72 | n/a | return flush_accumulator(acc); |
|---|
| 73 | n/a | } |
|---|
| 74 | n/a | |
|---|
| 75 | n/a | PyObject * |
|---|
| 76 | n/a | _PyAccu_FinishAsList(_PyAccu *acc) |
|---|
| 77 | n/a | { |
|---|
| 78 | n/a | int ret; |
|---|
| 79 | n/a | PyObject *res; |
|---|
| 80 | n/a | |
|---|
| 81 | n/a | ret = flush_accumulator(acc); |
|---|
| 82 | n/a | Py_CLEAR(acc->small); |
|---|
| 83 | n/a | if (ret) { |
|---|
| 84 | n/a | Py_CLEAR(acc->large); |
|---|
| 85 | n/a | return NULL; |
|---|
| 86 | n/a | } |
|---|
| 87 | n/a | res = acc->large; |
|---|
| 88 | n/a | acc->large = NULL; |
|---|
| 89 | n/a | return res; |
|---|
| 90 | n/a | } |
|---|
| 91 | n/a | |
|---|
| 92 | n/a | PyObject * |
|---|
| 93 | n/a | _PyAccu_Finish(_PyAccu *acc) |
|---|
| 94 | n/a | { |
|---|
| 95 | n/a | PyObject *list, *res; |
|---|
| 96 | n/a | if (acc->large == NULL) { |
|---|
| 97 | n/a | list = acc->small; |
|---|
| 98 | n/a | acc->small = NULL; |
|---|
| 99 | n/a | } |
|---|
| 100 | n/a | else { |
|---|
| 101 | n/a | list = _PyAccu_FinishAsList(acc); |
|---|
| 102 | n/a | if (!list) |
|---|
| 103 | n/a | return NULL; |
|---|
| 104 | n/a | } |
|---|
| 105 | n/a | res = join_list_unicode(list); |
|---|
| 106 | n/a | Py_DECREF(list); |
|---|
| 107 | n/a | return res; |
|---|
| 108 | n/a | } |
|---|
| 109 | n/a | |
|---|
| 110 | n/a | void |
|---|
| 111 | n/a | _PyAccu_Destroy(_PyAccu *acc) |
|---|
| 112 | n/a | { |
|---|
| 113 | n/a | Py_CLEAR(acc->small); |
|---|
| 114 | n/a | Py_CLEAR(acc->large); |
|---|
| 115 | n/a | } |
|---|