diff options
Diffstat (limited to 'hash.c')
| -rw-r--r-- | hash.c | 3081 |
1 files changed, 1620 insertions, 1461 deletions
@@ -35,6 +35,7 @@ #include "internal/hash.h" #include "internal/object.h" #include "internal/proc.h" +#include "internal/st.h" #include "internal/symbol.h" #include "internal/thread.h" #include "internal/time.h" @@ -44,19 +45,32 @@ #include "ruby/util.h" #include "ruby_assert.h" #include "symbol.h" -#include "transient_heap.h" #include "ruby/thread_native.h" #include "ruby/ractor.h" #include "vm_sync.h" +#include "builtin.h" + +/* Flags of RHash + * + * 1: RHASH_PASS_AS_KEYWORDS + * The hash is flagged as Ruby 2 keywords hash. + * 2: RHASH_PROC_DEFAULT + * The hash has a default proc (rather than a default value). + * 3: RHASH_ST_TABLE_FLAG + * The hash uses a ST table (rather than an AR table). + * 4-7: RHASH_AR_TABLE_SIZE_MASK + * The size of the AR table. + * 8-11: RHASH_AR_TABLE_BOUND_MASK + * The bounds of the AR table. + * 13-19: RHASH_LEV_MASK + * The iterational level of the hash. Used to prevent modifications + * to the hash during iteration. + */ #ifndef HASH_DEBUG #define HASH_DEBUG 0 #endif -#if HASH_DEBUG -#include "gc.h" -#endif - #define SET_DEFAULT(hash, ifnone) ( \ FL_UNSET_RAW(hash, RHASH_PROC_DEFAULT), \ RHASH_SET_IFNONE(hash, ifnone)) @@ -85,6 +99,7 @@ static VALUE rb_hash_s_try_convert(VALUE, VALUE); * 2. Insert WBs */ +/* :nodoc: */ VALUE rb_hash_freeze(VALUE hash) { @@ -92,6 +107,7 @@ rb_hash_freeze(VALUE hash) } VALUE rb_cHash; +VALUE rb_cHash_empty_frozen; static VALUE envtbl; static ID id_hash, id_flatten_bang; @@ -106,7 +122,7 @@ rb_hash_set_ifnone(VALUE hash, VALUE ifnone) return hash; } -static int +int rb_any_cmp(VALUE a, VALUE b) { if (a == b) return 0; @@ -164,7 +180,7 @@ any_hash(VALUE a, st_index_t (*other_func)(VALUE)) hnum = rb_hash_start(hnum); } else { - hnum = RSYMBOL(a)->hashval; + hnum = RSHIFT(RSYMBOL(a)->hashval, 1); } break; case T_FIXNUM: @@ -193,10 +209,26 @@ any_hash(VALUE a, st_index_t (*other_func)(VALUE)) return (long)hnum; } +VALUE rb_obj_hash(VALUE obj); +VALUE rb_vm_call0(rb_execution_context_t *ec, VALUE recv, ID id, int argc, const VALUE *argv, const rb_callable_method_entry_t *cme, int kw_splat); + static st_index_t obj_any_hash(VALUE obj) { - VALUE hval = rb_check_funcall_basic_kw(obj, id_hash, rb_mKernel, 0, 0, 0); + VALUE hval = Qundef; + VALUE klass = CLASS_OF(obj); + if (klass) { + const rb_callable_method_entry_t *cme = rb_callable_method_entry(klass, id_hash); + if (cme && METHOD_ENTRY_BASIC(cme)) { + // Optimize away the frame push overhead if it's the default Kernel#hash + if (cme->def->type == VM_METHOD_TYPE_CFUNC && cme->def->body.cfunc.func == (rb_cfunc_t)rb_obj_hash) { + hval = rb_obj_hash(obj); + } + else if (RBASIC_CLASS(cme->defined_class) == rb_mKernel) { + hval = rb_vm_call0(GET_EC(), obj, id_hash, 0, 0, cme, 0); + } + } + } if (UNDEF_P(hval)) { hval = rb_exec_recursive_outer_mid(hash_recursive, obj, 0, id_hash); @@ -221,7 +253,7 @@ obj_any_hash(VALUE obj) return FIX2LONG(hval); } -static st_index_t +st_index_t rb_any_hash(VALUE a) { return any_hash(a, obj_any_hash); @@ -289,40 +321,35 @@ objid_hash(VALUE obj) #endif } -/** +/* * call-seq: - * obj.hash -> integer - * - * Generates an Integer hash value for this object. This function must have the - * property that <code>a.eql?(b)</code> implies <code>a.hash == b.hash</code>. - * - * The hash value is used along with #eql? by the Hash class to determine if - * two objects reference the same hash key. Any hash value that exceeds the - * capacity of an Integer will be truncated before being used. + * hash -> integer * - * The hash value for an object may not be identical across invocations or - * implementations of Ruby. If you need a stable identifier across Ruby - * invocations and implementations you will need to generate one with a custom - * method. + * Returns the integer hash value for +self+; + * has the property that if <tt>foo.eql?(bar)</tt> + * then <tt>foo.hash == bar.hash</tt>. * - * Certain core classes such as Integer use built-in hash calculations and - * do not call the #hash method when used as a hash key. + * \Class Hash uses both #hash and #eql? to determine whether two objects + * used as hash keys are to be treated as the same key. + * A hash value that exceeds the capacity of an Integer is truncated before being used. * - * When implementing your own #hash based on multiple values, the best - * practice is to combine the class and any values using the hash code of an - * array: + * Many core classes override method Object#hash; + * other core classes (e.g., Integer) calculate the hash internally, + * and do not call the #hash method when used as a hash key. * - * For example: + * When implementing #hash for a user-defined class, + * best practice is to use Array#hash with the class name and the values + * that are important in the instance; + * this takes advantage of that method's logic for safely and efficiently + * generating a hash value: * * def hash * [self.class, a, b, c].hash * end * - * The reason for this is that the Array#hash method already has logic for - * safely and efficiently combining multiple hash values. - *-- - * \private - *++ + * The hash value may differ among invocations or implementations of Ruby. + * If you need stable hash-like identifiers across Ruby invocations and implementations, + * use a custom method to generate them. */ VALUE rb_obj_hash(VALUE obj) @@ -361,16 +388,17 @@ const struct st_hash_type rb_hashtype_ident = { rb_ident_hash, }; +#define RHASH_IDENTHASH_P(hash) (RHASH_TYPE(hash) == &identhash) +#define RHASH_STRING_KEY_P(hash, key) (!RHASH_IDENTHASH_P(hash) && (rb_obj_class(key) == rb_cString)) + typedef st_index_t st_hash_t; /* * RHASH_AR_TABLE_P(h): - * * as.ar == NULL or - * as.ar points ar_table. - * * as.ar is allocated by transient heap or xmalloc. + * RHASH_AR_TABLE points to ar_table. * * !RHASH_AR_TABLE_P(h): - * * as.st points st_table. + * RHASH_ST_TABLE points st_table. */ #define RHASH_AR_TABLE_MAX_BOUND RHASH_AR_TABLE_MAX_SIZE @@ -378,22 +406,6 @@ typedef st_index_t st_hash_t; #define RHASH_AR_TABLE_REF(hash, n) (&RHASH_AR_TABLE(hash)->pairs[n]) #define RHASH_AR_CLEARED_HINT 0xff -typedef struct ar_table_pair_struct { - VALUE key; - VALUE val; -} ar_table_pair; - -typedef struct ar_table_struct { - /* 64bit CPU: 8B * 2 * 8 = 128B */ - ar_table_pair pairs[RHASH_AR_TABLE_MAX_SIZE]; -} ar_table; - -size_t -rb_hash_ar_table_size(void) -{ - return sizeof(ar_table); -} - static inline st_hash_t ar_do_hash(st_data_t key) { @@ -409,13 +421,13 @@ ar_do_hash_hint(st_hash_t hash_value) static inline ar_hint_t ar_hint(VALUE hash, unsigned int index) { - return RHASH(hash)->ar_hint.ary[index]; + return RHASH_AR_TABLE(hash)->ar_hint.ary[index]; } static inline void ar_hint_set_hint(VALUE hash, unsigned int index, ar_hint_t hint) { - RHASH(hash)->ar_hint.ary[index] = hint; + RHASH_AR_TABLE(hash)->ar_hint.ary[index] = hint; } static inline void @@ -463,49 +475,23 @@ ar_set_entry(VALUE hash, unsigned int index, st_data_t key, st_data_t val, st_ha ((unsigned int)((RBASIC(h)->flags >> RHASH_AR_TABLE_BOUND_SHIFT) & \ (RHASH_AR_TABLE_BOUND_MASK >> RHASH_AR_TABLE_BOUND_SHIFT))) -#define RHASH_AR_TABLE_BOUND(h) (HASH_ASSERT(RHASH_AR_TABLE_P(h)), \ - RHASH_AR_TABLE_BOUND_RAW(h)) - #define RHASH_ST_TABLE_SET(h, s) rb_hash_st_table_set(h, s) #define RHASH_TYPE(hash) (RHASH_AR_TABLE_P(hash) ? &objhash : RHASH_ST_TABLE(hash)->type) #define HASH_ASSERT(expr) RUBY_ASSERT_MESG_WHEN(HASH_DEBUG, expr, #expr) -#if HASH_DEBUG -#define hash_verify(hash) hash_verify_(hash, __FILE__, __LINE__) - -void -rb_hash_dump(VALUE hash) +static inline unsigned int +RHASH_AR_TABLE_BOUND(VALUE h) { - rb_obj_info_dump(hash); - - if (RHASH_AR_TABLE_P(hash)) { - unsigned i, n = 0, bound = RHASH_AR_TABLE_BOUND(hash); - - fprintf(stderr, " size:%u bound:%u\n", - RHASH_AR_TABLE_SIZE(hash), RHASH_AR_TABLE_BOUND(hash)); - - for (i=0; i<bound; i++) { - st_data_t k, v; - - if (!ar_cleared_entry(hash, i)) { - char b1[0x100], b2[0x100]; - ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); - k = pair->key; - v = pair->val; - fprintf(stderr, " %d key:%s val:%s hint:%02x\n", i, - rb_raw_obj_info(b1, 0x100, k), - rb_raw_obj_info(b2, 0x100, v), - ar_hint(hash, i)); - n++; - } - else { - fprintf(stderr, " %d empty\n", i); - } - } - } + HASH_ASSERT(RHASH_AR_TABLE_P(h)); + const unsigned int bound = RHASH_AR_TABLE_BOUND_RAW(h); + HASH_ASSERT(bound <= RHASH_AR_TABLE_MAX_SIZE); + return bound; } +#if HASH_DEBUG +#define hash_verify(hash) hash_verify_(hash, __FILE__, __LINE__) + static VALUE hash_verify_(VALUE hash, const char *file, int line) { @@ -535,13 +521,6 @@ hash_verify_(VALUE hash, const char *file, int line) HASH_ASSERT(RHASH_AR_TABLE_BOUND_RAW(hash) == 0); } -#if USE_TRANSIENT_HEAP - if (RHASH_TRANSIENT_P(hash)) { - volatile st_data_t MAYBE_UNUSED(key) = RHASH_AR_TABLE_REF(hash, 0)->key; /* read */ - HASH_ASSERT(RHASH_AR_TABLE(hash) != NULL); - HASH_ASSERT(rb_transient_heap_managed_ptr_p(RHASH_AR_TABLE(hash))); - } -#endif return hash; } @@ -550,69 +529,30 @@ hash_verify_(VALUE hash, const char *file, int line) #endif static inline int -RHASH_TABLE_NULL_P(VALUE hash) -{ - if (RHASH(hash)->as.ar == NULL) { - HASH_ASSERT(RHASH_AR_TABLE_P(hash)); - return TRUE; - } - else { - return FALSE; - } -} - -static inline int RHASH_TABLE_EMPTY_P(VALUE hash) { return RHASH_SIZE(hash) == 0; } -int -rb_hash_ar_table_p(VALUE hash) -{ - if (FL_TEST_RAW((hash), RHASH_ST_TABLE_FLAG)) { - HASH_ASSERT(RHASH(hash)->as.st != NULL); - return FALSE; - } - else { - return TRUE; - } -} - -ar_table * -rb_hash_ar_table(VALUE hash) -{ - HASH_ASSERT(RHASH_AR_TABLE_P(hash)); - return RHASH(hash)->as.ar; -} +#define RHASH_SET_ST_FLAG(h) FL_SET_RAW(h, RHASH_ST_TABLE_FLAG) +#define RHASH_UNSET_ST_FLAG(h) FL_UNSET_RAW(h, RHASH_ST_TABLE_FLAG) -st_table * -rb_hash_st_table(VALUE hash) +static void +hash_st_table_init(VALUE hash, const struct st_hash_type *type, st_index_t size) { - HASH_ASSERT(!RHASH_AR_TABLE_P(hash)); - return RHASH(hash)->as.st; + st_init_existing_table_with_size(RHASH_ST_TABLE(hash), type, size); + RHASH_SET_ST_FLAG(hash); } void rb_hash_st_table_set(VALUE hash, st_table *st) { HASH_ASSERT(st != NULL); - FL_SET_RAW((hash), RHASH_ST_TABLE_FLAG); - RHASH(hash)->as.st = st; -} + RHASH_SET_ST_FLAG(hash); -static void -hash_ar_table_set(VALUE hash, ar_table *ar) -{ - HASH_ASSERT(RHASH_AR_TABLE_P(hash)); - HASH_ASSERT((RHASH_TRANSIENT_P(hash) && ar == NULL) ? FALSE : TRUE); - RHASH(hash)->as.ar = ar; - hash_verify(hash); + *RHASH_ST_TABLE(hash) = *st; } -#define RHASH_SET_ST_FLAG(h) FL_SET_RAW(h, RHASH_ST_TABLE_FLAG) -#define RHASH_UNSET_ST_FLAG(h) FL_UNSET_RAW(h, RHASH_ST_TABLE_FLAG) - static inline void RHASH_AR_TABLE_BOUND_SET(VALUE h, st_index_t n) { @@ -667,27 +607,7 @@ RHASH_AR_TABLE_CLEAR(VALUE h) RBASIC(h)->flags &= ~RHASH_AR_TABLE_SIZE_MASK; RBASIC(h)->flags &= ~RHASH_AR_TABLE_BOUND_MASK; - hash_ar_table_set(h, NULL); -} - -static ar_table* -ar_alloc_table(VALUE hash) -{ - ar_table *tab = (ar_table*)rb_transient_heap_alloc(hash, sizeof(ar_table)); - - if (tab != NULL) { - RHASH_SET_TRANSIENT_FLAG(hash); - } - else { - RHASH_UNSET_TRANSIENT_FLAG(hash); - tab = (ar_table*)ruby_xmalloc(sizeof(ar_table)); - } - - RHASH_AR_TABLE_SIZE_SET(hash, 0); - RHASH_AR_TABLE_BOUND_SET(hash, 0); - hash_ar_table_set(hash, tab); - - return tab; + memset(RHASH_AR_TABLE(h), 0, sizeof(ar_table)); } NOINLINE(static int ar_equal(VALUE x, VALUE y)); @@ -702,7 +622,7 @@ static unsigned ar_find_entry_hint(VALUE hash, ar_hint_t hint, st_data_t key) { unsigned i, bound = RHASH_AR_TABLE_BOUND(hash); - const ar_hint_t *hints = RHASH(hash)->ar_hint.ary; + const ar_hint_t *hints = RHASH_AR_TABLE(hash)->ar_hint.ary; /* if table is NULL, then bound also should be 0 */ @@ -749,92 +669,83 @@ ar_find_entry(VALUE hash, st_hash_t hash_value, st_data_t key) } static inline void -ar_free_and_clear_table(VALUE hash) +hash_ar_free_and_clear_table(VALUE hash) { - ar_table *tab = RHASH_AR_TABLE(hash); + RHASH_AR_TABLE_CLEAR(hash); - if (tab) { - if (RHASH_TRANSIENT_P(hash)) { - RHASH_UNSET_TRANSIENT_FLAG(hash); - } - else { - ruby_xfree(RHASH_AR_TABLE(hash)); - } - RHASH_AR_TABLE_CLEAR(hash); - } HASH_ASSERT(RHASH_AR_TABLE_SIZE(hash) == 0); HASH_ASSERT(RHASH_AR_TABLE_BOUND(hash) == 0); - HASH_ASSERT(RHASH_TRANSIENT_P(hash) == 0); } -static void -ar_try_convert_table(VALUE hash) -{ - if (!RHASH_AR_TABLE_P(hash)) return; - - const unsigned size = RHASH_AR_TABLE_SIZE(hash); +void rb_st_add_direct_with_hash(st_table *tab, st_data_t key, st_data_t value, st_hash_t hash); // st.c - st_table *new_tab; - st_index_t i; +enum ar_each_key_type { + ar_each_key_copy, + ar_each_key_cmp, + ar_each_key_insert, +}; - if (size < RHASH_AR_TABLE_MAX_SIZE) { - return; +static inline int +ar_each_key(ar_table *ar, int max, enum ar_each_key_type type, st_data_t *dst_keys, st_table *new_tab, st_hash_t *hashes) +{ + for (int i = 0; i < max; i++) { + ar_table_pair *pair = &ar->pairs[i]; + + switch (type) { + case ar_each_key_copy: + dst_keys[i] = pair->key; + break; + case ar_each_key_cmp: + if (dst_keys[i] != pair->key) return 1; + break; + case ar_each_key_insert: + if (UNDEF_P(pair->key)) continue; // deleted entry + rb_st_add_direct_with_hash(new_tab, pair->key, pair->val, hashes[i]); + break; + } } - new_tab = st_init_table_with_size(&objhash, size * 2); - - for (i = 0; i < RHASH_AR_TABLE_MAX_BOUND; i++) { - ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); - st_add_direct(new_tab, pair->key, pair->val); - } - ar_free_and_clear_table(hash); - RHASH_ST_TABLE_SET(hash, new_tab); - return; + return 0; } static st_table * ar_force_convert_table(VALUE hash, const char *file, int line) { - st_table *new_tab; - if (RHASH_ST_TABLE_P(hash)) { return RHASH_ST_TABLE(hash); } - - if (RHASH_AR_TABLE(hash)) { - unsigned i, bound = RHASH_AR_TABLE_BOUND(hash); - -#if defined(RHASH_CONVERT_TABLE_DEBUG) && RHASH_CONVERT_TABLE_DEBUG - rb_obj_info_dump(hash); - fprintf(stderr, "force_convert: %s:%d\n", file, line); - RB_DEBUG_COUNTER_INC(obj_hash_force_convert); -#endif - - new_tab = st_init_table_with_size(&objhash, RHASH_AR_TABLE_SIZE(hash)); - - for (i = 0; i < bound; i++) { - if (ar_cleared_entry(hash, i)) continue; - - ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); - st_add_direct(new_tab, pair->key, pair->val); - } - ar_free_and_clear_table(hash); - } else { - new_tab = st_init_table(&objhash); - } - RHASH_ST_TABLE_SET(hash, new_tab); - - return new_tab; -} + ar_table *ar = RHASH_AR_TABLE(hash); + st_hash_t hashes[RHASH_AR_TABLE_MAX_SIZE]; + unsigned int bound, size; + + // prepare hash values + do { + st_data_t keys[RHASH_AR_TABLE_MAX_SIZE]; + bound = RHASH_AR_TABLE_BOUND(hash); + size = RHASH_AR_TABLE_SIZE(hash); + ar_each_key(ar, bound, ar_each_key_copy, keys, NULL, NULL); + + for (unsigned int i = 0; i < bound; i++) { + // do_hash calls #hash method and it can modify hash object + hashes[i] = UNDEF_P(keys[i]) ? 0 : ar_do_hash(keys[i]); + } -static ar_table * -hash_ar_table(VALUE hash) -{ - if (RHASH_TABLE_NULL_P(hash)) { - ar_alloc_table(hash); + // check if modified + if (UNLIKELY(!RHASH_AR_TABLE_P(hash))) return RHASH_ST_TABLE(hash); + if (UNLIKELY(RHASH_AR_TABLE_BOUND(hash) != bound)) continue; + if (UNLIKELY(ar_each_key(ar, bound, ar_each_key_cmp, keys, NULL, NULL))) continue; + } while (0); + + // make st + st_table tab; + st_table *new_tab = &tab; + st_init_existing_table_with_size(new_tab, &objhash, size); + ar_each_key(ar, bound, ar_each_key_insert, NULL, new_tab, hashes); + hash_ar_free_and_clear_table(hash); + RHASH_ST_TABLE_SET(hash, new_tab); + return RHASH_ST_TABLE(hash); } - return RHASH_AR_TABLE(hash); } static int @@ -887,7 +798,6 @@ ar_add_direct_with_hash(VALUE hash, st_data_t key, st_data_t val, st_hash_t hash else { if (UNLIKELY(bin >= RHASH_AR_TABLE_MAX_BOUND)) { bin = ar_compact_table(hash); - hash_ar_table(hash); } HASH_ASSERT(bin < RHASH_AR_TABLE_MAX_BOUND); @@ -898,6 +808,14 @@ ar_add_direct_with_hash(VALUE hash, st_data_t key, st_data_t val, st_hash_t hash } } +static void +ensure_ar_table(VALUE hash) +{ + if (!RHASH_AR_TABLE_P(hash)) { + rb_raise(rb_eRuntimeError, "hash representation was changed during iteration"); + } +} + static int ar_general_foreach(VALUE hash, st_foreach_check_callback_func *func, st_update_callback_func *replace, st_data_t arg) { @@ -908,7 +826,10 @@ ar_general_foreach(VALUE hash, st_foreach_check_callback_func *func, st_update_c if (ar_cleared_entry(hash, i)) continue; ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); - enum st_retval retval = (*func)(pair->key, pair->val, arg, 0); + st_data_t key = (st_data_t)pair->key; + st_data_t val = (st_data_t)pair->val; + enum st_retval retval = (*func)(key, val, arg, 0); + ensure_ar_table(hash); /* pair may be not valid here because of theap */ switch (retval) { @@ -919,14 +840,13 @@ ar_general_foreach(VALUE hash, st_foreach_check_callback_func *func, st_update_c return 0; case ST_REPLACE: if (replace) { - VALUE key = pair->key; - VALUE val = pair->val; - retval = (*replace)(&key, &val, arg, TRUE); - - // TODO: pair should be same as pair before. - ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); - pair->key = key; - pair->val = val; + (*replace)(&key, &val, arg, TRUE); + + // Pair should not have moved + HASH_ASSERT(pair == RHASH_AR_TABLE_REF(hash, i)); + + pair->key = (VALUE)key; + pair->val = (VALUE)val; } break; case ST_DELETE: @@ -983,6 +903,7 @@ ar_foreach_check(VALUE hash, st_foreach_check_callback_func *func, st_data_t arg hint = ar_hint(hash, i); retval = (*func)(key, pair->val, arg, 0); + ensure_ar_table(hash); hash_verify(hash); switch (retval) { @@ -991,7 +912,7 @@ ar_foreach_check(VALUE hash, st_foreach_check_callback_func *func, st_data_t arg if (pair->key == never) break; ret = ar_find_entry_hint(hash, hint, key); if (ret == RHASH_AR_TABLE_MAX_BOUND) { - retval = (*func)(0, 0, arg, 1); + (*func)(0, 0, arg, 1); return 2; } } @@ -1032,7 +953,6 @@ ar_update(VALUE hash, st_data_t key, existing = (bin != RHASH_AR_TABLE_MAX_BOUND) ? TRUE : FALSE; } else { - hash_ar_table(hash); /* allocate ltbl if needed */ existing = FALSE; } @@ -1044,6 +964,7 @@ ar_update(VALUE hash, st_data_t key, old_key = key; retval = (*func)(&key, &value, arg, existing); /* pair can be invalid here because of theap */ + ensure_ar_table(hash); switch (retval) { case ST_CONTINUE: @@ -1081,8 +1002,6 @@ ar_insert(VALUE hash, st_data_t key, st_data_t value) return -1; } - hash_ar_table(hash); /* prepare ltbl */ - bin = ar_find_entry(hash, hash_value, key); if (bin == RHASH_AR_TABLE_MAX_BOUND) { if (RHASH_AR_TABLE_SIZE(hash) >= RHASH_AR_TABLE_MAX_SIZE) { @@ -1090,7 +1009,6 @@ ar_insert(VALUE hash, st_data_t key, st_data_t value) } else if (bin >= RHASH_AR_TABLE_MAX_BOUND) { bin = ar_compact_table(hash); - hash_ar_table(hash); } HASH_ASSERT(bin < RHASH_AR_TABLE_MAX_BOUND); @@ -1225,44 +1143,16 @@ static ar_table* ar_copy(VALUE hash1, VALUE hash2) { ar_table *old_tab = RHASH_AR_TABLE(hash2); + ar_table *new_tab = RHASH_AR_TABLE(hash1); - if (old_tab != NULL) { - ar_table *new_tab = RHASH_AR_TABLE(hash1); - if (new_tab == NULL) { - new_tab = (ar_table*) rb_transient_heap_alloc(hash1, sizeof(ar_table)); - if (new_tab != NULL) { - RHASH_SET_TRANSIENT_FLAG(hash1); - } - else { - RHASH_UNSET_TRANSIENT_FLAG(hash1); - new_tab = (ar_table*)ruby_xmalloc(sizeof(ar_table)); - } - } - *new_tab = *old_tab; - RHASH(hash1)->ar_hint.word = RHASH(hash2)->ar_hint.word; - RHASH_AR_TABLE_BOUND_SET(hash1, RHASH_AR_TABLE_BOUND(hash2)); - RHASH_AR_TABLE_SIZE_SET(hash1, RHASH_AR_TABLE_SIZE(hash2)); - hash_ar_table_set(hash1, new_tab); - - rb_gc_writebarrier_remember(hash1); - return new_tab; - } - else { - RHASH_AR_TABLE_BOUND_SET(hash1, RHASH_AR_TABLE_BOUND(hash2)); - RHASH_AR_TABLE_SIZE_SET(hash1, RHASH_AR_TABLE_SIZE(hash2)); - - if (RHASH_TRANSIENT_P(hash1)) { - RHASH_UNSET_TRANSIENT_FLAG(hash1); - } - else if (RHASH_AR_TABLE(hash1)) { - ruby_xfree(RHASH_AR_TABLE(hash1)); - } + *new_tab = *old_tab; + RHASH_AR_TABLE(hash1)->ar_hint.word = RHASH_AR_TABLE(hash2)->ar_hint.word; + RHASH_AR_TABLE_BOUND_SET(hash1, RHASH_AR_TABLE_BOUND(hash2)); + RHASH_AR_TABLE_SIZE_SET(hash1, RHASH_AR_TABLE_SIZE(hash2)); - hash_ar_table_set(hash1, NULL); + rb_gc_writebarrier_remember(hash1); - rb_gc_writebarrier_remember(hash1); - return