diff options
Diffstat (limited to 'enum.c')
| -rw-r--r-- | enum.c | 3617 |
1 files changed, 2346 insertions, 1271 deletions
@@ -9,20 +9,39 @@ **********************************************************************/ -#include "ruby/encoding.h" +#include "id.h" #include "internal.h" +#include "internal/compar.h" +#include "internal/enum.h" +#include "internal/hash.h" +#include "internal/imemo.h" +#include "internal/numeric.h" +#include "internal/object.h" +#include "internal/proc.h" +#include "internal/rational.h" +#include "internal/re.h" #include "ruby/util.h" -#include "id.h" +#include "ruby_assert.h" #include "symbol.h" -#include "transient_heap.h" - -#include <assert.h> VALUE rb_mEnumerable; static ID id_next; -static ID id_div; - +static ID id__alone; +static ID id__separator; +static ID id_chunk_categorize; +static ID id_chunk_enumerable; +static ID id_sliceafter_enum; +static ID id_sliceafter_pat; +static ID id_sliceafter_pred; +static ID id_slicebefore_enumerable; +static ID id_slicebefore_sep_pat; +static ID id_slicebefore_sep_pred; +static ID id_slicewhen_enum; +static ID id_slicewhen_inverted; +static ID id_slicewhen_pred; + +#define id_div idDiv #define id_each idEach #define id_eqq idEqq #define id_cmp idCmp @@ -46,9 +65,9 @@ static VALUE enum_yield(int argc, VALUE ary) { if (argc > 1) - return rb_yield_force_blockarg(ary); + return rb_yield_force_blockarg(ary); if (argc == 1) - return rb_yield(ary); + return rb_yield(ary); return rb_yield_values2(0, 0); } @@ -58,9 +77,9 @@ enum_yield_array(VALUE ary) long len = RARRAY_LEN(ary); if (len > 1) - return rb_yield_force_blockarg(ary); + return rb_yield_force_blockarg(ary); if (len == 1) - return rb_yield(RARRAY_AREF(ary, 0)); + return rb_yield(RARRAY_AREF(ary, 0)); return rb_yield_values2(0, 0); } @@ -71,7 +90,23 @@ grep_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); if (RTEST(rb_funcallv(memo->v1, id_eqq, 1, &i)) == RTEST(memo->u3.value)) { - rb_ary_push(memo->v2, i); + rb_ary_push(memo->v2, i); + } + return Qnil; +} + +static VALUE +grep_regexp_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) +{ + struct MEMO *memo = MEMO_CAST(args); + VALUE converted_element, match; + ENUM_WANT_SVALUE(); + + /* In case element can't be converted to a Symbol or String: not a match (don't raise) */ + converted_element = SYMBOL_P(i) ? i : rb_check_string_type(i); + match = NIL_P(converted_element) ? Qfalse : rb_reg_match_p(memo->v1, i, 0); + if (match == memo->u3.value) { + rb_ary_push(memo->v2, i); } return Qnil; } @@ -83,64 +118,93 @@ grep_iter_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); if (RTEST(rb_funcallv(memo->v1, id_eqq, 1, &i)) == RTEST(memo->u3.value)) { - rb_ary_push(memo->v2, enum_yield(argc, i)); + rb_ary_push(memo->v2, enum_yield(argc, i)); } return Qnil; } +static VALUE +enum_grep0(VALUE obj, VALUE pat, VALUE test) +{ + VALUE ary = rb_ary_new(); + struct MEMO *memo = rb_imemo_memo_new(pat, ary, test); + rb_block_call_func_t fn; + if (rb_block_given_p()) { + fn = grep_iter_i; + } + else if (RB_TYPE_P(pat, T_REGEXP) && + LIKELY(rb_method_basic_definition_p(CLASS_OF(pat), idEqq))) { + fn = grep_regexp_i; + } + else { + fn = grep_i; + } + rb_block_call(obj, id_each, 0, 0, fn, (VALUE)memo); + + return ary; +} + /* - * call-seq: - * enum.grep(pattern) -> array - * enum.grep(pattern) { |obj| block } -> array + * call-seq: + * grep(pattern) -> array + * grep(pattern) {|element| ... } -> array + * + * Returns an array of objects based elements of +self+ that match the given pattern. + * + * With no block given, returns an array containing each element + * for which <tt>pattern === element</tt> is +true+: + * + * a = ['foo', 'bar', 'car', 'moo'] + * a.grep(/ar/) # => ["bar", "car"] + * (1..10).grep(3..8) # => [3, 4, 5, 6, 7, 8] + * ['a', 'b', 0, 1].grep(Integer) # => [0, 1] * - * Returns an array of every element in <i>enum</i> for which - * <code>Pattern === element</code>. If the optional <em>block</em> is - * supplied, each matching element is passed to it, and the block's - * result is stored in the output array. + * With a block given, + * calls the block with each matching element and returns an array containing each + * object returned by the block: * - * (1..100).grep 38..44 #=> [38, 39, 40, 41, 42, 43, 44] - * c = IO.constants - * c.grep(/SEEK/) #=> [:SEEK_SET, :SEEK_CUR, :SEEK_END] - * res = c.grep(/SEEK/) { |v| IO.const_get(v) } - * res #=> [0, 1, 2] + * a = ['foo', 'bar', 'car', 'moo'] + * a.grep(/ar/) {|element| element.upcase } # => ["BAR", "CAR"] * + * Related: #grep_v. */ static VALUE enum_grep(VALUE obj, VALUE pat) { - VALUE ary = rb_ary_new(); - struct MEMO *memo = MEMO_NEW(pat, ary, Qtrue); - - rb_block_call(obj, id_each, 0, 0, rb_block_given_p() ? grep_iter_i : grep_i, (VALUE)memo); - - return ary; + return enum_grep0(obj, pat, Qtrue); } /* - * call-seq: - * enum.grep_v(pattern) -> array - * enum.grep_v(pattern) { |obj| block } -> array + * call-seq: + * grep_v(pattern) -> array + * grep_v(pattern) {|element| ... } -> array + * + * Returns an array of objects based on elements of +self+ + * that <em>don't</em> match the given pattern. + * + * With no block given, returns an array containing each element + * for which <tt>pattern === element</tt> is +false+: + * + * a = ['foo', 'bar', 'car', 'moo'] + * a.grep_v(/ar/) # => ["foo", "moo"] + * (1..10).grep_v(3..8) # => [1, 2, 9, 10] + * ['a', 'b', 0, 1].grep_v(Integer) # => ["a", "b"] * - * Inverted version of Enumerable#grep. - * Returns an array of every element in <i>enum</i> for which - * not <code>Pattern === element</code>. + * With a block given, + * calls the block with each non-matching element and returns an array containing each + * object returned by the block: * - * (1..10).grep_v 2..5 #=> [1, 6, 7, 8, 9, 10] - * res =(1..10).grep_v(2..5) { |v| v * 2 } - * res #=> [2, 12, 14, 16, 18, 20] + * a = ['foo', 'bar', 'car', 'moo'] + * a.grep_v(/ar/) {|element| element.upcase } # => ["FOO", "MOO"] * + * Related: #grep. */ static VALUE enum_grep_v(VALUE obj, VALUE pat) { - VALUE ary = rb_ary_new(); - struct MEMO *memo = MEMO_NEW(pat, ary, Qfalse); - - rb_block_call(obj, id_each, 0, 0, rb_block_given_p() ? grep_iter_i : grep_i, (VALUE)memo); - - return ary; + return enum_grep0(obj, pat, Qfalse); } #define COUNT_BIGNUM IMEMO_FL_USER0 @@ -150,13 +214,13 @@ static void imemo_count_up(struct MEMO *memo) { if (memo->flags & COUNT_BIGNUM) { - MEMO_V3_SET(memo, rb_int_succ(memo->u3.value)); + MEMO_V3_SET(memo, rb_int_succ(memo->u3.value)); } else if (++memo->u3.cnt == 0) { - /* overflow */ - unsigned long buf[2] = {0, 1}; - MEMO_V3_SET(memo, rb_big_unpack(buf, 2)); - memo->flags |= COUNT_BIGNUM; + /* overflow */ + unsigned long buf[2] = {0, 1}; + MEMO_V3_SET(memo, rb_big_unpack(buf, 2)); + memo->flags |= COUNT_BIGNUM; } } @@ -164,10 +228,10 @@ static VALUE imemo_count_value(struct MEMO *memo) { if (memo->flags & COUNT_BIGNUM) { - return memo->u3.value; + return memo->u3.value; } else { - return ULONG2NUM(memo->u3.cnt); + return ULONG2NUM(memo->u3.cnt); } } @@ -179,7 +243,7 @@ count_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) ENUM_WANT_SVALUE(); if (rb_equal(i, memo->v1)) { - imemo_count_up(memo); + imemo_count_up(memo); } return Qnil; } @@ -190,7 +254,7 @@ count_iter_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) struct MEMO *memo = MEMO_CAST(memop); if (RTEST(rb_yield_values2(argc, argv))) { - imemo_count_up(memo); + imemo_count_up(memo); } return Qnil; } @@ -205,20 +269,28 @@ count_all_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) } /* - * call-seq: - * enum.count -> int - * enum.count(item) -> int - * enum.count { |obj| block } -> int + * call-seq: + * count -> integer + * count(object) -> integer + * count {|element| ... } -> integer + * + * Returns the count of elements, based on an argument or block criterion, if given. + * + * With no argument and no block given, returns the number of elements: * - * Returns the number of items in +enum+ through enumeration. - * If an argument is given, the number of items in +enum+ that - * are equal to +item+ are counted. If a block is given, it - * counts the number of elements yielding a true value. + * [0, 1, 2].count # => 3 + * {foo: 0, bar: 1, baz: 2}.count # => 3 * - * ary = [1, 2, 4, 2] - * ary.count #=> 4 - * ary.count(2) #=> 2 - * ary.count{ |x| x%2==0 } #=> 3 + * With argument +object+ given, + * returns the number of elements that are <tt>==</tt> to +object+: + * + * [0, 1, 2, 1].count(1) # => 2 + * + * With a block given, calls the block with each element + * and returns the number of elements for which the block returns a truthy value: + * + * [0, 1, 2, 3].count {|element| element < 2} # => 2 + * {foo: 0, bar: 1, baz: 2}.count {|key, value| value < 2} # => 2 * */ @@ -230,66 +302,78 @@ enum_count(int argc, VALUE *argv, VALUE obj) rb_block_call_func *func; if (argc == 0) { - if (rb_block_given_p()) { - func = count_iter_i; - } - else { - func = count_all_i; - } + if (rb_block_given_p()) { + func = count_iter_i; + } + else { + func = count_all_i; + } } else { - rb_scan_args(argc, argv, "1", &item); - if (rb_block_given_p()) { - rb_warn("given block not used"); - } + rb_scan_args(argc, argv, "1", &item); + if (rb_block_given_p()) { + rb_warn("given block not used"); + } func = count_i; } - memo = MEMO_NEW(item, 0, 0); + memo = rb_imemo_memo_new(item, 0, 0); rb_block_call(obj, id_each, 0, 0, func, (VALUE)memo); return imemo_count_value(memo); } +NORETURN(static void found(VALUE i, VALUE memop)); +static void +found(VALUE i, VALUE memop) +{ + struct MEMO *memo = MEMO_CAST(memop); + MEMO_V1_SET(memo, i); + memo->u3.cnt = 1; + rb_iter_break(); +} + +static VALUE +find_i_fast(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) +{ + if (RTEST(rb_yield_values2(argc, argv))) { + ENUM_WANT_SVALUE(); + found(i, memop); + } + return Qnil; +} + static VALUE find_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) { ENUM_WANT_SVALUE(); if (RTEST(enum_yield(argc, i))) { - struct MEMO *memo = MEMO_CAST(memop); - MEMO_V1_SET(memo, i); - memo->u3.cnt = 1; - rb_iter_break(); + found(i, memop); } return Qnil; } /* - * call-seq: - * enum.detect(ifnone = nil) { |obj| block } -> obj or nil - * enum.find(ifnone = nil) { |obj| block } -> obj or nil - * enum.detect(ifnone = nil) -> an_enumerator - * enum.find(ifnone = nil) -> an_enumerator + * call-seq: + * find(if_none_proc = nil) {|element| ... } -> object or nil + * find(if_none_proc = nil) -> enumerator + * + * Returns the first element for which the block returns a truthy value. * - * Passes each entry in <i>enum</i> to <em>block</em>. Returns the - * first for which <em>block</em> is not false. If no - * object matches, calls <i>ifnone</i> and returns its result when it - * is specified, or returns <code>nil</code> otherwise. + * With a block given, calls the block with successive elements of the collection; + * returns the first element for which the block returns a truthy value: * - * If no block is given, an enumerator is returned instead. + * (0..9).find {|element| element > 2} # => 3 * - * (1..100).detect #=> #<Enumerator: 1..100:detect> - * (1..100).find #=> #<Enumerator: 1..100:find> + * If no such element is found, calls +if_none_proc+ and returns its return value. * - * (1..10).detect { |i| i % 5 == 0 && i % 7 == 0 } #=> nil - * (1..10).find { |i| i % 5 == 0 && i % 7 == 0 } #=> nil - * (1..10).detect(-> {0}) { |i| i % 5 == 0 && i % 7 == 0 } #=> 0 - * (1..10).find(-> {0}) { |i| i % 5 == 0 && i % 7 == 0 } #=> 0 - * (1..100).detect { |i| i % 5 == 0 && i % 7 == 0 } #=> 35 - * (1..100).find { |i| i % 5 == 0 && i % 7 == 0 } #=> 35 + * (0..9).find(proc {false}) {|element| element > 12} # => false + * {foo: 0, bar: 1, baz: 2}.find {|key, value| key.start_with?