Mercurial > hg > xemacs-beta
view src/elhash.c @ 314:341dac730539 r21-0b55
Import from CVS: tag r21-0b55
author | cvs |
---|---|
date | Mon, 13 Aug 2007 10:44:22 +0200 |
parents | c5d627a313b1 |
children | 8626e4521993 |
line wrap: on
line source
/* Lisp interface to hash tables. Copyright (C) 1992, 1993, 1994 Free Software Foundation, Inc. Copyright (C) 1995, 1996 Ben Wing. Copyright (C) 1997 Free Software Foundation, Inc. This file is part of XEmacs. XEmacs is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2, or (at your option) any later version. XEmacs is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with XEmacs; see the file COPYING. If not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ /* Synched up with: Not in FSF. */ #include <config.h> #include "lisp.h" #include "hash.h" #include "elhash.h" #include "bytecode.h" EXFUN (Fmake_weak_hashtable, 2); EXFUN (Fmake_key_weak_hashtable, 2); EXFUN (Fmake_value_weak_hashtable, 2); Lisp_Object Qhashtablep, Qhashtable; Lisp_Object Qweak, Qkey_weak, Qvalue_weak, Qnon_weak; #define LISP_OBJECTS_PER_HENTRY (sizeof (hentry) / sizeof (Lisp_Object))/* 2 */ struct hashtable { struct lcrecord_header header; unsigned int fullness; unsigned long (*hash_function) (CONST void *); int (*test_function) (CONST void *, CONST void *); Lisp_Object zero_entry; Lisp_Object harray; enum hashtable_type type; /* whether and how this hashtable is weak */ Lisp_Object next_weak; /* Used to chain together all of the weak hashtables. Don't mark through this. */ }; static Lisp_Object Vall_weak_hashtables; static Lisp_Object mark_hashtable (Lisp_Object obj, void (*markobj) (Lisp_Object)) { struct hashtable *table = XHASHTABLE (obj); if (table->type != HASHTABLE_NONWEAK) { /* If the table is weak, we don't want to mark the keys and values (we scan over them after everything else has been marked, and mark or remove them as necessary). Note that we will mark the table->harray itself at the same time; it's hard to mark that here without also marking its contents. */ return Qnil; } ((markobj) (table->zero_entry)); return table->harray; } /* Equality of hashtables. Two hashtables are equal when they are of the same type and test function, they have the same number of elements, and for each key in hashtable, the values are `equal'. This is similar to Common Lisp `equalp' of hashtables, with the difference that CL requires the keys to be compared with the test function, which we don't do. Doing that would require consing, and consing is bad idea in `equal'. Anyway, our method should provide the same result -- if the keys are not equal according to test function, then Fgethash() in hashtable_equal_mapper() will fail. */ struct hashtable_equal_closure { int depth; int equal; Lisp_Object other_table; }; static int hashtable_equal_mapper (CONST void *key, void *contents, void *arg) { struct hashtable_equal_closure *closure = (struct hashtable_equal_closure *)arg; Lisp_Object keytem, valuetem; Lisp_Object value_in_other; CVOID_TO_LISP (keytem, key); CVOID_TO_LISP (valuetem, contents); /* Look up the key in the other hashtable, and compare the values. */ value_in_other = Fgethash (keytem, closure->other_table, Qunbound); if (UNBOUNDP (value_in_other) || !internal_equal (valuetem, value_in_other, closure->depth)) { /* Give up. */ closure->equal = 0; return 1; } return 0; } static int hashtable_equal (Lisp_Object t1, Lisp_Object t2, int depth) { struct hashtable_equal_closure closure; struct hashtable *table1 = XHASHTABLE (t1); struct hashtable *table2 = XHASHTABLE (t2); /* The objects are `equal' if they are of the same type, so return 0 if types or test functions are not the same. Obviously, the number of elements must be equal, too. #### table->fullness is broken, so we cannot use it. */ if ((table1->test_function != table2->test_function) || (table1->type != table2->type) /*|| (table1->fullness != table2->fullness))*/ ) return 0; closure.depth = depth + 1; closure.equal = 1; closure.other_table = t2; elisp_maphash (hashtable_equal_mapper, t1, &closure); return