Index: Include/stringobject.h =================================================================== RCS file: /cvsroot/python/python/dist/src/Include/stringobject.h,v retrieving revision 2.34 diff -c -r2.34 stringobject.h *** Include/stringobject.h 24 May 2002 19:01:57 -0000 2.34 --- Include/stringobject.h 10 Aug 2002 09:13:20 -0000 *************** *** 25,31 **** */ /* Caching the hash (ob_shash) saves recalculation of a string's hash value. ! Interning strings (ob_sinterned) tries to ensure that only one string object with a given value exists, so equality tests can be one pointer comparison. This is generally restricted to strings that "look like" Python identifiers, although the intern() builtin can be used to force --- 25,31 ---- */ /* Caching the hash (ob_shash) saves recalculation of a string's hash value. ! Interning strings (ob_sstate) tries to ensure that only one string object with a given value exists, so equality tests can be one pointer comparison. This is generally restricted to strings that "look like" Python identifiers, although the intern() builtin can be used to force *************** *** 35,44 **** typedef struct { PyObject_VAR_HEAD long ob_shash; ! PyObject *ob_sinterned; char ob_sval[1]; } PyStringObject; extern DL_IMPORT(PyTypeObject) PyBaseString_Type; extern DL_IMPORT(PyTypeObject) PyString_Type; --- 35,48 ---- typedef struct { PyObject_VAR_HEAD long ob_shash; ! char ob_sstate; char ob_sval[1]; } PyStringObject; + #define SSTATE_NOT_INTERNED 0 + #define SSTATE_INTERNED_MORTAL 1 + #define SSTATE_INTERNED_IMMORTAL 2 + extern DL_IMPORT(PyTypeObject) PyBaseString_Type; extern DL_IMPORT(PyTypeObject) PyString_Type; *************** *** 61,69 **** --- 65,83 ---- extern DL_IMPORT(PyObject *) _PyString_FormatLong(PyObject*, int, int, int, char**, int*); + extern DL_IMPORT(void) PyString_Intern(PyObject **); extern DL_IMPORT(void) PyString_InternInPlace(PyObject **); extern DL_IMPORT(PyObject *) PyString_InternFromString(const char *); extern DL_IMPORT(void) _Py_ReleaseInternedStrings(void); + + /* Use only if you know it's a string */ + #define PyString_CHECK_INTERNED(op) (((PyStringObject *)(op))->ob_sstate) + + /* As safe as PyString_Intern, but faster */ + #define PyString_INTERN(op) do {\ + if (PyString_CheckExact(op) && !PyString_CheckInterned(*(op))) \ + PyString_Intern(op); \ + } while (0); /* Macro, trading safety for speed */ #define PyString_AS_STRING(op) (((PyStringObject *)(op))->ob_sval) Index: Objects/dictobject.c =================================================================== RCS file: /cvsroot/python/python/dist/src/Objects/dictobject.c,v retrieving revision 2.128 diff -c -r2.128 dictobject.c *** Objects/dictobject.c 17 Jul 2002 16:30:37 -0000 2.128 --- Objects/dictobject.c 10 Aug 2002 09:13:39 -0000 *************** *** 511,525 **** } mp = (dictobject *)op; if (PyString_CheckExact(key)) { ! if (((PyStringObject *)key)->ob_sinterned != NULL) { ! key = ((PyStringObject *)key)->ob_sinterned; ! hash = ((PyStringObject *)key)->ob_shash; ! } ! else { ! hash = ((PyStringObject *)key)->ob_shash; ! if (hash == -1) ! hash = PyObject_Hash(key); ! } } else { hash = PyObject_Hash(key); --- 511,519 ---- } mp = (dictobject *)op; if (PyString_CheckExact(key)) { ! hash = ((PyStringObject *)key)->ob_shash; ! if (hash == -1) ! hash = PyObject_Hash(key); } else { hash = PyObject_Hash(key); *************** *** 1891,1897 **** kv = PyString_FromString(key); if (kv == NULL) return -1; ! PyString_InternInPlace(&kv); /* XXX Should we really? */ err = PyDict_SetItem(v, kv, item); Py_DECREF(kv); return err; --- 1885,1891 ---- kv = PyString_FromString(key); if (kv == NULL) return -1; ! PyString_Intern(&kv); err = PyDict_SetItem(v, kv, item); Py_DECREF(kv); return err; Index: Objects/stringobject.c =================================================================== RCS file: /cvsroot/python/python/dist/src/Objects/stringobject.c,v retrieving revision 2.175 diff -c -r2.175 stringobject.c *** Objects/stringobject.c 6 Aug 2002 16:58:21 -0000 2.175 --- Objects/stringobject.c 10 Aug 2002 09:13:47 -0000 *************** *** 15,20 **** --- 15,31 ---- static PyStringObject *characters[UCHAR_MAX + 1]; static PyStringObject *nullstring; + /* This dictionary holds all interned strings. Note that references to + strings in this dictionary are *not* counted in the string's ob_refcnt. + When the interned string reaches a refcnt of 0 the string deallocation + function will delete the reference from this dictionary. + + Another way to look at this is that to say that the actual reference count + of a string is: s->ob_refcnt + s->ob_sstate?2:0 + */ + static PyObject *interned; + + /* For both PyString_FromString() and PyString_FromStringAndSize(), the parameter `size' denotes number of characters to allocate, not counting any *************** *** 69,88 **** return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sinterned = NULL; if (str != NULL) memcpy(op->ob_sval, str, size); op->ob_sval[size] = '\0'; /* share short strings */ if (size == 0) { PyObject *t = (PyObject *)op; ! PyString_InternInPlace(&t); op = (PyStringObject *)t; nullstring = op; Py_INCREF(op); } else if (size == 1 && str != NULL) { PyObject *t = (PyObject *)op; ! PyString_InternInPlace(&t); op = (PyStringObject *)t; characters[*str & UCHAR_MAX] = op; Py_INCREF(op); --- 80,99 ---- return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sstate = SSTATE_NOT_INTERNED; if (str != NULL) memcpy(op->ob_sval, str, size); op->ob_sval[size] = '\0'; /* share short strings */ if (size == 0) { PyObject *t = (PyObject *)op; ! PyString_Intern(&t); op = (PyStringObject *)t; nullstring = op; Py_INCREF(op); } else if (size == 1 && str != NULL) { PyObject *t = (PyObject *)op; ! PyString_Intern(&t); op = (PyStringObject *)t; characters[*str & UCHAR_MAX] = op; Py_INCREF(op); *************** *** 125,142 **** return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sinterned = NULL; memcpy(op->ob_sval, str, size+1); /* share short strings */ if (size == 0) { PyObject *t = (PyObject *)op; ! PyString_InternInPlace(&t); op = (PyStringObject *)t; nullstring = op; Py_INCREF(op); } else if (size == 1) { PyObject *t = (PyObject *)op; ! PyString_InternInPlace(&t); op = (PyStringObject *)t; characters[*str & UCHAR_MAX] = op; Py_INCREF(op); --- 136,153 ---- return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sstate = SSTATE_NOT_INTERNED; memcpy(op->ob_sval, str, size+1); /* share short strings */ if (size == 0) { PyObject *t = (PyObject *)op; ! PyString_Intern(&t); op = (PyStringObject *)t; nullstring = op; Py_INCREF(op); } else if (size == 1) { PyObject *t = (PyObject *)op; ! PyString_Intern(&t); op = (PyStringObject *)t; characters[*str & UCHAR_MAX] = op; Py_INCREF(op); *************** *** 486,491 **** --- 497,518 ---- static void string_dealloc(PyObject *op) { + switch (PyString_CHECK_INTERNED(op)) { + case SSTATE_NOT_INTERNED: + break; + + case SSTATE_INTERNED_MORTAL: + /* revive dead object temporarily for DelItem */ + op->ob_refcnt = 3; + PyDict_DelItem(interned, op); + break; + + case SSTATE_INTERNED_IMMORTAL: + Py_FatalError("Immortal interned string died."); + + default: + Py_FatalError("Inconsistent interned string state."); + } op->ob_type->tp_free(op); } *************** *** 