('b') } # => [:bar, 1] + * {foo: 0, bar: 1, baz: 2}.find(proc {[]}) {|key, value| key.start_with?('c') } # => [] + * + * With no block given, returns an Enumerator. * */ - static VALUE enum_find(int argc, VALUE *argv, VALUE obj) { @@ -298,13 +382,16 @@ enum_find(int argc, VALUE *argv, VALUE obj) if_none = rb_check_arity(argc, 0, 1) ? argv[0] : Qnil; RETURN_ENUMERATOR(obj, argc, argv); - memo = MEMO_NEW(Qundef, 0, 0); - rb_block_call(obj, id_each, 0, 0, find_i, (VALUE)memo); + memo = rb_imemo_memo_new(Qundef, 0, 0); + if (rb_block_pair_yield_optimizable()) + rb_block_call2(obj, id_each, 0, 0, find_i_fast, (VALUE)memo, RB_BLOCK_NO_USE_PACKED_ARGS); + else + rb_block_call2(obj, id_each, 0, 0, find_i, (VALUE)memo, RB_BLOCK_NO_USE_PACKED_ARGS); if (memo->u3.cnt) { - return memo->v1; + return memo->v1; } if (!NIL_P(if_none)) { - return rb_funcallv(if_none, id_call, 0, 0); + return rb_funcallv(if_none, id_call, 0, 0); } return Qnil; } @@ -317,8 +404,8 @@ find_index_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) ENUM_WANT_SVALUE(); if (rb_equal(i, memo->v2)) { - MEMO_V1_SET(memo, imemo_count_value(memo)); - rb_iter_break(); + MEMO_V1_SET(memo, imemo_count_value(memo)); + rb_iter_break(); } imemo_count_up(memo); return Qnil; @@ -330,29 +417,34 @@ find_index_iter_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memop)) struct MEMO *memo = MEMO_CAST(memop); if (RTEST(rb_yield_values2(argc, argv))) { - MEMO_V1_SET(memo, imemo_count_value(memo)); - rb_iter_break(); + MEMO_V1_SET(memo, imemo_count_value(memo)); + rb_iter_break(); } imemo_count_up(memo); return Qnil; } /* - * call-seq: - * enum.find_index(value) -> int or nil - * enum.find_index { |obj| block } -> int or nil - * enum.find_index -> an_enumerator + * call-seq: + * find_index(object) -> integer or nil + * find_index {|element| ... } -> integer or nil + * find_index -> enumerator + * + * Returns the index of the first element that meets a specified criterion, + * or +nil+ if no such element is found. * - * Compares each entry in <i>enum</i> with <em>value</em> or passes - * to <em>block</em>. Returns the index for the first for which the - * evaluated value is non-false. If no object matches, returns - * <code>nil</code> + * With argument +object+ given, + * returns the index of the first element that is <tt>==</tt> +object+: * - * If neither block nor argument is given, an enumerator is returned instead. + * ['a', 'b', 'c', 'b'].find_index('b') # => 1 * - * (1..10).find_index { |i| i % 5 == 0 && i % 7 == 0 } #=> nil - * (1..100).find_index { |i| i % 5 == 0 && i % 7 == 0 } #=> 34 - * (1..100).find_index(50) #=> 49 + * With a block given, calls the block with successive elements; + * returns the first element for which the block returns a truthy value: + * + * ['a', 'b', 'c', 'b'].find_index {|element| element.start_with?('b') } # => 1 + * {foo: 0, bar: 1, baz: 2}.find_index {|key, value| value > 1 } # => 2 + * + * With no argument and no block given, returns an Enumerator. * */ @@ -368,14 +460,14 @@ enum_find_index(int argc, VALUE *argv, VALUE obj) func = find_index_iter_i; } else { - rb_scan_args(argc, argv, "1", &condition_value); - if (rb_block_given_p()) { - rb_warn("given block not used"); - } + rb_scan_args(argc, argv, "1", &condition_value); + if (rb_block_given_p()) { + rb_warn("given block not used"); + } func = find_index_i; } - memo = MEMO_NEW(Qnil, condition_value, 0); + memo = rb_imemo_memo_new(Qnil, condition_value, 0); rb_block_call(obj, id_each, 0, 0, func, (VALUE)memo); return memo->v1; } @@ -386,7 +478,7 @@ find_all_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) ENUM_WANT_SVALUE(); if (RTEST(enum_yield(argc, i))) { - rb_ary_push(ary, i); + rb_ary_push(ary, i); } return Qnil; } @@ -416,29 +508,23 @@ enum_size_over_p(VALUE obj, long n) } /* - * call-seq: - * enum.find_all { |obj| block } -> array - * enum.select { |obj| block } -> array - * enum.filter { |obj| block } -> array - * enum.find_all -> an_enumerator - * enum.select -> an_enumerator - * enum.filter -> an_enumerator - * - * Returns an array containing all elements of +enum+ - * for which the given +block+ returns a true value. - * - * If no block is given, an Enumerator is returned instead. + * call-seq: + * select {|element| ... } -> array + * select -> enumerator * + * Returns an array containing elements selected by the block. * - * (1..10).find_all { |i| i % 3 == 0 } #=> [3, 6, 9] + * With a block given, calls the block with successive elements; + * returns an array of those elements for which the block returns a truthy value: * - * [1,2,3,4,5].select { |num| num.even? } #=> [2, 4] + * (0..9).select {|element| element % 3 == 0 } # => [0, 3, 6, 9] + * a = {foo: 0, bar: 1, baz: 2}.select {|key, value| key.start_with?('b') } + * a # => {:bar=>1, :baz=>2} * - * [:foo, :bar].filter { |x| x == :foo } #=> [:foo] + * With no block given, returns an Enumerator. * - * See also Enumerable#reject, Enumerable#grep. + * Related: #reject. */ - static VALUE enum_find_all(VALUE obj) { @@ -465,17 +551,19 @@ filter_map_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) } /* - * call-seq: - * enum.filter_map { |obj| block } -> array - * enum.filter_map -> an_enumerator + * call-seq: + * filter_map {|element| ... } -> array + * filter_map -> enumerator * - * Returns a new array containing the truthy results (everything except - * +false+ or +nil+) of running the +block+ for every element in +enum+. + * Returns an array containing truthy elements returned by the block. * - * If no block is given, an Enumerator is returned instead. + * With a block given, calls the block with successive elements; + * returns an array containing each truthy value returned by the block: * + * (0..9).filter_map {|i| i * 2 if i.even? } # => [0, 4, 8, 12, 16] + * {foo: 0, bar: 1, baz: 2}.filter_map {|key, value| key if value.even? } # => [:foo, :baz] * - * (1..10).filter_map { |i| i * 2 if i.even? } #=> [4, 8, 12, 16, 20] + * When no block given, returns an Enumerator. * */ static VALUE @@ -498,26 +586,27 @@ reject_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) ENUM_WANT_SVALUE(); if (!RTEST(enum_yield(argc, i))) { - rb_ary_push(ary, i); + rb_ary_push(ary, i); } return Qnil; } /* - * call-seq: - * enum.reject { |obj| block } -> array - * enum.reject -> an_enumerator + * call-seq: + * reject {|element| ... } -> array + * reject -> enumerator * - * Returns an array for all elements of +enum+ for which the given - * +block+ returns <code>false</code>. + * Returns an array of objects rejected by the block. * - * If no block is given, an Enumerator is returned instead. + * With a block given, calls the block with successive elements; + * returns an array of those elements for which the block returns +nil+ or +false+: * - * (1..10).reject { |i| i % 3 == 0 } #=> [1, 2, 4, 5, 7, 8, 10] + * (0..9).reject {|i| i * 2 if i.even? } # => [1, 3, 5, 7, 9] + * {foo: 0, bar: 1, baz: 2}.reject {|key, value| key if value.odd? } # => {:foo=>0, :baz=>2} * - * [1, 2, 3, 4, 5].reject { |num| num.even? } #=> [1, 3, 5] + * When no block given, returns an Enumerator. * - * See also Enumerable#find_all. + * Related: #select. */ static VALUE @@ -544,29 +633,27 @@ collect_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) static VALUE collect_all(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) { - rb_thread_check_ints(); rb_ary_push(ary, rb_enum_values_pack(argc, argv)); return Qnil; } /* - * call-seq: - * enum.collect { |obj| block } -> array - * enum.map { |obj| block } -> array - * enum.collect -> an_enumerator - * enum.map -> an_enumerator + * call-seq: + * map {|element| ... } -> array + * map -> enumerator + * + * Returns an array of objects returned by the block. * - * Returns a new array with the results of running <em>block</em> once - * for every element in <i>enum</i>. + * With a block given, calls the block with successive elements; + * returns an array of the objects returned by the block: * - * If no block is given, an enumerator is returned instead. + * (0..4).map {|i| i*i } # => [0, 1, 4, 9, 16] + * {foo: 0, bar: 1, baz: 2}.map {|key, value| value*2} # => [0, 2, 4] * - * (1..4).map { |i| i*i } #=> [1, 4, 9, 16] - * (1..4).collect { "cat" } #=> ["cat", "cat", "cat", "cat"] + * With no block given, returns an Enumerator. * */ - static VALUE enum_collect(VALUE obj) { @@ -591,31 +678,33 @@ flat_map_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) tmp = rb_check_array_type(i); if (NIL_P(tmp)) { - rb_ary_push(ary, i); + rb_ary_push(ary, i); } else { - rb_ary_concat(ary, tmp); + rb_ary_concat(ary, tmp); } return Qnil; } /* - * call-seq: - * enum.flat_map { |obj| block } -> array - * enum.collect_concat { |obj| block } -> array - * enum.flat_map -> an_enumerator - * enum.collect_concat -> an_enumerator + * call-seq: + * flat_map {|element| ... } -> array + * flat_map -> enumerator * - * Returns a new array with the concatenated results of running - * <em>block</em> once for every element in <i>enum</i>. + * Returns an array of flattened objects returned by the block. * - * If no block is given, an enumerator is returned instead. + * With a block given, calls the block with successive elements; + * returns a flattened array of objects returned by the block: * - * [1, 2, 3, 4].flat_map { |e| [e, -e] } #=> [1, -1, 2, -2, 3, -3, 4, -4] - * [[1, 2], [3, 4]].flat_map { |e| e + [100] } #=> [1, 2, 100, 3, 4, 100] + * [0, 1, 2, 3].flat_map {|element| -element } # => [0, -1, -2, -3] + * [0, 1, 2, 3].flat_map {|element| [element, -element] } # => [0, 0, 1, -1, 2, -2, 3, -3] + * [[0, 1], [2, 3]].flat_map {|e| e + [100] } # => [0, 1, 100, 2, 3, 100] + * {foo: 0, bar: 1, baz: 2}.flat_map {|key, value| [key, value] } # => [:foo, 0, :bar, 1, :baz, 2] * + * With no block given, returns an Enumerator. + * + * Alias: #collect_concat. */ - static VALUE enum_flat_map(VALUE obj) { @@ -631,68 +720,68 @@ enum_flat_map(VALUE obj) /* * call-seq: - * enum.to_a(*args) -> array - * enum.entries(*args) -> array + * to_a(*args) -> array * - * Returns an array containing the items in <i>enum</i>. + * Returns an array containing the items in +self+: * - * (1..7).to_a #=> [1, 2, 3, 4, 5, 6, 7] - * { 'a'=>1, 'b'=>2, 'c'=>3 }.to_a #=> [["a", 1], ["b", 2], ["c", 3]] + * (0..4).to_a # => [0, 1, 2, 3, 4] * - * require 'prime' - * Prime.entries 10 #=> [2, 3, 5, 7] */ static VALUE enum_to_a(int argc, VALUE *argv, VALUE obj) { VALUE ary = rb_ary_new(); - rb_block_call(obj, id_each, argc, argv, collect_all, ary); - OBJ_INFECT(ary, obj); + rb_block_call_kw(obj, id_each, argc, argv, collect_all, ary, RB_PASS_CALLED_KEYWORDS); return ary; } static VALUE -enum_hashify(VALUE obj, int argc, const VALUE *argv, rb_block_call_func *iter) +enum_hashify_into(VALUE obj, int argc, const VALUE *argv, rb_block_call_func *iter, VALUE hash) { - VALUE hash = rb_hash_new(); rb_block_call(obj, id_each, argc, argv, iter, hash); - OBJ_INFECT(hash, obj); return hash; } static VALUE +enum_hashify(VALUE obj, int argc, const VALUE *argv, rb_block_call_func *iter) +{ + return enum_hashify_into(obj, argc, argv, iter, rb_hash_new()); +} + +static VALUE enum_to_h_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, hash)) { ENUM_WANT_SVALUE(); - rb_thread_check_ints(); return rb_hash_set_pair(hash, i); } static VALUE enum_to_h_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, hash)) { - rb_thread_check_ints(); return rb_hash_set_pair(hash, rb_yield_values2(argc, argv)); } /* * call-seq: - * enum.to_h(*args) -> hash - * enum.to_h(*args) {...