old_tab; - } + return new_tab; } static void @@ -1278,32 +1168,32 @@ ar_clear(VALUE hash) } } -#if USE_TRANSIENT_HEAP -void -rb_hash_transient_heap_evacuate(VALUE hash, int promote) +static void +hash_st_free(VALUE hash) { - if (RHASH_TRANSIENT_P(hash)) { - ar_table *new_tab; - ar_table *old_tab = RHASH_AR_TABLE(hash); + HASH_ASSERT(RHASH_ST_TABLE_P(hash)); - if (UNLIKELY(old_tab == NULL)) { - return; - } - HASH_ASSERT(old_tab != NULL); - if (! promote) { - new_tab = rb_transient_heap_alloc(hash, sizeof(ar_table)); - if (new_tab == NULL) promote = true; - } - if (promote) { - new_tab = ruby_xmalloc(sizeof(ar_table)); - RHASH_UNSET_TRANSIENT_FLAG(hash); - } - *new_tab = *old_tab; - hash_ar_table_set(hash, new_tab); + st_table *tab = RHASH_ST_TABLE(hash); + + xfree(tab->bins); + xfree(tab->entries); +} + +static void +hash_st_free_and_clear_table(VALUE hash) +{ + hash_st_free(hash); + + RHASH_ST_CLEAR(hash); +} + +void +rb_hash_free(VALUE hash) +{ + if (RHASH_ST_TABLE_P(hash)) { + hash_st_free(hash); } - hash_verify(hash); } -#endif typedef int st_foreach_func(st_data_t, st_data_t, st_data_t); @@ -1371,7 +1261,6 @@ hash_ar_foreach_iter(st_data_t key, st_data_t value, st_data_t argp, int error) if (error) return ST_STOP; int status = (*arg->func)((VALUE)key, (VALUE)value, arg->arg); - /* TODO: rehash check? rb_raise(rb_eRuntimeError, "rehash occurred during iteration"); */ return hash_iter_status_check(status); } @@ -1383,89 +1272,84 @@ hash_foreach_iter(st_data_t key, st_data_t value, st_data_t argp, int error) if (error) return ST_STOP; - st_table *tbl = RHASH_ST_TABLE(arg->hash); int status = (*arg->func)((VALUE)key, (VALUE)value, arg->arg); - if (RHASH_ST_TABLE(arg->hash) != tbl) { - rb_raise(rb_eRuntimeError, "rehash occurred during iteration"); - } - return hash_iter_status_check(status); } -static int +static unsigned long iter_lev_in_ivar(VALUE hash) { VALUE levval = rb_ivar_get(hash, id_hash_iter_lev); HASH_ASSERT(FIXNUM_P(levval)); - return FIX2INT(levval); + long lev = FIX2LONG(levval); + HASH_ASSERT(lev >= 0); + return (unsigned long)lev; } void rb_ivar_set_internal(VALUE obj, ID id, VALUE val); static void -iter_lev_in_ivar_set(VALUE hash, int lev) +iter_lev_in_ivar_set(VALUE hash, unsigned long lev) { - rb_ivar_set_internal(hash, id_hash_iter_lev, INT2FIX(lev)); + HASH_ASSERT(lev >= RHASH_LEV_MAX); + HASH_ASSERT(POSFIXABLE(lev)); /* POSFIXABLE means fitting to long */ + rb_ivar_set_internal(hash, id_hash_iter_lev, LONG2FIX((long)lev)); } -static int +static inline unsigned long iter_lev_in_flags(VALUE hash) { - unsigned int u = (unsigned int)((RBASIC(hash)->flags >> RHASH_LEV_SHIFT) & RHASH_LEV_MAX); - return (int)u; + return (unsigned long)((RBASIC(hash)->flags >> RHASH_LEV_SHIFT) & RHASH_LEV_MAX); } -static int -RHASH_ITER_LEV(VALUE hash) +static inline void +iter_lev_in_flags_set(VALUE hash, unsigned long lev) { - int lev = iter_lev_in_flags(hash); + HASH_ASSERT(lev <= RHASH_LEV_MAX); + RBASIC(hash)->flags = ((RBASIC(hash)->flags & ~RHASH_LEV_MASK) | ((VALUE)lev << RHASH_LEV_SHIFT)); +} - if (lev == RHASH_LEV_MAX) { - return iter_lev_in_ivar(hash); - } - else { - return lev; - } +static inline bool +hash_iterating_p(VALUE hash) +{ + return iter_lev_in_flags(hash) > 0; } static void hash_iter_lev_inc(VALUE hash) { - int lev = iter_lev_in_flags(hash); + unsigned long lev = iter_lev_in_flags(hash); if (lev == RHASH_LEV_MAX) { - lev = iter_lev_in_ivar(hash); - iter_lev_in_ivar_set(hash, lev+1); + lev = iter_lev_in_ivar(hash) + 1; + if (!POSFIXABLE(lev)) { /* paranoiac check */ + rb_raise(rb_eRuntimeError, "too much nested iterations"); + } } else { lev += 1; - RBASIC(hash)->flags = ((RBASIC(hash)->flags & ~RHASH_LEV_MASK) | ((VALUE)lev << RHASH_LEV_SHIFT)); - if (lev == RHASH_LEV_MAX) { - iter_lev_in_ivar_set(hash, lev); - } + iter_lev_in_flags_set(hash, lev); + if (lev < RHASH_LEV_MAX) return; } + iter_lev_in_ivar_set(hash, lev); } static void hash_iter_lev_dec(VALUE hash) { - int lev = iter_lev_in_flags(hash); + unsigned long lev = iter_lev_in_flags(hash); if (lev == RHASH_LEV_MAX) { lev = iter_lev_in_ivar(hash); - HASH_ASSERT(lev > 0); - iter_lev_in_ivar_set(hash, lev-1); + if (lev > RHASH_LEV_MAX) { + iter_lev_in_ivar_set(hash, lev-1); + return; + } + rb_attr_delete(hash, id_hash_iter_lev); } - else { - HASH_ASSERT(lev > 0); - RBASIC(hash)->flags = ((RBASIC(hash)->flags & ~RHASH_LEV_MASK) | ((lev-1) << RHASH_LEV_SHIFT)); + else if (lev == 0) { + rb_raise(rb_eRuntimeError, "iteration level underflow"); } -} - -static VALUE -hash_foreach_ensure_rollback(VALUE hash) -{ - hash_iter_lev_inc(hash); - return 0; + iter_lev_in_flags_set(hash, lev - 1); } static VALUE @@ -1475,6 +1359,7 @@ hash_foreach_ensure(VALUE hash) return 0; } +/* This does not manage iteration level */ int rb_hash_stlike_foreach(VALUE hash, st_foreach_callback_func *func, st_data_t arg) { @@ -1486,6 +1371,7 @@ rb_hash_stlike_foreach(VALUE hash, st_foreach_callback_func *func, st_data_t arg } } +/* This does not manage iteration level */ int rb_hash_stlike_foreach_with_replace(VALUE hash, st_foreach_check_callback_func *func, st_update_callback_func *replace, st_data_t arg) { @@ -1536,11 +1422,23 @@ rb_hash_foreach(VALUE hash, rb_foreach_func *func, VALUE farg) hash_verify(hash); } +void rb_st_compact_table(st_table *tab); + +static void +compact_after_delete(VALUE hash) +{ + if (!hash_iterating_p(hash) && RHASH_ST_TABLE_P(hash)) { + rb_st_compact_table(RHASH_ST_TABLE(hash)); + } +} + static VALUE -hash_alloc_flags(VALUE klass, VALUE flags, VALUE ifnone) +hash_alloc_flags(VALUE klass, VALUE flags, VALUE ifnone, bool st) { const VALUE wb = (RGENGC_WB_PROTECTED_HASH ? FL_WB_PROTECTED : 0); - NEWOBJ_OF(hash, struct RHash, klass, T_HASH | wb | flags); + const size_t size = sizeof(struct RHash) + (st ? sizeof(st_table) : sizeof(ar_table)); + + NEWOBJ_OF(hash, struct RHash, klass, T_HASH | wb | flags, size, 0); RHASH_SET_IFNONE((VALUE)hash, ifnone); @@ -1550,7 +1448,8 @@ hash_alloc_flags(VALUE klass, VALUE flags, VALUE ifnone) static VALUE hash_alloc(VALUE klass) { - return hash_alloc_flags(klass, 0, Qnil); + /* Allocate to be able to fit both st_table and ar_table. */ + return hash_alloc_flags(klass, 0, Qnil, sizeof(st_table) > sizeof(ar_table)); } static VALUE @@ -1576,19 +1475,16 @@ copy_compare_by_id(VALUE hash, VALUE basis) return hash; } -MJIT_FUNC_EXPORTED VALUE +VALUE rb_hash_new_with_size(st_index_t size) { - VALUE ret = rb_hash_new(); - if (size == 0) { - /* do nothing */ - } - else if (size <= RHASH_AR_TABLE_MAX_SIZE) { - ar_alloc_table(ret); - } - else { - RHASH_ST_TABLE_SET(ret, st_init_table_with_size(&objhash, size)); + bool st = size > RHASH_AR_TABLE_MAX_SIZE; + VALUE ret = hash_alloc_flags(rb_cHash, 0, Qnil, st); + + if (st) { + hash_st_table_init(ret, &objhash, size); } + return ret; } @@ -1601,11 +1497,36 @@ rb_hash_new_capa(long capa) static VALUE hash_copy(VALUE ret, VALUE hash) { - if (!RHASH_EMPTY_P(hash)) { - if (RHASH_AR_TABLE_P(hash)) + if (rb_hash_compare_by_id_p(hash)) { + rb_gc_register_pinning_obj(ret); + } + + if (RHASH_AR_TABLE_P(hash)) { + if (RHASH_AR_TABLE_P(ret)) { ar_copy(ret, hash); - else if (RHASH_ST_TABLE_P(hash)) - RHASH_ST_TABLE_SET(ret, st_copy(RHASH_ST_TABLE(hash))); + } + else { + st_table *tab = RHASH_ST_TABLE(ret); + st_init_existing_table_with_size(tab, &objhash, RHASH_AR_TABLE_SIZE(hash)); + + int bound = RHASH_AR_TABLE_BOUND(hash); + for (int i = 0; i < bound; i++) { + if (ar_cleared_entry(hash, i)) continue; + + ar_table_pair *pair = RHASH_AR_TABLE_REF(hash, i); + st_add_direct(tab, pair->key, pair->val); + RB_OBJ_WRITTEN(ret, Qundef, pair->key); + RB_OBJ_WRITTEN(ret, Qundef, pair->val); + } + } + } + else { + HASH_ASSERT(sizeof(st_table) <= sizeof(ar_table)); + + RHASH_SET_ST_FLAG(ret); + st_replace(RHASH_ST_TABLE(ret), RHASH_ST_TABLE(hash)); + + rb_gc_writebarrier_remember(ret); } return ret; } @@ -1613,13 +1534,21 @@ hash_copy(VALUE ret, VALUE hash) static VALUE hash_dup_with_compare_by_id(VALUE hash) { - return hash_copy(copy_compare_by_id(rb_hash_new(), hash), hash); + VALUE dup = hash_alloc_flags(rb_cHash, 0, Qnil, RHASH_ST_TABLE_P(hash)); + if (RHASH_ST_TABLE_P(hash)) { + RHASH_SET_ST_FLAG(dup); + } + else { + RHASH_UNSET_ST_FLAG(dup); + } + + return hash_copy(dup, hash); } static VALUE hash_dup(VALUE hash, VALUE klass, VALUE flags) { - return hash_copy(hash_alloc_flags(klass, flags, RHASH_IFNONE(hash)), + return hash_copy(hash_alloc_flags(klass, flags, RHASH_IFNONE(hash), !RHASH_EMPTY_P(hash) && RHASH_ST_TABLE_P(hash)), hash); } @@ -1627,14 +1556,14 @@ VALUE rb_hash_dup(VALUE hash) { const VALUE flags = RBASIC(hash)->flags; - VALUE ret = hash_dup(hash, rb_obj_class(hash), - flags & (FL_EXIVAR|RHASH_PROC_DEFAULT)); - if (flags & FL_EXIVAR) - rb_copy_generic_ivar(ret, hash); + VALUE ret = hash_dup(hash, rb_obj_class(hash), flags & RHASH_PROC_DEFAULT); + + rb_copy_generic_ivar(ret, hash); + return ret; } -MJIT_FUNC_EXPORTED VALUE +VALUE rb_hash_resurrect(VALUE hash) { VALUE ret = hash_dup(hash, rb_cHash, 0); @@ -1647,7 +1576,7 @@ rb_hash_modify_check(VALUE hash) rb_check_frozen(hash); } -MJIT_FUNC_EXPORTED struct st_table * +struct st_table * rb_hash_tbl_raw(VALUE hash, const char *file, int line) { return ar_force_convert_table(hash, file, line); @@ -1708,7 +1637,7 @@ rb_hash_stlike_update(VALUE hash, st_data_t key, st_update_callback_func *func, if (RHASH_AR_TABLE_P(hash)) { int result = ar_update(hash, key, func, arg); if (result == -1) { - ar_try_convert_table(hash); + ar_force_convert_table(hash, __FILE__, __LINE__); } else { return result; @@ -1753,26 +1682,26 @@ tbl_update(VALUE hash, VALUE key, tbl_update_func func, st_data_t optional_arg) .func = func, .hash = hash, .key = key, - .value = (VALUE)optional_arg, + .value = 0 }; int ret = rb_hash_stlike_update(hash, key, tbl_update_modify, (st_data_t)&arg); /* write barrier */ RB_OBJ_WRITTEN(hash, Qundef, arg.key); - RB_OBJ_WRITTEN(hash, Qundef, arg.value); + if (arg.value) RB_OBJ_WRITTEN(hash, Qundef, arg.value); return ret; } -#define UPDATE_CALLBACK(iter_lev, func) ((iter_lev) > 0 ? func##_noinsert : func##_insert) +#define UPDATE_CALLBACK(iter_p, func) ((iter_p) ? func##_noinsert : func##_insert) -#define RHASH_UPDATE_ITER(h, iter_lev, key, func, a) do { \ - tbl_update((h), (key), UPDATE_CALLBACK((iter_lev), func), (st_data_t)(a)); \ +#define RHASH_UPDATE_ITER(h, iter_p, key, func, a) do { \ + tbl_update((h), (key), UPDATE_CALLBACK(iter_p, func), (st_data_t)(a)); \ } while (0) #define RHASH_UPDATE(hash, key, func, arg) \ - RHASH_UPDATE_ITER(hash, RHASH_ITER_LEV(hash), key, func, arg) + RHASH_UPDATE_ITER(hash, hash_iterating_p(hash), key, func, arg) static void set_proc_default(VALUE hash, VALUE proc) @@ -1790,90 +1719,71 @@ set_proc_default(VALUE hash, VALUE proc) RHASH_SET_IFNONE(hash, proc); } -/* - * call-seq: - * Hash.new(default_value = nil) -> new_hash - * Hash.new {|hash, key| ... } -> new_hash - * - * Returns a new empty \Hash object. - * - * The initial default value and initial default proc for the new hash - * depend on which form above was used. See {Default Values}[rdoc-ref:Hash@Default+Values]. - * - * If neither an argument nor a block given, - * initializes both the default value and the default proc to <tt>nil</tt>: - * h = Hash.new - * h.default # => nil - * h.default_proc # => nil - * - * If argument <tt>default_value</tt> given but no block given, - * initializes the default value to the given <tt>default_value</tt> - * and the default proc to <tt>nil</tt>: - * h = Hash.new(false) - * h.default # => false - * h.default_proc # => nil - * - * If a block given but no argument, stores the block as the default proc - * and sets the default value to <tt>nil</tt>: - * h = Hash.new {|hash, key| "Default value for #{key}" } - * h.default # => nil - * h.default_proc.class # => Proc - * h[:nosuch] # => "Default value for nosuch" - */ - static VALUE -rb_hash_initialize(int argc, VALUE *argv, VALUE hash) +rb_hash_init(rb_execution_context_t *ec, VALUE hash, VALUE capa_value, VALUE ifnone_unset, VALUE ifnone, VALUE block) { - VALUE ifnone; - rb_hash_modify(hash); - if (rb_block_given_p()) { - rb_check_arity(argc, 0, 0); - ifnone = rb_block_proc(); - SET_PROC_DEFAULT(hash, ifnone); + + if (capa_value != INT2FIX(0)) { + long capa = NUM2LONG(capa_value); + if (capa > 0 && RHASH_SIZE(hash) == 0 && RHASH_AR_TABLE_P(hash)) { + hash_st_table_init(hash, &objhash, capa); + } + } + + if (!NIL_P(block)) { + if (ifnone_unset != Qtrue) { + rb_check_arity(1, 0, 0); + } + else { + SET_PROC_DEFAULT(hash, block); + } } else { - rb_check_arity(argc, 0, 1); - ifnone = argc == 0 ? Qnil : argv[0]; - RHASH_SET_IFNONE(hash, ifnone); + RHASH_SET_IFNONE(hash, ifnone_unset == Qtrue ? Qnil : ifnone); } + hash_verify(hash); return hash; } +static VALUE rb_hash_to_a(VALUE hash); + /* * call-seq: * Hash[] -> new_empty_hash - * Hash[hash] -> new_hash + * Hash[other_hash] -> new_hash * Hash[ [*2_element_arrays] ] -> new_hash * Hash[*objects] -> new_hash * * Returns a new \Hash object populated with the given objects, if any. * See Hash::new. * - * With no argument, returns a new empty \Hash. + * With no argument given, returns a new empty hash. * - * When the single given argument is a \Hash, returns a new \Hash - * populated with the entries from the given \Hash, excluding the - * default value or proc. + * With a single argument +other_hash+ given that is a hash, + * returns a new hash initialized with the entries from that hash + * (but not with its +default+ or +default_proc+): * * h = {foo: 0, bar: 1, baz: 2} - * Hash[h] # => {:foo=>0, :bar=>1, :baz=>2} + * Hash[h] # => {foo: 0, bar: 1, baz: 2} * - * When the single given argument is an \Array of 2-element Arrays, - * returns a new \Hash object wherein each 2-element array forms a + * With a single argument +2_element_arrays+ given that is an array of 2-element arrays, + * returns a new hash wherein each given 2-element array forms a * key-value entry: * - * Hash[ [ [:foo, 0], [:bar, 1] ] ] # => {:foo=>0, :bar=>1} + * Hash[ [ [:foo, 0], [:bar, 1] ] ] # => {foo: 0, bar: 1} * - * When the argument count is an even number; - * returns a new \Hash object wherein each successive pair of arguments - * has become a key-value entry: + * With an even number of arguments +objects+ given, + * returns a new hash wherein each successive pair of arguments + * is a key-value entry: * - * Hash[:foo, 0, :bar, 1] # => {:foo=>0, :bar=>1} + * Hash[:foo, 0, :bar, 1] # => {foo: 0, bar: 1} * - * Raises an exception if the argument list does not conform to any + * Raises ArgumentError if the argument list does not conform to any * of the above. + * + * See also {Methods for Creating a Hash}[rdoc-ref:Hash@Methods+for+Creating+a+Hash]. */ static VALUE @@ -1884,12 +1794,23 @@ rb_hash_s_create(int argc, VALUE *argv, VALUE klass) if (argc == 1) { tmp = rb_hash_s_try_convert(Qnil, argv[0]); if (!NIL_P(tmp)) { - hash = hash_alloc(klass); - hash_copy(hash, tmp); - return hash; + if (!RHASH_EMPTY_P(tmp) && rb_hash_compare_by_id_p(tmp)) { + /* hash_copy for non-empty hash will copy compare_by_identity + flag, but we don't want it copied. Work around by + converting hash to flattened array and using that. */ + tmp = rb_hash_to_a(tmp); + } + else { + hash = hash_alloc(klass); + if (!RHASH_EMPTY_P(tmp)) + hash_copy(hash, tmp); + return hash; + } + } + else { + tmp = rb_check_array_type(argv[0]); } - tmp = rb_check_array_type(argv[0]); if (!NIL_P(tmp)) { long i; @@ -1927,7 +1848,7 @@ rb_hash_s_create(int argc, VALUE *argv, VALUE klass) return hash; } -MJIT_FUNC_EXPORTED VALUE +VALUE rb_to_hash_type(VALUE hash) { return rb_convert_type_with_id(hash, T_HASH, "Hash", idTo_hash); @@ -1942,16 +1863,15 @@ rb_check_hash_type(VALUE hash) /* * call-seq: - * Hash.try_convert(obj) -> obj, new_hash, or nil - * - * If +obj+ is a \Hash object, returns +obj+. + * Hash.try_convert(object) -> object, new_hash, or nil * - * Otherwise if +obj+ responds to <tt>:to_hash</tt>, - * calls <tt>obj.to_hash</tt> and returns the result. + * If +object+ is a hash, returns +object+. * - * Returns +nil+ if +obj+ does not respond to <tt>:to_hash</tt> + * Otherwise if +object+ responds to +:to_hash+, + * calls <tt>object.to_hash</tt>; + * returns the result if it is a hash, or raises TypeError if not. * - * Raises an exception unless <tt>obj.to_hash</tt> returns a \Hash object. + * Otherwise if +object+ does not respond to +:to_hash+, returns +nil+. */ static VALUE rb_hash_s_try_convert(VALUE dummy, VALUE hash) @@ -2000,9 +1920,12 @@ static VALUE rb_hash_s_ruby2_keywords_hash(VALUE dummy, VALUE hash) { Check_Type(hash, T_HASH); - hash = rb_hash_dup(hash); - RHASH(hash)->basic.flags |= RHASH_PASS_AS_KEYWORDS; - return hash; + VALUE tmp = rb_hash_dup(hash); + if (RHASH_EMPTY_P(hash) && rb_hash_compare_by_id_p(hash)) { + rb_hash_compare_by_id(tmp); + } + RHASH(tmp)->basic.flags |= RHASH_PASS_AS_KEYWORDS; + return tmp; } struct rehash_arg { @@ -2024,10 +1947,11 @@ rb_hash_rehash_i(VALUE key, VALUE value, VALUE arg) /* * call-seq: - * hash.rehash -> self + * rehash -> self * - * Rebuilds the hash table by recomputing the hash index for each key; + * Rebuilds the hash table for +self+ by recomputing the hash index for each key; * returns <tt>self</tt>. + * Calling this method ensures that the hash table is valid. * * The hash table becomes invalid if the hash value of a key * has changed after the entry was created. @@ -2040,25 +1964,27 @@ rb_hash_rehash(VALUE hash) VALUE tmp; st_table *tbl; - if (RHASH_ITER_LEV(hash) > 0) { + if (hash_iterating_p(hash)) { rb_raise(rb_eRuntimeError, "rehash during iteration"); } rb_hash_modify_check(hash); if (RHASH_AR_TABLE_P(hash)) { tmp = hash_alloc(0); - ar_alloc_table(tmp); rb_hash_foreach(hash, rb_hash_rehash_i, (VALUE)tmp); - ar_free_and_clear_table(hash); + + hash_ar_free_and_clear_table(hash); ar_copy(hash, tmp); - ar_free_and_clear_table(tmp); } else if (RHASH_ST_TABLE_P(hash)) { st_table *old_tab = RHASH_ST_TABLE(hash); tmp = hash_alloc(0); - tbl = st_init_table_with_size(old_tab->type, old_tab->num_entries); - RHASH_ST_TABLE_SET(tmp, tbl); + + hash_st_table_init(tmp, old_tab->type, old_tab->num_entries); + tbl = RHASH_ST_TABLE(tmp); + rb_hash_foreach(hash, rb_hash_rehash_i, (VALUE)tmp); - st_free_table(old_tab); + + hash_st_free(hash); RHASH_ST_TABLE_SET(hash, tbl); RHASH_ST_CLEAR(tmp); } @@ -2073,7 +1999,7 @@ call_default_proc(VALUE proc, VALUE hash, VALUE key) return rb_proc_call_with_block(proc, 2, args, Qnil); } -static bool +bool rb_hash_default_unredefined(VALUE hash) { VALUE klass = RBASIC_CLASS(hash); @@ -2118,7 +2044,7 @@ hash_stlike_lookup(VALUE hash, st_data_t key, st_data_t *pval) } } -MJIT_FUNC_EXPORTED int +int rb_hash_stlike_lookup(VALUE hash, st_data_t key, st_data_t *pval) { return hash_stlike_lookup(hash, key, pval); @@ -2126,16 +2052,19 @@ rb_hash_stlike_lookup(VALUE hash, st_data_t key, st_data_t *pval) /* * call-seq: - * hash[key] -> value + * self[key] -> object * - * Returns the value associated with the given +key+, if found: - * h = {foo: 0, bar: 1, baz: 2} - * h[:foo] # => 0 + * Searches for a hash key equivalent to the given +key+; + * see {Hash Key Equivalence}[rdoc-ref:Hash@Hash+Key+Equivalence]. * - * If +key+ is not found, returns a default value - * (see {Default Values}[rdoc-ref:Hash@Default+Values]): - * h = {foo: 0, bar: 1, baz: 2} - * h[:nosuch] # => nil + * If the key is found, returns its value: + * + * {foo: 0, bar: 1, baz: 2} + * h[:bar] # => 1 + * + * Otherwise, returns a default value (see {Hash Default}[rdoc-ref:Hash@Hash+Default]). + * + * Related: #[]=; see also {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ VALUE @@ -2172,25 +2101,28 @@ rb_hash_lookup(VALUE hash, VALUE key) /* * call-seq: - * hash.fetch(key) -> object - * hash.fetch(key, default_value) -> object - * hash.fetch(key) {|key| ... } -> object + * fetch(key) -> object + * fetch(key, default_value) -> object + * fetch(key) {|key| ... } -> object + * + * With no block given, returns the value for the given +key+, if found; * - * Returns the value for the given +key+, if found. * h = {foo: 0, bar: 1, baz: 2} - * h.fetch(:bar) # => 1 + * h.fetch(:bar) # => 