| Guido van Rossum | a330996 | 1993-07-28 09:05:47 +0000 | [diff] [blame] | 1 | #ifndef Py_DICTOBJECT_H | 
 | 2 | #define Py_DICTOBJECT_H | 
 | 3 | #ifdef __cplusplus | 
 | 4 | extern "C" { | 
 | 5 | #endif | 
 | 6 |  | 
| Guido van Rossum | f70e43a | 1991-02-19 12:39:46 +0000 | [diff] [blame] | 7 |  | 
| Guido van Rossum | 2ec9031 | 1997-05-13 21:23:32 +0000 | [diff] [blame] | 8 | /* Dictionary object type -- mapping from hashable object to object */ | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 9 |  | 
| Raymond Hettinger | 930427b | 2003-05-03 06:51:59 +0000 | [diff] [blame] | 10 | /* The distribution includes a separate file, Objects/dictnotes.txt, | 
| Tim Peters | 9b10f7e | 2006-05-30 04:16:25 +0000 | [diff] [blame] | 11 |    describing explorations into dictionary design and optimization. | 
| Raymond Hettinger | 930427b | 2003-05-03 06:51:59 +0000 | [diff] [blame] | 12 |    It covers typical dictionary use patterns, the parameters for | 
 | 13 |    tuning dictionaries, and several ideas for possible optimizations. | 
 | 14 | */ | 
 | 15 |  | 
| Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 16 | /* | 
 | 17 | There are three kinds of slots in the table: | 
 | 18 |  | 
 | 19 | 1. Unused.  me_key == me_value == NULL | 
 | 20 |    Does not hold an active (key, value) pair now and never did.  Unused can | 
 | 21 |    transition to Active upon key insertion.  This is the only case in which | 
 | 22 |    me_key is NULL, and is each slot's initial state. | 
 | 23 |  | 
 | 24 | 2. Active.  me_key != NULL and me_key != dummy and me_value != NULL | 
 | 25 |    Holds an active (key, value) pair.  Active can transition to Dummy upon | 
 | 26 |    key deletion.  This is the only case in which me_value != NULL. | 
 | 27 |  | 
 | 28 | 3. Dummy.  me_key == dummy and me_value == NULL | 
 | 29 |    Previously held an active (key, value) pair, but that was deleted and an | 
 | 30 |    active pair has not yet overwritten the slot.  Dummy can transition to | 
 | 31 |    Active upon key insertion.  Dummy slots cannot be made Unused again | 
 | 32 |    (cannot have me_key set to NULL), else the probe sequence in case of | 
 | 33 |    collision would have no way to know they were once active. | 
 | 34 |  | 
 | 35 | Note: .popitem() abuses the me_hash field of an Unused or Dummy slot to | 
 | 36 | hold a search finger.  The me_hash field of Unused or Dummy slots has no | 
 | 37 | meaning otherwise. | 
 | 38 | */ | 
 | 39 |  | 
 | 40 | /* PyDict_MINSIZE is the minimum size of a dictionary.  This many slots are | 
 | 41 |  * allocated directly in the dict object (in the ma_smalltable member). | 
 | 42 |  * It must be a power of 2, and at least 4.  8 allows dicts with no more | 
 | 43 |  * than 5 active entries to live in ma_smalltable (and so avoid an | 
 | 44 |  * additional malloc); instrumentation suggested this suffices for the | 
 | 45 |  * majority of dicts (consisting mostly of usually-small instance dicts and | 
 | 46 |  * usually-small dicts created to pass keyword arguments). | 
 | 47 |  */ | 
 | 48 | #define PyDict_MINSIZE 8 | 
 | 49 |  | 
 | 50 | typedef struct { | 
| Tim Peters | 9b10f7e | 2006-05-30 04:16:25 +0000 | [diff] [blame] | 51 | 	/* Cached hash code of me_key.  Note that hash codes are C longs. | 
 | 52 | 	 * We have to use Py_ssize_t instead because dict_popitem() abuses | 
 | 53 | 	 * me_hash to hold a search finger. | 
 | 54 | 	 */ | 
 | 55 | 	Py_ssize_t me_hash; | 
| Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 56 | 	PyObject *me_key; | 
 | 57 | 	PyObject *me_value; | 
| Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 58 | } PyDictEntry; | 
 | 59 |  | 
 | 60 | /* | 