} -> hash + * to_h(*args) -> hash + * to_h(*args) {|element| ... } -> hash + * + * When +self+ consists of 2-element arrays, + * returns a hash each of whose entries is the key-value pair + * formed from one of those arrays: * - * Returns the result of interpreting <i>enum</i> as a list of - * <tt>[key, value]</tt> pairs. + * [[:foo, 0], [:bar, 1], [:baz, 2]].to_h # => {:foo=>0, :bar=>1, :baz=>2} * - * %i[hello world].each_with_index.to_h - * # => {:hello => 0, :world => 1} + * When a block is given, the block is called with each element of +self+; + * the block should return a 2-element array which becomes a key-value pair + * in the returned hash: * - * If a block is given, the results of the block on each element of - * the enum will be used as pairs. + * (0..3).to_h {|i| [i, i ** 2]} # => {0=>0, 1=>1, 2=>4, 3=>9} * - * (1..5).to_h {|x| [x, x ** 2]} - * #=> {1=>1, 2=>4, 3=>9, 4=>16, 5=>25} + * Raises an exception if an element of +self+ is not a 2-element array, + * and a block is not passed. */ static VALUE @@ -709,11 +798,11 @@ inject_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, p)) ENUM_WANT_SVALUE(); - if (memo->v1 == Qundef) { - MEMO_V1_SET(memo, i); + if (UNDEF_P(memo->v1)) { + MEMO_V1_SET(memo, i); } else { - MEMO_V1_SET(memo, rb_yield_values(2, memo->v1, i)); + MEMO_V1_SET(memo, rb_yield_values(2, memo->v1, i)); } return Qnil; } @@ -726,18 +815,18 @@ inject_op_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, p)) ENUM_WANT_SVALUE(); - if (memo->v1 == Qundef) { - MEMO_V1_SET(memo, i); + if (UNDEF_P(memo->v1)) { + MEMO_V1_SET(memo, i); } else if (SYMBOL_P(name = memo->u3.value)) { - const ID mid = SYM2ID(name); - MEMO_V1_SET(memo, rb_funcallv(memo->v1, mid, 1, &i)); + const ID mid = SYM2ID(name); + MEMO_V1_SET(memo, rb_funcallv_public(memo->v1, mid, 1, &i)); } else { - VALUE args[2]; - args[0] = name; - args[1] = i; - MEMO_V1_SET(memo, rb_f_send(numberof(args), args, memo->v1)); + VALUE args[2]; + args[0] = name; + args[1] = i; + MEMO_V1_SET(memo, rb_f_send(numberof(args), args, memo->v1)); } return Qnil; } @@ -750,9 +839,9 @@ ary_inject_op(VALUE ary, VALUE init, VALUE op) long i, n; if (RARRAY_LEN(ary) == 0) - return init == Qundef ? Qnil : init; + return UNDEF_P(init) ? Qnil : init; - if (init == Qundef) { + if (UNDEF_P(init)) { v = RARRAY_AREF(ary, 0); i = 1; if (RARRAY_LEN(ary) == 1) @@ -765,20 +854,20 @@ ary_inject_op(VALUE ary, VALUE init, VALUE op) id = SYM2ID(op); if (id == idPLUS) { - if (RB_INTEGER_TYPE_P(v) && - rb_method_basic_definition_p(rb_cInteger, idPLUS) && - rb_obj_respond_to(v, idPLUS, FALSE)) { + if (RB_INTEGER_TYPE_P(v) && + rb_method_basic_definition_p(rb_cInteger, idPLUS) && + rb_obj_respond_to(v, idPLUS, FALSE)) { n = 0; for (; i < RARRAY_LEN(ary); i++) { e = RARRAY_AREF(ary, i); if (FIXNUM_P(e)) { n += FIX2LONG(e); /* should not overflow long type */ if (!FIXABLE(n)) { - v = rb_big_plus(ULONG2NUM(n), v); + v = rb_big_plus(LONG2NUM(n), v); n = 0; } } - else if (RB_TYPE_P(e, T_BIGNUM)) + else if (RB_BIGNUM_TYPE_P(e)) v = rb_big_plus(e, v); else goto not_integer; @@ -801,48 +890,152 @@ ary_inject_op(VALUE ary, VALUE init, VALUE op) /* * call-seq: - * enum.inject(initial, sym) -> obj - * enum.inject(sym) -> obj - * enum.inject(initial) { |memo, obj| block } -> obj - * enum.inject { |memo, obj| block } -> obj - * enum.reduce(initial, sym) -> obj - * enum.reduce(sym) -> obj - * enum.reduce(initial) { |memo, obj| block } -> obj - * enum.reduce { |memo, obj| block } -> obj - * - * Combines all elements of <i>enum</i> by applying a binary - * operation, specified by a block or a symbol that names a - * method or operator. - * - * The <i>inject</i> and <i>reduce</i> methods are aliases. There - * is no performance benefit to either. - * - * If you specify a block, then for each element in <i>enum</i> - * the block is passed an accumulator value (<i>memo</i>) and the element. - * If you specify a symbol instead, then each element in the collection - * will be passed to the named method of <i>memo</i>. - * In either case, the result becomes the new value for <i>memo</i>. - * At the end of the iteration, the final value of <i>memo</i> is the - * return value for the method. - * - * If you do not explicitly specify an <i>initial</i> value for <i>memo</i>, - * then the first element of collection is used as the initial value - * of <i>memo</i>. - * - * - * # Sum some numbers - * (5..10).reduce(:+) #=> 45 - * # Same using a block and inject - * (5..10).inject { |sum, n| sum + n } #=> 45 - * # Multiply some numbers - * (5..10).reduce(1, :*) #=> 151200 - * # Same using a block - * (5..10).inject(1) { |product, n| product * n } #=> 151200 - * # find the longest word - * longest = %w{ cat sheep bear }.inject do |memo, word| - * memo.length > word.length ? memo : word - * end - * longest #=> "sheep" + * inject(symbol) -> object + * inject(initial_value, symbol) -> object + * inject {|memo, value| ... } -> object + * inject(initial_value) {|memo, value| ... } -> object + * + * Returns the result of applying a reducer to an initial value and + * the first element of the Enumerable. It then takes the result and applies the + * function to it and the second element of the collection, and so on. The + * return value is the result returned by the final call to the function. + * + * You can think of + * + * [ a, b, c, d ].inject(i) { |r, v| fn(r, v) } + * + * as being + * + * fn(fn(fn(fn(i, a), b), c), d) + * + * In a way the +inject+ function _injects_ the function + * between the elements of the enumerable. + * + * +inject+ is aliased as +reduce+. You use it when you want to + * _reduce_ a collection to a single value. + * + * <b>The Calling Sequences</b> + * + * Let's start with the most verbose: + * + * enum.inject(initial_value) do |result, next_value| + * # do something with +result+ and +next_value+ + * # the value returned by the block becomes the + * # value passed in to the next iteration + * # as +result+ + * end + * + * For example: + * + * product = [ 2, 3, 4 ].inject(1) do |result, next_value| + * result * next_value + * end + * product #=> 24 + * + * When this runs, the block is first called with +1+ (the initial value) and + * +2+ (the first element of the array). The block returns <tt>1*2</tt>, so on + * the next iteration the block is called with +2+ (the previous result) and + * +3+. The block returns +6+, and is called one last time with +6+ and +4+. + * The result of the block, +24+ becomes the value returned by +inject+. This + * code returns the product of the elements in the enumerable. + * + * <b>First Shortcut: Default Initial value</b> + * + * In the case of the previous example, the initial value, +1+, wasn't really + * necessary: the calculation of the product of a list of numbers is self-contained. + * + * In these circumstances, you can omit the +initial_value+ parameter. +inject+ + * will then initially call the block with the first element of the collection + * as the +result+ parameter and the second element as the +next_value+. + * + * [ 2, 3, 4 ].inject do |result, next_value| + * result * next_value + * end + * + * This shortcut is convenient, but can only be used when the block produces a result + * which can be passed back to it as a first parameter. + * + * Here's an example where that's not the case: it returns a hash where the keys are words + * and the values are the number of occurrences of that word in the enumerable. + * + * freqs = File.read("README.md") + * .scan(/\w{2,}/) + * .reduce(Hash.new(0)) do |counts, word| + * counts[word] += 1 + * counts + * end + * freqs #=> {"Actions"=>4, + * "Status"=>5, + * "MinGW"=>3, + * "https"=>27, + * "github"=>10, + * "com"=>15, ... + * + * Note that the last line of the block is just the word +counts+. This ensures the + * return value of the block is the result that's being calculated. + * + * <b>Second Shortcut: a Reducer function</b> + * + * A <i>reducer function</i> is a function that takes a partial result and the next value, + * returning the next partial result. The block that is given to +inject+ is a reducer. + * + * You can also write a reducer as a function and pass the name of that function + * (as a symbol) to +inject+. However, for this to work, the function + * + * 1. Must be defined on the type of the result value + * 2. Must accept a single parameter, the next value in the collection, and + * 3. Must return an updated result which will also implement the function. + * + * Here's an example that adds elements to a string. The two calls invoke the functions + * String#concat and String#+ on the result so far, passing it the next value. + * + * s = [ "cat", " ", "dog" ].inject("", :concat) + * s #=> "cat dog" + * s = [ "cat", " ", "dog" ].inject("The result is:", :+) + * s #=> "The result is: cat dog" + * + * Here's a more complex example when the result object maintains + * state of a different type to the enumerable elements. + * + * class Turtle + * + * def initialize + * @x = @y = 0 + * end + * + * def move(dir) + * case dir + * when "n" then @y += 1 + * when "s" then @y -= 1 + * when "e" then @x += 1 + * when "w" then @x -= 1 + * end + * self + * end + * end + * + * position = "nnneesw".chars.reduce(Turtle.new, :move) + * position #=>> #<Turtle:0x00000001052f4698 @y=2, @x=1> + * + * <b>Third Shortcut: Reducer With no Initial Value</b> + * + * If your reducer returns a value that it can accept as a parameter, then you + * don't have to pass in an initial value. Here <tt>:*</tt> is the name of the + * _times_ function: + * + * product = [ 2, 3, 4 ].inject(:*) + * product # => 24 + * + * String concatenation again: + * + * s = [ "cat", " ", "dog" ].inject(:+) + * s #=> "cat dog" + * + * And an example that converts a hash to an array of two-element subarrays. + * + * nested = {foo: 0, bar: 1}.inject([], :push) + * nested # => [[:foo, 0], [:bar, 1]] + * * */ static VALUE @@ -852,28 +1045,36 @@ enum_inject(int argc, VALUE *argv, VALUE obj) VALUE init, op; rb_block_call_func *iter = inject_i; ID id; + int num_args; + + if (rb_block_given_p()) { + num_args = rb_scan_args(argc, argv, "02", &init, &op); + } + else { + num_args = rb_scan_args(argc, argv, "11", &init, &op); + } - switch (rb_scan_args(argc, argv, "02", &init, &op)) { + switch (num_args) { case 0: - init = Qundef; - break; + init = Qundef; + break; case 1: - if (rb_block_given_p()) { - break; - } - id = rb_check_id(&init); - op = id ? ID2SYM(id) : init; - init = Qundef; - iter = inject_op_i; - break; + if (rb_block_given_p()) { + break; + } + id = rb_check_id(&init); + op = id ? ID2SYM(id) : init; + init = Qundef; + iter = inject_op_i; + break; case 2: - if (rb_block_given_p()) { - rb_warning("given block not used"); - } - id = rb_check_id(&op); - if (id) op = ID2SYM(id); - iter = inject_op_i; - break; + if (rb_block_given_p()) { + rb_warning("given block not used"); + } + id = rb_check_id(&op); + if (id) op = ID2SYM(id); + iter = inject_op_i; + break; } if (iter == inject_op_i && @@ -883,9 +1084,9 @@ enum_inject(int argc, VALUE *argv, VALUE obj) return ary_inject_op(obj, init, op); } - memo = MEMO_NEW(init, Qnil, op); + memo = rb_imemo_memo_new(init, Qnil, op); rb_block_call(obj, id_each, 0, 0, iter, (VALUE)memo); - if (memo->v1 == Qundef) return Qnil; + if (UNDEF_P(memo->v1)) return Qnil; return memo->v1; } @@ -897,10 +1098,10 @@ partition_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, arys)) ENUM_WANT_SVALUE(); if (RTEST(enum_yield(argc, i))) { - ary = memo->v1; + ary = memo->v1; } else { - ary = memo->v2; + ary = memo->v2; } rb_ary_push(ary, i); return Qnil; @@ -908,16 +1109,29 @@ partition_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, arys)) /* * call-seq: - * enum.partition { |obj| block } -> [ true_array, false_array ] - * enum.partition -> an_enumerator + * partition {|element| ... } -> [true_array, false_array] + * partition -> enumerator + * + * With a block given, returns an array of two arrays: * - * Returns two arrays, the first containing the elements of - * <i>enum</i> for which the block evaluates to true, the second - * containing the rest. + * - The first having those elements for which the block returns a truthy value. + * - The other having all other elements. * - * If no block is given, an enumerator is returned instead. + * Examples: * - * (1..6).partition { |v| v.even? } #=> [[2, 4, 6], [1, 3, 5]] + * p = (1..4).partition {|i| i.even? } + * p # => [[2, 4], [1, 3]] + * p = ('a'..'d').partition {|c| c < 'c' } + * p # => [["a", "b"], ["c", "d"]] + * h = {foo: 0, bar: 1, baz: 2, bat: 3} + * p = h.partition {|key, value| key.start_with?