1 * - * If +key+ is not found and no block was given, - * returns +default_value+: - * {}.fetch(:nosuch, :default) # => :default + * If the key is not found, returns +default_value+, if given, + * or raises KeyError otherwise: * - * If +key+ is not found and a block was given, - * yields +key+ to the block and returns the block's return value: - * {}.fetch(:nosuch) {|key| "No key #{key}"} # => "No key nosuch" + * h.fetch(:nosuch, :default) # => :default + * h.fetch(:nosuch) # Raises KeyError. * - * Raises KeyError if neither +default_value+ nor a block was given. + * With a block given, calls the block with +key+ and returns the block's return value: + * + * {}.fetch(:nosuch) {|key| "No key #{key}"} # => "No key nosuch" * * Note that this method does not use the values of either #default or #default_proc. + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -2237,12 +2169,12 @@ rb_hash_fetch(VALUE hash, VALUE key) /* * call-seq: - * hash.default -> object - * hash.default(key) -> object + * default -> object + * default(key) -> object * * Returns the default value for the given +key+. * The returned value will be determined either by the default proc or by the default value. - * See {Default Values}[rdoc-ref:Hash@Default+Values]. + * See {Hash Default}[rdoc-ref:Hash@Hash+Default]. * * With no argument, returns the current default value: * h = {} @@ -2271,7 +2203,7 @@ rb_hash_default(int argc, VALUE *argv, VALUE hash) /* * call-seq: - * hash.default = value -> object + * default = value -> object * * Sets the default value to +value+; returns +value+: * h = {} @@ -2279,10 +2211,10 @@ rb_hash_default(int argc, VALUE *argv, VALUE hash) * h.default = false # => false * h.default # => false * - * See {Default Values}[rdoc-ref:Hash@Default+Values]. + * See {Hash Default}[rdoc-ref:Hash@Hash+Default]. */ -static VALUE +VALUE rb_hash_set_default(VALUE hash, VALUE ifnone) { rb_hash_modify_check(hash); @@ -2292,10 +2224,10 @@ rb_hash_set_default(VALUE hash, VALUE ifnone) /* * call-seq: - * hash.default_proc -> proc or nil + * default_proc -> proc or nil * * Returns the default proc for +self+ - * (see {Default Values}[rdoc-ref:Hash@Default+Values]): + * (see {Hash Default}[rdoc-ref:Hash@Hash+Default]): * h = {} * h.default_proc # => nil * h.default_proc = proc {|hash, key| "Default value for #{key}" } @@ -2313,10 +2245,10 @@ rb_hash_default_proc(VALUE hash) /* * call-seq: - * hash.default_proc = proc -> proc + * default_proc = proc -> proc * - * Sets the default proc for +self+ to +proc+: - * (see {Default Values}[rdoc-ref:Hash@Default+Values]): + * Sets the default proc for +self+ to +proc+ + * (see {Hash Default}[rdoc-ref:Hash@Hash+Default]): * h = {} * h.default_proc # => nil * h.default_proc = proc { |hash, key| "Default value for #{key}" } @@ -2360,15 +2292,18 @@ key_i(VALUE key, VALUE value, VALUE arg) /* * call-seq: - * hash.key(value) -> key or nil + * key(value) -> key or nil * * Returns the key for the first-found entry with the given +value+ * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]): + * * h = {foo: 0, bar: 2, baz: 2} * h.key(0) # => :foo * h.key(2) # => :bar * - * Returns +nil+ if so such value is found. + * Returns +nil+ if no such value is found. + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -2433,29 +2368,36 @@ rb_hash_delete(VALUE hash, VALUE key) /* * call-seq: - * hash.delete(key) -> value or nil - * hash.delete(key) {|key| ... } -> object + * delete(key) -> value or nil + * delete(key) {|key| ... } -> object * - * Deletes the entry for the given +key+ and returns its associated value. + * If an entry for the given +key+ is found, + * deletes the entry and returns its associated value; + * otherwise returns +nil+ or calls the given block. + * + * With no block given and +key+ found, deletes the entry and returns its value: * - * If no block is given and +key+ is found, deletes the entry and returns the associated value: * h = {foo: 0, bar: 1, baz: 2} * h.delete(:bar) # => 1 - * h # => {:foo=>0, :baz=>2} + * h # => {foo: 0, baz: 2} + * + * With no block given and +key+ not found, returns +nil+. * - * If no block given and +key+ is not found, returns +nil+. + * With a block given and +key+ found, ignores the block, + * deletes the entry, and returns its value: * - * If a block is given and +key+ is found, ignores the block, - * deletes the entry, and returns the associated value: * h = {foo: 0, bar: 1, baz: 2} * h.delete(:baz) { |key| raise 'Will never happen'} # => 2 - * h # => {:foo=>0, :bar=>1} + * h # => {foo: 0, bar: 1} * - * If a block is given and +key+ is not found, + * With a block given and +key+ not found, * calls the block and returns the block's return value: + * * h = {foo: 0, bar: 1, baz: 2} * h.delete(:nosuch) { |key| "Key #{key} not found" } # => "Key nosuch not found" - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2467,6 +2409,7 @@ rb_hash_delete_m(VALUE hash, VALUE key) val = rb_hash_delete_entry(hash, key); if (!UNDEF_P(val)) { + compact_after_delete(hash); return val; } else { @@ -2496,16 +2439,18 @@ shift_i_safe(VALUE key, VALUE value, VALUE arg) /* * call-seq: - * hash.shift -> [key, value] or nil + * shift -> [key, value] or nil + * + * Removes and returns the first entry of +self+ as a 2-element array; + * see {Entry Order}[rdoc-ref:Hash@Entry+Order]: * - * Removes the first hash entry - * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]); - * returns a 2-element \Array containing the removed key and value: * h = {foo: 0, bar: 1, baz: 2} * h.shift # => [:foo, 0] - * h # => {:bar=>1, :baz=>2} + * h # => {bar: 1, baz: 2} * - * Returns nil if the hash is empty. + * Returns +nil+ if +self+ is empty. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2516,7 +2461,7 @@ rb_hash_shift(VALUE hash) rb_hash_modify_check(hash); if (RHASH_AR_TABLE_P(hash)) { var.key = Qundef; - if (RHASH_ITER_LEV(hash) == 0) { + if (!hash_iterating_p(hash)) { if (ar_shift(hash, &var.key, &var.val)) { return rb_assoc_new(var.key, var.val); } @@ -2531,7 +2476,7 @@ rb_hash_shift(VALUE hash) } if (RHASH_ST_TABLE_P(hash)) { var.key = Qundef; - if (RHASH_ITER_LEV(hash) == 0) { + if (!hash_iterating_p(hash)) { if (st_shift(RHASH_ST_TABLE(hash), &var.key, &var.val)) { return rb_assoc_new(var.key, var.val); } @@ -2565,19 +2510,19 @@ hash_enum_size(VALUE hash, VALUE args, VALUE eobj) /* * call-seq: - * hash.delete_if {|key, value| ... } -> self - * hash.delete_if -> new_enumerator + * delete_if {|key, value| ... } -> self + * delete_if -> new_enumerator * - * If a block given, calls the block with each key-value pair; - * deletes each entry for which the block returns a truthy value; - * returns +self+: - * h = {foo: 0, bar: 1, baz: 2} - * h.delete_if {|key, value| value > 0 } # => {:foo=>0} + * With a block given, calls the block with each key-value pair, + * deletes each entry for which the block returns a truthy value, + * and returns +self+: * - * If no block given, returns a new \Enumerator: * h = {foo: 0, bar: 1, baz: 2} - * e = h.delete_if # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:delete_if> - * e.each { |key, value| value > 0 } # => {:foo=>0} + * h.delete_if {|key, value| value > 0 } # => {foo: 0} + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ VALUE @@ -2587,26 +2532,28 @@ rb_hash_delete_if(VALUE hash) rb_hash_modify_check(hash); if (!RHASH_TABLE_EMPTY_P(hash)) { rb_hash_foreach(hash, delete_if_i, hash); + compact_after_delete(hash); } return hash; } /* * call-seq: - * hash.reject! {|key, value| ... } -> self or nil - * hash.reject! -> new_enumerator + * reject! {|key, value| ... } -> self or nil + * reject! -> new_enumerator * - * Returns +self+, whose remaining entries are those - * for which the block returns +false+ or +nil+: - * h = {foo: 0, bar: 1, baz: 2} - * h.reject! {|key, value| value < 2 } # => {:baz=>2} + * With a block given, calls the block with each entry's key and value; + * removes the entry from +self+ if the block returns a truthy value. * - * Returns +nil+ if no entries are removed. + * Return +self+ if any entries were removed, +nil+ otherwise: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.reject! # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:reject!> - * e.each {|key, value| key.start_with?('b') } # => {:foo=>0} + * h.reject! {|key, value| value < 2 } # => {baz: 2} + * h.reject! {|key, value| value < 2 } # => nil + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2625,20 +2572,21 @@ rb_hash_reject_bang(VALUE hash) /* * call-seq: - * hash.reject {|key, value| ... } -> new_hash - * hash.reject -> new_enumerator + * reject {|key, value| ... } -> new_hash + * reject -> new_enumerator * - * Returns a new \Hash object whose entries are all those - * from +self+ for which the block returns +false+ or +nil+: - * h = {foo: 0, bar: 1, baz: 2} - * h1 = h.reject {|key, value| key.start_with?('b') } - * h1 # => {:foo=>0} + * With a block given, returns a copy of +self+ with zero or more entries removed; + * calls the block with each key-value pair; + * excludes the entry in the copy if the block returns a truthy value, + * includes it otherwise: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.reject # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:reject> - * h1 = e.each {|key, value| key.start_with?('b') } - * h1 # => {:foo=>0} + * h.reject {|key, value| key.start_with?('b') } + * # => {foo: 0} + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2650,19 +2598,22 @@ rb_hash_reject(VALUE hash) result = hash_dup_with_compare_by_id(hash); if (!RHASH_EMPTY_P(hash)) { rb_hash_foreach(result, delete_if_i, result); + compact_after_delete(result); } return result; } /* * call-seq: - * hash.slice(*keys) -> new_hash + * slice(*keys) -> new_hash + * + * Returns a new hash containing the entries from +self+ for the given +keys+; + * ignores any keys that are not found: * - * Returns a new \Hash object containing the entries for the given +keys+: * h = {foo: 0, bar: 1, baz: 2} - * h.slice(:baz, :foo) # => {:baz=>2, :foo=>0} + * h.slice(:baz, :foo, :nosuch) # => {baz: 2, foo: 0} * - * Any given +keys+ that are not found are ignored. + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2688,13 +2639,16 @@ rb_hash_slice(int argc, VALUE *argv, VALUE hash) /* * call-seq: - * hsh.except(*keys) -> a_hash + * except(*keys) -> new_hash + * + * Returns a copy of +self+ that excludes entries for the given +keys+; + * any +keys+ that are not found are ignored: * - * Returns a new \Hash excluding entries for the given +keys+: - * h = { a: 100, b: 200, c: 300 } - * h.except(:a) #=> {:b=>200, :c=>300} + * h = {foo:0, bar: 1, baz: 2} # => {:foo=>0, :bar=>1, :baz=>2} + * h.except(:baz, :foo) # => {:bar=>1} + * h.except(:bar, :nosuch) # => {:foo=>0, :baz=>2} * - * Any given +keys+ that are not found are ignored. + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2709,21 +2663,26 @@ rb_hash_except(int argc, VALUE *argv, VALUE hash) key = argv[i]; rb_hash_delete(result, key); } + compact_after_delete(result); return result; } /* * call-seq: - * hash.values_at(*keys) -> new_array + * values_at(*keys) -> new_array + * + * Returns a new array containing values for the given +keys+: * - * Returns a new \Array containing values for the given +keys+: * h = {foo: 0, bar: 1, baz: 2} * h.values_at(:baz, :foo) # => [2, 0] * - * The {default values}[rdoc-ref:Hash@Default+Values] are returned - * for any keys that are not found: + * The {hash default}[rdoc-ref:Hash@Hash+Default] is returned + * for each key that is not found: + * * h.values_at(:hello, :foo) # => [nil, 0] + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -2740,22 +2699,26 @@ rb_hash_values_at(int argc, VALUE *argv, VALUE hash) /* * call-seq: - * hash.fetch_values(*keys) -> new_array - * hash.fetch_values(*keys) {|key| ... } -> new_array + * fetch_values(*keys) -> new_array + * fetch_values(*keys) {|key| ... } -> new_array + * + * When all given +keys+ are found, + * returns a new array containing the values associated with the given +keys+: * - * Returns a new \Array containing the values associated with the given keys *keys: * h = {foo: 0, bar: 1, baz: 2} * h.fetch_values(:baz, :foo) # => [2, 0] * - * Returns a new empty \Array if no arguments given. + * When any given +keys+ are not found and a block is given, + * calls the block with each unfound key and uses the block's return value + * as the value for that key: * - * When a block is given, calls the block with each missing key, - * treating the block's return value as the value for that key: - * h = {foo: 0, bar: 1, baz: 2} - * values = h.fetch_values(:bar, :foo, :bad, :bam) {|key| key.to_s} - * values # => [1, 0, "bad", "bam"] + * h.fetch_values(:bar, :foo, :bad, :bam) {|key| key.to_s} + * # => [1, 0, "bad", "bam"] + * + * When any given +keys+ are not found and no block is given, + * raises KeyError. * - * When no block is given, raises an exception if any given key is not found. + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -2782,19 +2745,18 @@ keep_if_i(VALUE key, VALUE value, VALUE hash) /* * call-seq: - * hash.select {|key, value| ... } -> new_hash - * hash.select -> new_enumerator + * select {|key, value| ... } -> new_hash + * select -> new_enumerator * - * Hash#filter is an alias for Hash#select. + * With a block given, calls the block with each entry's key and value; + * returns a new hash whose entries are those for which the block returns a truthy value: * - * Returns a new \Hash object whose entries are those for which the block returns a truthy value: * h = {foo: 0, bar: 1, baz: 2} - * h.select {|key, value| value < 2 } # => {:foo=>0, :bar=>1} + * h.select {|key, value| value < 2 } # => {foo: 0, bar: 1} * - * Returns a new \Enumerator if no block given: - * h = {foo: 0, bar: 1, baz: 2} - * e = h.select # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:select> - * e.each {|key, value| value < 2 } # => {:foo=>0, :bar=>1} + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2806,27 +2768,29 @@ rb_hash_select(VALUE hash) result = hash_dup_with_compare_by_id(hash); if (!RHASH_EMPTY_P(hash)) { rb_hash_foreach(result, keep_if_i, result); + compact_after_delete(result); } return result; } /* * call-seq: - * hash.select! {|key, value| ... } -> self or nil - * hash.select! -> new_enumerator + * select! {|key, value| ... } -> self or nil + * select! -> new_enumerator * - * Hash#filter! is an alias for Hash#select!. + * With a block given, calls the block with each entry's key and value; + * removes from +self+ each entry for which the block returns +false+ or +nil+. + * + * Returns +self+ if any entries were removed, +nil+ otherwise: * - * Returns +self+, whose entries are those for which the block returns a truthy value: * h = {foo: 0, bar: 1, baz: 2} - * h.select! {|key, value| value < 2 } => {:foo=>0, :bar=>1} + * h.select! {|key, value| value < 2 } # => {foo: 0, bar: 1} + * h.select! {|key, value| value < 2 } # => nil * - * Returns +nil+ if no entries were removed. * - * Returns a new \Enumerator if no block given: - * h = {foo: 0, bar: 1, baz: 2} - * e = h.select! # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:select!> - * e.each { |key, value| value < 2 } # => {:foo=>0, :bar=>1} + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2845,19 +2809,19 @@ rb_hash_select_bang(VALUE hash) /* * call-seq: - * hash.keep_if {|key, value| ... } -> self - * hash.keep_if -> new_enumerator + * keep_if {|key, value| ... } -> self + * keep_if -> new_enumerator * - * Calls the block for each key-value pair; + * With a block given, calls the block for each key-value pair; * retains the entry if the block returns a truthy value; - * otherwise deletes the entry; returns +self+. - * h = {foo: 0, bar: 1, baz: 2} - * h.keep_if { |key, value| key.start_with?('b') } # => {:bar=>1, :baz=>2} + * otherwise deletes the entry; returns +self+: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.keep_if # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:keep_if> - * e.each { |key, value| key.start_with?('b') } # => {:bar=>1, :baz=>2} + * h.keep_if { |key, value| key.start_with?('b') } # => {bar: 1, baz: 2} + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -2879,9 +2843,11 @@ clear_i(VALUE key, VALUE value, VALUE dummy) /* * call-seq: - * hash.clear -> self + * clear -> self * - * Removes all hash entries; returns +self+. + * Removes all entries from +self+; returns emptied +self+. + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ VALUE @@ -2889,7 +2855,7 @@ rb_hash_clear(VALUE hash) { rb_hash_modify_check(hash); - if (RHASH_ITER_LEV(hash) > 0) { + if (hash_iterating_p(hash)) { rb_hash_foreach(hash, clear_i, 0); } else if (RHASH_AR_TABLE_P(hash)) { @@ -2897,6 +2863,7 @@ rb_hash_clear(VALUE hash) } else { st_clear(RHASH_ST_TABLE(hash)); + compact_after_delete(hash); } return hash; @@ -2912,7 +2879,7 @@ hash_aset(st_data_t *key, st_data_t *val, struct update_arg *arg, int existing) VALUE rb_hash_key_str(VALUE key) { - if (!RB_FL_ANY_RAW(key, FL_EXIVAR) && RBASIC_CLASS(key) == rb_cString) { + if (!rb_obj_gen_fields_p(key) && RBASIC_CLASS(key) == rb_cString) { return rb_fstring(key); } else { @@ -2934,59 +2901,68 @@ NOINSERT_UPDATE_CALLBACK(hash_aset_str) /* * call-seq: - * hash[key] = value -> value - * hash.store(key, value) + * self[key] = object -> object * - * Hash#store is an alias for Hash#[]=. - - * Associates the given +value+ with the given +key+; returns +value+. + * Associates the given +object+ with the given +key+; returns +object+. + * + * Searches for a hash key equivalent to the given +key+; + * see {Hash Key Equivalence}[rdoc-ref:Hash@Hash+Key+Equivalence]. * - * If the given +key+ exists, replaces its value with the given +value+; + * If the key is found, replaces its value with the given +object+; * the ordering is not affected * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]): + * * h = {foo: 0, bar: 1} * h[:foo] = 2 # => 2 - * h.store(:bar, 3) # => 3 - * h # => {:foo=>2, :bar=>3} + * h[:foo] # => 2 * - * If +key+ does not exist, adds the +key+ and +value+; + * If +key+ is not found, creates a new entry for the given +key+ and +object+; * the new entry is last in the order * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]): + * * h = {foo: 0, bar: 1} * h[:baz] = 2 # => 2 - * h.store(:bat, 3) # => 3 - * h # => {:foo=>0, :bar=>1, :baz=>2, :bat=>3} + * h[:baz] # => 2 + * h # => {foo: 0, bar: 1, baz: 2} + * + * Related: #[]; see also {Methods for Assigning}[rdoc-ref:Hash@Methods+for+Assigning]. */ VALUE rb_hash_aset(VALUE hash, VALUE key, VALUE val) { - int iter_lev = RHASH_ITER_LEV(hash); + bool iter_p = hash_iterating_p(hash); rb_hash_modify(hash); - if (RHASH_TABLE_NULL_P(hash)) { - if (iter_lev > 0) no_new_key(); - ar_alloc_table(hash); - } - - if (RHASH_TYPE(hash) == &identhash || rb_obj_class(key) != rb_cString) { - RHASH_UPDATE_ITER(hash, iter_lev, key, hash_aset, val); + if (!RHASH_STRING_KEY_P(hash, key)) { + RHASH_UPDATE_ITER(hash, iter_p, key, hash_aset, val); } else { - RHASH_UPDATE_ITER(hash, iter_lev, key, hash_aset_str, val); + RHASH_UPDATE_ITER(hash, iter_p, key, hash_aset_str, val); } return val; } /* * call-seq: - * hash.replace(other_hash) -> self + * replace(other_hash) -> self * * Replaces the entire contents of +self+ with the contents of +other_hash+; * returns +self+: + * * h = {foo: 0, bar: 1, baz: 2} - * h.replace({bat: 3, bam: 4}) # => {:bat=>3, :bam=>4} + * h.replace({bat: 3, bam: 4}) # => {bat: 3, bam: 4} + * + * Also replaces the default value or proc of +self+ with the default value + * or proc of +other_hash+. + * + * h = {} + * other = Hash.new(:ok) + * h.replace(other) + * h.default # => :ok + * + * Related: see {Methods for Assigning}[rdoc-ref:Hash@Methods+for+Assigning]. */ static VALUE @@ -2994,7 +2970,7 @@ rb_hash_replace(VALUE hash, VALUE hash2) { rb_hash_modify_check(hash); if (hash == hash2) return hash; - if (RHASH_ITER_LEV(hash) > 0) { + if (hash_iterating_p(hash)) { rb_raise(rb_eRuntimeError, "can't replace hash during iteration"); } hash2 = to_hash(hash2); @@ -3002,32 +2978,26 @@ rb_hash_replace(VALUE hash, VALUE hash2) COPY_DEFAULT(hash, hash2); if (RHASH_AR_TABLE_P(hash)) { - ar_free_and_clear_table(hash); + hash_ar_free_and_clear_table(hash); } else { - st_free_table(RHASH_ST_TABLE(hash)); - RHASH_ST_CLEAR(hash); - } - hash_copy(hash, hash2); - if (RHASH_EMPTY_P(hash2) && RHASH_ST_TABLE_P(hash2)) { - /* ident hash */ - RHASH_ST_TABLE_SET(hash, st_init_table_with_size(RHASH_TYPE(hash2), 0)); + hash_st_free_and_clear_table(hash); } - rb_gc_writebarrier_remember(hash); + hash_copy(hash, hash2); return hash; } /* * call-seq: - * hash.length -> integer - * hash.size -> integer + * size -> integer * * Returns the count of entries in +self+: - * {foo: 0, bar: 1, baz: 2}.length # => 3 * - * Hash#length is an alias for Hash#size. + * {foo: 0, bar: 1, baz: 2}.size # => 3 + * + * Related: see {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ VALUE @@ -3044,14 +3014,17 @@ rb_hash_size_num(VALUE hash) /* * call-seq: - * hash.empty? -> true or false + * empty? -> true or false * * Returns +true+ if there are no hash entries, +false+ otherwise: + * * {}.empty? # => true - * {foo: 0, bar: 1, baz: 2}.empty? # => false + * {foo: 0}.empty? # => false + * + * Related: see {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ -static VALUE +VALUE rb_hash_empty_p(VALUE hash) { return RBOOL(RHASH_EMPTY_P(hash)); @@ -3066,26 +3039,22 @@ each_value_i(VALUE key, VALUE value, VALUE _) /* * call-seq: - * hash.each_value {|value| ... } -> self - * hash.each_value -> new_enumerator + * each_value {|value| ... } -> self + * each_value -> new_enumerator * - * Calls the given block with each value; returns +self+: - * h = {foo: 0, bar: 1, baz: 2} - * h.each_value {|value| puts value } # => {:foo=>0, :bar=>1, :baz=>2} - * Output: - * 0 - * 1 - * 2 + * With a block given, calls the block with each value; returns +self+: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.each_value # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:each_value> - * h1 = e.each {|value| puts value } - * h1 # => {:foo=>0, :bar=>1, :baz=>2} + * h.each_value {|value| puts value } # => {foo: 0, bar: 1, baz: 2} + * * Output: * 0 * 1 * 2 + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Iterating}[rdoc-ref:Hash@Methods+for+Iterating]. */ static VALUE @@ -3105,26 +3074,22 @@ each_key_i(VALUE key, VALUE value, VALUE _) /* * call-seq: - * hash.each_key {|key| ... } -> self - * hash.each_key -> new_enumerator + * each_key {|key| ... } -> self + * each_key -> new_enumerator * - * Calls the given block with each key; returns +self+: - * h = {foo: 0, bar: 1, baz: 2} - * h.each_key {|key| puts key } # => {:foo=>0, :bar=>1, :baz=>2} - * Output: - * foo - * bar - * baz + * With a block given, calls the block with each key; returns +self+: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.each_key # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:each_key> - * h1 = e.each {|key| puts key } - * h1 # => {:foo=>0, :bar=>1, :baz=>2} + * h.each_key {|key| puts key } # => {foo: 0, bar: 1, baz: 2} + * * Output: * foo * bar * baz + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Iterating}[rdoc-ref:Hash@Methods+for+Iterating]. */ static VALUE rb_hash_each_key(VALUE hash) @@ -3153,30 +3118,23 @@ each_pair_i_fast(VALUE key, VALUE value, VALUE _) /* * call-seq: - * hash.each {|key, value| ... } -> self - * hash.each_pair {|key, value| ... } -> self - * hash.each -> new_enumerator - * hash.each_pair -> new_enumerator + * each_pair {|key, value| ... } -> self + * each_pair -> new_enumerator * - * Hash#each is an alias for Hash#each_pair. - - * Calls the given block with each key-value pair; returns +self+: - * h = {foo: 0, bar: 1, baz: 2} - * h.each_pair {|key, value| puts "#{key}: #{value}"} # => {:foo=>0, :bar=>1, :baz=>2} - * Output: - * foo: 0 - * bar: 1 - * baz: 2 + * With a block given, calls the block with each key-value pair; returns +self+: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.each_pair # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:each_pair> - * h1 = e.each {|key, value| puts "#{key}: #{value}"} - * h1 # => {:foo=>0, :bar=>1, :baz=>2} + * h.each_pair {|key, value| puts "#{key}: #{value}"} # => {foo: 0, bar: 1, baz: 2} + * * Output: + * * foo: 0 * bar: 1 * baz: 2 + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Iterating}[rdoc-ref:Hash@Methods+for+Iterating]. */ static VALUE @@ -3222,40 +3180,91 @@ transform_keys_i(VALUE key, VALUE value, VALUE result) /* * call-seq: - * hash.transform_keys {|key| ... } -> new_hash - * hash.transform_keys(hash2) -> new_hash - * hash.transform_keys(hash2) {|other_key| ...} -> new_hash - * hash.transform_keys -> new_enumerator + * transform_keys {|old_key| ... } -> new_hash + * transform_keys(other_hash) -> new_hash + * transform_keys(other_hash) {|old_key| ...} -> new_hash + * transform_keys -> new_enumerator + * + * With an argument, a block, or both given, + * derives a new hash +new_hash+ from +self+, the argument, and/or the block; + * all, some, or none of its keys may be different from those in +self+. * - * Returns a new \Hash object; each entry has: - * * A key provided by the block. - * * The value from +self+. + * With a block given and no argument, + * +new_hash+ has keys determined only by the block. * - * An optional hash argument can be provided to map keys to new keys. - * Any key not given will be mapped using the provided block, - * or remain the same if no block is given. + * For each key/value pair <tt>old_key/value</tt> in +self+, calls the block with +old_key+; + * the block's return value becomes +new_key+; + * sets <tt>new_hash[new_key] = value</tt>; + * a duplicate key overwrites: * - * Transform keys: * h = {foo: 0, bar: 1, baz: 2} - * h1 = h.transform_keys {|key| key.to_s } - * h1 # => {"foo"=>0, "bar"=>1, "baz"=>2} + * h.transform_keys {|old_key| old_key.to_s } + * # => {"foo" => 0, "bar" => 1, "baz" => 2} + * h.transform_keys {|old_key| 'xxx' } + * # => {"xxx" => 2} * - * h.transform_keys(foo: :bar, bar: :foo) - * #=> {bar: 0, foo: 1, baz: 2} + * With argument +other_hash+ given and no block, + * +new_hash+ may have new keys provided by +other_hash+ + * and unchanged keys provided by +self+. * - * h.transform_keys(foo: :hello, &:to_s) - * #=> {:hello=>0, "bar"=>1, "baz"=>2} + * For each key/value pair <tt>old_key/old_value</tt> in +self+, + * looks for key +old_key+ in +other_hash+: * - * Overwrites values for duplicate keys: - * h = {foo: 0, bar: 1, baz: 2} - * h1 = h.transform_keys {|key| :bat } - * h1 # => {:bat=>2} + * - If +old_key+ is found, its value <tt>other_hash[old_key]</tt> is taken as +new_key+; + * sets <tt>new_hash[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys(baz: :BAZ, bar: :BAR, foo: :FOO) + * # => {FOO: 0, BAR: 1, BAZ: 2} + * h.transform_keys(baz: :FOO, bar: :FOO, foo: :FOO) + * # => {FOO: 2} + * + * - If +old_key+ is not found, + * sets <tt>new_hash[old_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys({}) + * # => {foo: 0, bar: 1, baz: 2} + * h.transform_keys(baz: :foo) + * # => {foo: 2, bar: 1} + * + * Unused keys in +other_hash+ are ignored: * - * Returns a new \Enumerator if no block given: * h = {foo: 0, bar: 1, baz: 2} - * e = h.transform_keys # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:transform_keys> - * h1 = e.each { |key| key.to_s } - * h1 # => {"foo"=>0, "bar"=>1, "baz"=>2} + * h.transform_keys(bat: 3) + * # => {foo: 0, bar: 1, baz: 2} + * + * With both argument +other_hash+ and a block given, + * +new_hash+ has new keys specified by +other_hash+ or by the block, + * and unchanged keys provided by +self+. + * + * For each pair +old_key+ and +value+ in +self+: + * + * - If +other_hash+ has key +old_key+ (with value +new_key+), + * does not call the block for that key; + * sets <tt>new_hash[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys(baz: :BAZ, bar: :BAR, foo: :FOO) {|key| fail 'Not called' } + * # => {FOO: 0, BAR: 1, BAZ: 2} + * + * - If +other_hash+ does not have key +old_key+, + * calls the block with +old_key+ and takes its return value as +new_key+; + * sets <tt>new_hash[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys(baz: :BAZ) {|key| key.to_s.reverse } + * # => {"oof" => 0, "rab" => 1, BAZ: 2} + * h.transform_keys(baz: :BAZ) {|key| 'ook' } + * # => {"ook" => 1, BAZ: 2} + * + * With no argument and no block given, returns a new Enumerator. + * + * Related: see {Methods for Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values]. */ static VALUE rb_hash_transform_keys(int argc, VALUE *argv, VALUE hash) @@ -3289,13 +3298,97 @@ static int flatten_i(VALUE key, VALUE val, VALUE ary); /* * call-seq: - * hash.transform_keys! {|key| ... } -> self - * hash.transform_keys!(hash2) -> self - * hash.transform_keys!(hash2) {|other_key| ...} -> self - * hash.transform_keys! -> new_enumerator + * transform_keys! {|old_key| ... } -> self + * transform_keys!(other_hash) -> self + * transform_keys!(other_hash) {|old_key| ...} -> self + * transform_keys! -> new_enumerator + * + * With an argument, a block, or both given, + * derives keys from the argument, the block, and +self+; + * all, some, or none of the keys in +self+ may be changed. + * + * With a block given and no argument, + * derives keys only from the block; + * all, some, or none of the keys in +self+ may be changed. + * + * For each key/value pair <tt>old_key/value</tt> in +self+, calls the block with +old_key+; + * the block's return value becomes +new_key+; + * removes the entry for +old_key+: <tt>self.delete(old_key)</tt>; + * sets <tt>self[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys! {|old_key| old_key.to_s } + * # => {"foo" => 0, "bar" => 1, "baz" => 2} + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys! {|old_key| 'xxx' } + * # => {"xxx" => 2} + * + * With argument +other_hash+ given and no block, + * derives keys for +self+ from +other_hash+ and +self+; + * all, some, or none of the keys in +self+ may be changed. + * + * For each key/value pair <tt>old_key/old_value</tt> in +self+, + * looks for key +old_key+ in +other_hash+: + * + * - If +old_key+ is found, takes value <tt>other_hash[old_key]</tt> as +new_key+; + * removes the entry for +old_key+: <tt>self.delete(old_key)</tt>; + * sets <tt>self[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :BAZ, bar: :BAR, foo: :FOO) + * # => {FOO: 0, BAR: 1, BAZ: 2} + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :FOO, bar: :FOO, foo: :FOO) + * # => {FOO: 2} + * + * - If +old_key+ is not found, does nothing: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!({}) + * # => {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :foo) + * # => {foo: 2, bar: 1} + * + * Unused keys in +other_hash+ are ignored: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(bat: 3) + * # => {foo: 0, bar: 1, baz: 2} + * + * With both argument +other_hash+ and a block given, + * derives keys from +other_hash+, the block, and +self+; + * all, some, or none of the keys in +self+ may be changed. * - * Same as Hash#transform_keys but modifies the receiver in place - * instead of returning a new hash. + * For each pair +old_key+ and +value+ in +self+: + * + * - If +other_hash+ has key +old_key+ (with value +new_key+), + * does not call the block for that key; + * removes the entry for +old_key+: <tt>self.delete(old_key)</tt>; + * sets <tt>self[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :BAZ, bar: :BAR, foo: :FOO) {|key| fail 'Not called' } + * # => {FOO: 0, BAR: 1, BAZ: 2} + * + * - If +other_hash+ does not have key +old_key+, + * calls the block with +old_key+ and takes its return value as +new_key+; + * removes the entry for +old_key+: <tt>self.delete(old_key)</tt>; + * sets <tt>self[new_key] = value</tt>; + * a duplicate key overwrites: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :BAZ) {|key| key.to_s.reverse } + * # => {"oof" => 0, "rab" => 1, BAZ: 2} + * h = {foo: 0, bar: 1, baz: 2} + * h.transform_keys!(baz: :BAZ) {|key| 'ook' } + * # => {"ook" => 1, BAZ: 2} + * + * With no argument and no block given, returns a new Enumerator. + * + * Related: see {Methods for Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values]. */ static VALUE rb_hash_transform_keys_bang(int argc, VALUE *argv, VALUE hash) @@ -3342,6 +3435,7 @@ rb_hash_transform_keys_bang(int argc, VALUE *argv, VALUE hash) rb_ary_clear(pairs); rb_hash_clear(new_keys); } + compact_after_delete(hash); return hash; } @@ -3361,25 +3455,37 @@ transform_values_foreach_replace(st_data_t *key, st_data_t *value, st_data_t arg return ST_CONTINUE; } +static VALUE +transform_values_call(VALUE hash) +{ + rb_hash_stlike_foreach_with_replace(hash, transform_values_foreach_func, transform_values_foreach_replace, hash); + return hash; +} + +static void +transform_values(VALUE hash) +{ + hash_iter_lev_inc(hash); + rb_ensure(transform_values_call, hash, hash_foreach_ensure, hash); +} + /* * call-seq: - * hash.transform_values {|value| ... } -> new_hash - * hash.transform_values -> new_enumerator + * transform_values {|value| ... } -> new_hash + * transform_values -> new_enumerator * - * Returns a new \Hash object; each entry has: - * * A key from +self+. - * * A value provided by the block. + * With a block given, returns a new hash +new_hash+; + * for each pair +key+/+value+ in +self+, + * calls the block with +value+ and captures its return as +new_value+; + * adds to +new_hash+ the entry +key+/+new_value+: * - * Transform values: * h = {foo: 0, bar: 1, baz: 2} * h1 = h.transform_values {|value| value * 100} - * h1 # => {:foo=>0, :bar=>100, :baz=>200} + * h1 # => {foo: 0, bar: 100, baz: 200} * - * Returns a new \Enumerator if no block given: - * h = {foo: 0, bar: 1, baz: 2} - * e = h.transform_values # => #<Enumerator: {:foo=>0, :bar=>1, :baz=>2}:transform_values> - * h1 = e.each { |value| value * 100} - * h1 # => {:foo=>0, :bar=>100, :baz=>200} + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values]. */ static VALUE rb_hash_transform_values(VALUE hash) @@ -3391,7 +3497,8 @@ rb_hash_transform_values(VALUE hash) SET_DEFAULT(result, Qnil); if (!RHASH_EMPTY_P(hash)) { - rb_hash_stlike_foreach_with_replace(result, transform_values_foreach_func, transform_values_foreach_replace, result); + transform_values(result); + compact_after_delete(result); } return result; @@ -3399,18 +3506,24 @@ rb_hash_transform_values(VALUE hash) /* * call-seq: - * hash.transform_values! {|value| ... } -> self - * hash.transform_values! -> new_enumerator + * transform_values! {|old_value| ... } -> self + * transform_values! -> new_enumerator * - * Returns +self+, whose keys are unchanged, and whose values are determined by the given block. - * h = {foo: 0, bar: 1, baz: 2} - * h.transform_values! {|value| value * 100} # => {:foo=>0, :bar=>100, :baz=>200} * - * Returns a new \Enumerator if no block given: + * With a block given, changes the values of +self+ as determined by the block; + * returns +self+. + * + * For each entry +key+/+old_value+ in +self+, + * calls the block with +old_value+, + * captures its return value as +new_value+, + * and sets <tt>self[key] = new_value</tt>: + * * h = {foo: 0, bar: 1, baz: 2} - * e = h.transform_values! # => #<Enumerator: {:foo=>0, :bar=>100, :baz=>200}:transform_values!> - * h1 = e.each {|value| value * 100} - * h1 # => {:foo=>0, :bar=>100, :baz=>200} + * h.transform_values! {|value| value * 100} # => {foo: 0, bar: 100, baz: 200} + * + * With no block given, returns a new Enumerator. + * + * Related: see {Methods for Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values]. */ static VALUE rb_hash_transform_values_bang(VALUE hash) @@ -3419,7 +3532,7 @@ rb_hash_transform_values_bang(VALUE hash) rb_hash_modify_check(hash); if (!RHASH_TABLE_EMPTY_P(hash)) { - rb_hash_stlike_foreach_with_replace(hash, transform_values_foreach_func, transform_values_foreach_replace, hash); + transform_values(hash); } return hash; @@ -3434,12 +3547,15 @@ to_a_i(VALUE key, VALUE value, VALUE ary) /* * call-seq: - * hash.to_a -> new_array + * to_a -> new_array + * + * Returns all elements of +self+ as an array of 2-element arrays; + * each nested array contains a key-value pair from +self+: * - * Returns a new \Array of 2-element \Array objects; - * each nested \Array contains a key-value pair from +self+: * h = {foo: 0, bar: 1, baz: 2} * h.to_a # => [[:foo, 0], [:bar, 1], [:baz, 2]] + * + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE @@ -3453,20 +3569,65 @@ rb_hash_to_a(VALUE hash) return ary; } +static bool +symbol_key_needs_quote(VALUE str) +{ + long len = RSTRING_LEN(str); + if (len == 0 || !rb_str_symname_p(str)) return true; + const char *s = RSTRING_PTR(str); + char first = s[0]; + if (first == '@' || first == '$' || first == '!') return true; + if (!at_char_boundary(s, s + len - 1, RSTRING_END(str), rb_enc_get(str))) return false; + switch (s[len - 1]) { + case '+': + case '-': + case '*': + case '/': + case '`': + case '%': + case '^': + case '&': + case '|': + case ']': + case '<': + case '=': + case '>': + case '~': + case '@': + return true; + default: + return false; + } +} + static int inspect_i(VALUE key, VALUE value, VALUE str) { VALUE str2; - str2 = rb_inspect(key); + bool is_symbol = SYMBOL_P(key); + bool quote = false; + if (is_symbol) { + str2 = rb_sym2str(key); + quote = symbol_key_needs_quote(str2); + } + else { + str2 = rb_inspect(key); + } if (RSTRING_LEN(str) > 1) { rb_str_buf_cat_ascii(str, ", "); } else { rb_enc_copy(str, str2); } - rb_str_buf_append(str, str2); - rb_str_buf_cat_ascii(str, "=>"); + if (quote) { + rb_str_buf_append(str, rb_str_inspect(str2)); + } + else { + rb_str_buf_append(str, str2); + } + + rb_str_buf_cat_ascii(str, is_symbol ? ": " : " => "); str2 = rb_inspect(value); rb_str_buf_append(str, str2); @@ -3488,13 +3649,14 @@ inspect_hash(VALUE hash, VALUE dummy, int recur) /* * call-seq: - * hash.inspect -> new_string + * inspect -> new_string + * + * Returns a new string containing the hash entries: * - * Returns a new \String containing the hash entries: * h = {foo: 0, bar: 1, baz: 2} - * h.inspect # => "{:foo=>0, :bar=>1, :baz=>2}" + * h.inspect # => "{foo: 0, bar: 1, baz: 2}" * - * Hash#to_s is an alias for Hash#inspect. + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE @@ -3507,9 +3669,11 @@ rb_hash_inspect(VALUE hash) /* * call-seq: - * hash.to_hash -> self + * to_hash -> self * * Returns +self+. + * + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE rb_hash_to_hash(VALUE hash) @@ -3552,21 +3716,22 @@ rb_hash_to_h_block(VALUE hash) /* * call-seq: - * hash.to_h -> self or new_hash - * hash.to_h {|key, value| ... } -> new_hash - * - * For an instance of \Hash, returns +self+. + * to_h {|key, value| ... } -> new_hash + * to_h -> self or new_hash * - * For a subclass of \Hash, returns a new \Hash - * containing the content of +self+. + * With a block given, returns a new hash whose content is based on the block; + * the block is called with each entry's key and value; + * the block should return a 2-element array + * containing the key and value to be included in the returned array: * - * When a block is given, returns a new \Hash object - * whose content is based on the block; - * the block should return a 2-element \Array object - * specifying the key-value pair to be included in the returned \Array: * h = {foo: 0, bar: 1, baz: 2} - * h1 = h.to_h {|key, value| [value, key] } - * h1 # => {0=>:foo, 1=>:bar, 2=>:baz} + * h.to_h {|key, value| [value, key] } + * # => {0 => :foo, 1 => :bar, 2 => :baz} + * + * With no block given, returns +self+ if +self+ is an instance of +Hash+; + * if +self+ is a subclass of +Hash+, returns a new hash containing the content of +self+. + * + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE @@ -3591,14 +3756,17 @@ keys_i(VALUE key, VALUE value, VALUE ary) /* * call-seq: - * hash.keys -> new_array + * keys -> new_array + * + * Returns a new array containing all keys in +self+: * - * Returns a new \Array containing all keys in +self+: * h = {foo: 0, bar: 1, baz: 2} * h.keys # => [:foo, :bar, :baz] + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ -MJIT_FUNC_EXPORTED VALUE +VALUE rb_hash_keys(VALUE hash) { st_index_t size = RHASH_SIZE(hash); @@ -3607,7 +3775,7 @@ rb_hash_keys(VALUE hash) if (size == 0) return keys; if (ST_DATA_COMPATIBLE_P(VALUE)) { - RARRAY_PTR_USE_TRANSIENT(keys, ptr, { + RARRAY_PTR_USE(keys, ptr, { if (RHASH_AR_TABLE_P(hash)) { size = ar_keys(hash, ptr, size); } @@ -3635,11 +3803,14 @@ values_i(VALUE key, VALUE value, VALUE ary) /* * call-seq: - * hash.values -> new_array + * values -> new_array + * + * Returns a new array containing all values in +self+: * - * Returns a new \Array containing all values in +self+: * h = {foo: 0, bar: 1, baz: 2} * h.values # => [0, 1, 2] + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ VALUE @@ -3654,20 +3825,19 @@ rb_hash_values(VALUE hash) if (ST_DATA_COMPATIBLE_P(VALUE)) { if (RHASH_AR_TABLE_P(hash)) { rb_gc_writebarrier_remember(values); - RARRAY_PTR_USE_TRANSIENT(values, ptr, { + RARRAY_PTR_USE(values, ptr, { size = ar_values(hash, ptr, size); }); } else if (RHASH_ST_TABLE_P(hash)) { st_table *table = RHASH_ST_TABLE(hash); rb_gc_writebarrier_remember(values); - RARRAY_PTR_USE_TRANSIENT(values, ptr, { + RARRAY_PTR_USE(values, ptr, { size = st_values(table, ptr, size); }); } rb_ary_set_len(values, size); } - else { rb_hash_foreach(hash, values_i, values); } @@ -3677,17 +3847,18 @@ rb_hash_values(VALUE hash) /* * call-seq: - * hash.include?