closure.equal; } /* Printing hashtables. This is non-trivial, because we use a readable structure-style syntax for hashtables. This means that a typical hashtable will be readably printed in the form of: #s(hashtable size 2 data (key1 value1 key2 value2)) The supported keywords are `type' (non-weak (or nil), weak, key-weak and value-weak), `test' (eql (or nil), eq or equal), `size' (a natnum or nil) and `data' (a list). If `print-readably' is non-nil, then a simpler syntax is used; for instance: #<hashtable size 2/13 data (key1 value1 key2 value2) 0x874d> The data is truncated to four pairs, and the rest is shown with `...'. This printer does not cons. */ struct print_hashtable_data_closure { EMACS_INT count; /* Used to implement truncation for non-readable printing, as well as to avoid the unnecessary space at the beginning. */ Lisp_Object printcharfun; }; static int print_hashtable_data_mapper (CONST void *key, void *contents, void *arg) { Lisp_Object keytem, valuetem; struct print_hashtable_data_closure *closure = (struct print_hashtable_data_closure *)arg; if (closure->count < 4 || print_readably) { CVOID_TO_LISP (keytem, key); CVOID_TO_LISP (valuetem, contents); if (closure->count) write_c_string (" ", closure->printcharfun); print_internal (keytem, closure->printcharfun, 1); write_c_string (" ", closure->printcharfun); print_internal (valuetem, closure->printcharfun, 1); } ++closure->count; return 0; } /* Print the data of the hashtable. This maps through a Lisp hashtable and prints key/value pairs using PRINTCHARFUN. */ static void print_hashtable_data (Lisp_Object hashtable, Lisp_Object printcharfun) { struct print_hashtable_data_closure closure; closure.count = 0; closure.printcharfun = printcharfun; write_c_string (" data (", printcharfun); elisp_maphash (print_hashtable_data_mapper, hashtable, &closure); write_c_string ((!print_readably && closure.count > 4) ? " ...)" : ")", printcharfun); } /* Needed for tests. */ static int lisp_object_eql_equal (CONST void *x1, CONST void *x2); static int lisp_object_equal_equal (CONST void *x1, CONST void *x2); static void print_hashtable (Lisp_Object obj, Lisp_Object printcharfun, int escapeflag) { struct hashtable *table = XHASHTABLE (obj); char buf[128]; write_c_string (print_readably ? "#s(hashtable" : "#<hashtable", printcharfun); if (table->type != HASHTABLE_NONWEAK) { sprintf (buf, " type %s", (table->type == HASHTABLE_WEAK ? "weak" : table->type == HASHTABLE_KEY_WEAK ? "key-weak" : table->type == HASHTABLE_VALUE_WEAK ? "value-weak" : "you-d-better-not-see-this")); write_c_string (buf, printcharfun); } /* These checks have a kludgy look to them, but they are safe. Due to nature of hashing, you cannot use arbitrary test functions anyway. */ if (!table->test_function) write_c_string (" test eq", printcharfun); else if (table->test_function == lisp_object_equal_equal) write_c_string (" test equal", printcharfun); else if (table->test_function == lisp_object_eql_equal) DO_NOTHING; else abort (); if (table->fullness || !print_readably) { if (print_readably) sprintf (buf, " size %u", table->fullness); else sprintf (buf, " size %u/%ld", table->fullness, XVECTOR_LENGTH (table->harray) / LISP_OBJECTS_PER_HENTRY); write_c_string (buf, printcharfun); } if (table->fullness) print_hashtable_data (obj, printcharfun); if (print_readably) write_c_string (")", printcharfun); else { sprintf (buf, " 0x%x>", table->header.uid); write_c_string (buf, printcharfun); } } DEFINE_LRECORD_IMPLEMENTATION ("hashtable", hashtable, mark_hashtable, print_hashtable, 0, /* #### Implement hashtable_hash()! */ hashtable_equal, 0, struct hashtable); /* Pretty reading of hashtables. Here we use the existing structures mechanism (which is, unfortunately, pretty cumbersome) for validating and instantiating the hashtables. The idea is that the side-effect of reading a #s(hashtable PLIST) object is creation of a hashtable with desired properties, and that the hashtable is returned. */ /* Validation functions: each keyword provides its own validation function. The errors should maybe be continuable, but it is unclear how this would cope with ERRB. */ static int hashtable_type_validate (Lisp_Object keyword, Lisp_Object value, Error_behavior errb) { if (!