('b') } + * p # => [[[:bar, 1], [:baz, 2], [:bat, 3]], [[:foo, 0]]] + * p = h.partition {|key, value| value < 2 } + * p # => [[[:foo, 0], [:bar, 1]], [[:baz, 2], [:bat, 3]]] + * + * With no block given, returns an Enumerator. + * + * Related: Enumerable#group_by. * */ @@ -928,7 +1142,7 @@ enum_partition(VALUE obj) RETURN_SIZED_ENUMERATOR(obj, 0, 0, enum_size); - memo = MEMO_NEW(rb_ary_new(), rb_ary_new(), 0); + memo = rb_imemo_memo_new(rb_ary_new(), rb_ary_new(), 0); rb_block_call(obj, id_each, 0, 0, partition_i, (VALUE)memo); return rb_assoc_new(memo->v1, memo->v2); @@ -945,27 +1159,34 @@ group_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, hash)) group = enum_yield(argc, i); values = rb_hash_aref(hash, group); if (!RB_TYPE_P(values, T_ARRAY)) { - values = rb_ary_new3(1, i); - rb_hash_aset(hash, group, values); + values = rb_ary_new3(1, i); + rb_hash_aset(hash, group, values); } else { - rb_ary_push(values, i); + rb_ary_push(values, i); } return Qnil; } /* * call-seq: - * enum.group_by { |obj| block } -> a_hash - * enum.group_by -> an_enumerator + * group_by {|element| ... } -> hash + * group_by -> enumerator + * + * With a block given returns a hash: + * + * - Each key is a return value from the block. + * - Each value is an array of those elements for which the block returned that key. * - * Groups the collection by result of the block. Returns a hash where the - * keys are the evaluated result from the block and the values are - * arrays of elements in the collection that correspond to the key. + * Examples: * - * If no block is given an enumerator is returned. + * g = (1..6).group_by {|i| i%3 } + * g # => {1=>[1, 4], 2=>[2, 5], 0=>[3, 6]} + * h = {foo: 0, bar: 1, baz: 0, bat: 1} + * g = h.group_by {|key, value| value } + * g # => {0=>[[:foo, 0], [:baz, 0]], 1=>[[:bar, 1], [:bat, 1]]} * - * (1..6).group_by { |i| i%3 } #=> {0=>[3, 6], 1=>[1, 4], 2=>[2, 5]} + * With no block given, returns an Enumerator. * */ @@ -977,47 +1198,105 @@ enum_group_by(VALUE obj) return enum_hashify(obj, 0, 0, group_by_i); } -static void -tally_up(VALUE hash, VALUE group) +static int +tally_up(st_data_t *group, st_data_t *value, st_data_t arg, int existing) { - VALUE tally = rb_hash_aref(hash, group); - if (NIL_P(tally)) { + VALUE tally = (VALUE)*value; + VALUE hash = (VALUE)arg; + if (!existing) { tally = INT2FIX(1); } else if (FIXNUM_P(tally) && tally < INT2FIX(FIXNUM_MAX)) { tally += INT2FIX(1) & ~FIXNUM_FLAG; } else { + Check_Type(tally, T_BIGNUM); tally = rb_big_plus(tally, INT2FIX(1)); + RB_OBJ_WRITTEN(hash, Qundef, tally); + } + *value = (st_data_t)tally; + return ST_CONTINUE; +} + +static VALUE +rb_enum_tally_up(VALUE hash, VALUE group) +{ + if (!rb_hash_stlike_update(hash, group, tally_up, (st_data_t)hash)) { + RB_OBJ_WRITTEN(hash, Qundef, group); } - rb_hash_aset(hash, group, tally); + return hash; } static VALUE tally_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, hash)) { ENUM_WANT_SVALUE(); - tally_up(hash, i); + rb_enum_tally_up(hash, i); return Qnil; } /* * call-seq: - * enum.tally -> a_hash + * tally(hash = {}) -> hash + * + * When argument +hash+ is not given, + * returns a new hash whose keys are the distinct elements in +self+; + * each integer value is the count of occurrences of each element: + * + * %w[a b c b c a c b].tally # => {"a"=>2, "b"=>3, "c"=>3} + * + * When argument +hash+ is given, + * returns +hash+, possibly augmented; for each element +ele+ in +self+: + * + * - Adds it as a key with a zero value if that key does not already exist: + * + * hash[ele] = 0 unless hash.include?(ele) + * + * - Increments the value of key +ele+: + * + * hash[ele] += 1 + * + * This is useful for accumulating tallies across multiple enumerables: * - * Tallys the collection. Returns a hash where the keys are the - * elements and the values are numbers of elements in the collection - * that correspond to the key. + * h = {} # => {} + * %w[a c d b c a].tally(h) # => {"a"=>2, "c"=>2, "d"=>1, "b"=>1} + * %w[b a z].tally(h) # => {"a"=>3, "c"=>2, "d"=>1, "b"=>2, "z"=>1} + * %w[b a m].tally(h) # => {"a"=>4, "c"=>2, "d"=>1, "b"=>3, "z"=>1, "m"=>1} + * + * The key to be added or found for an element depends on the class of +self+; + * see {Enumerable in Ruby Classes}[rdoc-ref:Enumerable@Enumerable+in+Ruby+Classes]. + * + * Examples: + * + * - Array (and certain array-like classes): + * the key is the element (as above). + * - Hash (and certain hash-like classes): + * the key is the 2-element array formed from the key-value pair: + * + * h = {} # => {} + * {foo: 'a', bar: 'b'}.tally(h) # => {[:foo, "a"]=>1, [:bar, "b"]=>1} + * {foo: 'c', bar: 'd'}.tally(h) # => {[:foo, "a"]=>1, [:bar, "b"]=>1, [:foo, "c"]=>1, [:bar, "d"]=>1} + * {foo: 'a', bar: 'b'}.tally(h) # => {[:foo, "a"]=>2, [:bar, "b"]=>2, [:foo, "c"]=>1, [:bar, "d"]=>1} + * {foo: 'c', bar: 'd'}.tally(h) # => {[:foo, "a"]=>2, [:bar, "b"]=>2, [:foo, "c"]=>2, [:bar, "d"]=>2} * - * ["a", "b", "c", "b"].tally #=> {"a"=>1, "b"=>2, "c"=>1} */ static VALUE -enum_tally(VALUE obj) +enum_tally(int argc, VALUE *argv, VALUE obj) { - return enum_hashify(obj, 0, 0, tally_i); + VALUE hash; + if (rb_check_arity(argc, 0, 1)) { + hash = rb_to_hash_type(argv[0]); + rb_check_frozen(hash); + } + else { + hash = rb_hash_new(); + } + + return enum_hashify_into(obj, 0, 0, tally_i, hash); } +NORETURN(static VALUE first_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, params))); static VALUE first_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, params)) { @@ -1034,18 +1313,26 @@ static VALUE enum_take(VALUE obj, VALUE n); /* * call-seq: - * enum.first -> obj or nil - * enum.first(n) -> an_array + * first -> element or nil + * first(n) -> array * - * Returns the first element, or the first +n+ elements, of the enumerable. - * If the enumerable is empty, the first form returns <code>nil</code>, and the - * second form returns an empty array. + * Returns the first element or elements. * - * %w[foo bar baz].first #=> "foo" - * %w[foo bar baz].first(2) #=> ["foo", "bar"] - * %w[foo bar baz].first(10) #=> ["foo", "bar", "baz"] - * [].first #=> nil - * [].first(10) #=> [] + * With no argument, returns the first element, or +nil+ if there is none: + * + * (1..4).first # => 1 + * %w[a b c].first # => "a" + * {foo: 1, bar: 1, baz: 2}.first # => [:foo, 1] + * [].first # => nil + * + * With integer argument +n+, returns an array + * containing the first +n+ elements that exist: + * + * (1..4).first(2) # => [1, 2] + * %w[a b c d].first(3) # => ["a", "b", "c"] + * %w[a b c d].first(50) # => ["a", "b", "c", "d"] + * {foo: 1, bar: 1, baz: 2}.first(2) # => [[:foo, 1], [:bar, 1]] + * [].first(2) # => [] * */ @@ -1055,37 +1342,44 @@ enum_first(int argc, VALUE *argv, VALUE obj) struct MEMO *memo; rb_check_arity(argc, 0, 1); if (argc > 0) { - return enum_take(obj, argv[0]); + return enum_take(obj, argv[0]); } else { - memo = MEMO_NEW(Qnil, 0, 0); - rb_block_call(obj, id_each, 0, 0, first_i, (VALUE)memo); - return memo->v1; + memo = rb_imemo_memo_new(Qnil, 0, 0); + rb_block_call(obj, id_each, 0, 0, first_i, (VALUE)memo); + return memo->v1; } } - /* * call-seq: - * enum.sort -> array - * enum.sort { |a, b| block } -> array + * sort -> array + * sort {|a, b| ... } -> array + * + * Returns an array containing the sorted elements of +self+. + * The ordering of equal elements is indeterminate and may be unstable. * - * Returns an array containing the items in <i>enum</i> sorted. + * With no block given, the sort compares + * using the elements' own method <tt>#<=></tt>: * - * Comparisons for the sort will be done using the items' own - * <code><=></code> operator or using an optional code block. + * %w[b c a d].sort # => ["a", "b", "c", "d"] + * {foo: 0, bar: 1, baz: 2}.sort # => [[:bar, 1], [:baz, 2], [:foo, 0]] * - * The block must implement a comparison between +a+ and +b+ and return - * an integer less than 0 when +b+ follows +a+, +0+ when +a+ and +b+ - * are equivalent, or an integer greater than 0 when +a+ follows +b+. + * With a block given, comparisons in the block determine the ordering. + * The block is called with two elements +a+ and +b+, and must return: * - * The result is not guaranteed to be stable. When the comparison of two - * elements returns +0+, the order of the elements is unpredictable. + * - A negative integer if <tt>a < b</tt>. + * - Zero if <tt>a == b</tt>. + * - A positive integer if <tt>a > b</tt>. * - * %w(rhea kea flea).sort #=> ["flea", "kea", "rhea"] - * (1..10).sort { |a, b| b <=> a } #=> [10, 9, 8, 7, 6, 5, 4, 3, 2, 1] + * Examples: * - * See also Enumerable#sort_by. It implements a Schwartzian transform + * a = %w[b c a d] + * a.sort {|a, b| b <=> a } # => ["d", "c", "b", "a"] + * h = {foo: 0, bar: 1, baz: 2} + * h.sort {|a, b| b <=> a } # => [[:foo, 0], [:baz, 2], [:bar, 1]] + * + * See also #sort_by. It implements a Schwartzian transform * which is useful when key computation or comparison is expensive. */ @@ -1096,10 +1390,12 @@ enum_sort(VALUE obj) } #define SORT_BY_BUFSIZE 16 +#define SORT_BY_UNIFORMED(num, flo, fix) (((num&1)<<2)|((flo&1)<<1)|fix) struct sort_by_data { const VALUE ary; const VALUE buf; - long n; + uint8_t n; + uint8_t primitive_uniformed; }; static VALUE @@ -1114,18 +1410,23 @@ sort_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _data)) v = enum_yield(argc, i); if (RBASIC(ary)->klass) { - rb_raise(rb_eRuntimeError, "sort_by reentered"); + rb_raise(rb_eRuntimeError, "sort_by reentered"); } if (RARRAY_LEN(data->buf) != SORT_BY_BUFSIZE*2) { - rb_raise(rb_eRuntimeError, "sort_by reentered"); + rb_raise(rb_eRuntimeError, "sort_by reentered"); } + if (data->primitive_uniformed) { + data->primitive_uniformed &= SORT_BY_UNIFORMED((FIXNUM_P(v)) || (RB_FLOAT_TYPE_P(v)), + RB_FLOAT_TYPE_P(v), + FIXNUM_P(v)); + } RARRAY_ASET(data->buf, data->n*2, v); RARRAY_ASET(data->buf, data->n*2+1, i); data->n++; if (data->n == SORT_BY_BUFSIZE) { - rb_ary_concat(ary, data->buf); - data->n = 0; + rb_ary_concat(ary, data->buf); + data->n = 0; } return Qnil; } @@ -1133,36 +1434,212 @@ sort_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _data)) static int sort_by_cmp(const void *ap, const void *bp, void *data) { - struct cmp_opt_data cmp_opt = { 0, 0 }; VALUE a; VALUE b; VALUE ary = (VALUE)data; if (RBASIC(ary)->klass) { - rb_raise(rb_eRuntimeError, "sort_by reentered"); + rb_raise(rb_eRuntimeError, "sort_by reentered"); } a = *(VALUE *)ap; b = *(VALUE *)bp; - return OPTIMIZED_CMP(a, b, cmp_opt); + return OPTIMIZED_CMP(a, b); } + +/* + This is parts of uniform sort +*/ + +#define uless rb_uniform_is_less +#define UNIFORM_SWAP(a,b)\ + do{struct rb_uniform_sort_data tmp = a; a = b; b = tmp;} while(0) + +struct rb_uniform_sort_data { + VALUE v; + VALUE i; +}; + +static inline bool +rb_uniform_is_less(VALUE a, VALUE b) +{ + + if (FIXNUM_P(a) && FIXNUM_P(b)) { + return (SIGNED_VALUE)a < (SIGNED_VALUE)b; + } + else if (FIXNUM_P(a)) { + RUBY_ASSERT(RB_FLOAT_TYPE_P(b)); + return rb_float_cmp(b, a) > 0; + } + else { + RUBY_ASSERT(RB_FLOAT_TYPE_P(a)); + return rb_float_cmp(a, b) < 0; + } +} + +static inline bool +rb_uniform_is_larger(VALUE a, VALUE b) +{ + + if (FIXNUM_P(a) && FIXNUM_P(b)) { + return (SIGNED_VALUE)a > (SIGNED_VALUE)b; + } + else if (FIXNUM_P(a)) { + RUBY_ASSERT(RB_FLOAT_TYPE_P(b)); + return rb_float_cmp(b, a) < 0; + } + else { + RUBY_ASSERT(RB_FLOAT_TYPE_P(a)); + return rb_float_cmp(a, b) > 0; + } +} + +#define med3_val(a,b,c) (uless(a,b)?