728,734 **** return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sinterned = NULL; memcpy(op->ob_sval, a->ob_sval, (int) a->ob_size); memcpy(op->ob_sval + a->ob_size, b->ob_sval, (int) b->ob_size); op->ob_sval[size] = '\0'; --- 755,761 ---- return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sstate = SSTATE_NOT_INTERNED; memcpy(op->ob_sval, a->ob_sval, (int) a->ob_size); memcpy(op->ob_sval + a->ob_size, b->ob_sval, (int) b->ob_size); op->ob_sval[size] = '\0'; *************** *** 771,777 **** return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sinterned = NULL; for (i = 0; i < size; i += a->ob_size) memcpy(op->ob_sval+i, a->ob_sval, (int) a->ob_size); op->ob_sval[size] = '\0'; --- 798,804 ---- return PyErr_NoMemory(); PyObject_INIT_VAR(op, &PyString_Type, size); op->ob_shash = -1; ! op->ob_sstate = SSTATE_NOT_INTERNED; for (i = 0; i < size; i += a->ob_size) memcpy(op->ob_sval+i, a->ob_sval, (int) a->ob_size); op->ob_sval[size] = '\0'; *************** *** 936,944 **** if (a->ob_shash != -1) return a->ob_shash; - if (a->ob_sinterned != NULL) - return (a->ob_shash = - ((PyStringObject *)(a->ob_sinterned))->ob_shash); len = a->ob_size; p = (unsigned char *) a->ob_sval; x = *p << 7; --- 963,968 ---- *************** *** 2910,2917 **** memcpy(PyString_AS_STRING(pnew), PyString_AS_STRING(tmp), n+1); ((PyStringObject *)pnew)->ob_shash = ((PyStringObject *)tmp)->ob_shash; ! ((PyStringObject *)pnew)->ob_sinterned = ! ((PyStringObject *)tmp)->ob_sinterned; } Py_DECREF(tmp); return pnew; --- 2934,2940 ---- memcpy(PyString_AS_STRING(pnew), PyString_AS_STRING(tmp), n+1); ((PyStringObject *)pnew)->ob_shash = ((PyStringObject *)tmp)->ob_shash; ! ((PyStringObject *)pnew)->ob_sstate = SSTATE_NOT_INTERNED; } Py_DECREF(tmp); return pnew; *************** *** 3820,3891 **** return NULL; } - - - /* This dictionary will leak at PyString_Fini() time. That's acceptable - * because PyString_Fini() specifically frees interned strings that are - * only referenced by this dictionary. The CVS log entry for revision 2.45 - * says: - * - * Change the Fini function to only remove otherwise unreferenced - * strings from the interned table. There are references in - * hard-to-find static variables all over the interpreter, and it's not - * worth trying to get rid of all those; but "uninterning" isn't fair - * either and may cause subtle failures later -- so we have to keep them - * in the interned table. - */ - static PyObject *interned; - void ! PyString_InternInPlace(PyObject **p) { register PyStringObject *s = (PyStringObject *)(*p); PyObject *t; if (s == NULL || !PyString_Check(s)) ! Py_FatalError("PyString_InternInPlace: strings only please!"); ! if ((t = s->ob_sinterned) != NULL) { ! if (t == (PyObject *)s) ! return; ! Py_INCREF(t); ! *p = t; ! Py_DECREF(s); return; - } if (interned == NULL) { interned = PyDict_New(); if (interned == NULL) return; } if ((t = PyDict_GetItem(interned, (PyObject *)s)) != NULL) { Py_INCREF(t); ! *p = s->ob_sinterned = t; ! Py_DECREF(s); return; } ! /* Ensure that only true string objects appear in the intern dict, ! and as the value of ob_sinterned. */ ! if (PyString_CheckExact(s)) { ! t = (PyObject *)s; ! if (PyDict_SetItem(interned, t, t) == 0) { ! s->ob_sinterned = t; ! return; ! } ! } ! else { t = PyString_FromStringAndSize(PyString_AS_STRING(s), PyString_GET_SIZE(s)); ! if (t != NULL) { ! if (PyDict_SetItem(interned, t, t) == 0) { ! *p = s->ob_sinterned = t; ! Py_DECREF(s); ! return; ! } ! Py_DECREF(t); } } PyErr_Clear(); } PyObject * PyString_InternFromString(const char *cp) --- 3843,3906 ---- return