(NILP (value) || EQ (value, Qnon_weak) || EQ (value, Qweak) || EQ (value, Qkey_weak) || EQ (value, Qvalue_weak))) { maybe_signal_simple_error ("Invalid hashtable type", value, Qhashtable, errb); return 0; } return 1; } static int hashtable_test_validate (Lisp_Object keyword, Lisp_Object value, Error_behavior errb) { if (!(NILP (value) || EQ (value, Qeq) || EQ (value, Qeql) || EQ (value, Qequal))) { maybe_signal_simple_error ("Invalid hashtable test", value, Qhashtable, errb); return 0; } return 1; } static int hashtable_size_validate (Lisp_Object keyword, Lisp_Object value, Error_behavior errb) { if (!NATNUMP (value)) { maybe_signal_error (Qwrong_type_argument, list2 (Qnatnump, value), Qhashtable, errb); return 0; } return 1; } static int hashtable_data_validate (Lisp_Object keyword, Lisp_Object value, Error_behavior errb) { int num = 0; Lisp_Object tail; /* #### Doesn't respect ERRB! */ EXTERNAL_LIST_LOOP (tail, value) { ++num; QUIT; } if (num & 1) { maybe_signal_simple_error ("Hashtable data must have alternating keyword/value pairs", value, Qhashtable, errb); return 0; } return 1; } /* The actual instantiation of hashtable. This does practically no error checking, because it relies on the fact that the paranoid functions above have error-checked everything to the last details. If this assumption is wrong, we will get a crash immediately (with error-checking compiled in), and we'll know if there is a bug in the structure mechanism. So there. */ static Lisp_Object hashtable_instantiate (Lisp_Object plist) { /* I'm not sure whether this can GC, but better safe than sorry. */ Lisp_Object hashtab = Qnil; Lisp_Object type = Qnil, test = Qnil, size = Qnil, data = Qnil; struct gcpro gcpro1; GCPRO1 (hashtab); while (!NILP (plist)) { Lisp_Object key, value; key = XCAR (plist); plist = XCDR (plist); value = XCAR (plist); plist = XCDR (plist); if (EQ (key, Qtype)) type = value; else if (EQ (key, Qtest)) test = value; else if (EQ (key, Qsize)) size = value; else if (EQ (key, Qdata)) data = value; else abort (); } if (NILP (type)) type = Qnon_weak; if (NILP (size)) /* Divide by two, because data is a plist. */ size = make_int (XINT (Flength (data)) / 2); /* Create the hashtable. */ if (EQ (type, Qnon_weak)) hashtab = Fmake_hashtable (size, test); else if (EQ (type, Qweak)) hashtab = Fmake_weak_hashtable (size, test); else if (EQ (type, Qkey_weak)) hashtab = Fmake_key_weak_hashtable (size, test); else if (EQ (type, Qvalue_weak)) hashtab = Fmake_value_weak_hashtable (size, test); else abort (); /* And fill it with data. */ while (!NILP (data)) { Lisp_Object key, value; key = XCAR (data); data = XCDR (data); value = XCAR (data); data = XCDR (data); Fputhash (key, value, hashtab); } UNGCPRO; return hashtab; } /* Initialize the hashtable as a structure type. This is called from emacs.c. */ void structure_type_create_hashtable (void) { struct structure_type *st; st = define_structure_type (Qhashtable, 0, hashtable_instantiate); define_structure_type_keyword (st, Qtype, hashtable_type_validate); define_structure_type_keyword (st, Qtest, hashtable_test_validate); define_structure_type_keyword (st, Qsize, hashtable_size_validate); define_structure_type_keyword (st, Qdata, hashtable_data_validate); } /* Basic conversion and allocation functions. */ /* Create a C hashtable from the data in the Lisp hashtable. The actual vector is not copied, nor are the keys or values copied. */ static void ht_copy_to_c (struct hashtable *ht, c_hashtable c_table) { int len = XVECTOR_LENGTH (ht->harray); c_table->harray = (hentry *) XVECTOR_DATA (ht->harray); c_table->zero_set = (!GC_UNBOUNDP (ht->zero_entry)); c_table->zero_entry = LISP_TO_VOID (ht->zero_entry); #ifndef LRECORD_VECTOR if (len < 0) { /* #### if alloc.c mark_object() changes, this must change too. */ /* barf gag retch. When a vector is marked, its len is made less than 0. In the