(uless(b,c)?b:uless(c,a)?a:c):(uless(c,b)?b:uless(a,c)?a:c)) + +static void +rb_uniform_insertionsort_2(struct rb_uniform_sort_data* ptr_begin, + struct rb_uniform_sort_data* ptr_end) +{ + if ((ptr_end - ptr_begin) < 2) return; + struct rb_uniform_sort_data tmp, *j, *k, + *index = ptr_begin+1; + for (; index < ptr_end; index++) { + tmp = *index; + j = k = index; + if (uless(tmp.v, ptr_begin->v)) { + while (ptr_begin < j) { + *j = *(--k); + j = k; + } + } + else { + while (uless(tmp.v, (--k)->v)) { + *j = *k; + j = k; + } + } + *j = tmp; + } +} + +static inline void +rb_uniform_heap_down_2(struct rb_uniform_sort_data* ptr_begin, + size_t offset, size_t len) +{ + size_t c; + struct rb_uniform_sort_data tmp = ptr_begin[offset]; + while ((c = (offset<<1)+1) <= len) { + if (c < len && uless(ptr_begin[c].v, ptr_begin[c+1].v)) { + c++; + } + if (!uless(tmp.v, ptr_begin[c].v)) break; + ptr_begin[offset] = ptr_begin[c]; + offset = c; + } + ptr_begin[offset] = tmp; +} + +static void +rb_uniform_heapsort_2(struct rb_uniform_sort_data* ptr_begin, + struct rb_uniform_sort_data* ptr_end) +{ + size_t n = ptr_end - ptr_begin; + if (n < 2) return; + + for (size_t offset = n>>1; offset > 0;) { + rb_uniform_heap_down_2(ptr_begin, --offset, n-1); + } + for (size_t offset = n-1; offset > 0;) { + UNIFORM_SWAP(*ptr_begin, ptr_begin[offset]); + rb_uniform_heap_down_2(ptr_begin, 0, --offset); + } +} + + +static void +rb_uniform_quicksort_intro_2(struct rb_uniform_sort_data* ptr_begin, + struct rb_uniform_sort_data* ptr_end, size_t d) +{ + + if (ptr_end - ptr_begin <= 16) { + rb_uniform_insertionsort_2(ptr_begin, ptr_end); + return; + } + if (d == 0) { + rb_uniform_heapsort_2(ptr_begin, ptr_end); + return; + } + + VALUE x = med3_val(ptr_begin->v, + ptr_begin[(ptr_end - ptr_begin)>>1].v, + ptr_end[-1].v); + struct rb_uniform_sort_data *i = ptr_begin; + struct rb_uniform_sort_data *j = ptr_end-1; + + do { + while (uless(i->v, x)) i++; + while (uless(x, j->v)) j--; + if (i <= j) { + UNIFORM_SWAP(*i, *j); + i++; + j--; + } + } while (i <= j); + j++; + if (ptr_end - j > 1) rb_uniform_quicksort_intro_2(j, ptr_end, d-1); + if (i - ptr_begin > 1) rb_uniform_quicksort_intro_2(ptr_begin, i, d-1); +} + +/** + * Direct primitive data compare sort. Implement with intro sort. + * @param[in] ptr_begin The begin address of target rb_ary's raw pointer. + * @param[in] ptr_end The end address of target rb_ary's raw pointer. +**/ +static void +rb_uniform_intro_sort_2(struct rb_uniform_sort_data* ptr_begin, + struct rb_uniform_sort_data* ptr_end) +{ + size_t n = ptr_end - ptr_begin; + size_t d = CHAR_BIT * sizeof(n) - nlz_intptr(n) - 1; + bool sorted_flag = true; + + for (struct rb_uniform_sort_data* ptr = ptr_begin+1; ptr < ptr_end; ptr++) { + if (rb_uniform_is_larger((ptr-1)->v, (ptr)->v)) { + sorted_flag = false; + break; + } + } + + if (sorted_flag) { + return; + } + rb_uniform_quicksort_intro_2(ptr_begin, ptr_end, d<<1); +} + +#undef uless + + /* * call-seq: - * enum.sort_by { |obj| block } -> array - * enum.sort_by -> an_enumerator + * sort_by {|element| ... } -> array + * sort_by -> enumerator * - * Sorts <i>enum</i> using a set of keys generated by mapping the - * values in <i>enum</i> through the given block. + * With a block given, returns an array of elements of +self+, + * sorted according to the value returned by the block for each element. + * The ordering of equal elements is indeterminate and may be unstable. * - * The result is not guaranteed to be stable. When two keys are equal, - * the order of the corresponding elements is unpredictable. + * Examples: * - * If no block is given, an enumerator is returned instead. + * a = %w[xx xxx x xxxx] + * a.sort_by {|s| s.size } # => ["x", "xx", "xxx", "xxxx"] + * a.sort_by {|s| -s.size } # => ["xxxx", "xxx", "xx", "x"] + * h = {foo: 2, bar: 1, baz: 0} + * h.sort_by{|key, value| value } # => [[:baz, 0], [:bar, 1], [:foo, 2]] + * h.sort_by{|key, value| key } # => [[:bar, 1], [:baz, 0], [:foo, 2]] * - * %w{apple pear fig}.sort_by { |word| word.length } - * #=> ["fig", "pear", "apple"] + * With no block given, returns an Enumerator. * * The current implementation of #sort_by generates an array of * tuples containing the original collection element and the mapped @@ -1237,48 +1714,59 @@ enum_sort_by(VALUE obj) RETURN_SIZED_ENUMERATOR(obj, 0, 0, enum_size); if (RB_TYPE_P(obj, T_ARRAY) && RARRAY_LEN(obj) <= LONG_MAX/2) { - ary = rb_ary_new2(RARRAY_LEN(obj)*2); + ary = rb_ary_new2(RARRAY_LEN(obj)*2); } else { - ary = rb_ary_new(); + ary = rb_ary_new(); } RBASIC_CLEAR_CLASS(ary); - buf = rb_ary_tmp_new(SORT_BY_BUFSIZE*2); + buf = rb_ary_hidden_new(SORT_BY_BUFSIZE*2); rb_ary_store(buf, SORT_BY_BUFSIZE*2-1, Qnil); - memo = MEMO_NEW(0, 0, 0); - OBJ_INFECT(memo, obj); + memo = rb_imemo_memo_new(0, 0, 0); data = (struct sort_by_data *)&memo->v1; RB_OBJ_WRITE(memo, &data->ary, ary); RB_OBJ_WRITE(memo, &data->buf, buf); data->n = 0; + data->primitive_uniformed = SORT_BY_UNIFORMED((CMP_OPTIMIZABLE(FLOAT) && CMP_OPTIMIZABLE(INTEGER)), + CMP_OPTIMIZABLE(FLOAT), + CMP_OPTIMIZABLE(INTEGER)); rb_block_call(obj, id_each, 0, 0, sort_by_i, (VALUE)memo); ary = data->ary; buf = data->buf; if (data->n) { - rb_ary_resize(buf, data->n*2); - rb_ary_concat(ary, buf); + rb_ary_resize(buf, data->n*2); + rb_ary_concat(ary, buf); } if (RARRAY_LEN(ary) > 2) { - RARRAY_PTR_USE(ary, ptr, - ruby_qsort(ptr, RARRAY_LEN(ary)/2, 2*sizeof(VALUE), - sort_by_cmp, (void *)ary)); + if (data->primitive_uniformed) { + RARRAY_PTR_USE(ary, ptr, + rb_uniform_intro_sort_2((struct rb_uniform_sort_data*)ptr, + (struct rb_uniform_sort_data*)(ptr + RARRAY_LEN(ary)))); + } + else { + RARRAY_PTR_USE(ary, ptr, + ruby_qsort(ptr, RARRAY_LEN(ary)/2, 2*sizeof(VALUE), + sort_by_cmp, (void *)ary)); + } } if (RBASIC(ary)->klass) { - rb_raise(rb_eRuntimeError, "sort_by reentered"); + rb_raise(rb_eRuntimeError, "sort_by reentered"); } for (i=1; i<RARRAY_LEN(ary); i+=2) { - RARRAY_ASET(ary, i/2, RARRAY_AREF(ary, i)); + RARRAY_ASET(ary, i/2, RARRAY_AREF(ary, i)); } rb_ary_resize(ary, RARRAY_LEN(ary)/2); RBASIC_SET_CLASS_RAW(ary, rb_cArray); - OBJ_INFECT(ary, memo); return ary; } #define ENUMFUNC(name) argc ? name##_eqq : rb_block_given_p() ? name##_iter_i : name##_i -#define MEMO_ENUM_NEW(v1) (rb_check_arity(argc, 0, 1), MEMO_NEW((v1), (argc ? *argv : 0), 0)) +#define ENUM_BLOCK_CALL(name) \ + rb_block_call2(obj, id_each, 0, 0, ENUMFUNC(name), (VALUE)memo, rb_block_given_p() && rb_block_pair_yield_optimizable() ? RB_BLOCK_NO_USE_PACKED_ARGS : 0); + +#define MEMO_ENUM_NEW(v1) (rb_check_arity(argc, 0, 1), rb_imemo_memo_new((v1), (argc ? *argv : 0), 0)) #define DEFINE_ENUMFUNCS(name) \ static VALUE enum_##name##_func(VALUE result, struct MEMO *memo); \ @@ -1314,33 +1802,55 @@ enum_##name##_func(VALUE result, struct MEMO *memo) DEFINE_ENUMFUNCS(all) { if (!RTEST(result)) { - MEMO_V1_SET(memo, Qfalse); - rb_iter_break(); + MEMO_V1_SET(memo, Qfalse); + rb_iter_break(); } return Qnil; } /* * call-seq: - * enum.all? [{ |obj| block } ] -> true or false - * enum.all?(pattern) -> true or false - * - * Passes each element of the collection to the given block. The method - * returns <code>true</code> if the block never returns - * <code>false</code> or <code>nil</code>. If the block is not given, - * Ruby adds an implicit block of <code>{ |obj| obj }</code> which will - * cause #all? to return +true+ when none of the collection members are - * +false+ or +nil+. - * - * If instead a pattern is supplied, the method returns whether - * <code>pattern === element</code> for every collection member. - * - * %w[ant bear cat].all? { |word| word.length >= 3 } #=> true - * %w[ant bear cat].all? { |word| word.length >= 4 } #=> false - * %w[ant bear cat].all?(/t/) #=> false - * [1, 2i, 3.14].all?(Numeric) #=> true - * [nil, true, 99].all? #=> false - * [].all? #=> true + * all? -> true or false + * all?(pattern) -> true or false + * all? {|element| ... } -> true or false + * + * Returns whether every element meets a given criterion. + * + * If +self+ has no element, returns +true+ and argument or block + * are not used. + * + * With no argument and no block, + * returns whether every element is truthy: + * + * (1..4).all? # => true + * %w[a b c d].all? # => true + * [1, 2, nil].all? # => false + * ['a','b', false].all? # => false + * [].all? # => true + * + * With argument +pattern+ and no block, + * returns whether for each element +element+, + * <tt>pattern === element</tt>: + * + * (1..4).all?(Integer) # => true + * (1..4).all?(Numeric) # => true + * (1..4).all?(Float) # => false + * %w[bar baz bat bam].all?(/ba/) # => true + * %w[bar baz bat bam].all?(/bar/) # => false + * %w[bar baz bat bam].all?('ba') # => false + * {foo: 0, bar: 1, baz: 2}.all?(Array) # => true + * {foo: 0, bar: 1, baz: 2}.all?(Hash) # => false + * [].all?(Integer) # => true + * + * With a block given, returns whether the block returns a truthy value + * for every element: + * + * (1..4).all? {|element| element < 5 } # => true + * (1..4).all? {|element| element < 4 } # => false + * {foo: 0, bar: 1, baz: 2}.all? {|key, value| value < 3 } # => true + * {foo: 0, bar: 1, baz: 2}.all? {|key, value| value < 2 } # => false + * + * Related: #any?, #none? #one?. * */ @@ -1349,41 +1859,61 @@ enum_all(int argc, VALUE *argv, VALUE obj) { struct MEMO *memo = MEMO_ENUM_NEW(Qtrue); WARN_UNUSED_BLOCK(argc); - rb_block_call(obj, id_each, 0, 0, ENUMFUNC(all), (VALUE)memo); + ENUM_BLOCK_CALL(all); return memo->v1; } DEFINE_ENUMFUNCS(any) { if (RTEST(result)) { - MEMO_V1_SET(memo, Qtrue); - rb_iter_break(); + MEMO_V1_SET(memo, Qtrue); + rb_iter_break(); } return Qnil; } /* * call-seq: - * enum.any? [{ |obj| block }] -> true or false - * enum.any?(pattern) -> true or false - * - * Passes each element of the collection to the given block. The method - * returns <code>true</code> if the block ever returns a value other - * than <code>false</code> or <code>nil</code>. If the block is not - * given, Ruby adds an implicit block of <code>{ |obj| obj }</code> that - * will cause #any? to return +true+ if at least one of the collection - * members is not +false+ or +nil+. - * - * If instead a pattern is supplied, the method returns whether - * <code>pattern === element</code> for any collection member. - * - * %w[ant bear cat].any? { |word| word.length >= 3 } #=> true - * %w[ant bear cat].any? { |word| word.length >= 4 } #=> true - * %w[ant bear cat].any?(/d/) #=> false - * [nil, true, 99].any?(Integer) #=> true - * [nil, true, 99].any? #=> true - * [].any? #=> false - * + * any? -> true or false + * any?(pattern) -> true or false + * any? {|element| ... } -> true or false + * + * Returns whether any element meets a given criterion. + * + * If +self+ has no element, returns +false+ and argument or block + * are not used. + * + * With no argument and no block, + * returns whether any element is truthy: + * + * (1..4).any? # => true + * %w[a b c d].any? # => true + * [1, false, nil].any? # => true + * [].any? # => false + * + * With argument +pattern+ and no block, + * returns whether for any element +element+, + * <tt>pattern === element</tt>: + * + * [nil, false, 0].any?