(key) -> true or false - * hash.has_key?(key) -> true or false - * hash.key?(key) -> true or false - * hash.member?(key) -> true or false - - * Methods #has_key?, #key?, and #member? are aliases for \#include?. + * include?(key) -> true or false + * + * Returns whether +key+ is a key in +self+: * - * Returns +true+ if +key+ is a key in +self+, otherwise +false+. + * h = {foo: 0, bar: 1, baz: 2} + * h.include?(:bar) # => true + * h.include?(:BAR) # => false + * + * Related: {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ -MJIT_FUNC_EXPORTED VALUE +VALUE rb_hash_has_key(VALUE hash, VALUE key) { return RBOOL(hash_stlike_lookup(hash, key, NULL)); @@ -3707,12 +3878,11 @@ rb_hash_search_value(VALUE key, VALUE value, VALUE arg) /* * call-seq: - * hash.has_value?(value) -> true or false - * hash.value?(value) -> true or false + * has_value?(value) -> true or false * - * Method #value? is an alias for \#has_value?. + * Returns whether +value+ is a value in +self+. * - * Returns +true+ if +value+ is a value in +self+, otherwise +false+. + * Related: {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ static VALUE @@ -3809,21 +3979,25 @@ hash_equal(VALUE hash1, VALUE hash2, int eql) /* * call-seq: - * hash == object -> true or false + * self == other -> true or false * - * Returns +true+ if all of the following are true: - * * +object+ is a \Hash object. - * * +hash+ and +object+ have the same keys (regardless of order). - * * For each key +key+, <tt>hash[key] == object[key]</tt>. + * Returns whether all of the following are true: * - * Otherwise, returns +false+. + * - +other+ is a +Hash+ object (or can be converted to one). + * - +self+ and +other+ have the same keys (regardless of order). + * - For each key +key+, <tt>self[key] == other[key]</tt>. * - * Equal: - * h1 = {foo: 0, bar: 1, baz: 2} - * h2 = {foo: 0, bar: 1, baz: 2} - * h1 == h2 # => true - * h3 = {baz: 2, bar: 1, foo: 0} - * h1 == h3 # => true + * Examples: + * + * h = {foo: 0, bar: 1} + * h == {foo: 0, bar: 1} # => true # Equal entries (same order) + * h == {bar: 1, foo: 0} # => true # Equal entries (different order). + * h == 1 # => false # Object not a hash. + * h == {} # => false # Different number of entries. + * h == {foo: 0, bar: 1} # => false # Different key. + * h == {foo: 0, bar: 1} # => false # Different value. + * + * Related: see {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ static VALUE @@ -3834,21 +4008,23 @@ rb_hash_equal(VALUE hash1, VALUE hash2) /* * call-seq: - * hash.eql? object -> true or false + * eql?(object) -> true or false * * Returns +true+ if all of the following are true: - * * +object+ is a \Hash object. - * * +hash+ and +object+ have the same keys (regardless of order). - * * For each key +key+, <tt>h[key] eql? object[key]</tt>. + * + * - The given +object+ is a +Hash+ object. + * - +self+ and +object+ have the same keys (regardless of order). + * - For each key +key+, <tt>self[key].eql?(object[key])</tt>. * * Otherwise, returns +false+. * - * Equal: * h1 = {foo: 0, bar: 1, baz: 2} * h2 = {foo: 0, bar: 1, baz: 2} * h1.eql? h2 # => true * h3 = {baz: 2, bar: 1, foo: 0} * h1.eql? h3 # => true + * + * Related: see {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ static VALUE @@ -3871,16 +4047,19 @@ hash_i(VALUE key, VALUE val, VALUE arg) /* * call-seq: - * hash.hash -> an_integer + * hash -> an_integer * - * Returns the \Integer hash-code for the hash. + * Returns the integer hash-code for the hash. + * + * Two hashes have the same hash-code if their content is the same + * (regardless of order): * - * Two \Hash objects have the same hash-code if their content is the same - * (regardless or order): * h1 = {foo: 0, bar: 1, baz: 2} * h2 = {baz: 2, bar: 1, foo: 0} * h2.hash == h1.hash # => true * h2.eql? h1 # => true + * + * Related: see {Methods for Querying}[rdoc-ref:Hash@Methods+for+Querying]. */ static VALUE @@ -3905,17 +4084,21 @@ rb_hash_invert_i(VALUE key, VALUE value, VALUE hash) /* * call-seq: - * hash.invert -> new_hash + * invert -> new_hash + * + * Returns a new hash with each key-value pair inverted: * - * Returns a new \Hash object with the each key-value pair inverted: * h = {foo: 0, bar: 1, baz: 2} * h1 = h.invert * h1 # => {0=>:foo, 1=>:bar, 2=>:baz} * - * Overwrites any repeated new keys: + * Overwrites any repeated new keys * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]): + * * h = {foo: 0, bar: 0, baz: 0} * h.invert # => {0=>:baz} + * + * Related: see {Methods for Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values]. */ static VALUE @@ -3928,109 +4111,133 @@ rb_hash_invert(VALUE hash) } static int -rb_hash_update_callback(st_data_t *key, st_data_t *value, struct update_arg *arg, int existing) -{ - *value = arg->arg; - return ST_CONTINUE; -} - -NOINSERT_UPDATE_CALLBACK(rb_hash_update_callback) - -static int rb_hash_update_i(VALUE key, VALUE value, VALUE hash) { - RHASH_UPDATE(hash, key, rb_hash_update_callback, value); + rb_hash_aset(hash, key, value); return ST_CONTINUE; } +struct update_call_args { + VALUE hash, newvalue, *argv; + int argc; + bool block_given; + bool iterating; +}; + static int rb_hash_update_block_callback(st_data_t *key, st_data_t *value, struct update_arg *arg, int existing) { - st_data_t newvalue = arg->arg; + VALUE k = (VALUE)*key, v = (VALUE)*value; + struct update_call_args *ua = (void *)arg->arg; + VALUE newvalue = ua->newvalue, hash = arg->hash; if (existing) { - newvalue = (st_data_t)rb_yield_values(3, (VALUE)*key, (VALUE)*value, (VALUE)newvalue); + hash_iter_lev_inc(hash); + ua->iterating = true; + newvalue = rb_yield_values(3, k, v, newvalue); + hash_iter_lev_dec(hash); + ua->iterating = false; } - *value = newvalue; + else if (RHASH_STRING_KEY_P(hash, k) && !RB_OBJ_FROZEN(k)) { + *key = (st_data_t)rb_hash_key_str(k); + } + *value = (st_data_t)newvalue; return ST_CONTINUE; } NOINSERT_UPDATE_CALLBACK(rb_hash_update_block_callback) static int -rb_hash_update_block_i(VALUE key, VALUE value, VALUE hash) +rb_hash_update_block_i(VALUE key, VALUE value, VALUE args) { - RHASH_UPDATE(hash, key, rb_hash_update_block_callback, value); + struct update_call_args *ua = (void *)args; + ua->newvalue = value; + RHASH_UPDATE(ua->hash, key, rb_hash_update_block_callback, args); return ST_CONTINUE; } +static VALUE +rb_hash_update_call(VALUE args) +{ + struct update_call_args *arg = (void *)args; + + for (int i = 0; i < arg->argc; i++){ + VALUE hash = to_hash(arg->argv[i]); + if (arg->block_given) { + rb_hash_foreach(hash, rb_hash_update_block_i, args); + } + else { + rb_hash_foreach(hash, rb_hash_update_i, arg->hash); + } + } + return arg->hash; +} + +static VALUE +rb_hash_update_ensure(VALUE args) +{ + struct update_call_args *ua = (void *)args; + if (ua->iterating) hash_iter_lev_dec(ua->hash); + return Qnil; +} + /* * call-seq: - * hash.merge! -> self - * hash.merge!(*other_hashes) -> self - * hash.merge!(*other_hashes) { |key, old_value, new_value| ... } -> self + * update(*other_hashes) -> self + * update(*other_hashes) { |key, old_value, new_value| ... } -> self * - * Merges each of +other_hashes+ into +self+; returns +self+. + * Updates values and/or adds entries to +self+; returns +self+. * - * Each argument in +other_hashes+ must be a \Hash. + * Each argument +other_hash+ in +other_hashes+ must be a hash. * - * \Method #update is an alias for \#merge!. + * With no block given, for each successive entry +key+/+new_value+ in each successive +other_hash+: * - * With arguments and no block: - * * Returns +self+, after the given hashes are merged into it. - * * The given hashes are merged left to right. - * * Each new entry is added at the end. - * * Each duplicate-key entry's value overwrites the previous value. + * - If +key+ is in +self+, sets <tt>self[key] = new_value</tt>, whose position is unchanged: * - * Example: - * h = {foo: 0, bar: 1, baz: 2} - * h1 = {bat: 3, bar: 4} - * h2 = {bam: 5, bat:6} - * h.merge!(h1, h2) # => {:foo=>0, :bar=>4, :baz=>2, :bat=>6, :bam=>5} + * h0 = {foo: 0, bar: 1, baz: 2} + * h1 = {bar: 3, foo: -1} + * h0.update(h1) # => {foo: -1, bar: 3, baz: 2} * - * With arguments and a block: - * * Returns +self+, after the given hashes are merged. - * * The given hashes are merged left to right. - * * Each new-key entry is added at the end. - * * For each duplicate key: - * * Calls the block with the key and the old and new values. - * * The block's return value becomes the new value for the entry. + * - If +key+ is not in +self+, adds the entry at the end of +self+: * - * Example: - * h = {foo: 0, bar: 1, baz: 2} - * h1 = {bat: 3, bar: 4} - * h2 = {bam: 5, bat:6} - * h3 = h.merge!(h1, h2) { |key, old_value, new_value| old_value + new_value } - * h3 # => {:foo=>0, :bar=>5, :baz=>2, :bat=>9, :bam=>5} + * h = {foo: 0, bar: 1, baz: 2} + * h.update({bam: 3, bah: 4}) # => {foo: 0, bar: 1, baz: 2, bam: 3, bah: 4} * - * With no arguments: - * * Returns +self+, unmodified. - * * The block, if given, is ignored. + * With a block given, for each successive entry +key+/+new_value+ in each successive +other_hash+: * - * Example: - * h = {foo: 0, bar: 1, baz: 2} - * h.merge # => {:foo=>0, :bar=>1, :baz=>2} - * h1 = h.merge! { |key, old_value, new_value| raise 'Cannot happen' } - * h1 # => {:foo=>0, :bar=>1, :baz=>2} + * - If +key+ is in +self+, fetches +old_value+ from <tt>self[key]</tt>, + * calls the block with +key+, +old_value+, and +new_value+, + * and sets <tt>self[key] = new_value</tt>, whose position is unchanged : + * + * season = {AB: 75, H: 20, HR: 3, SO: 17, W: 11, HBP: 3} + * today = {AB: 3, H: 1, W: 1} + * yesterday = {AB: 4, H: 2, HR: 1} + * season.update(yesterday, today) {|key, old_value, new_value| old_value + new_value } + * # => {AB: 82, H: 23, HR: 4, SO: 17, W: 12, HBP: 3} + * + * - If +key+ is not in +self+, adds the entry at the end of +self+: + * + * h = {foo: 0, bar: 1, baz: 2} + * h.update({bat: 3}) { fail 'Cannot happen' } + * # => {foo: 0, bar: 1, baz: 2, bat: 3} + * + * Related: see {Methods for Assigning}[rdoc-ref:Hash@Methods+for+Assigning]. */ static VALUE rb_hash_update(int argc, VALUE *argv, VALUE self) { - int i; - bool block_given = rb_block_given_p(); + struct update_call_args args = { + .hash = self, + .argv = argv, + .argc = argc, + .block_given = rb_block_given_p(), + .iterating = false, + }; + VALUE arg = (VALUE)&args; rb_hash_modify(self); - for (i = 0; i < argc; i++){ - VALUE hash = to_hash(argv[i]); - if (block_given) { - rb_hash_foreach(hash, rb_hash_update_block_i, self); - } - else { - rb_hash_foreach(hash, rb_hash_update_i, self); - } - } - return self; + return rb_ensure(rb_hash_update_call, arg, rb_hash_update_ensure, arg); } struct update_func_arg { @@ -4084,53 +4291,48 @@ rb_hash_update_by(VALUE hash1, VALUE hash2, rb_hash_update_func *func) /* * call-seq: - * hash.merge -> copy_of_self - * hash.merge(*other_hashes) -> new_hash - * hash.merge(*other_hashes) { |key, old_value, new_value| ... } -> new_hash - * - * Returns the new \Hash formed by merging each of +other_hashes+ - * into a copy of +self+. + * merge(*other_hashes) -> new_hash + * merge(*other_hashes) { |key, old_value, new_value| ... } -> new_hash * - * Each argument in +other_hashes+ must be a \Hash. + * Each argument +other_hash+ in +other_hashes+ must be a hash. * - * --- + * With arguments +other_hashes+ given and no block, + * returns the new hash formed by merging each successive +other_hash+ + * into a copy of +self+; + * returns that copy; + * for each successive entry in +other_hash+: * - * With arguments and no block: - * * Returns the new \Hash object formed by merging each successive - * \Hash in +other_hashes+ into +self+. - * * Each new-key entry is added at the end. - * * Each duplicate-key entry's value overwrites the previous value. + * - For a new key, the entry is added at the end of +self+. + * - For duplicate key, the entry overwrites the entry in +self+, + * whose position is unchanged. * * Example: + * * h = {foo: 0, bar: 1, baz: 2} * h1 = {bat: 3, bar: 4} * h2 = {bam: 5, bat:6} - * h.merge(h1, h2) # => {:foo=>0, :bar=>4, :baz=>2, :bat=>6, :bam=>5} + * h.merge(h1, h2) # => {foo: 0, bar: 4, baz: 2, bat: 6, bam: 5} + * + * With arguments +other_hashes+ and a block given, behaves as above + * except that for a duplicate key + * the overwriting entry takes it value not from the entry in +other_hash+, + * but instead from the block: * - * With arguments and a block: - * * Returns a new \Hash object that is the merge of +self+ and each given hash. - * * The given hashes are merged left to right. - * * Each new-key entry is added at the end. - * * For each duplicate key: - * * Calls the block with the key and the old and new values. - * * The block's return value becomes the new value for the entry. + * - The block is called with the duplicate key and the values + * from both +self+ and +other_hash+. + * - The block's return value becomes the new value for the entry in +self+. * * Example: + * * h = {foo: 0, bar: 1, baz: 2} * h1 = {bat: 3, bar: 4} * h2 = {bam: 5, bat:6} - * h3 = h.merge(h1, h2) { |key, old_value, new_value| old_value + new_value } - * h3 # => {:foo=>0, :bar=>5, :baz=>2, :bat=>9, :bam=>5} + * h.merge(h1, h2) { |key, old_value, new_value| old_value + new_value } + * # => {foo: 0, bar: 5, baz: 2, bat: 9, bam: 5} * - * With no arguments: - * * Returns a copy of +self+. - * * The block, if given, is ignored. + * With no arguments, returns a copy of +self+; the block, if given, is ignored. * - * Example: - * h = {foo: 0, bar: 1, baz: 2} - * h.merge # => {:foo=>0, :bar=>1, :baz=>2} - * h1 = h.merge { |key, old_value, new_value| raise 'Cannot happen' } - * h1 # => {:foo=>0, :bar=>1, :baz=>2} + * Related: see {Methods for Assigning}[rdoc-ref:Hash@Methods+for+Assigning]. */ static VALUE @@ -4145,24 +4347,17 @@ assoc_cmp(VALUE a, VALUE b) return !RTEST(rb_equal(a, b)); } -static VALUE -lookup2_call(VALUE arg) -{ - VALUE *args = (VALUE *)arg; - return rb_hash_lookup2(args[0], args[1], Qundef); -} - -struct reset_hash_type_arg { - VALUE hash; - const struct st_hash_type *orighash; +struct assoc_arg { + st_table *tbl; + st_data_t key; }; static VALUE -reset_hash_type(VALUE arg) +assoc_lookup(VALUE arg) { - struct reset_hash_type_arg *p = (struct reset_hash_type_arg *)arg; - HASH_ASSERT(RHASH_ST_TABLE_P(p->hash)); - RHASH_ST_TABLE(p->hash)->type = p->orighash; + struct assoc_arg *p = (struct assoc_arg*)arg; + st_data_t data; + if (st_lookup(p->tbl, p->key, &data)) return (VALUE)data; return Qundef; } @@ -4180,42 +4375,46 @@ assoc_i(VALUE key, VALUE val, VALUE arg) /* * call-seq: - * hash.assoc(key) -> new_array or nil + * assoc(key) -> entry or nil + * + * If the given +key+ is found, returns its entry as a 2-element array + * containing that key and its value: * - * If the given +key+ is found, returns a 2-element \Array containing that key and its value: * h = {foo: 0, bar: 1, baz: 2} * h.assoc(:bar) # => [:bar, 1] * - * Returns +nil+ if key +key+ is not found. + * Returns +nil+ if the key is not found. + * + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE rb_hash_assoc(VALUE hash, VALUE key) { - st_table *table; - const struct st_hash_type *orighash; VALUE args[2]; if (RHASH_EMPTY_P(hash)) return Qnil; - ar_force_convert_table(hash, __FILE__, __LINE__); - HASH_ASSERT(RHASH_ST_TABLE_P(hash)); - table = RHASH_ST_TABLE(hash); - orighash = table->type; - - if (orighash != &identhash) { - VALUE value; - struct reset_hash_type_arg ensure_arg; - struct st_hash_type assochash; - - assochash.compare = assoc_cmp; - assochash.hash = orighash->hash; - table->type = &assochash; - args[0] = hash; - args[1] = key; - ensure_arg.hash = hash; - ensure_arg.orighash = orighash; - value = rb_ensure(lookup2_call, (VALUE)&args, reset_hash_type, (VALUE)&ensure_arg); + if (RHASH_ST_TABLE_P(hash) && !RHASH_IDENTHASH_P(hash)) { + VALUE value = Qundef; + st_table assoctable = *RHASH_ST_TABLE(hash); + assoctable.type = &(struct st_hash_type){ + .compare = assoc_cmp, + .hash = assoctable.type->hash, + }; + VALUE arg = (VALUE)&(struct assoc_arg){ + .tbl = &assoctable, + .key = (st_data_t)key, + }; + + if (RB_OBJ_FROZEN(hash)) { + value = assoc_lookup(arg); + } + else { + hash_iter_lev_inc(hash); + value = rb_ensure(assoc_lookup, arg, hash_foreach_ensure, hash); + } + hash_verify(hash); if (!UNDEF_P(value)) return rb_assoc_new(key, value); } @@ -4239,15 +4438,18 @@ rassoc_i(VALUE key, VALUE val, VALUE arg) /* * call-seq: - * hash.rassoc(value) -> new_array or nil + * rassoc(value) -> new_array or nil + * + * Searches +self+ for the first entry whose value is <tt>==</tt> to the given +value+; + * see {Entry Order}[rdoc-ref:Hash@Entry+Order]. + * + * If the entry is found, returns its key and value as a 2-element array; + * returns +nil+ if not found: * - * Returns a new 2-element \Array consisting of the key and value - * of the first-found entry whose value is <tt>==</tt> to value - * (see {Entry Order}[rdoc-ref:Hash@Entry+Order]): * h = {foo: 0, bar: 1, baz: 1} * h.rassoc(1) # => [:bar, 1] * - * Returns +nil+ if no such value found. + * Related: see {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -4275,33 +4477,38 @@ flatten_i(VALUE key, VALUE val, VALUE ary) /* * call-seq: - * hash.flatten -> new_array - * hash.flatten(level) -> new_array + * flatten(depth = 1) -> new_array + * + * With positive integer +depth+, + * returns a new array that is a recursive flattening of +self+ to the given +depth+. + * + * At each level of recursion: * - * Returns a new \Array object that is a 1-dimensional flattening of +self+. + * - Each element whose value is an array is "flattened" (that is, replaced by its individual array elements); + * see Array#flatten. + * - Each element whose value is not an array is unchanged. + * even if the value is an object that has instance method flatten (such as a hash). * - * --- + * Examples; note that entry <tt>foo: {bar: 1, baz: 2}</tt> is never flattened. * - * By default, nested Arrays are not flattened: - * h = {foo: 0, bar: [:bat, 3], baz: 2} - * h.flatten # => [:foo, 0, :bar, [:bat, 3], :baz, 2] + * h = {foo: {bar: 1, baz: 2}, bat: [:bam, [:bap, [:bah]]]} + * h.flatten(1) # => [:foo, {:bar=>1, :baz=>2}, :bat, [:bam, [:bap, [:bah]]]] + * h.flatten(2) # => [:foo, {:bar=>1, :baz=>2}, :bat, :bam, [:bap, [:bah]]] + * h.flatten(3) # => [:foo, {:bar=>1, :baz=>2}, :bat, :bam, :bap, [:bah]] + * h.flatten(4) # => [:foo, {:bar=>1, :baz=>2}, :bat, :bam, :bap, :bah] + * h.flatten(5) # => [:foo, {:bar=>1, :baz=>2}, :bat, :bam, :bap, :bah] * - * Takes the depth of recursive flattening from \Integer argument +level+: - * h = {foo: 0, bar: [:bat, [:baz, [:bat, ]]]} - * h.flatten(1) # => [:foo, 0, :bar, [:bat, [:baz, [:bat]]]] - * h.flatten(2) # => [:foo, 0, :bar, :bat, [:baz, [:bat]]] - * h.flatten(3) # => [:foo, 0, :bar, :bat, :baz, [:bat]] - * h.flatten(4) # => [:foo, 0, :bar, :bat, :baz, :bat] + * With negative integer +depth+, + * flattens all levels: * - * When +level+ is negative, flattens all nested Arrays: - * h = {foo: 0, bar: [:bat, [:baz, [:bat, ]]]} - * h.flatten(-1) # => [:foo, 0, :bar, :bat, :baz, :bat] - * h.flatten(-2) # => [:foo, 0, :bar, :bat, :baz, :bat] + * h.flatten(-1) # => [:foo, {:bar=>1, :baz=>2}, :bat, :bam, :bap, :bah] * - * When +level+ is zero, returns the equivalent of #to_a : - * h = {foo: 0, bar: [:bat, 