prune_weak_hashtables() stage, we are called on vectors that are like this, and we must be able to deal. */ assert (gc_in_progress); len = -1 - len; } #endif c_table->size = len/LISP_OBJECTS_PER_HENTRY; c_table->fullness = ht->fullness; c_table->hash_function = ht->hash_function; c_table->test_function = ht->test_function; XSETHASHTABLE (c_table->elisp_table, ht); } static void ht_copy_from_c (c_hashtable c_table, struct hashtable *ht) { struct Lisp_Vector dummy; /* C is truly hateful */ void *vec_addr = ((char *) c_table->harray - ((char *) &(dummy.contents[0]) - (char *) &dummy)); XSETVECTOR (ht->harray, vec_addr); if (c_table->zero_set) VOID_TO_LISP (ht->zero_entry, c_table->zero_entry); else ht->zero_entry = Qunbound; ht->fullness = c_table->fullness; } static struct hashtable * allocate_hashtable (void) { struct hashtable *table = alloc_lcrecord_type (struct hashtable, lrecord_hashtable); table->harray = Qnil; table->zero_entry = Qunbound; table->fullness = 0; table->hash_function = 0; table->test_function = 0; return table; } void * elisp_hvector_malloc (unsigned int bytes, Lisp_Object table) { Lisp_Object new_vector; struct hashtable *ht = XHASHTABLE (table); assert (bytes > XVECTOR_LENGTH (ht->harray) * sizeof (Lisp_Object)); new_vector = make_vector ((bytes / sizeof (Lisp_Object)), Qnull_pointer); return (void *) XVECTOR_DATA (new_vector); } void elisp_hvector_free (void *ptr, Lisp_Object table) { struct hashtable *ht = XHASHTABLE (table); #if defined (USE_ASSERTIONS) || defined (DEBUG_XEMACS) Lisp_Object current_vector = ht->harray; #endif assert (((void *) XVECTOR_DATA (current_vector)) == ptr); ht->harray = Qnil; /* Let GC do its job */ } DEFUN ("hashtablep", Fhashtablep, 1, 1, 0, /* Return t if OBJ is a hashtable, else nil. */ (obj)) { return HASHTABLEP (obj) ? Qt : Qnil; } #if 0 /* I don't think these are needed any more. If using the general lisp_object_equal_*() functions causes efficiency problems, these can be resurrected. --ben */ /* equality and hash functions for Lisp strings */ int lisp_string_equal (CONST void *x1, CONST void *x2) { /* This is wrong anyway. You can't use strcmp() on Lisp strings, because they can contain zero characters. */ Lisp_Object str1, str2; CVOID_TO_LISP (str1, x1); CVOID_TO_LISP (str2, x2); return !strcmp ((char *) XSTRING_DATA (str1), (char *) XSTRING_DATA (str2)); } unsigned long lisp_string_hash (CONST void *x) { Lisp_Object str; CVOID_TO_LISP (str, x); return hash_string (XSTRING_DATA (str), XSTRING_LENGTH (str)); } #endif /* 0 */ static int lisp_object_eql_equal (CONST void *x1, CONST void *x2) { Lisp_Object obj1, obj2; CVOID_TO_LISP (obj1, x1); CVOID_TO_LISP (obj2, x2); return FLOATP (obj1) ? internal_equal (obj1, obj2, 0) : EQ (obj1, obj2); } static unsigned long lisp_object_eql_hash (CONST void *x) { Lisp_Object obj; CVOID_TO_LISP (obj, x); if (FLOATP (obj)) return internal_hash (obj, 0); else return LISP_HASH (obj); } static int lisp_object_equal_equal (CONST void *x1, CONST void *x2) { Lisp_Object obj1, obj2; CVOID_TO_LISP (obj1, x1); CVOID_TO_LISP (obj2, x2); return internal_equal (obj1, obj2, 0); } static unsigned long lisp_object_equal_hash (CONST void *x) { Lisp_Object obj; CVOID_TO_LISP (obj, x); return internal_hash (obj, 0); } Lisp_Object make_lisp_hashtable (int size, enum hashtable_type type, enum hashtable_test_fun test) { Lisp_Object result; struct hashtable *table = allocate_hashtable (); table->harray = make_vector ((compute_harray_size (size) * LISP_OBJECTS_PER_HENTRY), Qnull_pointer); switch (test) { case HASHTABLE_EQ: table->test_function = NULL; table->hash_function = NULL; break; case HASHTABLE_EQL: table->test_function = lisp_object_eql_equal; table->hash_function = lisp_object_eql_hash; break; case HASHTABLE_EQUAL: table->test_function = lisp_object_equal_equal; table->hash_function = lisp_object_equal_hash; break; default: abort (); } table->type = type; XSETHASHTABLE (result, table); if (table->type != HASHTABLE_NONWEAK) { table->next_weak = Vall_weak_hashtables; Vall_weak_hashtables = result; } else table->next_weak = Qunbound; return result; } static enum hashtable_test_fun decode_hashtable_test_fun (Lisp_Object sym) { if (NILP (sym)) return HASHTABLE_EQL; if (EQ (sym, Qeq)) return HASHTABLE_EQ; if (EQ (sym, Qequal)) return HASHTABLE_EQUAL; if (EQ (sym, Qeql)) return HASHTABLE_EQL; signal_simple_error ("Invalid hashtable test function", sym); return HASHTABLE_EQ; /* not reached */ } DEFUN ("make-hashtable", Fmake_hashtable, 1, 2, 0, /* Return a new hashtable object of initial size SIZE. Comparison between keys is done with TEST-FUN, which must be one of `eq', `eql', or `equal'. The default is `eql'; i.e. two keys must be the same object (or have the same floating-point value, for floats) to be considered equivalent. See also `make-weak-hashtable', `make-key-weak-hashtable', and `make-value-weak-hashtable'. */ (size, test_fun)) { CHECK_NATNUM (size); return make_lisp_hashtable (XINT (size), HASHTABLE_NONWEAK, decode_hashtable_test_fun (test_fun)); } DEFUN ("copy-hashtable", Fcopy_hashtable, 1, 1, 0, /* Return a new hashtable containing the same keys and values as HASHTABLE. The keys and values will not themselves be copied. */ (hashtable)) { struct _C_hashtable old_htbl; struct _C_hashtable new_htbl; struct hashtable *old_ht; struct hashtable *new_ht; Lisp_Object result; CHECK_HASHTABLE (hashtable); old_ht = XHASHTABLE (hashtable); ht_copy_to_c (old_ht, &old_htbl); /* we can't just call Fmake_hashtable() here because that will make a table that is slightly larger than the one we're trying to copy, which will make copy_hash() blow up. */ new_ht = allocate_hashtable (); new_ht->fullness = 0; new_ht->zero_entry = Qunbound; new_ht->hash_function = old_ht->hash_function; new_ht->test_function = old_ht->test_function; new_ht->harray = Fmake_vector (Flength (old_ht->harray), Qnull_pointer); ht_copy_to_c (new_ht, &new_htbl); copy_hash (&new_htbl, &old_htbl); ht_copy_from_c (&new_htbl, new_ht); new_ht->type = old_ht->type; XSETHASHTABLE (result, new_ht); if (UNBOUNDP (old_ht->next_weak)) new_ht->next_weak = Qunbound; else { new_ht->next_weak = Vall_weak_hashtables; Vall_weak_hashtables = result; } return result; } DEFUN ("gethash", Fgethash, 2, 3, 0, /* Find hash value for KEY in HASHTABLE. If there is no corresponding value, return DEFAULT (defaults to nil). */ (key, hashtable, default_)) { CONST void *vval; struct _C_hashtable htbl; if (!gc_in_progress) CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); if (gethash (LISP_TO_VOID (key), &htbl, &vval)) { Lisp_Object val; CVOID_TO_LISP (val, vval); return val; } else return default_; } DEFUN ("remhash", Fremhash, 2, 2, 0, /* Remove hash value for KEY in HASHTABLE. */ (key, hashtable)) { struct _C_hashtable htbl; CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); remhash (LISP_TO_VOID (key), &htbl); ht_copy_from_c (&htbl, XHASHTABLE (hashtable)); return Qnil; } DEFUN ("puthash", Fputhash, 3, 3, 0, /* Hash KEY to VAL in HASHTABLE. */ (key, val, hashtable)) { struct hashtable *ht; void *vkey = LISP_TO_VOID (key); CHECK_HASHTABLE (hashtable); ht = XHASHTABLE (hashtable); if (!vkey) ht->zero_entry = val; else { struct gcpro gcpro1, gcpro2, gcpro3; struct _C_hashtable htbl; ht_copy_to_c (XHASHTABLE (hashtable), &htbl); GCPRO3 (key, val, hashtable); puthash (vkey, LISP_TO_VOID (val), &htbl); ht_copy_from_c (&htbl, XHASHTABLE (hashtable)); UNGCPRO; } return val; } DEFUN ("clrhash", Fclrhash, 1, 1, 0, /* Remove all entries from HASHTABLE. */ (hashtable)) { struct _C_hashtable htbl; CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); clrhash (&htbl); ht_copy_from_c (&htbl, XHASHTABLE (hashtable)); return Qnil; } DEFUN ("hashtable-fullness", Fhashtable_fullness, 1, 1, 0, /* Return number of entries in HASHTABLE. */ (hashtable)) { struct _C_hashtable htbl; CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); return make_int (htbl.fullness); } DEFUN ("hashtable-type", Fhashtable_type, 1, 1, 0, /* Return type of HASHTABLE. This can be one of `non-weak', `weak', `key-weak' and `value-weak'. */ (hashtable)) { CHECK_HASHTABLE (hashtable); switch (XHASHTABLE (hashtable)->type) { case HASHTABLE_WEAK: return Qweak; case HASHTABLE_KEY_WEAK: return Qkey_weak; case HASHTABLE_VALUE_WEAK: return Qvalue_weak; default: return Qnon_weak; } } DEFUN ("hashtable-test-function", Fhashtable_test_function, 1, 1, 0, /* Return test function of HASHTABLE. This can be one of `eq', `eql' or `equal'. */ (hashtable)) { int (*fun) (CONST void *, CONST void *); CHECK_HASHTABLE (hashtable); fun = XHASHTABLE (hashtable)->test_function; if (fun == lisp_object_eql_equal) return Qeql; else if (fun == lisp_object_equal_equal) return Qequal; else return Qeq; } static void verify_function (Lisp_Object function, CONST char *description) { /* #### Unused DESCRIPTION? */ if (SYMBOLP (function)) { if (NILP (function)) return; else function = indirect_function (function, 1); } if (SUBRP (function) || COMPILED_FUNCTIONP (function)) return; else if (CONSP (function)) { Lisp_Object funcar = XCAR (function); if ((SYMBOLP (funcar)) && (EQ (funcar, Qlambda) || EQ (funcar, Qautoload))) return; } signal_error (Qinvalid_function, list1 (function)); } static int lisp_maphash_function (CONST void *void_key, void *void_val, void *void_fn) { /* This function can GC */ Lisp_Object key, val, fn; CVOID_TO_LISP (key, void_key); VOID_TO_LISP (val, void_val); VOID_TO_LISP (fn, void_fn); call2 (fn, key, val); return 0; } DEFUN ("maphash", Fmaphash, 2, 2, 0, /* Map FUNCTION over entries in HASHTABLE, calling it with two args, each key and value in the table. */ (function, hashtable)) { struct _C_hashtable htbl; struct gcpro gcpro1, gcpro2; verify_function (function, GETTEXT ("hashtable mapping function")); CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); GCPRO2 (hashtable, function); maphash (lisp_maphash_function, &htbl, LISP_TO_VOID (function)); UNGCPRO; return Qnil; } /* This function is for mapping a *C* function over the elements of a lisp hashtable. */ void elisp_maphash (int (*function) (CONST void *key, void *contents, void *extra_arg), Lisp_Object hashtable, void *closure) { struct _C_hashtable htbl; if (!gc_in_progress) CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); maphash (function, &htbl, closure); } void elisp_map_remhash (remhash_predicate function, Lisp_Object hashtable, void *closure) { struct _C_hashtable htbl; if (!gc_in_progress) CHECK_HASHTABLE (hashtable); ht_copy_to_c (XHASHTABLE (hashtable), &htbl); map_remhash (function, &htbl, closure); ht_copy_from_c (&htbl, XHASHTABLE (hashtable)); } #if 0 void elisp_table_op (Lisp_Object table, generic_hashtable_op op, void *arg1, void *arg2, void *arg3) { struct _C_hashtable htbl; CHECK_HASHTABLE (table); ht_copy_to_c (XHASHTABLE (table), &htbl); (*op) (&htbl, arg1, arg2, arg3); ht_copy_from_c (&htbl, XHASHTABLE (table)); } #endif /* 0 */ DEFUN ("make-weak-hashtable", Fmake_weak_hashtable, 1, 2, 0, /* Return a new fully weak hashtable object of initial size SIZE. A weak hashtable is one whose pointers do not count as GC referents: for any key-value pair in the hashtable, if the only remaining pointer to either the key or the value is in a weak hash table, then the pair will be removed from the table, and the key and value collected. A non-weak hash table (or any other pointer) would prevent the object from being collected. You can also create semi-weak hashtables; see `make-key-weak-hashtable' and `make-value-weak-hashtable'. */ (size, test_fun)) { CHECK_NATNUM (size); return make_lisp_hashtable (XINT (size), HASHTABLE_WEAK, decode_hashtable_test_fun (test_fun)); } DEFUN ("make-key-weak-hashtable", Fmake_key_weak_hashtable, 1, 2, 0, /* Return a new key-weak hashtable object of initial size SIZE. A key-weak hashtable is similar to a fully-weak hashtable (see `make-weak-hashtable') except that a key-value pair will be removed only if the key remains unmarked outside of weak hashtables. The pair will remain in the hashtable if the key is pointed to by something other than a weak hashtable, even if the value is not. */ (size, test_fun)) { CHECK_NATNUM (size); return make_lisp_hashtable (XINT (size), HASHTABLE_KEY_WEAK, decode_hashtable_test_fun (test_fun)); } DEFUN ("make-value-weak-hashtable", Fmake_value_weak_hashtable, 1, 2, 0, /* Return a new value-weak hashtable object of initial size SIZE. A value-weak hashtable is similar to a fully-weak hashtable (see `make-weak-hashtable') except that a key-value pair will be removed only if the value remains unmarked outside of weak hashtables. The pair will remain in the hashtable if the value is pointed to by something other than a weak hashtable, even if the key is not. */ (size, test_fun)) { CHECK_NATNUM (size); return make_lisp_hashtable (XINT (size), HASHTABLE_VALUE_WEAK, decode_hashtable_test_fun (test_fun)); } struct marking_closure { int (*obj_marked_p) (Lisp_Object); void (*markobj) (Lisp_Object); enum hashtable_type type; int did_mark; }; static int marking_mapper (CONST void *key, void *contents, void *closure) { Lisp_Object keytem, valuetem; struct marking_closure *fmh = (struct marking_closure *) closure; /* This function is called over each pair in the hashtable. We complete the marking for semi-weak hashtables. */ CVOID_TO_LISP (keytem, key); CVOID_TO_LISP (valuetem, contents); switch (fmh->type) { case HASHTABLE_KEY_WEAK: if ((fmh->obj_marked_p) (keytem) && !(fmh->obj_marked_p) (valuetem)) { (fmh->markobj) (valuetem); fmh->did_mark = 1; } break; case HASHTABLE_VALUE_WEAK: if ((fmh->obj_marked_p) (valuetem) && !(fmh->obj_marked_p) (keytem)) { (fmh->markobj) (keytem); fmh->did_mark = 1; } break; case HASHTABLE_KEY_CAR_WEAK: if (!CONSP (keytem) || (fmh->obj_marked_p) (XCAR (keytem))) { if (!(fmh->obj_marked_p) (keytem)) { (fmh->markobj) (keytem); fmh->did_mark = 1; } if (!(fmh->obj_marked_p) (valuetem)) { (fmh->markobj) (valuetem); fmh->did_mark = 1; } } break; case HASHTABLE_VALUE_CAR_WEAK: if (!CONSP (valuetem) || (fmh->obj_marked_p) (XCAR (valuetem))) { if (!(fmh->obj_marked_p) (keytem)) { (fmh->markobj) (keytem); fmh->did_mark = 1; } if (!(fmh->obj_marked_p) (valuetem)) { (fmh->markobj) (valuetem); fmh->did_mark = 1; } } break; default: abort (); /* Huh? */ } return 0; } int finish_marking_weak_hashtables (int (*obj_marked_p) (Lisp_Object), void (*markobj) (Lisp_Object)) { Lisp_Object rest; int did_mark = 0; for (rest = Vall_weak_hashtables; !GC_NILP (rest); rest = XHASHTABLE (rest)->next_weak) { enum hashtable_type type; if (! ((*obj_marked_p) (rest))) /* The hashtable is probably garbage. Ignore it. */ continue; type = XHASHTABLE (rest)->type; if (type == HASHTABLE_KEY_WEAK || type == HASHTABLE_VALUE_WEAK || type == HASHTABLE_KEY_CAR_WEAK || type == HASHTABLE_VALUE_CAR_WEAK) { struct marking_closure fmh; fmh.obj_marked_p = obj_marked_p; fmh.markobj = markobj; fmh.type = type; fmh.did_mark = 0; /* Now, scan over all the pairs. For all pairs that are half-marked, we may need to mark the other half if we're keeping this pair. */ elisp_maphash (marking_mapper, rest, &fmh); if (fmh.did_mark) did_mark = 1; } /* #### If alloc.c mark_object changes, this must change also... */ { /* Now mark the vector itself. (We don't need to call markobj here because we know that everything *in* it is already marked, we just need to prevent the vector itself from disappearing.) (The remhash above has taken care of zero_entry.) */ struct Lisp_Vector *ptr = XVECTOR (XHASHTABLE (rest)->harray); #ifdef LRECORD_VECTOR if (! MARKED_RECORD_P(XHASHTABLE(rest)->harray)) { MARK_RECORD_HEADER(&(ptr->header.lheader)); did_mark = 1; } #else int len = vector_length (ptr); if (len >= 0) { ptr->size = -1 - len; did_mark = 1; } #endif /* else it's already marked (remember, this function is iterated until marking stops) */ } } return did_mark; } struct pruning_closure { int (*obj_marked_p) (Lisp_Object); }; static int pruning_mapper (CONST void *key, CONST void *contents, void *closure) { Lisp_Object keytem, valuetem; struct pruning_closure *fmh = (struct pruning_closure *) closure; /* This function is called over each pair in the hashtable. We remove the pairs that aren't completely marked (everything that is going to stay ought to have been marked already by the finish_marking