(Integer) # => true + * [nil, false, 0].any?(Numeric) # => true + * [nil, false, 0].any?(Float) # => false + * %w[bar baz bat bam].any?(/m/) # => true + * %w[bar baz bat bam].any?(/foo/) # => false + * %w[bar baz bat bam].any?('ba') # => false + * {foo: 0, bar: 1, baz: 2}.any?(Array) # => true + * {foo: 0, bar: 1, baz: 2}.any?(Hash) # => false + * [].any?(Integer) # => false + * + * With a block given, returns whether the block returns a truthy value + * for any element: + * + * (1..4).any? {|element| element < 2 } # => true + * (1..4).any? {|element| element < 1 } # => false + * {foo: 0, bar: 1, baz: 2}.any? {|key, value| value < 1 } # => true + * {foo: 0, bar: 1, baz: 2}.any? {|key, value| value < 0 } # => false + * + * Related: #all?, #none?, #one?. */ static VALUE @@ -1391,20 +1921,20 @@ enum_any(int argc, VALUE *argv, VALUE obj) { struct MEMO *memo = MEMO_ENUM_NEW(Qfalse); WARN_UNUSED_BLOCK(argc); - rb_block_call(obj, id_each, 0, 0, ENUMFUNC(any), (VALUE)memo); + ENUM_BLOCK_CALL(any); return memo->v1; } DEFINE_ENUMFUNCS(one) { if (RTEST(result)) { - if (memo->v1 == Qundef) { - MEMO_V1_SET(memo, Qtrue); - } - else if (memo->v1 == Qtrue) { - MEMO_V1_SET(memo, Qfalse); - rb_iter_break(); - } + if (UNDEF_P(memo->v1)) { + MEMO_V1_SET(memo, Qtrue); + } + else if (memo->v1 == Qtrue) { + MEMO_V1_SET(memo, Qfalse); + rb_iter_break(); + } } return Qnil; } @@ -1434,11 +1964,10 @@ cmpint_reenter_check(struct nmin_data *data, VALUE val) static int nmin_cmp(const void *ap, const void *bp, void *_data) { - struct cmp_opt_data cmp_opt = { 0, 0 }; struct nmin_data *data = (struct nmin_data *)_data; VALUE a = *(const VALUE *)ap, b = *(const VALUE *)bp; #define rb_cmpint(cmp, a, b) rb_cmpint(cmpint_reenter_check(data, (cmp)), a, b) - return OPTIMIZED_CMP(a, b, cmp_opt); + return OPTIMIZED_CMP(a, b); #undef rb_cmpint } @@ -1466,7 +1995,7 @@ nmin_filter(struct nmin_data *data) long i, j; if (data->curlen <= data->n) - return; + return; n = data->n; beg = RARRAY_PTR(data->buf); @@ -1486,46 +2015,46 @@ nmin_filter(struct nmin_data *data) } while (0) while (1) { - long pivot_index = left + (right-left)/2; - long num_pivots = 1; - - SWAP(pivot_index, right); - pivot_index = right; - - store_index = left; - i = left; - while (i <= right-num_pivots) { - int c = data->cmpfunc(GETPTR(i), GETPTR(pivot_index), data); - if (data->rev) - c = -c; - if (c == 0) { - SWAP(i, right-num_pivots); - num_pivots++; - continue; - } - if (c < 0) { - SWAP(i, store_index); - store_index++; - } - i++; - } - j = store_index; - for (i = right; right-num_pivots < i; i--) { - if (i <= j) - break; - SWAP(j, i); - j++; - } - - if (store_index <= n && n <= store_index+num_pivots) - break; - - if (n < store_index) { - right = store_index-1; - } - else { - left = store_index+num_pivots; - } + long pivot_index = left + (right-left)/2; + long num_pivots = 1; + + SWAP(pivot_index, right); + pivot_index = right; + + store_index = left; + i = left; + while (i <= right-num_pivots) { + int c = data->cmpfunc(GETPTR(i), GETPTR(pivot_index), data); + if (data->rev) + c = -c; + if (c == 0) { + SWAP(i, right-num_pivots); + num_pivots++; + continue; + } + if (c < 0) { + SWAP(i, store_index); + store_index++; + } + i++; + } + j = store_index; + for (i = right; right-num_pivots < i; i--) { + if (i <= j) + break; + SWAP(j, i); + j++; + } + + if (store_index <= n && n <= store_index+num_pivots) + break; + + if (n < store_index) { + right = store_index-1; + } + else { + left = store_index+num_pivots; + } } #undef GETPTR #undef SWAP @@ -1536,7 +2065,7 @@ nmin_filter(struct nmin_data *data) } static VALUE -nmin_i(VALUE i, VALUE *_data, int argc, VALUE *argv) +nmin_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _data)) { struct nmin_data *data = (struct nmin_data *)_data; VALUE cmpv; @@ -1544,11 +2073,11 @@ nmin_i(VALUE i, VALUE *_data, int argc, VALUE *argv) ENUM_WANT_SVALUE(); if (data->by) - cmpv = enum_yield(argc, i); + cmpv = enum_yield(argc, i); else - cmpv = i; + cmpv = i; - if (data->limit != Qundef) { + if (!UNDEF_P(data->limit)) { int c = data->cmpfunc(&cmpv, &data->limit, data); if (data->rev) c = -c; @@ -1557,13 +2086,13 @@ nmin_i(VALUE i, VALUE *_data, int argc, VALUE *argv) } if (data->by) - rb_ary_push(data->buf, cmpv); + rb_ary_push(data->buf, cmpv); rb_ary_push(data->buf, i); data->curlen++; if (data->curlen == data->bufmax) { - nmin_filter(data); + nmin_filter(data); } return Qnil; @@ -1584,28 +2113,28 @@ rb_nmin_run(VALUE obj, VALUE num, int by, int rev, int ary) rb_raise(rb_eArgError, "too big size"); data.bufmax = data.n * 4; data.curlen = 0; - data.buf = rb_ary_tmp_new(data.bufmax * (by ? 2 : 1)); + data.buf = rb_ary_hidden_new(data.bufmax * (by ? 2 : 1)); data.limit = Qundef; data.cmpfunc = by ? nmin_cmp : rb_block_given_p() ? nmin_block_cmp : - nmin_cmp; + nmin_cmp; data.rev = rev; data.by = by; if (ary) { - long i; - for (i = 0; i < RARRAY_LEN(obj); i++) { - VALUE args[1]; - args[0] = RARRAY_AREF(obj, i); - nmin_i(obj, (VALUE*)&data, 1, args); - } + long i; + for (i = 0; i < RARRAY_LEN(obj); i++) { + VALUE args[1]; + args[0] = RARRAY_AREF(obj, i); + nmin_i(obj, (VALUE)&data, 1, args, Qundef); + } } else { - rb_block_call(obj, id_each, 0, 0, nmin_i, (VALUE)&data); + rb_block_call(obj, id_each, 0, 0, nmin_i, (VALUE)&data); } nmin_filter(&data); result = data.buf; if (by) { - long i; + long i; RARRAY_PTR_USE(result, ptr, { ruby_qsort(ptr, RARRAY_LEN(result)/2, @@ -1615,7 +2144,7 @@ rb_nmin_run(VALUE obj, VALUE num, int by, int rev, int ary) ptr[i/2] = ptr[i]; } }); - rb_ary_resize(result, RARRAY_LEN(result)/2); + rb_ary_resize(result, RARRAY_LEN(result)/2); } else { RARRAY_PTR_USE(result, ptr, { @@ -1633,26 +2162,45 @@ rb_nmin_run(VALUE obj, VALUE num, int by, int rev, int ary) /* * call-seq: - * enum.one? [{ |obj| block }] -> true or false - * enum.one?(pattern) -> true or false - * - * Passes each element of the collection to the given block. The method - * returns <code>true</code> if the block returns <code>true</code> - * exactly once. If the block is not given, <code>one?</code> will return - * <code>true</code> only if exactly one of the collection members is - * true. - * - * If instead a pattern is supplied, the method returns whether - * <code>pattern === element</code> for exactly one collection member. - * - * %w{ant bear cat}.one? { |word| word.length == 4 } #=> true - * %w{ant bear cat}.one? { |word| word.length > 4 } #=> false - * %w{ant bear cat}.one? { |word| word.length < 4 } #=> false - * %w{ant bear cat}.one?(/t/) #=> false - * [ nil, true, 99 ].one? #=> false - * [ nil, true, false ].one? #=> true - * [ nil, true, 99 ].one?(Integer) #=> true - * [].one? #=> false + * one? -> true or false + * one?(pattern) -> true or false + * one? {|element| ... } -> true or false + * + * Returns whether exactly one element meets a given criterion. + * + * With no argument and no block, + * returns whether exactly one element is truthy: + * + * (1..1).one? # => true + * [1, nil, false].one? # => true + * (1..4).one? # => false + * {foo: 0}.one? # => true + * {foo: 0, bar: 1}.one? # => false + * [].one? # => false + * + * With argument +pattern+ and no block, + * returns whether for exactly one element +element+, + * <tt>pattern === element</tt>: + * + * [nil, false, 0].one?(Integer) # => true + * [nil, false, 0].one?(Numeric) # => true + * [nil, false, 0].one?(Float) # => false + * %w[bar baz bat bam].one?(/m/) # => true + * %w[bar baz bat bam].one?(/foo/) # => false + * %w[bar baz bat bam].one?('ba') # => false + * {foo: 0, bar: 1, baz: 2}.one?(Array) # => false + * {foo: 0}.one?(Array) # => true + * [].one?(Integer) # => false + * + * With a block given, returns whether the block returns a truthy value + * for exactly one element: + * + * (1..4).one? {|element| element < 2 } # => true + * (1..4).one? {|element| element < 1 } # => false + * {foo: 0, bar: 1, baz: 2}.one? {|key, value| value < 1 } # => true + * {foo: 0, bar: 1, baz: 2}.one? {|key, value| value < 2 } # => false + * + * Related: #none?, #all?, #any?. * */ static VALUE @@ -1662,42 +2210,60 @@ enum_one(int argc, VALUE *argv, VALUE obj) VALUE result; WARN_UNUSED_BLOCK(argc); - rb_block_call(obj, id_each, 0, 0, ENUMFUNC(one), (VALUE)memo); + ENUM_BLOCK_CALL(one); result = memo->v1; - if (result == Qundef) return Qfalse; + if (UNDEF_P(result)) return Qfalse; return result; } DEFINE_ENUMFUNCS(none) { if (RTEST(result)) { - MEMO_V1_SET(memo, Qfalse); - rb_iter_break(); + MEMO_V1_SET(memo, Qfalse); + rb_iter_break(); } return Qnil; } /* * call-seq: - * enum.none? [{ |obj| block }] -> true or false - * enum.none?(pattern) -> true or false - * - * Passes each element of the collection to the given block. The method - * returns <code>true</code> if the block never returns <code>true</code> - * for all elements. If the block is not given, <code>none?</code> will return - * <code>true</code> only if none of the collection members is true. - * - * If instead a pattern is supplied, the method returns whether - * <code>pattern === element</code> for none of the collection members. - * - * %w{ant bear cat}.none? { |word| word.length == 5 } #=> true - * %w{ant bear cat}.none? { |word| word.length >= 4 } #=> false - * %w{ant bear cat}.none?(/d/) #=> true - * [1, 3.14, 42].none?(Float) #=> false - * [].none? #=> true - * [nil].none? #=> true - * [nil, false].none? #=> true - * [nil, false, true].none? #=> false + * none? -> true or false + * none?(pattern) -> true or false + * none? {|element| ... } -> true or false + * + * Returns whether no element meets a given criterion. + * + * With no argument and no block, + * returns whether no element is truthy: + * + * (1..4).none? # => false + * [nil, false].none? # => true + * {foo: 0}.none? # => false + * {foo: 0, bar: 1}.none? # => false + * [].none? # => true + * + * With argument +pattern+ and no block, + * returns whether for no element +element+, + * <tt>pattern === element</tt>: + * + * [nil, false, 1.1].none?(Integer) # => true + * %w[bar baz bat bam].none?(/m/) # => false + * %w[bar baz bat bam].none?(/foo/) # => true + * %w[bar baz bat bam].none?('ba') # => true + * {foo: 0, bar: 1, baz: 2}.none?(Hash) # => true + * {foo: 0}.none?(Array) # => false + * [].none?