3], baz: 2} - * h.flatten(0) # => [[:foo, 0], [:bar, [:bat, 3]], [:baz, 2]] - * h.flatten(0) == h.to_a # => true + * With +depth+ zero, + * returns the equivalent of #to_a: + * + * h.flatten(0) # => [[:foo, {:bar=>1, :baz=>2}], [:bat, [:bam, [:bap, [:bah]]]]] + * + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE @@ -4346,44 +4553,46 @@ delete_if_nil(VALUE key, VALUE value, VALUE hash) return ST_CONTINUE; } -static int -set_if_not_nil(VALUE key, VALUE value, VALUE hash) -{ - if (!NIL_P(value)) { - rb_hash_aset(hash, key, value); - } - return ST_CONTINUE; -} - /* * call-seq: - * hash.compact -> new_hash + * compact -> new_hash * * Returns a copy of +self+ with all +nil+-valued entries removed: + * * h = {foo: 0, bar: nil, baz: 2, bat: nil} - * h1 = h.compact - * h1 # => {:foo=>0, :baz=>2} + * h.compact # => {foo: 0, baz: 2} + * + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE rb_hash_compact(VALUE hash) { - VALUE result = rb_hash_new(); + VALUE result = rb_hash_dup(hash); if (!RHASH_EMPTY_P(hash)) { - rb_hash_foreach(hash, set_if_not_nil, result); + rb_hash_foreach(result, delete_if_nil, result); + compact_after_delete(result); + } + else if (rb_hash_compare_by_id_p(hash)) { + result = rb_hash_compare_by_id(result); } return result; } /* * call-seq: - * hash.compact! -> self or nil + * compact! -> self or nil + * + * If +self+ contains any +nil+-valued entries, + * returns +self+ with all +nil+-valued entries removed; + * returns +nil+ otherwise: * - * Returns +self+ with all its +nil+-valued entries removed (in place): * h = {foo: 0, bar: nil, baz: 2, bat: nil} - * h.compact! # => {:foo=>0, :baz=>2} + * h.compact! + * h # => {foo: 0, baz: 2} + * h.compact! # => nil * - * Returns +nil+ if no entries were removed. + * Related: see {Methods for Deleting}[rdoc-ref:Hash@Methods+for+Deleting]. */ static VALUE @@ -4400,34 +4609,30 @@ rb_hash_compact_bang(VALUE hash) return Qnil; } -static st_table *rb_init_identtable_with_size(st_index_t size); - /* * call-seq: - * hash.compare_by_identity -> self + * compare_by_identity -> self * - * Sets +self+ to consider only identity in comparing keys; - * two keys are considered the same only if they are the same object; - * returns +self+. + * Sets +self+ to compare keys using _identity_ (rather than mere _equality_); + * returns +self+: + * + * By default, two keys are considered to be the same key + * if and only if they are _equal_ objects (per method #eql?): * - * By default, these two object are considered to be the same key, - * so +s1+ will overwrite +s0+: - * s0 = 'x' - * s1 = 'x' * h = {} - * h.compare_by_identity? # => false - * h[s0] = 0 - * h[s1] = 1 + * h['x'] = 0 + * h['x'] = 1 # Overwrites. * h # => {"x"=>1} * - * After calling \#compare_by_identity, the keys are considered to be different, - * and therefore do not overwrite each other: - * h = {} - * h.compare_by_identity # => {} - * h.compare_by_identity? # => true - * h[s0] = 0 - * h[s1] = 1 - * h # => {"x"=>0, "x"=>1} + * When this method has been called, two keys are considered to be the same key + * if and only if they are the _same_ object: + * + * h.compare_by_identity + * h['x'] = 2 # Does not overwrite. + * h # => {"x"=>1, "x"=>2} + * + * Related: #compare_by_identity?; + * see also {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ VALUE @@ -4439,38 +4644,66 @@ rb_hash_compare_by_id(VALUE hash) if (rb_hash_compare_by_id_p(hash)) return hash; rb_hash_modify_check(hash); - ar_force_convert_table(hash, __FILE__, __LINE__); - HASH_ASSERT(RHASH_ST_TABLE_P(hash)); + if (hash_iterating_p(hash)) { + rb_raise(rb_eRuntimeError, "compare_by_identity during iteration"); + } + + if (RHASH_TABLE_EMPTY_P(hash)) { + // Fast path: There's nothing to rehash, so we don't need a `tmp` table. + // We're most likely an AR table, so this will need an allocation. + ar_force_convert_table(hash, __FILE__, __LINE__); + HASH_ASSERT(RHASH_ST_TABLE_P(hash)); + + RHASH_ST_TABLE(hash)->type = &identhash; + } + else { + // Slow path: Need to rehash the members of `self` into a new + // `tmp` table using the new `identhash` compare/hash functions. + tmp = hash_alloc(0); + hash_st_table_init(tmp, &identhash, RHASH_SIZE(hash)); + identtable = RHASH_ST_TABLE(tmp); + + rb_hash_foreach(hash, rb_hash_rehash_i, (VALUE)tmp); + rb_hash_free(hash); + + // We know for sure `identtable` is an st table, + // so we can skip `ar_force_convert_table` here. + RHASH_ST_TABLE_SET(hash, identtable); + RHASH_ST_CLEAR(tmp); + } - tmp = hash_alloc(0); - identtable = rb_init_identtable_with_size(RHASH_SIZE(hash)); - RHASH_ST_TABLE_SET(tmp, identtable); - rb_hash_foreach(hash, rb_hash_rehash_i, (VALUE)tmp); - st_free_table(RHASH_ST_TABLE(hash)); - RHASH_ST_TABLE_SET(hash, identtable); - RHASH_ST_CLEAR(tmp); + rb_gc_register_pinning_obj(hash); return hash; } /* * call-seq: - * hash.compare_by_identity? -> true or false + * compare_by_identity? -> true or false * - * Returns +true+ if #compare_by_identity has been called, +false+ otherwise. + * Returns whether #compare_by_identity has been called: + * + * h = {} + * h.compare_by_identity? # => false + * h.compare_by_identity + * h.compare_by_identity? # => true + * + * Related: #compare_by_identity; + * see also {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ -MJIT_FUNC_EXPORTED VALUE +VALUE rb_hash_compare_by_id_p(VALUE hash) { - return RBOOL(RHASH_ST_TABLE_P(hash) && RHASH_ST_TABLE(hash)->type == &identhash); + return RBOOL(RHASH_IDENTHASH_P(hash)); } VALUE rb_ident_hash_new(void) { VALUE hash = rb_hash_new(); - RHASH_ST_TABLE_SET(hash, st_init_table(&identhash)); + hash_st_table_init(hash, &identhash, 0); + rb_gc_register_pinning_obj(hash); return hash; } @@ -4478,7 +4711,8 @@ VALUE rb_ident_hash_new_with_size(st_index_t size) { VALUE hash = rb_hash_new(); - RHASH_ST_TABLE_SET(hash, st_init_table_with_size(&identhash, size)); + hash_st_table_init(hash, &identhash, size); + rb_gc_register_pinning_obj(hash); return hash; } @@ -4488,12 +4722,6 @@ rb_init_identtable(void) return st_init_table(&identhash); } -static st_table * -rb_init_identtable_with_size(st_index_t size) -{ - return st_init_table_with_size(&identhash, size); -} - static int any_p_i(VALUE key, VALUE value, VALUE arg) { @@ -4529,31 +4757,42 @@ any_p_i_pattern(VALUE key, VALUE value, VALUE arg) /* * call-seq: - * hash.any? -> true or false - * hash.any?(object) -> true or false - * hash.any? {|key, value| ... } -> true or false + * any? -> true or false + * any?(entry) -> true or false + * any? {|key, value| ... } -> true or false * * Returns +true+ if any element satisfies a given criterion; * +false+ otherwise. * + * If +self+ has no element, returns +false+ and argument or block are not used; + * otherwise behaves as below. + * * With no argument and no block, - * returns +true+ if +self+ is non-empty; +false+ if empty. + * returns +true+ if +self+ is non-empty, +false+ otherwise. * - * With argument +object+ and no block, + * With argument +entry+ and no block, * returns +true+ if for any key +key+ - * <tt>h.assoc(key) == object</tt>: + * <tt>self.assoc(key) == entry</tt>, +false+ otherwise: + * * h = {foo: 0, bar: 1, baz: 2} + * h.assoc(:bar) # => [:bar, 1] * h.any?([:bar, 1]) # => true * h.any?([:bar, 0]) # => false - * h.any?([:baz, 1]) # => false * - * With no argument and a block, + * With no argument and a block given, * calls the block with each key-value pair; - * returns +true+ if the block returns any truthy value, + * returns +true+ if the block returns a truthy value, * +false+ otherwise: + * * h = {foo: 0, bar: 1, baz: 2} * h.any? {|key, value| value < 3 } # => true * h.any? {|key, value| value > 3 } # => false + * + * With both argument +entry+ and a block given, + * issues a warning and ignores the block. + * + * Related: Enumerable#any? (which this method overrides); + * see also {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -4587,31 +4826,37 @@ rb_hash_any_p(int argc, VALUE *argv, VALUE hash) /* * call-seq: - * hash.dig(key, *identifiers) -> object + * dig(key, *identifiers) -> object + * + * Finds and returns an object found in nested objects, + * as specified by +key+ and +identifiers+. * - * Finds and returns the object in nested objects - * that is specified by +key+ and +identifiers+. * The nested objects may be instances of various classes. * See {Dig Methods}[rdoc-ref:dig_methods.rdoc]. * - * Nested Hashes: + * Nested hashes: + * * h = {foo: {bar: {baz: 2}}} - * h.dig(:foo) # => {:bar=>{:baz=>2}} - * h.dig(:foo, :bar) # => {:baz=>2} + * h.dig(:foo) # => {bar: {baz: 2}} + * h.dig(:foo, :bar) # => {baz: 2} * h.dig(:foo, :bar, :baz) # => 2 * h.dig(:foo, :bar, :BAZ) # => nil * - * Nested Hashes and Arrays: + * Nested hashes and arrays: + * * h = {foo: {bar: [:a, :b, :c]}} * h.dig(:foo, :bar, 2) # => :c * - * This method will use the {default values}[rdoc-ref:Hash@Default+Values] - * for keys that are not present: + * If no such object is found, + * returns the {hash default}[rdoc-ref:Hash@Hash+Default]: + * * h = {foo: {bar: [:a, :b, :c]}} * h.dig(:hello) # => nil * h.default_proc = -> (hash, _key) { hash } - * h.dig(:hello, :world) # => h - * h.dig(:hello, :world, :foo, :bar, 2) # => :c + * h.dig(:hello, :world) + * # => {:foo=>{:bar=>[:a, :b, :c]}} + * + * Related: {Methods for Fetching}[rdoc-ref:Hash@Methods+for+Fetching]. */ static VALUE @@ -4646,14 +4891,21 @@ hash_le(VALUE hash1, VALUE hash2) /* * call-seq: - * hash <= other_hash -> true or false + * self <= other -> true or false * - * Returns +true+ if +hash+ is a subset of +other_hash+, +false+ otherwise: - * h1 = {foo: 0, bar: 1} - * h2 = {foo: 0, bar: 1, baz: 2} - * h1 <= h2 # => true - * h2 <= h1 # => false - * h1 <= h1 # => true + * Returns whether the entries of +self+ are a subset of the entries of +other+: + * + * h0 = {foo: 0, bar: 1} + * h1 = {foo: 0, bar: 1, baz: 2} + * h0 <= h0 # => true + * h0 <= h1 # => true + * h1 <= h0 # => false + * + * See {Hash Inclusion}[rdoc-ref:language/hash_inclusion.rdoc]. + * + * Raises TypeError if +other_hash+ is not a hash and cannot be converted to a hash. + * + * Related: see {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ static VALUE rb_hash_le(VALUE hash, VALUE other) @@ -4665,14 +4917,23 @@ rb_hash_le(VALUE hash, VALUE other) /* * call-seq: - * hash < other_hash -> true or false + * self < other -> true or false * - * Returns +true+ if +hash+ is a proper subset of +other_hash+, +false+ otherwise: - * h1 = {foo: 0, bar: 1} - * h2 = {foo: 0, bar: 1, baz: 2} - * h1 < h2 # => true - * h2 < h1 # => false - * h1 < h1 # => false + * Returns whether the entries of +self+ are a proper subset of the entries of +other+: + * + * h = {foo: 0, bar: 1} + * h < {foo: 0, bar: 1, baz: 2} # => true # Proper subset. + * h < {baz: 2, bar: 1, foo: 0} # => true # Order may differ. + * h < h # => false # Not a proper subset. + * h < {bar: 1, foo: 0} # => false # Not a proper subset. + * h < {foo: 0, bar: 1, baz: 2} # => false # Different key. + * h < {foo: 0, bar: 1, baz: 2} # => false # Different value. + * + * See {Hash Inclusion}[rdoc-ref:language/hash_inclusion.rdoc]. + * + * Raises TypeError if +other_hash+ is not a hash and cannot be converted to a hash. + * + * Related: see {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ static VALUE rb_hash_lt(VALUE hash, VALUE other) @@ -4684,14 +4945,21 @@ rb_hash_lt(VALUE hash, VALUE other) /* * call-seq: - * hash >= other_hash -> true or false + * self >= other -> true or false * - * Returns +true+ if +hash+ is a superset of +other_hash+, +false+ otherwise: - * h1 = {foo: 0, bar: 1, baz: 2} - * h2 = {foo: 0, bar: 1} - * h1 >= h2 # => true - * h2 >= h1 # => false - * h1 >= h1 # => true + * Returns whether the entries of +self+ are a superset of the entries of +other+: + * + * h0 = {foo: 0, bar: 1, baz: 2} + * h1 = {foo: 0, bar: 1} + * h0 >= h1 # => true + * h0 >= h0 # => true + * h1 >= h0 # => false + * + * See {Hash Inclusion}[rdoc-ref:language/hash_inclusion.rdoc]. + * + * Raises TypeError if +other_hash+ is not a hash and cannot be converted to a hash. + * + * Related: see {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ static VALUE rb_hash_ge(VALUE hash, VALUE other) @@ -4703,14 +4971,23 @@ rb_hash_ge(VALUE hash, VALUE other) /* * call-seq: - * hash > other_hash -> true or false + * self > other -> true or false * - * Returns +true+ if +hash+ is a proper superset of +other_hash+, +false+ otherwise: - * h1 = {foo: 0, bar: 1, baz: 2} - * h2 = {foo: 0, bar: 1} - * h1 > h2 # => true - * h2 > h1 # => false - * h1 > h1 # => false + * Returns whether the entries of +self+ are a proper superset of the entries of +other+: + * + * h = {foo: 0, bar: 1, baz: 2} + * h > {foo: 0, bar: 1} # => true # Proper superset. + * h > {bar: 1, foo: 0} # => true # Order may differ. + * h > h # => false # Not a proper superset. + * h > {baz: 2, bar: 1, foo: 0} # => false # Not a proper superset. + * h > {foo: 0, bar: 1} # => false # Different key. + * h > {foo: 0, bar: 1} # => false # Different value. + * + * See {Hash Inclusion}[rdoc-ref:language/hash_inclusion.rdoc]. + * + * Raises TypeError if +other_hash+ is not a hash and cannot be converted to a hash. + * + * Related: see {Methods for Comparing}[rdoc-ref:Hash@Methods+for+Comparing]. */ static VALUE rb_hash_gt(VALUE hash, VALUE other) @@ -4729,15 +5006,18 @@ hash_proc_call(RB_BLOCK_CALL_FUNC_ARGLIST(key, hash)) /* * call-seq: - * hash.to_proc -> proc + * to_proc -> proc + * + * Returns a Proc object that maps a key to its value: * - * Returns a \Proc object that maps a key to its value: * h = {foo: 0, bar: 1, baz: 2} * proc = h.to_proc * proc.class # => Proc * proc.call(:foo) # => 0 * proc.call(:bar) # => 1 * proc.call(:nosuch) # => nil + * + * Related: see {Methods for Converting}[rdoc-ref:Hash@Methods+for+Converting]. */ static VALUE rb_hash_to_proc(VALUE hash) @@ -4745,6 +5025,7 @@ rb_hash_to_proc(VALUE hash) return rb_func_lambda_new(hash_proc_call, hash, 1, 1); } +/* :nodoc: */ static VALUE rb_hash_deconstruct_keys(VALUE hash, VALUE keys) { @@ -4754,10 +5035,8 @@ rb_hash_deconstruct_keys(VALUE hash, VALUE keys) static int add_new_i(st_data_t *key, st_data_t *val, st_data_t arg, int existing) { - VALUE *args = (VALUE *)arg; if (existing) return ST_STOP; - RB_OBJ_WRITTEN(args[0], Qundef, (VALUE)*key); - RB_OBJ_WRITE(args[0], (VALUE *)val, args[1]); + *val = arg; return ST_CONTINUE; } @@ -4769,23 +5048,25 @@ int rb_hash_add_new_element(VALUE hash, VALUE key, VALUE val) { st_table *tbl; - int ret = 0; - VALUE args[2]; - args[0] = hash; - args[1] = val; + int ret = -1; if (RHASH_AR_TABLE_P(hash)) { - hash_ar_table(hash); - - ret = ar_update(hash, (st_data_t)key, add_new_i, (st_data_t)args); - if (ret != -1) { - return ret; + ret = ar_update(hash, (st_data_t)key, add_new_i, (st_data_t)val); + if (ret == -1) { + ar_force_convert_table(hash, __FILE__, __LINE__); } - ar_try_convert_table(hash); } - tbl = RHASH_TBL_RAW(hash); - return st_update(tbl, (st_data_t)key, add_new_i, (st_data_t)args); + if (ret == -1) { + tbl = RHASH_TBL_RAW(hash); + ret = st_update(tbl, (st_data_t)key, add_new_i, (st_data_t)val); + } + if (!ret) { + // Newly inserted + RB_OBJ_WRITTEN(hash, Qundef, key); + RB_OBJ_WRITTEN(hash, Qundef, val); + } + return ret; } static st_data_t @@ -4815,15 +5096,6 @@ rb_hash_bulk_insert(long argc, const VALUE *argv, VALUE hash) if (argc > 0) { st_index_t size = argc / 2; - if (RHASH_TABLE_NULL_P(hash)) { - if (size <= RHASH_AR_TABLE_MAX_SIZE) { - hash_ar_table(hash); - } - else { - RHASH_TBL_RAW(hash); - } - } - if (RHASH_AR_TABLE_P(hash) && (RHASH_AR_TABLE_SIZE(hash) + size <= RHASH_AR_TABLE_MAX_SIZE)) { ar_bulk_insert(hash, argc, argv); @@ -4861,8 +5133,7 @@ extern char **environ; #define ENVNMATCH(s1, s2, n) (memcmp((s1), (s2), (n)) == 0) #endif -#define ENV_LOCK() RB_VM_LOCK_ENTER() -#define ENV_UNLOCK() RB_VM_LOCK_LEAVE() +#define ENV_LOCKING() RB_VM_LOCKING() static inline rb_encoding * env_encoding(void) @@ -4884,28 +5155,27 @@ env_enc_str_new(const char *ptr, long len, rb_encoding *enc) } static VALUE -env_str_new(const char *ptr, long len) +env_str_new(const char *ptr, long len, rb_encoding *enc) { - return env_enc_str_new(ptr, len, env_encoding()); + return env_enc_str_new(ptr, len, enc); } static VALUE -env_str_new2(const char *ptr) +env_str_new2(const char *ptr, rb_encoding *enc) { if (!ptr) return Qnil; - return env_str_new(ptr, strlen(ptr)); + return env_str_new(ptr, strlen(ptr), enc); } static VALUE getenv_with_lock(const char *name) { VALUE ret; - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { const char *val = getenv(name); - ret = env_str_new2(val); + ret = env_str_new2(val, enc); } - ENV_UNLOCK(); return ret; } @@ -4914,11 +5184,9 @@ has_env_with_lock(const char *name) { const char *val; - ENV_LOCK(); - { + ENV_LOCKING() { val = getenv(name); } - ENV_UNLOCK(); return val ? true : false; } @@ -4926,9 +5194,7 @@ has_env_with_lock(const char *name) static const char TZ_ENV[] = "TZ"; static void * -get_env_cstr( - VALUE str, - const char *name) +get_env_cstr(VALUE str, const char *name) { char *var; rb_encoding *enc = rb_enc_get(str); @@ -4950,7 +5216,7 @@ static inline const char * env_name(volatile VALUE *s) { const char *name; - SafeStringValue(*s); + StringValue(*s); get_env_ptr(name, *s); return name; } @@ -4960,7 +5226,7 @@ env_name(volatile VALUE *s) static VALUE env_aset(VALUE nm, VALUE val); static void -reset_by_modified_env(const char *nam) +reset_by_modified_env(const char *nam, const char *val) { /* * ENV['TZ'] = nil has a special meaning. @@ -4969,7 +5235,7 @@ reset_by_modified_env(const char *nam) * This hack might works only on Linux glibc. */ if (ENVMATCH(nam, TZ_ENV)) { - ruby_reset_timezone(); + ruby_reset_timezone(val); } } @@ -4977,7 +5243,7 @@ static VALUE env_delete(VALUE name) { const char *nam = env_name(name); - reset_by_modified_env(nam); + reset_by_modified_env(nam, NULL); VALUE val = getenv_with_lock(nam); if (!NIL_P(val)) { @@ -5123,44 +5389,6 @@ envix(const char *nam) } #endif -#if defined(_WIN32) -static size_t -getenvsize(const WCHAR* p) -{ - const WCHAR* porg = p; - while (*p++) p += lstrlenW(p) + 1; - return p - porg + 1; -} - -static size_t -getenvblocksize(void) -{ -#ifdef _MAX_ENV - return _MAX_ENV; -#else - return 32767; -#endif -} - -static int -check_envsize(size_t n) -{ - if (_WIN32_WINNT < 0x0600 && rb_w32_osver() < 6) { - /* https://msdn.microsoft.com/en-us/library/windows/desktop/ms682653(v=vs.85).aspx */ - /* Windows Server 2003 and Windows XP: The maximum size of the - * environment block for the process is 32,767 characters. */ - WCHAR* p = GetEnvironmentStringsW(); - if (!p) return -1; /* never happen */ - n += getenvsize(p); - FreeEnvironmentStringsW(p); - if (n >= getenvblocksize()) { - return -1; - } - } - return 0; -} -#endif - #if defined(_WIN32) || \ (defined(__sun) && !(defined(HAVE_SETENV) && defined(HAVE_UNSETENV))) @@ -5186,9 +5414,6 @@ void ruby_setenv(const char *name, const char *value) { #if defined(_WIN32) -# if defined(MINGW_HAS_SECURE_API) || RUBY_MSVCRT_VERSION >= 80 -# define HAVE__WPUTENV_S 1 -# endif VALUE buf; WCHAR *wname; WCHAR *wvalue = 0; @@ -5199,43 +5424,30 @@ ruby_setenv(const char *name, const char *value) if (value) { int len2; len2 = MultiByteToWideChar(CP_UTF8, 0, value, -1, NULL, 0); - if (check_envsize((size_t)len + len2)) { /* len and len2 include '\0' */ - goto fail; /* 2 for '=' & '\0' */ - } wname = ALLOCV_N(WCHAR, buf, len + len2); wvalue = wname + len; MultiByteToWideChar(CP_UTF8, 0, name, -1, wname, len); MultiByteToWideChar(CP_UTF8, 0, value, -1, wvalue, len2); -#ifndef HAVE__WPUTENV_S - wname[len-1] = L'='; -#endif } else { wname = ALLOCV_N(WCHAR, buf, len + 1); MultiByteToWideChar(CP_UTF8, 0, name, -1, wname, len); wvalue = wname + len; *wvalue = L'\0'; -#ifndef HAVE__WPUTENV_S - wname[len-1] = L'='; -#endif } - ENV_LOCK(); - { -#ifndef HAVE__WPUTENV_S - failed = _wputenv(wname); -#else + ENV_LOCKING() { + /* Use _wputenv_s() instead of SetEnvironmentVariableW() to make sure + * special variables like "TZ" are interpret by libc. */ failed = _wputenv_s(wname, wvalue); -#endif } - ENV_UNLOCK(); ALLOCV_END(buf); /* even if putenv() failed, clean up and try to delete the * variable from the system area. */ if (!value || !