NULL; } void ! PyString_Intern(PyObject **p) { register PyStringObject *s = (PyStringObject *)(*p); PyObject *t; if (s == NULL || !PyString_Check(s)) ! Py_FatalError("PyString_Intern: strings only please!"); ! if (PyString_CHECK_INTERNED(s)) return; if (interned == NULL) { interned = PyDict_New(); + if (interned == NULL) return; + /* PyObject_GC_UnTrack(interned); ??? */ } if ((t = PyDict_GetItem(interned, (PyObject *)s)) != NULL) { Py_INCREF(t); ! Py_DECREF(*p); ! *p = t; return; } ! /* Ensure that only true string objects appear in the intern dict */ ! if (!PyString_CheckExact(s)) { t = PyString_FromStringAndSize(PyString_AS_STRING(s), PyString_GET_SIZE(s)); ! if (t == NULL) { ! PyErr_Clear(); ! return; } + } else { + t = (PyObject*) s; + Py_INCREF(t); + } + + if (PyDict_SetItem(interned, t, t) == 0) { + /* The two references in interned are not counted by + refcnt. The string deallocator will take care of this */ + ((PyObject *)t)->ob_refcnt-=2; + PyString_CHECK_INTERNED(t) = SSTATE_INTERNED_MORTAL; + Py_DECREF(*p); + *p = t; + return; } + Py_DECREF(t); PyErr_Clear(); } + void + PyString_InternInPlace(PyObject **p) + { + PyString_Intern(p); + if (PyString_CHECK_INTERNED(*p) != SSTATE_INTERNED_IMMORTAL) { + PyString_CHECK_INTERNED(*p) = SSTATE_INTERNED_IMMORTAL; + Py_INCREF(*p); + } + } + PyObject * PyString_InternFromString(const char *cp) *************** *** 3907,3934 **** } Py_XDECREF(nullstring); nullstring = NULL; - if (interned) { - int pos, changed; - PyObject *key, *value; - do { - changed = 0; - pos = 0; - while (PyDict_Next(interned, &pos, &key, &value)) { - if (key->ob_refcnt == 2 && key == value) { - PyDict_DelItem(interned, key); - changed = 1; - } - } - } while (changed); - } } void _Py_ReleaseInternedStrings(void) { ! if (interned) { ! fprintf(stderr, "releasing interned strings\n"); ! PyDict_Clear(interned); ! Py_DECREF(interned); ! interned = NULL; ! } } --- 3922,3948 ---- } Py_XDECREF(nullstring); nullstring = NULL; } void _Py_ReleaseInternedStrings(void) { ! PyObject *keys; ! PyStringObject *s; ! int i; ! ! if (interned == NULL) return; ! keys = PyMapping_Keys(interned); ! for (i = 0; i < PyMapping_Size(keys); i++) { ! s = (PyStringObject *) PyList_GET_ITEM(keys, i); ! /* Only immortal strings should be left in interned ! by now. Check and report ??? */ ! s->ob_sstate = SSTATE_INTERNED_MORTAL; ! s->ob_refcnt = 1; ! } ! Py_XDECREF(keys); /* Does the actual disposal */ ! if (PyDict_Size(interned) != 0) ! Py_FatalError("Error releasing interned string"); ! /* _PyObject_GC_TRACK(interned); ??? */ ! Py_DECREF(interned); ! interned = NULL; } Index: Python/compile.c =================================================================== RCS file: /cvsroot/python/python/dist/src/Python/compile.c,v retrieving revision 2.252 diff -c -r2.252 compile.c *** Python/compile.c 7 Aug 2002 15:18:57 -0000 2.252 --- Python/compile.c 10 Aug 2002 09:13:57 -0000 *************** *** 318,324 **** PyErr_BadInternalCall(); return -1; } ! PyString_InternInPlace(&PyTuple_GET_ITEM(tuple, i)); } return 0; } --- 318,324 ---- PyErr_BadInternalCall(); return -1; } ! PyString_Intern(&PyTuple_GET_ITEM(tuple, i)); } return 0; } *************** *** 358,364 **** continue; if (!all_name_chars((unsigned char *)PyString_AS_STRING(v))) continue; ! PyString_InternInPlace(&PyTuple_GET_ITEM(consts, i)); } co = PyObject_NEW(PyCodeObject, &PyCode_Type); if (co != NULL) { --- 358,364 ---- continue; if (!all_name_chars((unsigned char *)PyString_AS_STRING(v))) continue; ! PyString_Intern(&PyTuple_GET_ITEM(consts, i)); } co = PyObject_NEW(PyCodeObject, &PyCode_Type); if (co != NULL) {