(Integer) # => true + * + * With a block given, returns whether the block returns a truthy value + * for no element: + * + * (1..4).none? {|element| element < 1 } # => true + * (1..4).none? {|element| element < 2 } # => false + * {foo: 0, bar: 1, baz: 2}.none? {|key, value| value < 0 } # => true + * {foo: 0, bar: 1, baz: 2}.none? {|key, value| value < 1 } # => false + * + * Related: #one?, #all?, #any?. + * */ static VALUE enum_none(int argc, VALUE *argv, VALUE obj) @@ -1705,13 +2271,12 @@ enum_none(int argc, VALUE *argv, VALUE obj) struct MEMO *memo = MEMO_ENUM_NEW(Qtrue); WARN_UNUSED_BLOCK(argc); - rb_block_call(obj, id_each, 0, 0, ENUMFUNC(none), (VALUE)memo); + ENUM_BLOCK_CALL(none); return memo->v1; } struct min_t { VALUE min; - struct cmp_opt_data cmp_opt; }; static VALUE @@ -1721,13 +2286,13 @@ min_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); - if (memo->min == Qundef) { - memo->min = i; + if (UNDEF_P(memo->min)) { + memo->min = i; } else { - if (OPTIMIZED_CMP(i, memo->min, memo->cmp_opt) < 0) { - memo->min = i; - } + if (OPTIMIZED_CMP(i, memo->min) < 0) { + memo->min = i; + } } return Qnil; } @@ -1740,14 +2305,14 @@ min_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); - if (memo->min == Qundef) { - memo->min = i; + if (UNDEF_P(memo->min)) { + memo->min = i; } else { - cmp = rb_yield_values(2, i, memo->min); - if (rb_cmpint(cmp, i, memo->min) < 0) { - memo->min = i; - } + cmp = rb_yield_values(2, i, memo->min); + if (rb_cmpint(cmp, i, memo->min) < 0) { + memo->min = i; + } } return Qnil; } @@ -1755,33 +2320,66 @@ min_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) /* * call-seq: - * enum.min -> obj - * enum.min { |a, b| block } -> obj - * enum.min(n) -> array - * enum.min(n) { |a, b| block } -> array - * - * Returns the object in _enum_ with the minimum value. The - * first form assumes all objects implement Comparable; - * the second uses the block to return <em>a <=> b</em>. - * - * a = %w(albatross dog horse) - * a.min #=> "albatross" - * a.min { |a, b| a.length <=> b.length } #=> "dog" - * - * If the +n+ argument is given, minimum +n+ elements are returned - * as a sorted array. - * - * a = %w[albatross dog horse] - * a.min(2) #=> ["albatross", "dog"] - * a.min(2) {|a, b| a.length <=> b.length } #=> ["dog", "horse"] - * [5, 1, 3, 4, 2].min(3) #=> [1, 2, 3] + * min -> element + * min(n) -> array + * min {|a, b| ... } -> element + * min(n) {|a, b| ... } -> array + * + * Returns the element with the minimum element according to a given criterion. + * The ordering of equal elements is indeterminate and may be unstable. + * + * With no argument and no block, returns the minimum element, + * using the elements' own method <tt>#<=></tt> for comparison: + * + * (1..4).min # => 1 + * (-4..-1).min # => -4 + * %w[d c b a].min # => "a" + * {foo: 0, bar: 1, baz: 2}.min # => [:bar, 1] + * [].min # => nil + * + * With positive integer argument +n+ given, and no block, + * returns an array containing the first +n+ minimum elements that exist: + * + * (1..4).min(2) # => [1, 2] + * (-4..-1).min(2) # => [-4, -3] + * %w[d c b a].min(2) # => ["a", "b"] + * {foo: 0, bar: 1, baz: 2}.min(2) # => [[:bar, 1], [:baz, 2]] + * [].min(2) # => [] + * + * With a block given, the block determines the minimum elements. + * The block is called with two elements +a+ and +b+, and must return: + * + * - A negative integer if <tt>a < b</tt>. + * - Zero if <tt>a == b</tt>. + * - A positive integer if <tt>a > b</tt>. + * + * With a block given and no argument, + * returns the minimum element as determined by the block: + * + * %w[xxx x xxxx xx].min {|a, b| a.size <=> b.size } # => "x" + * h = {foo: 0, bar: 1, baz: 2} + * h.min {|pair1, pair2| pair1[1] <=> pair2[1] } # => [:foo, 0] + * [].min {|a, b| a <=> b } # => nil + * + * With a block given and positive integer argument +n+ given, + * returns an array containing the first +n+ minimum elements that exist, + * as determined by the block. + * + * %w[xxx x xxxx xx].min(2) {|a, b| a.size <=> b.size } # => ["x", "xx"] + * h = {foo: 0, bar: 1, baz: 2} + * h.min(2) {|pair1, pair2| pair1[1] <=> pair2[1] } + * # => [[:foo, 0], [:bar, 1]] + * [].min(2) {|a, b| a <=> b } # => [] + * + * Related: #min_by, #minmax, #max. + * */ static VALUE enum_min(int argc, VALUE *argv, VALUE obj) { VALUE memo; - struct min_t *m = NEW_CMP_OPT_MEMO(struct min_t, memo); + struct min_t *m = NEW_MEMO_FOR(struct min_t, memo); VALUE result; VALUE num; @@ -1789,22 +2387,19 @@ enum_min(int argc, VALUE *argv, VALUE obj) return rb_nmin_run(obj, num, 0, 0, 0); m->min = Qundef; - m->cmp_opt.opt_methods = 0; - m->cmp_opt.opt_inited = 0; if (rb_block_given_p()) { - rb_block_call(obj, id_each, 0, 0, min_ii, memo); + rb_block_call(obj, id_each, 0, 0, min_ii, memo); } else { - rb_block_call(obj, id_each, 0, 0, min_i, memo); + rb_block_call(obj, id_each, 0, 0, min_i, memo); } result = m->min; - if (result == Qundef) return Qnil; + if (UNDEF_P(result)) return Qnil; return result; } struct max_t { VALUE max; - struct cmp_opt_data cmp_opt; }; static VALUE @@ -1814,13 +2409,13 @@ max_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); - if (memo->max == Qundef) { - memo->max = i; + if (UNDEF_P(memo->max)) { + memo->max = i; } else { - if (OPTIMIZED_CMP(i, memo->max, memo->cmp_opt) > 0) { - memo->max = i; - } + if (OPTIMIZED_CMP(i, memo->max) > 0) { + memo->max = i; + } } return Qnil; } @@ -1833,47 +2428,80 @@ max_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); - if (memo->max == Qundef) { - memo->max = i; + if (UNDEF_P(memo->max)) { + memo->max = i; } else { - cmp = rb_yield_values(2, i, memo->max); - if (rb_cmpint(cmp, i, memo->max) > 0) { - memo->max = i; - } + cmp = rb_yield_values(2, i, memo->max); + if (rb_cmpint(cmp, i, memo->max) > 0) { + memo->max = i; + } } return Qnil; } /* * call-seq: - * enum.max -> obj - * enum.max { |a, b| block } -> obj - * enum.max(n) -> array - * enum.max(n) { |a, b| block } -> array - * - * Returns the object in _enum_ with the maximum value. The - * first form assumes all objects implement Comparable; - * the second uses the block to return <em>a <=> b</em>. - * - * a = %w(albatross dog horse) - * a.max #=> "horse" - * a.max { |a, b| a.length <=> b.length } #=> "albatross" - * - * If the +n+ argument is given, maximum +n+ elements are returned - * as an array, sorted in descending order. - * - * a = %w[albatross dog horse] - * a.max(2) #=> ["horse", "dog"] - * a.max(2) {|a, b| a.length <=> b.length } #=> ["albatross", "horse"] - * [5, 1, 3, 4, 2].max(3) #=> [5, 4, 3] + * max -> element + * max(n) -> array + * max {|a, b| ... } -> element + * max(n) {|a, b| ... } -> array + * + * Returns the element with the maximum element according to a given criterion. + * The ordering of equal elements is indeterminate and may be unstable. + * + * With no argument and no block, returns the maximum element, + * using the elements' own method <tt>#<=></tt> for comparison: + * + * (1..4).max # => 4 + * (-4..-1).max # => -1 + * %w[d c b a].max # => "d" + * {foo: 0, bar: 1, baz: 2}.max # => [:foo, 0] + * [].max # => nil + * + * With positive integer argument +n+ given, and no block, + * returns an array containing the first +n+ maximum elements that exist: + * + * (1..4).max(2) # => [4, 3] + * (-4..-1).max(2) # => [-1, -2] + * %w[d c b a].max(2) # => ["d", "c"] + * {foo: 0, bar: 1, baz: 2}.max(2) # => [[:foo, 0], [:baz, 2]] + * [].max(2) # => [] + * + * With a block given, the block determines the maximum elements. + * The block is called with two elements +a+ and +b+, and must return: + * + * - A negative integer if <tt>a < b</tt>. + * - Zero if <tt>a == b</tt>. + * - A positive integer if <tt>a > b</tt>. + * + * With a block given and no argument, + * returns the maximum element as determined by the block: + * + * %w[xxx x xxxx xx].max {|a, b| a.size <=> b.size } # => "xxxx" + * h = {foo: 0, bar: 1, baz: 2} + * h.max {|pair1, pair2| pair1[1] <=> pair2[1] } # => [:baz, 2] + * [].max {|a, b| a <=> b } # => nil + * + * With a block given and positive integer argument +n+ given, + * returns an array containing the first +n+ maximum elements that exist, + * as determined by the block. + * + * %w[xxx x xxxx xx].max(2) {|a, b| a.size <=> b.size } # => ["xxxx", "xxx"] + * h = {foo: 0, bar: 1, baz: 2} + * h.max(2) {|pair1, pair2| pair1[1] <=> pair2[1] } + * # => [[:baz, 2], [:bar, 1]] + * [].max(2) {|a, b| a <=> b } # => [] + * + * Related: #min, #minmax, #max_by. + * */ static VALUE enum_max(int argc, VALUE *argv, VALUE obj) { VALUE memo; - struct max_t *m = NEW_CMP_OPT_MEMO(struct max_t, memo); + struct max_t *m = NEW_MEMO_FOR(struct max_t, memo); VALUE result; VALUE num; @@ -1881,16 +2509,14 @@ enum_max(int argc, VALUE *argv, VALUE obj) return rb_nmin_run(obj, num, 0, 1, 0); m->max = Qundef; - m->cmp_opt.opt_methods = 0; - m->cmp_opt.opt_inited = 0; if (rb_block_given_p()) { - rb_block_call(obj, id_each, 0, 0, max_ii, (VALUE)memo); + rb_block_call(obj, id_each, 0, 0, max_ii, (VALUE)memo); } else { - rb_block_call(obj, id_each, 0, 0, max_i, (VALUE)memo); + rb_block_call(obj, id_each, 0, 0, max_i, (VALUE)memo); } result = m->max; - if (result == Qundef) return Qnil; + if (UNDEF_P(result)) return Qnil; return result; } @@ -1898,7 +2524,6 @@ struct minmax_t { VALUE min; VALUE max; VALUE last; - struct cmp_opt_data cmp_opt; }; static void @@ -1906,19 +2531,19 @@ minmax_i_update(VALUE i, VALUE j, struct minmax_t *memo) { int n; - if (memo->min == Qundef) { - memo->min = i; - memo->max = j; + if (UNDEF_P(memo->min)) { + memo->min = i; + memo->max = j; } else { - n = OPTIMIZED_CMP(i, memo->min, memo->cmp_opt); - if (n < 0) { - memo->min = i; - } - n = OPTIMIZED_CMP(j, memo->max, memo->cmp_opt); - if (n > 0) { - memo->max = j; - } + n = OPTIMIZED_CMP(i, memo->min); + if (n < 0) { + memo->min = i; + } + n = OPTIMIZED_CMP(j, memo->max); + if (n > 0) { + memo->max = j; + } } } @@ -1931,14 +2556,14 @@ minmax_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) ENUM_WANT_SVALUE(); - if (memo->last == Qundef) { + if (UNDEF_P(memo->last)) { memo->last = i; return Qnil; } j = memo->last; memo->last = Qundef; - n = OPTIMIZED_CMP(j, i, memo->cmp_opt); + n = OPTIMIZED_CMP(j, i); if (n == 0) i = j; else if (n < 0) { @@ -1958,19 +2583,19 @@ minmax_ii_update(VALUE i, VALUE j, struct minmax_t *memo) { int n; - if (memo->min == Qundef) { - memo->min = i; - memo->max = j; + if (UNDEF_P(memo->min)) { + memo->min = i; + memo->max = j; } else { - n = rb_cmpint(rb_yield_values(2, i, memo->min), i, memo->min); - if (n < 0) { - memo->min = i; - } - n = rb_cmpint(rb_yield_values(2, j, memo->max), j, memo->max); - if (n > 0) { - memo->max = j; - } + n = rb_cmpint(rb_yield_values(2, i, memo->min), i, memo->min); + if (n < 0) { + memo->min = i; + } + n = rb_cmpint(rb_yield_values(2, j, memo->max), j, memo->max); + if (n > 0) { + memo->max = j; + } } } @@ -1983,7 +2608,7 @@ minmax_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) ENUM_WANT_SVALUE(); - if (memo->last == Qundef) { + if (UNDEF_P(memo->last)) { memo->last = i; return Qnil; } @@ -2007,41 +2632,55 @@ minmax_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) /* * call-seq: - * enum.minmax -> [min, max] - * enum.minmax { |a, b| block } -> [min, max] + * minmax -> [minimum, maximum] + * minmax {|a, b| ... } -> [minimum, maximum] * - * Returns a two element array which contains the minimum and the - * maximum value in the enumerable. The first form assumes all - * objects implement Comparable; the second uses the - * block to return <em>a <=> b</em>. + * Returns a 2-element array containing the minimum and maximum elements + * according to a given criterion. + * The ordering of equal elements is indeterminate and may be unstable. + * + * With no argument and no block, returns the minimum and maximum elements, + * using the elements' own method <tt>#<=></tt> for comparison: + * + * (1..4).minmax # => [1, 4] + * (-4..