*value) { /* putenv() doesn't handle empty value */ - if (!SetEnvironmentVariable(name, value) && + if (!SetEnvironmentVariableW(wname, value ? wvalue : NULL) && GetLastError() != ERROR_ENVVAR_NOT_FOUND) goto fail; } if (failed) { @@ -5245,30 +5457,24 @@ ruby_setenv(const char *name, const char *value) #elif defined(HAVE_SETENV) && defined(HAVE_UNSETENV) if (value) { int ret; - ENV_LOCK(); - { + ENV_LOCKING() { ret = setenv(name, value, 1); } - ENV_UNLOCK(); - if (ret) rb_sys_fail_str(rb_sprintf("setenv(%s)", name)); + if (ret) rb_sys_fail_sprintf("setenv(%s)", name); } else { #ifdef VOID_UNSETENV - ENV_LOCK(); - { + ENV_LOCKING() { unsetenv(name); } - ENV_UNLOCK(); #else int ret; - ENV_LOCK(); - { + ENV_LOCKING() { ret = unsetenv(name); } - ENV_UNLOCK(); - if (ret) rb_sys_fail_str(rb_sprintf("unsetenv(%s)", name)); + if (ret) rb_sys_fail_sprintf("unsetenv(%s)", name); #endif } #elif defined __sun @@ -5285,12 +5491,11 @@ ruby_setenv(const char *name, const char *value) mem_size = len + strlen(value) + 2; mem_ptr = malloc(mem_size); if (mem_ptr == NULL) - rb_sys_fail_str(rb_sprintf("malloc(%"PRIuSIZE")", mem_size)); + rb_sys_fail_sprintf("malloc(%"PRIuSIZE")", mem_size); snprintf(mem_ptr, mem_size, "%s=%s", name, value); } - ENV_LOCK(); - { + ENV_LOCKING() { for (env_ptr = GET_ENVIRON(environ); (str = *env_ptr) != 0; ++env_ptr) { if (!strncmp(str, name, len) && str[len] == '=') { if (!in_origenv(str)) free(str); @@ -5299,27 +5504,23 @@ ruby_setenv(const char *name, const char *value) } } } - ENV_UNLOCK(); if (value) { int ret; - ENV_LOCK(); - { + ENV_LOCKING() { ret = putenv(mem_ptr); } - ENV_UNLOCK(); if (ret) { free(mem_ptr); - rb_sys_fail_str(rb_sprintf("putenv(%s)", name)); + rb_sys_fail_sprintf("putenv(%s)", name); } } #else /* WIN32 */ size_t len; int i; - ENV_LOCK(); - { + ENV_LOCKING() { i = envix(name); /* where does it go? */ if (environ == origenviron) { /* need we copy environment? */ @@ -5360,7 +5561,6 @@ ruby_setenv(const char *name, const char *value) finish:; } - ENV_UNLOCK(); #endif /* WIN32 */ } @@ -5375,8 +5575,6 @@ ruby_unsetenv(const char *name) * ENV[name] = value -> value * ENV.store(name, value) -> value * - * ENV.store is an alias for ENV.[]=. - * * Creates, updates, or deletes the named environment variable, returning the value. * Both +name+ and +value+ may be instances of String. * See {Valid Names and Values}[rdoc-ref:ENV@Valid+Names+and+Values]. @@ -5429,15 +5627,15 @@ env_aset(VALUE nm, VALUE val) env_delete(nm); return Qnil; } - SafeStringValue(nm); - SafeStringValue(val); + StringValue(nm); + StringValue(val); /* nm can be modified in `val.to_str`, don't get `name` before * check for `val` */ get_env_ptr(name, nm); get_env_ptr(value, val); ruby_setenv(name, value); - reset_by_modified_env(name); + reset_by_modified_env(name, value); return val; } @@ -5447,8 +5645,7 @@ env_keys(int raw) rb_encoding *enc = raw ? 0 : rb_locale_encoding(); VALUE ary = rb_ary_new(); - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); @@ -5462,7 +5659,6 @@ env_keys(int raw) } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return ary; } @@ -5492,8 +5688,7 @@ rb_env_size(VALUE ehash, VALUE args, VALUE eobj) char **env; long cnt = 0; - ENV_LOCK(); - { + ENV_LOCKING() { env = GET_ENVIRON(environ); for (; *env ; ++env) { if (strchr(*env, '=')) { @@ -5502,7 +5697,6 @@ rb_env_size(VALUE ehash, VALUE args, VALUE eobj) } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return LONG2FIX(cnt); } @@ -5543,20 +5737,19 @@ env_values(void) { VALUE ary = rb_ary_new(); - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); if (s) { - rb_ary_push(ary, env_str_new2(s+1)); + rb_ary_push(ary, env_str_new2(s+1, enc)); } env++; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return ary; } @@ -5617,7 +5810,7 @@ env_each_value(VALUE ehash) * ENV.each_pair { |name, value| block } -> ENV * ENV.each_pair -> an_enumerator * - * Yields each environment variable name and its value as a 2-element \Array: + * Yields each environment variable name and its value as a 2-element Array: * h = {} * ENV.each_pair { |name, value| h[name] = value } # => ENV * h # => {"bar"=>"1", "foo"=>"0"} @@ -5637,21 +5830,20 @@ env_each_pair(VALUE ehash) VALUE ary = rb_ary_new(); - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); if (s) { - rb_ary_push(ary, env_str_new(*env, s-*env)); - rb_ary_push(ary, env_str_new2(s+1)); + rb_ary_push(ary, env_str_new(*env, s-*env, enc)); + rb_ary_push(ary, env_str_new2(s+1, enc)); } env++; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); if (rb_block_pair_yield_optimizable()) { for (i=0; i<RARRAY_LEN(ary); i+=2) { @@ -5753,7 +5945,7 @@ env_delete_if(VALUE ehash) * Returns +nil+ in the Array for each name that is not an ENV name: * ENV.values_at('foo', 'bat', 'bar', 'bam') # => ["0", nil, "1", nil] * - * Returns an empty \Array if no names given. + * Returns an empty Array if no names given. * * Raises an exception if any name is invalid. * See {Invalid Names and Values}[rdoc-ref:ENV@Invalid+Names+and+Values]. @@ -5778,8 +5970,6 @@ env_values_at(int argc, VALUE *argv, VALUE _) * ENV.filter { |name, value| block } -> hash of name/value pairs * ENV.filter -> an_enumerator * - * ENV.filter is an alias for ENV.select. - * * Yields each environment variable name and its value as a 2-element Array, * returning a Hash of the names and values for which the block returns a truthy value: * ENV.replace('foo' => '0', 'bar' => '1', 'baz' => '2') @@ -5823,8 +6013,6 @@ env_select(VALUE ehash) * ENV.filter! { |name, value| block } -> ENV or nil * ENV.filter! -> an_enumerator * - * ENV.filter! is an alias for ENV.select!. - * * Yields each environment variable name and its value as a 2-element Array, * deleting each entry for which the block returns +false+ or +nil+, * and returning ENV if any deletions made, or +nil+ otherwise: @@ -5992,30 +6180,27 @@ env_to_s(VALUE _) static VALUE env_inspect(VALUE _) { - VALUE i; VALUE str = rb_str_buf_new2("{"); + rb_encoding *enc = env_encoding(); - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { - char *s = strchr(*env, '='); + const char *s = strchr(*env, '='); if (env != environ) { rb_str_buf_cat2(str, ", "); } if (s) { - rb_str_buf_cat2(str, "\""); - rb_str_buf_cat(str, *env, s-*env); - rb_str_buf_cat2(str, "\"=>"); - i = rb_inspect(rb_str_new2(s+1)); - rb_str_buf_append(str, i); + rb_str_buf_append(str, rb_str_inspect(env_enc_str_new(*env, s-*env, enc))); + rb_str_buf_cat2(str, " => "); + s++; + rb_str_buf_append(str, rb_str_inspect(env_enc_str_new(s, strlen(s), enc))); } env++; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); rb_str_buf_cat2(str, "}"); @@ -6036,20 +6221,19 @@ env_to_a(VALUE _) { VALUE ary = rb_ary_new(); - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); if (s) { - rb_ary_push(ary, rb_assoc_new(env_str_new(*env, s-*env), - env_str_new2(s+1))); + rb_ary_push(ary, rb_assoc_new(env_str_new(*env, s-*env, enc), + env_str_new2(s+1, enc))); } env++; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return ary; } @@ -6073,13 +6257,11 @@ env_size_with_lock(void) { int i = 0; - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (env[i]) i++; FREE_ENVIRON(environ); } - ENV_UNLOCK(); return i; } @@ -6115,15 +6297,13 @@ env_empty_p(VALUE _) { bool empty = true; - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); if (env[0] != 0) { empty = false; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return RBOOL(empty); } @@ -6135,8 +6315,6 @@ env_empty_p(VALUE _) * ENV.member?(name) -> true or false * ENV.key?(name) -> true or false * - * ENV.has_key?, ENV.member?, and ENV.key? are aliases for ENV.include?. - * * Returns +true+ if there is an environment variable with the given +name+: * ENV.replace('foo' => '0', 'bar' => '1') * ENV.include?('foo') # => true @@ -6214,8 +6392,7 @@ env_has_value(VALUE dmy, VALUE obj) VALUE ret = Qfalse; - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); @@ -6230,7 +6407,6 @@ env_has_value(VALUE dmy, VALUE obj) } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return ret; } @@ -6257,8 +6433,7 @@ env_rassoc(VALUE dmy, VALUE obj) VALUE result = Qnil; - ENV_LOCK(); - { + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { @@ -6275,7 +6450,6 @@ env_rassoc(VALUE dmy, VALUE obj) } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return result; } @@ -6299,18 +6473,18 @@ env_rassoc(VALUE dmy, VALUE obj) static VALUE env_key(VALUE dmy, VALUE value) { - SafeStringValue(value); + StringValue(value); VALUE str = Qnil; - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); if (s++) { long len = strlen(s); if (RSTRING_LEN(value) == len && strncmp(s, RSTRING_PTR(value), len) == 0) { - str = env_str_new(*env, s-*env-1); + str = env_str_new(*env, s-*env-1, enc); break; } } @@ -6318,7 +6492,6 @@ env_key(VALUE dmy, VALUE value) } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return str; } @@ -6328,20 +6501,19 @@ env_to_hash(void) { VALUE hash = rb_hash_new(); - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); while (*env) { char *s = strchr(*env, '='); if (s) { - rb_hash_aset(hash, env_str_new(*env, s-*env), - env_str_new2(s+1)); + rb_hash_aset(hash, env_str_new(*env, s-*env, enc), + env_str_new2(s+1, enc)); } env++; } FREE_ENVIRON(environ); } - ENV_UNLOCK(); return hash; } @@ -6385,7 +6557,7 @@ env_f_to_hash(VALUE _) * Each name/value pair in ENV is yielded to the block. * The block must return a 2-element Array (name/value pair) * that is added to the return Hash as a key and value: - * ENV.to_h { |name, value| [name.to_sym, value.to_i] } # => {:bar=>1, :foo=>0} + * ENV.to_h { |name, value| [name.to_sym, value.to_i] } # => {bar: 1, foo: 0} * Raises an exception if the block does not return an Array: * ENV.to_h { |name, value| name } # Raises TypeError (wrong element type String (expected array)) * Raises an exception if the block returns an Array of the wrong size: @@ -6481,21 +6653,20 @@ env_shift(VALUE _) VALUE result = Qnil; VALUE key = Qnil; - ENV_LOCK(); - { + rb_encoding *enc = env_encoding(); + ENV_LOCKING() { char **env = GET_ENVIRON(environ); if (*env) { const char *p = *env; char *s = strchr(p, '='); if (s) { - key = env_str_new(p, s-p); - VALUE val = env_str_new2(getenv(RSTRING_PTR(key))); + key = env_str_new(p, s-p, enc); + VALUE val = env_str_new2(getenv(RSTRING_PTR(key)), enc); result = rb_assoc_new(key, val); } } FREE_ENVIRON(environ); } - ENV_UNLOCK(); if (!NIL_P(key)) { env_delete(key); @@ -6617,8 +6788,6 @@ env_update_block_i(VALUE key, VALUE val, VALUE _) * ENV.merge!(*hashes) -> ENV * ENV.merge!(*hashes) { |name, env_val, hash_val| block } -> ENV * - * ENV.update is an alias for ENV.merge!. - * * Adds to ENV each key/value pair in the given +hash+; returns ENV: * ENV.replace('foo' => '0', 'bar' => '1') * ENV.merge!('baz' => '2', 'bat' => '3') # => {"bar"=>"1", "bat"=>"3", "baz"=>"2", "foo"=>"0"} @@ -6708,70 +6877,71 @@ static const rb_data_type_t env_data_type = { }; /* - * A \Hash maps each of its unique keys to a specific value. + * A \Hash object maps each of its unique keys to a specific value. + * + * A hash has certain similarities to an Array, but: * - * A \Hash has certain similarities to an \Array, but: - * - An \Array index is always an \Integer. - * - A \Hash key can be (almost) any object. + * - An array index is always an integer. + * - A hash key can be (almost) any object. * * === \Hash \Data Syntax * - * The older syntax for \Hash data uses the "hash rocket," <tt>=></tt>: + * The original syntax for a hash entry uses the "hash rocket," <tt>=></tt>: * * h = {:foo => 0, :bar => 1, :baz => 2} - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} * - * Alternatively, but only for a \Hash key that's a \Symbol, + * Alternatively, but only for a key that's a symbol, * you can use a newer JSON-style syntax, - * where each bareword becomes a \Symbol: + * where each bareword becomes a symbol: * * h = {foo: 0, bar: 1, baz: 2} - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} * - * You can also use a \String in place of a bareword: + * You can also use a string in place of a bareword: * * h = {'foo': 0, 'bar': 1, 'baz': 2} - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} * * And you can mix the styles: * * h = {foo: 0, :bar => 1, 'baz': 2} - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} * * But it's an error to try the JSON-style syntax - * for a key that's not a bareword or a String: + * for a key that's not a bareword or a string: * * # Raises SyntaxError (syntax error, unexpected ':', expecting =>): * h = {0: 'zero'} * - * Hash value can be omitted, meaning that value will be fetched from the context + * The value can be omitted, meaning that value will be fetched from the context * by the name of the key: * * x = 0 * y = 100 * h = {x:, y:} - * h # => {:x=>0, :y=>100} + * h # => {x: 0, y: 100} * * === Common Uses * - * You can use a \Hash to give names to objects: + * You can use a hash to give names to objects: * * person = {name: 'Matz', language: 'Ruby'} - * person # => {:name=>"Matz", :language=>"Ruby"} + * person # => {name: "Matz", language: "Ruby"} * - * You can use a \Hash to give names to method arguments: + * You can use a hash to give names to method arguments: * * def some_method(hash) * p hash * end - * some_method({foo: 0, bar: 1, baz: 2}) # => {:foo=>0, :bar=>1, :baz=>2} + * some_method({foo: 0, bar: 1, baz: 2}) # => {foo: 0, bar: 1, baz: 2} * - * Note: when the last argument in a method call is a \Hash, + * Note: when the last argument in a method call is a hash, * the curly braces may be omitted: * - * some_method(foo: 0, bar: 1, baz: 2) # => {:foo=>0, :bar=>1, :baz=>2} + * some_method(foo: 0, bar: 1, baz: 2) # => {foo: 0, bar: 1, baz: 2} * - * You can use a \Hash to initialize an object: + * You can use a hash to initialize an object: * * class Dev * attr_accessor :name, :language @@ -6789,63 +6959,55 @@ static const rb_data_type_t env_data_type = { * * - A {hash literal}[rdoc-ref:syntax/literals.rdoc@Hash+Literals]. * - * You can convert certain objects to Hashes with: + * You can convert certain objects to hashes with: * - * - \Method #Hash. + * - Method Kernel#Hash. * - * You can create a \Hash by calling method Hash.new. - * - * Create an empty Hash: + * You can create a hash by calling method Hash.new: * + * # Create an empty hash. * h = Hash.new * h # => {} * h.class # => Hash * - * You can create a \Hash by calling method Hash.[]. - * - * Create an empty Hash: + * You can create a hash by calling method Hash.[]: * + * # Create an empty hash. * h = Hash[] * h # => {} - * - * Create a \Hash with initial entries: - * + * # Create a hash with initial entries. * h = Hash[foo: 0, bar: 1, baz: 2] - * h # => {:foo=>0, :bar=>1, :baz=>2} - * - * You can create a \Hash by using its literal form (curly braces). + * h # => {foo: 0, bar: 1, baz: 2} * - * Create an empty \Hash: + * You can create a hash by using its literal form (curly braces): * + * # Create an empty hash. * h = {} * h # => {} - * - * Create a \Hash with initial entries: - * + * # Create a +Hash+ with initial entries. * h = {foo: 0, bar: 1, baz: 2} - * h # => {:foo=>0, :bar=>1, :baz=>2} - * + * h # => {foo: 0, bar: 1, baz: 2} * * === \Hash Value Basics * - * The simplest way to retrieve a \Hash value (instance method #[]): + * The simplest way to retrieve a hash value (instance method #[]): * * h = {foo: 0, bar: 1, baz: 2} * h[:foo] # => 0 * - * The simplest way to create or update a \Hash value (instance method #[]=): + * The simplest way to create or update a hash value (instance method #[]=): * * h = {foo: 0, bar: 1, baz: 2} * h[:bat] = 3 # => 3 - * h # => {:foo=>0, :bar=>1, :baz=>2, :bat=>3} + * h # => {foo: 0, bar: 1, baz: 2, bat: 3} * h[:foo] = 4 # => 4 - * h # => {:foo=>4, :bar=>1, :baz=>2, :bat=>3} + * h # => {foo: 4, bar: 1, baz: 2, bat: 3} * - * The simplest way to delete a \Hash entry (instance method #delete): + * The simplest way to delete a hash entry (instance method #delete): * * h = {foo: 0, bar: 1, baz: 2} * h.delete(:bar) # => 1 - * h # => {:foo=>0, :baz=>2} + * h # => {foo: 0, baz: 2} * * === Entry Order * @@ -6853,41 +7015,41 @@ static const rb_data_type_t env_data_type = { * * - Iterative methods such as <tt>each</tt>, <tt>each_key</tt>, <tt>each_pair</tt>, <tt>each_value</tt>. * - Other order-sensitive methods such as <tt>shift</tt>, <tt>keys</tt>, <tt>values</tt>. - * - The \String returned by method <tt>inspect</tt>. + * - The string returned by method <tt>inspect</tt>. * - * A new \Hash has its initial ordering per the given entries: + * A new hash has its initial ordering per the given entries: * * h = Hash[foo: 0, bar: 1] - * h # => {:foo=>0, :bar=>1} + * h # => {foo: 0, bar: 1} * * New entries are added at the end: * * h[:baz] = 2 - * h # => {:foo=>0, :bar=>1, :baz=>2} + * h # => {foo: 0, bar: 1, baz: 2} * * Updating a value does not affect the order: * * h[:baz] = 3 - * h # => {:foo=>0, :bar=>1, :baz=>3} + * h # => {foo: 0, bar: 1, baz: 3} * * But re-creating a deleted entry can affect the order: * * h.delete(:foo) * h[:foo] = 5 - * h # => {:bar=>1, :baz=>3, :foo=>5} + * h # => {bar: 1, baz: 3, foo: 5} * - * === \Hash Keys + * === +Hash+ Keys * - * ==== \Hash Key Equivalence + * ==== +Hash+ Key Equivalence * * Two objects are treated as the same \hash key when their <code>hash</code> value * is identical and the two objects are <code>eql?</code> to each other. * - * ==== Modifying an Active \Hash Key + * ==== Modifying an Active +Hash+ Key * - * Modifying a \Hash key while it is in use damages the hash's index. + * Modifying a +Hash+ key while it is in use damages the hash's index. * - * This \Hash has keys that are Arrays: + * This +Hash+ has keys that are Arrays: * * a0 = [ :foo, :bar ] * a1 = [ :baz, :bat ] @@ -6901,7 +7063,7 @@ static const rb_data_type_t env_data_type = { * a0[0] = :bam * a0.hash # => 1069447059 * - * And damages the \Hash index: + * And damages the +Hash+ index: * * h.include?(a0) # => false * h[a0] # => nil @@ -6912,9 +7074,9 @@ static const rb_data_type_t env_data_type = { * h.include?(a0) # => true * h[a0] # => 0 * - * A \String key is always safe. - * That's because an unfrozen \String - * passed as a key will be replaced by a duplicated and frozen \String: + * A String key is always safe. + * That's because an unfrozen String + * passed as a key will be replaced by a duplicated and frozen String: * * s = 'foo' * s.frozen? # => false @@ -6922,15 +7084,15 @@ static const rb_data_type_t env_data_type = { * first_key = h.keys.first * first_key.frozen? # => true * - * ==== User-Defined \Hash Keys + * ==== User-Defined +Hash+ Keys * - * To be useable as a \Hash key, objects must implement the methods <code>hash</code> and <code>eql?</code>. - * Note: this requirement does not apply if the \Hash uses #compare_by_identity since comparison will then + * To be usable as a +Hash+ key, objects must implement the methods <code>hash</code> and <code>eql?</code>. + * Note: this requirement does not apply if the +Hash+ uses #compare_by_identity since comparison will then * rely on the keys' object id instead of <code>hash</code> and <code>eql?</code>. * - * \Object defines basic implementation for <code>hash</code> and <code>eq?</code> that makes each object + * Object defines basic implementation for <code>hash</code> and <code>eq?</code> that makes each object * a distinct key. Typically, user-defined classes will want to override these methods to provide meaningful - * behavior, or for example inherit \Struct that has useful definitions for these. + * behavior, or for example inherit Struct that has useful definitions for these. * * A typical implementation of <code>hash</code> is based on the * object's data while <code>eql?