Antoine Pitrou | eeb7eea | 2011-10-06 18:57:27 +0200 | [diff] [blame] | 1 | /* Accumulator struct implementation */ |
| 2 | |
| 3 | #include "Python.h" |
| 4 | |
| 5 | static PyObject * |
| 6 | join_list_unicode(PyObject *lst) |
| 7 | { |
| 8 | /* return ''.join(lst) */ |
| 9 | PyObject *sep, *ret; |
| 10 | sep = PyUnicode_FromStringAndSize("", 0); |
| 11 | ret = PyUnicode_Join(sep, lst); |
| 12 | Py_DECREF(sep); |
| 13 | return ret; |
| 14 | } |
| 15 | |
| 16 | int |
| 17 | _PyAccu_Init(_PyAccu *acc) |
| 18 | { |
| 19 | /* Lazily allocated */ |
| 20 | acc->large = NULL; |
| 21 | acc->small = PyList_New(0); |
| 22 | if (acc->small == NULL) |
| 23 | return -1; |
| 24 | return 0; |
| 25 | } |
| 26 | |
| 27 | static int |
| 28 | flush_accumulator(_PyAccu *acc) |
| 29 | { |
| 30 | Py_ssize_t nsmall = PyList_GET_SIZE(acc->small); |
| 31 | if (nsmall) { |
| 32 | int ret; |
| 33 | PyObject *joined; |
| 34 | if (acc->large == NULL) { |
| 35 | acc->large = PyList_New(0); |
| 36 | if (acc->large == NULL) |
| 37 | return -1; |
| 38 | } |
| 39 | joined = join_list_unicode(acc->small); |
| 40 | if (joined == NULL) |
| 41 | return -1; |
| 42 | if (PyList_SetSlice(acc->small, 0, nsmall, NULL)) { |
| 43 | Py_DECREF(joined); |
| 44 | return -1; |
| 45 | } |
| 46 | ret = PyList_Append(acc->large, joined); |
| 47 | Py_DECREF(joined); |
| 48 | return ret; |
| 49 | } |
| 50 | return 0; |
| 51 | } |
| 52 | |
| 53 | int |
| 54 | _PyAccu_Accumulate(_PyAccu *acc, PyObject *unicode) |
| 55 | { |
| 56 | Py_ssize_t nsmall; |
| 57 | assert(PyUnicode_Check(unicode)); |
| 58 | |
| 59 | if (PyList_Append(acc->small, unicode)) |
| 60 | return -1; |
| 61 | nsmall = PyList_GET_SIZE(acc->small); |
| 62 | /* Each item in a list of unicode objects has an overhead (in 64-bit |
| 63 | * builds) of: |
| 64 | * - 8 bytes for the list slot |
| 65 | * - 56 bytes for the header of the unicode object |
| 66 | * that is, 64 bytes. 100000 such objects waste more than 6MB |
| 67 | * compared to a single concatenated string. |
| 68 | */ |
| 69 | if (nsmall < 100000) |
| 70 | return 0; |
| 71 | return flush_accumulator(acc); |
| 72 | } |
| 73 | |
| 74 | PyObject * |
| 75 | _PyAccu_FinishAsList(_PyAccu *acc) |
| 76 | { |
| 77 | int ret; |
| 78 | PyObject *res; |
| 79 | |
| 80 | ret = flush_accumulator(acc); |
| 81 | Py_CLEAR(acc->small); |
| 82 | if (ret) { |
| 83 | Py_CLEAR(acc->large); |
| 84 | return NULL; |
| 85 | } |
| 86 | res = acc->large; |
| 87 | acc->large = NULL; |
| 88 | return res; |
| 89 | } |
| 90 | |
| 91 | PyObject * |
| 92 | _PyAccu_Finish(_PyAccu *acc) |
| 93 | { |
| 94 | PyObject *list, *res; |
| 95 | if (acc->large == NULL) { |
| 96 | list = acc->small; |
| 97 | acc->small = NULL; |
| 98 | } |
| 99 | else { |
| 100 | list = _PyAccu_FinishAsList(acc); |
| 101 | if (!list) |
| 102 | return NULL; |
| 103 | } |
| 104 | res = join_list_unicode(list); |
| 105 | Py_DECREF(list); |
| 106 | return res; |
| 107 | } |
| 108 | |
| 109 | void |
| 110 | _PyAccu_Destroy(_PyAccu *acc) |
| 111 | { |
| 112 | Py_CLEAR(acc->small); |
| 113 | Py_CLEAR(acc->large); |
| 114 | } |