 | 61 | To ensure the lookup algorithm terminates, there must be at least one Unused | 
 | 62 | slot (NULL key) in the table. | 
 | 63 | The value ma_fill is the number of non-NULL keys (sum of Active and Dummy); | 
 | 64 | ma_used is the number of non-NULL, non-dummy keys (== the number of non-NULL | 
 | 65 | values == the number of Active items). | 
 | 66 | To avoid slowing down lookups on a near-full table, we resize the table when | 
 | 67 | it's two-thirds full. | 
 | 68 | */ | 
 | 69 | typedef struct _dictobject PyDictObject; | 
 | 70 | struct _dictobject { | 
 | 71 | 	PyObject_HEAD | 
| Tim Peters | 9b10f7e | 2006-05-30 04:16:25 +0000 | [diff] [blame] | 72 | 	Py_ssize_t ma_fill;  /* # Active + # Dummy */ | 
 | 73 | 	Py_ssize_t ma_used;  /* # Active */ | 
| Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 74 |  | 
 | 75 | 	/* The table contains ma_mask + 1 slots, and that's a power of 2. | 
 | 76 | 	 * We store the mask instead of the size because the mask is more | 
 | 77 | 	 * frequently needed. | 
 | 78 | 	 */ | 
| Tim Peters | 9b10f7e | 2006-05-30 04:16:25 +0000 | [diff] [blame] | 79 | 	Py_ssize_t ma_mask; | 
| Tim Peters | 6d6c1a3 | 2001-08-02 04:15:00 +0000 | [diff] [blame] | 80 |  | 
 | 81 | 	/* ma_table points to ma_smalltable for small tables, else to | 
 | 82 | 	 * additional malloc'ed memory.  ma_table is never NULL!  This rule | 
 | 83 | 	 * saves repeated runtime null-tests in the workhorse getitem and | 
 | 84 | 	 * setitem calls. | 
 | 85 | 	 */ | 
 | 86 | 	PyDictEntry *ma_table; | 
 | 87 | 	PyDictEntry *(*ma_lookup)(PyDictObject *mp, PyObject *key, long hash); | 
 | 88 | 	PyDictEntry ma_smalltable[PyDict_MINSIZE]; | 
 | 89 | }; | 
 | 90 |  | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 91 | PyAPI_DATA(PyTypeObject) PyDict_Type; | 
| Alexandre Vassalotti | 69eb516 | 2010-01-11 23:17:10 +0000 | [diff] [blame] | 92 | PyAPI_DATA(PyTypeObject) PyDictIterKey_Type; | 
 | 93 | PyAPI_DATA(PyTypeObject) PyDictIterValue_Type; | 
 | 94 | PyAPI_DATA(PyTypeObject) PyDictIterItem_Type; | 
 | 95 | PyAPI_DATA(PyTypeObject) PyDictKeys_Type; | 
 | 96 | PyAPI_DATA(PyTypeObject) PyDictItems_Type; | 
 | 97 | PyAPI_DATA(PyTypeObject) PyDictValues_Type; | 
| Guido van Rossum | 2ec9031 | 1997-05-13 21:23:32 +0000 | [diff] [blame] | 98 |  | 
| Neal Norwitz | ee3a1b5 | 2007-02-25 19:44:48 +0000 | [diff] [blame] | 99 | #define PyDict_Check(op) \ | 
| Christian Heimes | e93237d | 2007-12-19 02:37:44 +0000 | [diff] [blame] | 100 |                  PyType_FastSubclass(Py_TYPE(op), Py_TPFLAGS_DICT_SUBCLASS) | 
 | 101 | #define PyDict_CheckExact(op) (Py_TYPE(op) == &PyDict_Type) | 
| Alexandre Vassalotti | 69eb516 | 2010-01-11 23:17:10 +0000 | [diff] [blame] | 102 | #define PyDictKeys_Check(op) (Py_TYPE(op) == &PyDictKeys_Type) | 
 | 103 | #define PyDictItems_Check(op) (Py_TYPE(op) == &PyDictItems_Type) | 
 | 104 | #define PyDictValues_Check(op) (Py_TYPE(op) == &PyDictValues_Type) | 
 | 105 | /* This excludes Values, since they are not sets. */ | 
 | 106 | # define PyDictViewSet_Check(op) \ | 
 | 107 | 	(PyDictKeys_Check(op) || PyDictItems_Check(op)) | 
| Guido van Rossum | 2ec9031 | 1997-05-13 21:23:32 +0000 | [diff] [blame] | 108 |  | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 109 | PyAPI_FUNC(PyObject *) PyDict_New(void); | 
 | 110 | PyAPI_FUNC(PyObject *) PyDict_GetItem(PyObject *mp, PyObject *key); | 
 | 111 | PyAPI_FUNC(int) PyDict_SetItem(PyObject *mp, PyObject *key, PyObject *item); | 
 | 112 | PyAPI_FUNC(int) PyDict_DelItem(PyObject *mp, PyObject *key); | 
 | 113 | PyAPI_FUNC(void) PyDict_Clear(PyObject *mp); | 
 | 114 | PyAPI_FUNC(int) PyDict_Next( | 
| Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 115 | 	PyObject *mp, Py_ssize_t *pos, PyObject **key, PyObject **value); | 
| Raymond Hettinger | d6fc72a | 2007-02-19 02:03:19 +0000 | [diff] [blame] | 116 | PyAPI_FUNC(int) _PyDict_Next( | 
 | 117 | 	PyObject *mp, Py_ssize_t *pos, PyObject **key, PyObject **value, long *hash); | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 118 | PyAPI_FUNC(PyObject *) PyDict_Keys(PyObject *mp); | 
 | 119 | PyAPI_FUNC(PyObject *) PyDict_Values(PyObject *mp); | 
 | 120 | PyAPI_FUNC(PyObject *) PyDict_Items(PyObject *mp); | 
| Martin v. Löwis | 18e1655 | 2006-02-15 17:27:45 +0000 | [diff] [blame] | 121 | PyAPI_FUNC(Py_ssize_t) PyDict_Size(PyObject *mp); | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 122 | PyAPI_FUNC(PyObject *) PyDict_Copy(PyObject *mp); | 
| Raymond Hettinger | bc0f2ab | 2003-11-25 21:12:14 +0000 | [diff] [blame] | 123 | PyAPI_FUNC(int) PyDict_Contains(PyObject *mp, PyObject *key); | 
| Raymond Hettinger | d6fc72a | 2007-02-19 02:03:19 +0000 | [diff] [blame] | 124 | PyAPI_FUNC(int) _PyDict_Contains(PyObject *mp, PyObject *key, long hash); | 
| Raymond Hettinger | fd7ed40 | 2007-12-18 21:24:09 +0000 | [diff] [blame] | 125 | PyAPI_FUNC(PyObject *) _PyDict_NewPresized(Py_ssize_t minused); | 
| Antoine Pitrou | f8387af | 2009-03-23 18:41:45 +0000 | [diff] [blame] | 126 | PyAPI_FUNC(void) _PyDict_MaybeUntrack(PyObject *mp); | 
| Jeremy Hylton | a12c7a7 | 2000-03-30 22:27:31 +0000 | [diff] [blame] | 127 |  | 
| Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 128 | /* PyDict_Update(mp, other) is equivalent to PyDict_Merge(mp, other, 1). */ | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 129 | PyAPI_FUNC(int) PyDict_Update(PyObject *mp, PyObject *other); | 
| Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 130 |  | 
 | 131 | /* PyDict_Merge updates/merges from a mapping object (an object that | 
 | 132 |    supports PyMapping_Keys() and PyObject_GetItem()).  If override is true, | 
 | 133 |    the last occurrence of a key wins, else the first.  The Python | 
 | 134 |    dict.update(other) is equivalent to PyDict_Merge(dict, other, 1). | 
 | 135 | */ | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 136 | PyAPI_FUNC(int) PyDict_Merge(PyObject *mp, | 
| Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 137 | 				   PyObject *other, | 
 | 138 | 				   int override); | 
 | 139 |  | 
 | 140 | /* PyDict_MergeFromSeq2 updates/merges from an iterable object producing | 
 | 141 |    iterable objects of length 2.  If override is true, the last occurrence | 
 | 142 |    of a key wins, else the first.  The Python dict constructor dict(seq2) | 
 | 143 |    is equivalent to dict={}; PyDict_MergeFromSeq(dict, seq2, 1). | 
 | 144 | */ | 
| Mark Hammond | 91a681d | 2002-08-12 07:21:58 +0000 | [diff] [blame] | 145 | PyAPI_FUNC(int) PyDict_MergeFromSeq2(PyObject *d, | 
| Tim Peters | f582b82 | 2001-12-11 18:51:08 +0000 | [diff] [blame] | 146 | 					   PyObject *seq2, | 
 | 147 | 					   int override); | 
| Guido van Rossum | 2ec9031 | 1997-05-13 21:23:32 +0000 | [diff] [blame] | 148 |  | 
| Martin v. Löwis | 32b4a1b | 2002-12-11 13:21:12 +0000 | [diff] [blame] | 149 | PyAPI_FUNC(PyObject *) PyDict_GetItemString(PyObject *dp, const char *key); | 
 | 150 | PyAPI_FUNC(int) PyDict_SetItemString(PyObject *dp, const char *key, PyObject *item); | 
 | 151 | PyAPI_FUNC(int) PyDict_DelItemString(PyObject *dp, const char *key); | 
| Guido van Rossum | 85a5fbb | 1990-10-14 12:07:46 +0000 | [diff] [blame] | 152 |  | 
| Guido van Rossum | a330996 | 1993-07-28 09:05:47 +0000 | [diff] [blame] | 153 | #ifdef __cplusplus | 
 | 154 | } | 
 | 155 | #endif | 
 | 156 | #endif /* !Py_DICTOBJECT_H */ |