</code> is usually aliased to the overridden @@ -6953,7 +7115,7 @@ static const rb_data_type_t env_data_type = { * alias eql? == * * def hash - * @author.hash ^ @title.hash # XOR + * [self.class, @author, @title].hash * end * end * @@ -6967,99 +7129,98 @@ static const rb_data_type_t env_data_type = { * * reviews.length #=> 1 * - * === Default Values + * === Key Not Found? * - * The methods #[], #values_at and #dig need to return the value associated to a certain key. - * When that key is not found, that value will be determined by its default proc (if any) - * or else its default (initially `nil`). + * When a method tries to retrieve and return the value for a key and that key <i>is found</i>, + * the returned value is the value associated with the key. * - * You can retrieve the default value with method #default: + * But what if the key <i>is not found</i>? + * In that case, certain methods will return a default value while other will raise a \KeyError. * - * h = Hash.new - * h.default # => nil + * ==== Nil Return Value * - * You can set the default value by passing an argument to method Hash.new or - * with method #default= + * If you want +nil+ returned for a not-found key, you can call: * - * h = Hash.new(-1) - * h.default # => -1 - * h.default = 0 - * h.default # => 0 + * - #[](key) (usually written as <tt>#[key]</tt>. + * - #assoc(key). + * - #dig(key, *identifiers). + * - #values_at(*keys). * - * This default value is returned for #[], #values_at and #dig when a key is - * not found: + * You can override these behaviors for #[], #dig, and #values_at (but not #assoc); + * see {Hash Default}[rdoc-ref:Hash@Hash+Default]. * - * counts = {foo: 42} - * counts.default # => nil (default) - * counts[:foo] = 42 - * counts[:bar] # => nil - * counts.default = 0 - * counts[:bar] # => 0 - * counts.values_at(:foo, :bar, :baz) # => [42, 0, 0] - * counts.dig(:bar) # => 0 + * ==== \KeyError * - * Note that the default value is used without being duplicated. It is not advised to set - * the default value to a mutable object: + * If you want KeyError raised for a not-found key, you can call: * - * synonyms = Hash.new([]) - * synonyms[:hello] # => [] - * synonyms[:hello] << :hi # => [:hi], but this mutates the default! - * synonyms.default # => [:hi] - * synonyms[:world] << :universe - * synonyms[:world] # => [:hi, :universe], oops - * synonyms.keys # => [], oops + * - #fetch(key). + * - #fetch_values(*keys). * - * To use a mutable object as default, it is recommended to use a default proc + * ==== \Hash Default * - * ==== Default \Proc + * For certain methods (#[], #dig, and #values_at), + * the return value for a not-found key is determined by two hash properties: * - * When the default proc for a \Hash is set (i.e., not +nil+), - * the default value returned by method #[] is determined by the default proc alone. + * - <i>default value</i>: returned by method #default. + * - <i>default proc</i>: returned by method #default_proc. * - * You can retrieve the default proc with method #default_proc: + * In the simple case, both values are +nil+, + * and the methods return +nil+ for a not-found key; + * see {Nil Return Value}[rdoc-ref:Hash@Nil+Return+Value] above. * - * h = Hash.new - * h.default_proc # => nil + * Note that this entire section ("Hash Default"): * - * You can set the default proc by calling Hash.new with a block or - * calling the method #default_proc= + * - Applies _only_ to methods #[], #dig, and #values_at. + * - Does _not_ apply to methods #assoc, #fetch, or #fetch_values, + * which are not affected by the default value or default proc. * - * h = Hash.new { |hash, key| "Default value for #{key}" } - * h.default_proc.class # => Proc - * h.default_proc = proc { |hash, key| "Default value for #{key.inspect}" } - * h.default_proc.class # => Proc + * ===== Any-Key Default + * + * You can define an any-key default for a hash; + * that is, a value that will be returned for _any_ not-found key: + * + * - The value of #default_proc <i>must be</i> +nil+. + * - The value of #default (which may be any object, including +nil+) + * will be returned for a not-found key. * - * When the default proc is set (i.e., not +nil+) - * and method #[] is called with with a non-existent key, - * #[] calls the default proc with both the \Hash object itself and the missing key, - * then returns the proc's return value: + * You can set the default value when the hash is created with Hash.new and option +default_value+, + * or later with method #default=. * - * h = Hash.new { |hash, key| "Default value for #{key}" } - * h[:nosuch] # => "Default value for nosuch" + * Note: although the value of #default may be any object, + * it may not be a good idea to use a mutable object. * - * Note that in the example above no entry for key +:nosuch+ is created: + * ===== Per-Key Defaults * - * h.include?(:nosuch) # => false + * You can define a per-key default for a hash; + * that is, a Proc that will return a value based on the key itself. * - * However, the proc itself can add a new entry: + * You can set the default proc when the hash is created with Hash.new and a block, + * or later with method #default_proc=. * - * synonyms = Hash.new { |hash, key| hash[key] = [] } - * synonyms.include?(:hello) # => false - * synonyms[:hello] << :hi # => [:hi] - * synonyms[:world] << :universe # => [:universe] - * synonyms.keys # => [:hello, :world] + * Note that the proc can modify +self+, + * but modifying +self+ in this way is not thread-safe; + * multiple threads can concurrently call into the default proc + * for the same key. * - * Note that setting the default proc will clear the default value and vice versa. + * ==== \Method Default + * + * For two methods, you can specify a default value for a not-found key + * that has effect only for a single method call + * (and not for any subsequent calls): + * + * - For method #fetch, you can specify an any-key default: + * - For either method #fetch or method #fetch_values, + * you can specify a per-key default via a block. * * === What's Here * - * First, what's elsewhere. \Class \Hash: + * First, what's elsewhere. Class +Hash+: * * - Inherits from {class Object}[rdoc-ref:Object@What-27s+Here]. * - Includes {module Enumerable}[rdoc-ref:Enumerable@What-27s+Here], * which provides dozens of additional methods. * - * Here, class \Hash provides methods that are useful for: + * Here, class +Hash+ provides methods that are useful for: * * - {Creating a Hash}[rdoc-ref:Hash@Methods+for+Creating+a+Hash] * - {Setting Hash State}[rdoc-ref:Hash@Methods+for+Setting+Hash+State] @@ -7071,17 +7232,16 @@ static const rb_data_type_t env_data_type = { * - {Iterating}[rdoc-ref:Hash@Methods+for+Iterating] * - {Converting}[rdoc-ref:Hash@Methods+for+Converting] * - {Transforming Keys and Values}[rdoc-ref:Hash@Methods+for+Transforming+Keys+and+Values] - * - {And more....}[rdoc-ref:Hash@Other+Methods] * - * \Class \Hash also includes methods from module Enumerable. + * Class +Hash+ also includes methods from module Enumerable. * - * ==== Methods for Creating a \Hash + * ==== Methods for Creating a +Hash+ * * - ::[]: Returns a new hash populated with given objects. * - ::new: Returns a new empty hash. * - ::try_convert: Returns a new hash created from a given object. * - * ==== Methods for Setting \Hash State + * ==== Methods for Setting +Hash+ State * * - #compare_by_identity: Sets +self+ to consider only identity in comparing keys. * - #default=: Sets the default to a given value. @@ -7097,10 +7257,9 @@ static const rb_data_type_t env_data_type = { * - #empty?: Returns whether there are no entries. * - #eql?: Returns whether a given object is equal to +self+. * - #hash: Returns the integer hash code. - * - #has_value?: Returns whether a given object is a value in +self+. - * - #include?, #has_key?, #member?, #key?: Returns whether a given object is a key in +self+. - * - #length, #size: Returns the count of entries. - * - #value?: Returns whether a given object is a value in +self+. + * - #has_value? (aliased as #value?): Returns whether a given object is a value in +self+. + * - #include? (aliased as #has_key?, #member?, #key?): Returns whether a given object is a key in +self+. + * - #size (aliased as #length): Returns the count of entries. * * ==== Methods for Comparing * @@ -7108,7 +7267,7 @@ static const rb_data_type_t env_data_type = { * - #<=: Returns whether +self+ is a subset of a given object. * - #==: Returns whether a given object is equal to +self+. * - #>: Returns whether +self+ is a proper superset of a given object - * - #>=: Returns whether +self+ is a proper superset of a given object. + * - #>=: Returns whether +self+ is a superset of a given object. * * ==== Methods for Fetching * @@ -7121,16 +7280,16 @@ static const rb_data_type_t env_data_type = { * - #key: Returns the key for the first-found entry with a given value. * - #keys: Returns an array containing all keys in +self+. * - #rassoc: Returns a 2-element array consisting of the key and value - of the first-found entry having a given value. - * - #values: Returns an array containing all values in +self+/ + * of the first-found entry having a given value. + * - #values: Returns an array containing all values in +self+. * - #values_at: Returns an array containing values for given keys. * * ==== Methods for Assigning * - * - #[]=, #store: Associates a given key with a given value. + * - #[]= (aliased as #store): Associates a given key with a given value. * - #merge: Returns the hash formed by merging each given hash into a copy of +self+. - * - #merge!, #update: Merges each given hash into +self+. - * - #replace: Replaces the entire contents of +self+ with the contents of a given hash. + * - #update (aliased as #merge!): Merges each given hash into +self+. + * - #replace (aliased as #initialize_copy): Replaces the entire contents of +self+ with the contents of a given hash. * * ==== Methods for Deleting * @@ -7140,7 +7299,7 @@ static const rb_data_type_t env_data_type = { * - #compact!: Removes all +nil+-valued entries from +self+. * - #delete: Removes the entry for a given key. * - #delete_if: Removes entries selected by a given block. - * - #filter!, #select!: Keep only those entries selected by a given block. + * - #select! (aliased as #filter!): Keep only those entries selected by a given block. * - #keep_if: Keep only those entries selected by a given block. * - #reject!: Removes entries selected by a given block. * - #shift: Removes and returns the first entry. @@ -7149,36 +7308,34 @@ static const rb_data_type_t env_data_type = { * * - #compact: Returns a copy of +self+ with all +nil+-valued entries removed. * - #except: Returns a copy of +self+ with entries removed for specified keys. - * - #filter, #select: Returns a copy of +self+ with only those entries selected by a given block. + * - #select (aliased as #filter): Returns a copy of +self+ with only those entries selected by a given block. * - #reject: Returns a copy of +self+ with entries removed as specified by a given block. * - #slice: Returns a hash containing the entries for given keys. * * ==== Methods for Iterating - * - #each, #each_pair: Calls a given block with each key-value pair. + * - #each_pair (aliased as #each): Calls a given block with each key-value pair. * - #each_key: Calls a given block with each key. * - #each_value: Calls a given block with each value. * * ==== Methods for Converting * - * - #inspect, #to_s: Returns a new String containing the hash entries. + * - #flatten: Returns an array that is a 1-dimensional flattening of +self+. + * - #inspect (aliased as #to_s): Returns a new String containing the hash entries. * - #to_a: Returns a new array of 2-element arrays; * each nested array contains a key-value pair from +self+. - * - #to_h: Returns +self+ if a \Hash; - * if a subclass of \Hash, returns a \Hash containing the entries from +self+. + * - #to_h: Returns +self+ if a +Hash+; + * if a subclass of +Hash+, returns a +Hash+ containing the entries from +self+. * - #to_hash: Returns +self+. * - #to_proc: Returns a proc that maps a given key to its value. * * ==== Methods for Transforming Keys and Values * + * - #invert: Returns a hash with the each key-value pair inverted. * - #transform_keys: Returns a copy of +self+ with modified keys. * - #transform_keys!: Modifies keys in +self+ * - #transform_values: Returns a copy of +self+ with modified values. * - #transform_values!: Modifies values in +self+. * - * ==== Other Methods - * - #flatten: Returns an array that is a 1-dimensional flattening of +self+. - * - #invert: Returns a hash with the each key-value pair inverted. - * */ void @@ -7195,9 +7352,9 @@ Init_Hash(void) rb_define_alloc_func(rb_cHash, empty_hash_alloc); rb_define_singleton_method(rb_cHash, "[]", rb_hash_s_create, -1); rb_define_singleton_method(rb_cHash, "try_convert", rb_hash_s_try_convert, 1); - rb_define_method(rb_cHash, "initialize", rb_hash_initialize, -1); rb_define_method(rb_cHash, "initialize_copy", rb_hash_replace, 1); rb_define_method(rb_cHash, "rehash", rb_hash_rehash, 0); + rb_define_method(rb_cHash, "freeze", rb_hash_freeze, 0); rb_define_method(rb_cHash, "to_hash", rb_hash_to_hash, 0); rb_define_method(rb_cHash, "to_h", rb_hash_to_h, 0); @@ -7284,17 +7441,21 @@ Init_Hash(void) rb_define_singleton_method(rb_cHash, "ruby2_keywords_hash?", rb_hash_s_ruby2_keywords_hash_p, 1); rb_define_singleton_method(rb_cHash, "ruby2_keywords_hash", rb_hash_s_ruby2_keywords_hash, 1); + rb_cHash_empty_frozen = rb_hash_freeze(rb_hash_new()); + RB_OBJ_SET_SHAREABLE(rb_cHash_empty_frozen); + rb_vm_register_global_object(rb_cHash_empty_frozen); + /* Document-class: ENV * - * ENV is a hash-like accessor for environment variables. + * +ENV+ is a hash-like accessor for environment variables. * * === Interaction with the Operating System * - * The ENV object interacts with the operating system's environment variables: + * The +ENV+ object interacts with the operating system's environment variables: * - * - When you get the value for a name in ENV, the value is retrieved from among the current environment variables. - * - When you create or set a name-value pair in ENV, the name and value are immediately set in the environment variables. - * - When you delete a name-value pair in ENV, it is immediately deleted from the environment variables. + * - When you get the value for a name in +ENV+, the value is retrieved from among the current environment variables. + * - When you create or set a name-value pair in +ENV+, the name and value are immediately set in the environment variables. + * - When you delete a name-value pair in +ENV+, it is immediately deleted from the environment variables. * * === Names and Values * @@ -7344,33 +7505,33 @@ Init_Hash(void) * * === About Ordering * - * ENV enumerates its name/value pairs in the order found + * +ENV+ enumerates its name/value pairs in the order found * in the operating system's environment variables. - * Therefore the ordering of ENV content is OS-dependent, and may be indeterminate. + * Therefore the ordering of +ENV+ content is OS-dependent, and may be indeterminate. * * This will be seen in: - * - A Hash returned by an ENV method. - * - An Enumerator returned by an ENV method. + * - A Hash returned by an +ENV+ method. + * - An Enumerator returned by an +ENV+ method. * - An Array returned by ENV.keys, ENV.values, or ENV.to_a. * - The String returned by ENV.inspect. * - The Array returned by ENV.shift. * - The name returned by ENV.key. * * === About the Examples - * Some methods in ENV return ENV itself. Typically, there are many environment variables. - * It's not useful to display a large ENV in the examples here, - * so most example snippets begin by resetting the contents of ENV: - * - ENV.replace replaces ENV with a new collection of entries. - * - ENV.clear empties ENV. + * Some methods in +ENV+ return +ENV+ itself. Typically, there are many environment variables. + * It's not useful to display a large +ENV+ in the examples here, + * so most example snippets begin by resetting the contents of +ENV+: + * - ENV.replace replaces +ENV+ with a new collection of entries. + * - ENV.clear empties +ENV+. * - * == What's Here + * === What's Here * - * First, what's elsewhere. \Class \ENV: + * First, what's elsewhere. Class +ENV+: * * - Inherits from {class Object}[rdoc-ref:Object@What-27s+Here]. * - Extends {module Enumerable}[rdoc-ref:Enumerable@What-27s+Here], * - * Here, class \ENV provides methods that are useful for: + * Here, class +ENV+ provides methods that are useful for: * * - {Querying}[rdoc-ref:ENV@Methods+for+Querying] * - {Assigning}[rdoc-ref:ENV@Methods+for+Assigning] @@ -7379,26 +7540,26 @@ Init_Hash(void) * - {Converting}[rdoc-ref:ENV@Methods+for+Converting] * - {And more ....}[rdoc-ref:ENV@More+Methods] * - * === Methods for Querying + * ==== Methods for Querying * * - ::[]: Returns the value for the given environment variable name if it exists: - * - ::empty?: Returns whether \ENV is empty. - * - ::has_value?, ::value?: Returns whether the given value is in \ENV. + * - ::empty?: Returns whether +ENV+ is empty. + * - ::has_value?, ::value?: Returns whether the given value is in +ENV+. * - ::include?, ::has_key?, ::key?, ::member?: Returns whether the given name - is in \ENV. + is in +ENV+. * - ::key: Returns the name of the first entry with the given value. * - ::size, ::length: Returns the number of entries. * - ::value?: Returns whether any entry has the given value. * - * === Methods for Assigning + * ==== Methods for Assigning * * - ::[]=, ::store: Creates, updates, or deletes the named environment variable. - * - ::clear: Removes every environment variable; returns \ENV: - * - ::update, ::merge!: Adds to \ENV each key/value pair in the given hash. - * - ::replace: Replaces the entire content of the \ENV + * - ::clear: Removes every environment variable; returns +ENV+: + * - ::update, ::merge!: Adds to +ENV+ each key/value pair in the given hash. + * - ::replace: Replaces the entire content of the +ENV+ * with the name/value pairs in the given hash. * - * === Methods for Deleting + * ==== Methods for Deleting * * - ::delete: Deletes the named environment variable name if it exists. * - ::delete_if: Deletes entries selected by the block. @@ -7407,22 +7568,22 @@ Init_Hash(void) * - ::select!, ::filter!: Deletes entries selected by the block. * - ::shift: Removes and returns the first entry. * - * === Methods for Iterating + * ==== Methods for Iterating * * - ::each, ::each_pair: Calls the block with each name/value pair. * - ::each_key: Calls the block with each name. * - ::each_value: Calls the block with each value. * - * === Methods for Converting + * ==== Methods for Converting * * - ::assoc: Returns a 2-element array containing the name and value * of the named environment variable if it exists: - * - ::clone: Returns \ENV (and issues a warning). + * - ::clone: Raises an exception. * - ::except: Returns a hash of all name/value pairs except those given. * - ::fetch: Returns the value for the given name. - * - ::inspect: Returns the contents of \ENV as a string. - * - ::invert: Returns a hash whose keys are the ENV values, - and whose values are the corresponding ENV names. + * - ::inspect: Returns the contents of +ENV+ as a string. + * - ::invert: Returns a hash whose keys are the +ENV+ values, + and whose values are the corresponding +ENV+ names. * - ::keys: Returns an array of all names. * - ::rassoc: Returns the name and value of the first found entry * that has the given value. @@ -7436,11 +7597,11 @@ Init_Hash(void) * - ::values: Returns all values as an array. * - ::values_at: Returns an array of the values for the given name. * - * === More Methods + * ==== More Methods * * - ::dup: Raises an exception. * - ::freeze: Raises an exception. - * - ::rehash: Returns +nil+, without modifying \ENV. + * - ::rehash: Returns +nil+, without modifying +ENV+. * */ @@ -7451,8 +7612,7 @@ Init_Hash(void) origenviron = environ; envtbl = TypedData_Wrap_Struct(rb_cObject, &env_data_type, NULL); rb_extend_object(envtbl, rb_mEnumerable); - FL_SET_RAW(envtbl, RUBY_FL_SHAREABLE); - + RB_OBJ_SET_SHAREABLE(envtbl); rb_define_singleton_method(envtbl, "[]", rb_f_getenv, 1); rb_define_singleton_method(envtbl, "fetch", env_fetch, -1); @@ -7511,14 +7671,13 @@ Init_Hash(void) rb_undef_method(envtbl_class, "initialize_dup"); /* - * ENV is a Hash-like accessor for environment variables. + * +ENV+ is a Hash-like accessor for environment variables. * * See ENV (the class) for more details. */ rb_define_global_const("ENV", envtbl); - /* for callcc */ - ruby_register_rollback_func_for_ensure(hash_foreach_ensure, hash_foreach_ensure_rollback); - HASH_ASSERT(sizeof(ar_hint_t) * RHASH_AR_TABLE_MAX_SIZE == sizeof(VALUE)); } + +#include "hash.rbinc" |