stage). */ CVOID_TO_LISP (keytem, key); CVOID_TO_LISP (valuetem, contents); return ! ((*fmh->obj_marked_p) (keytem) && (*fmh->obj_marked_p) (valuetem)); } void prune_weak_hashtables (int (*obj_marked_p) (Lisp_Object)) { Lisp_Object rest, prev = Qnil; for (rest = Vall_weak_hashtables; !GC_NILP (rest); rest = XHASHTABLE (rest)->next_weak) { if (! ((*obj_marked_p) (rest))) { /* This table itself is garbage. Remove it from the list. */ if (GC_NILP (prev)) Vall_weak_hashtables = XHASHTABLE (rest)->next_weak; else XHASHTABLE (prev)->next_weak = XHASHTABLE (rest)->next_weak; } else { struct pruning_closure fmh; fmh.obj_marked_p = obj_marked_p; /* Now, scan over all the pairs. Remove all of the pairs in which the key or value, or both, is unmarked (depending on the type of weak hashtable). */ elisp_map_remhash (pruning_mapper, rest, &fmh); prev = rest; } } } /* Return a hash value for an array of Lisp_Objects of size SIZE. */ unsigned long internal_array_hash (Lisp_Object *arr, int size, int depth) { int i; unsigned long hash = 0; if (size <= 5) { for (i = 0; i < size; i++) hash = HASH2 (hash, internal_hash (arr[i], depth + 1)); return hash; } /* just pick five elements scattered throughout the array. A slightly better approach would be to offset by some noise factor from the points chosen below. */ for (i = 0; i < 5; i++) hash = HASH2 (hash, internal_hash (arr[i*size/5], depth + 1)); return hash; } /* Return a hash value for a Lisp_Object. This is for use when hashing objects with the comparison being `equal' (for `eq', you can just use the Lisp_Object itself as the hash value). You need to make a tradeoff between the speed of the hash function and how good the hashing is. In particular, the hash function needs to be FAST, so you can't just traipse down the whole tree hashing everything together. Most of the time, objects will differ in the first few elements you hash. Thus, we only go to a short depth (5) and only hash at most 5 elements out of a vector. Theoretically we could still take 5^5 time (a big big number) to compute a hash, but practically this won't ever happen. */ unsigned long internal_hash (Lisp_Object obj, int depth) { if (depth > 5) return 0; if (CONSP (obj)) { /* no point in worrying about tail recursion, since we're not going very deep */ return HASH2 (internal_hash (XCAR (obj), depth + 1), internal_hash (XCDR (obj), depth + 1)); } else if (STRINGP (obj)) return hash_string (XSTRING_DATA (obj), XSTRING_LENGTH (obj)); else if (VECTORP (obj)) { struct Lisp_Vector *v = XVECTOR (obj); return HASH2 (vector_length (v), internal_array_hash (v->contents, vector_length (v), depth + 1)); } else if (LRECORDP (obj)) { CONST struct lrecord_implementation *imp = XRECORD_LHEADER_IMPLEMENTATION (obj); if (imp->hash) return (imp->hash) (obj, depth); } return LISP_HASH (obj); } #if 0 xxDEFUN ("internal-hash-value", Finternal_hash_value, 1, 1, 0, /* Hash value of OBJECT. For debugging. The value is returned as (HIGH . LOW). */ (object)) { /* This function is pretty 32bit-centric. */ unsigned long hash = internal_hash (object, 0); return Fcons (hash >> 16, hash & 0xffff); } #endif /************************************************************************/ /* initialization */ /************************************************************************/ void syms_of_elhash (void) { DEFSUBR (Fmake_hashtable); DEFSUBR (Fcopy_hashtable); DEFSUBR (Fhashtablep); DEFSUBR (Fgethash); DEFSUBR (Fputhash); DEFSUBR (Fremhash); DEFSUBR (Fclrhash); DEFSUBR (Fmaphash); DEFSUBR (Fhashtable_fullness); DEFSUBR (Fhashtable_type); DEFSUBR (Fhashtable_test_function); DEFSUBR (Fmake_weak_hashtable); DEFSUBR (Fmake_key_weak_hashtable); DEFSUBR (Fmake_value_weak_hashtable); #if 0 DEFSUBR (Finternal_hash_value); #endif defsymbol (&Qhashtablep, "hashtablep"); defsymbol (&Qhashtable, "hashtable"); defsymbol (&Qweak, "weak"); defsymbol (&Qkey_weak, "key-weak"); defsymbol (&Qvalue_weak, "value-weak"); defsymbol (&Qnon_weak, "non-weak"); } void vars_of_elhash (void) { /* This must NOT be staticpro'd */ Vall_weak_hashtables = Qnil; }