-1).minmax # => [-4, -1] + * %w[d c b a].minmax # => ["a", "d"] + * {foo: 0, bar: 1, baz: 2}.minmax # => [[:bar, 1], [:foo, 0]] + * [].minmax # => [nil, nil] + * + * With a block given, returns the minimum and maximum elements + * as determined by the block: + * + * %w[xxx x xxxx xx].minmax {|a, b| a.size <=> b.size } # => ["x", "xxxx"] + * h = {foo: 0, bar: 1, baz: 2} + * h.minmax {|pair1, pair2| pair1[1] <=> pair2[1] } + * # => [[:foo, 0], [:baz, 2]] + * [].minmax {|a, b| a <=> b } # => [nil, nil] + * + * Related: #min, #max, #minmax_by. * - * a = %w(albatross dog horse) - * a.minmax #=> ["albatross", "horse"] - * a.minmax { |a, b| a.length <=> b.length } #=> ["dog", "albatross"] */ static VALUE enum_minmax(VALUE obj) { VALUE memo; - struct minmax_t *m = NEW_CMP_OPT_MEMO(struct minmax_t, memo); + struct minmax_t *m = NEW_MEMO_FOR(struct minmax_t, memo); m->min = Qundef; m->last = Qundef; - m->cmp_opt.opt_methods = 0; - m->cmp_opt.opt_inited = 0; if (rb_block_given_p()) { - rb_block_call(obj, id_each, 0, 0, minmax_ii, memo); - if (m->last != Qundef) - minmax_ii_update(m->last, m->last, m); + rb_block_call(obj, id_each, 0, 0, minmax_ii, memo); + if (!UNDEF_P(m->last)) + minmax_ii_update(m->last, m->last, m); } else { - rb_block_call(obj, id_each, 0, 0, minmax_i, memo); - if (m->last != Qundef) - minmax_i_update(m->last, m->last, m); + rb_block_call(obj, id_each, 0, 0, minmax_i, memo); + if (!UNDEF_P(m->last)) + minmax_i_update(m->last, m->last, m); } - if (m->min != Qundef) { - return rb_assoc_new(m->min, m->max); + if (!UNDEF_P(m->min)) { + return rb_assoc_new(m->min, m->max); } return rb_assoc_new(Qnil, Qnil); } @@ -2049,45 +2688,57 @@ enum_minmax(VALUE obj) static VALUE min_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) { - struct cmp_opt_data cmp_opt = { 0, 0 }; struct MEMO *memo = MEMO_CAST(args); VALUE v; ENUM_WANT_SVALUE(); v = enum_yield(argc, i); - if (memo->v1 == Qundef) { - MEMO_V1_SET(memo, v); - MEMO_V2_SET(memo, i); + if (UNDEF_P(memo->v1)) { + MEMO_V1_SET(memo, v); + MEMO_V2_SET(memo, i); } - else if (OPTIMIZED_CMP(v, memo->v1, cmp_opt) < 0) { - MEMO_V1_SET(memo, v); - MEMO_V2_SET(memo, i); + else if (OPTIMIZED_CMP(v, memo->v1) < 0) { + MEMO_V1_SET(memo, v); + MEMO_V2_SET(memo, i); } return Qnil; } /* * call-seq: - * enum.min_by {|obj| block } -> obj - * enum.min_by -> an_enumerator - * enum.min_by(n) {|obj| block } -> array - * enum.min_by(n) -> an_enumerator + * min_by {|element| ... } -> element + * min_by(n) {|element| ... } -> array + * min_by -> enumerator + * min_by(n) -> enumerator + * + * Returns the elements for which the block returns the minimum values. + * + * With a block given and no argument, + * returns the element for which the block returns the minimum value: + * + * (1..4).min_by {|element| -element } # => 4 + * %w[a b c d].min_by {|element| -element.ord } # => "d" + * {foo: 0, bar: 1, baz: 2}.min_by {|key, value| -value } # => [:baz, 2] + * [].min_by {|element| -element } # => nil * - * Returns the object in <i>enum</i> that gives the minimum - * value from the given block. + * With a block given and positive integer argument +n+ given, + * returns an array containing the +n+ elements + * for which the block returns minimum values: * - * If no block is given, an enumerator is returned instead. + * (1..4).min_by(2) {|element| -element } + * # => [4, 3] + * %w[a b c d].min_by(2) {|element| -element.ord } + * # => ["d", "c"] + * {foo: 0, bar: 1, baz: 2}.min_by(2) {|key, value| -value } + * # => [[:baz, 2], [:bar, 1]] + * [].min_by(2) {|element| -element } + * # => [] * - * a = %w(albatross dog horse) - * a.min_by { |x| x.length } #=> "dog" + * Returns an Enumerator if no block is given. * - * If the +n+ argument is given, minimum +n+ elements are returned - * as an array. These +n+ elements are sorted by the value from the - * given block. + * Related: #min, #minmax, #max_by. * - * a = %w[albatross dog horse] - * p a.min_by(2) {|x| x.length } #=> ["dog", "horse"] */ static VALUE @@ -2103,7 +2754,7 @@ enum_min_by(int argc, VALUE *argv, VALUE obj) if (argc && !NIL_P(num = argv[0])) return rb_nmin_run(obj, num, 1, 0, 0); - memo = MEMO_NEW(Qundef, Qnil, 0); + memo = rb_imemo_memo_new(Qundef, Qnil, 0); rb_block_call(obj, id_each, 0, 0, min_by_i, (VALUE)memo); return memo->v2; } @@ -2111,89 +2762,56 @@ enum_min_by(int argc, VALUE *argv, VALUE obj) static VALUE max_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) { - struct cmp_opt_data cmp_opt = { 0, 0 }; struct MEMO *memo = MEMO_CAST(args); VALUE v; ENUM_WANT_SVALUE(); v = enum_yield(argc, i); - if (memo->v1 == Qundef) { - MEMO_V1_SET(memo, v); - MEMO_V2_SET(memo, i); + if (UNDEF_P(memo->v1)) { + MEMO_V1_SET(memo, v); + MEMO_V2_SET(memo, i); } - else if (OPTIMIZED_CMP(v, memo->v1, cmp_opt) > 0) { - MEMO_V1_SET(memo, v); - MEMO_V2_SET(memo, i); + else if (OPTIMIZED_CMP(v, memo->v1) > 0) { + MEMO_V1_SET(memo, v); + MEMO_V2_SET(memo, i); } return Qnil; } /* * call-seq: - * enum.max_by {|obj| block } -> obj - * enum.max_by -> an_enumerator - * enum.max_by(n) {|obj| block } -> obj - * enum.max_by(n) -> an_enumerator - * - * Returns the object in <i>enum</i> that gives the maximum - * value from the given block. - * - * If no block is given, an enumerator is returned instead. - * - * a = %w(albatross dog horse) - * a.max_by { |x| x.length } #=> "albatross" - * - * If the +n+ argument is given, maximum +n+ elements are returned - * as an array. These +n+ elements are sorted by the value from the - * given block, in descending order. - * - * a = %w[albatross dog horse] - * a.max_by(2) {|x| x.length } #=> ["albatross", "horse"] - * - * enum.max_by(n) can be used to implement weighted random sampling. - * Following example implements and use Enumerable#wsample. - * - * module Enumerable - * # weighted random sampling. - * # - * # Pavlos S. Efraimidis, Paul G. Spirakis - * # Weighted random sampling with a reservoir - * # Information Processing Letters - * # Volume 97, Issue 5 (16 March 2006) - * def wsample(n) - * self.max_by(n) {|v| rand ** (1.0/yield(v)) } - * end - * end - * e = (-20..20).to_a*10000 - * a = e.wsample(20000) {|x| - * Math.exp(-(x/5.0)**2) # normal distribution - * } - * # a is 20000 samples from e. - * p a.length #=> 20000 - * h = a.group_by {|x| x } - * -10.upto(10) {|x| puts "*" * (h[x].length/30.0).to_i if h[x] } - * #=> * - * # *** - * # ****** - * # *********** - * # ****************** - * # ***************************** - * # ***************************************** - * # **************************************************** - * # *************************************************************** - * # ******************************************************************** - * # *********************************************************************** - * # *********************************************************************** - * # ************************************************************** - * # **************************************************** - * # *************************************** - * # *************************** - * # ****************** - * # *********** - * # ******* - * # *** - * # * + * max_by {|element| ... } -> element + * max_by(n) {|element| ... } -> array + * max_by -> enumerator + * max_by(n) -> enumerator + * + * Returns the elements for which the block returns the maximum values. + * + * With a block given and no argument, + * returns the element for which the block returns the maximum value: + * + * (1..4).max_by {|element| -element } # => 1 + * %w[a b c d].max_by {|element| -element.ord } # => "a" + * {foo: 0, bar: 1, baz: 2}.max_by {|key, value| -value } # => [:foo, 0] + * [].max_by {|element| -element } # => nil + * + * With a block given and positive integer argument +n+ given, + * returns an array containing the +n+ elements + * for which the block returns maximum values: + * + * (1..4).max_by(2) {|element| -element } + * # => [1, 2] + * %w[a b c d].max_by(2) {|element| -element.ord } + * # => ["a", "b"] + * {foo: 0, bar: 1, baz: 2}.max_by(2) {|key, value| -value } + * # => [[:foo, 0], [:bar, 1]] + * [].max_by(2) {|element| -element } + * # => [] + * + * Returns an Enumerator if no block is given. + * + * Related: #max, #minmax, #min_by. * */ @@ -2210,7 +2828,7 @@ enum_max_by(int argc, VALUE *argv, VALUE obj) if (argc && !NIL_P(num = argv[0])) return rb_nmin_run(obj, num, 1, 1, 0); - memo = MEMO_NEW(Qundef, Qnil, 0); + memo = rb_imemo_memo_new(Qundef, Qnil, 0); rb_block_call(obj, id_each, 0, 0, max_by_i, (VALUE)memo); return memo->v2; } @@ -2227,30 +2845,27 @@ struct minmax_by_t { static void minmax_by_i_update(VALUE v1, VALUE v2, VALUE i1, VALUE i2, struct minmax_by_t *memo) { - struct cmp_opt_data cmp_opt = { 0, 0 }; - - if (memo->min_bv == Qundef) { - memo->min_bv = v1; - memo->max_bv = v2; - memo->min = i1; - memo->max = i2; + if (UNDEF_P(memo->min_bv)) { + memo->min_bv = v1; + memo->max_bv = v2; + memo->min = i1; + memo->max = i2; } else { - if (OPTIMIZED_CMP(v1, memo->min_bv, cmp_opt) < 0) { - memo->min_bv = v1; - memo->min = i1; - } - if (OPTIMIZED_CMP(v2, memo->max_bv, cmp_opt) > 0) { - memo->max_bv = v2; - memo->max = i2; - } + if (OPTIMIZED_CMP(v1, memo->min_bv) < 0) { + memo->min_bv = v1; + memo->min = i1; + } + if (OPTIMIZED_CMP(v2, memo->max_bv) > 0) { + memo->max_bv = v2; + memo->max = i2; + } } } static VALUE minmax_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) { - struct cmp_opt_data cmp_opt = { 0, 0 }; struct minmax_by_t *memo = MEMO_FOR(struct minmax_by_t, _memo); VALUE vi, vj, j; int n; @@ -2259,7 +2874,7 @@ minmax_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) vi = enum_yield(argc, i); - if (memo->last_bv == Qundef) { + if (UNDEF_P(memo->last_bv)) { memo->last_bv = vi; memo->last = i; return Qnil; @@ -2268,7 +2883,7 @@ minmax_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) j = memo->last; memo->last_bv = Qundef; - n = OPTIMIZED_CMP(vj, vi, cmp_opt); + n = OPTIMIZED_CMP(vj, vi); if (n == 0) { i = j; vi = vj; @@ -2290,17 +2905,25 @@ minmax_by_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) /* * call-seq: - * enum.minmax_by { |obj| block } -> [min, max] - * enum.minmax_by -> an_enumerator + * minmax_by {|element| ... } -> [minimum, maximum] + * minmax_by -> enumerator + * + * Returns a 2-element array containing the elements + * for which the block returns minimum and maximum values: * - * Returns a two element array containing the objects in - * <i>enum</i> that correspond to the minimum and maximum values respectively - * from the given block. + * (1..4).minmax_by {|element| -element } + * # => [4, 1] + * %w[a b c d].minmax_by {|element| -element.ord } + * # => ["d", "a"] + * {foo: 0, bar: 1, baz: 2}.minmax_by {|key, value| -value } + * # => [[:baz, 2], [:foo, 0]] + * [].minmax_by {|element| -element } + * # => [nil, nil] * - * If no block is given, an enumerator is returned instead. + * Returns an Enumerator if no block is given. + * + * Related: #max_by, #minmax, #min_by. * - * a = %w(albatross dog horse) - * a.minmax_by { |x| x.length } #=> ["dog", "albatross"] */ static VALUE @@ -2318,7 +2941,7 @@ enum_minmax_by(VALUE obj) m->last_bv = Qundef; m->last = Qundef; rb_block_call(obj, id_each, 0, 0, minmax_by_i, memo); - if (m->last_bv != Qundef) + if (!UNDEF_P(m->last_bv)) minmax_by_i_update(m->last_bv, m->last_bv, m->last, m->last, m); m = MEMO_FOR(struct minmax_by_t, memo); return rb_assoc_new(m->min, m->max); @@ -2330,108 +2953,129 @@ member_i(RB_BLOCK_CALL_FUNC_ARGLIST(iter, args)) struct MEMO *memo = MEMO_CAST(args); if (rb_equal(rb_enum_values_pack(argc, argv), memo->v1)) { - MEMO_V2_SET(memo, Qtrue); - rb_iter_break(); + MEMO_V2_SET(memo, Qtrue); + rb_iter_break(); } return Qnil; } /* * call-seq: - * enum.include?(obj) -> true or false - * enum.member?(obj) -> true or false + * include?(object) -> true or false * - * Returns <code>true</code> if any member of <i>enum</i> equals - * <i>obj</i>. Equality is tested using <code>==</code>. + * Returns whether for any element <tt>object == element</tt>: * - * (1..10).include? 5 #=> true - * (1..10).include? 15 #=> false - * (1..10).member? 5 #=> true - * (1..10).member? 15 #=> false + * (1..4).include?(2) # => true + * (1..4).include?(5) # => false + * (1..4).include?('2') # => false + * %w[a b c d].include?('b') # => true + * %w[a b c d].include?('2') # => false + * {foo: 0, bar: 1, baz: 2}.include?(:foo) # => true + * {foo: 0, bar: 1, baz: 2}.include?('foo') # => false + * {foo: 0, bar: 1, baz: 2}.include?(0) # => false * */ static VALUE enum_member(VALUE obj, VALUE val) { - struct MEMO *memo = MEMO_NEW(val, Qfalse, 0); + struct MEMO *memo = rb_imemo_memo_new(val, Qfalse, 0); rb_block_call(obj, id_each, 0, 0, member_i, (VALUE)memo); return memo->v2; } static VALUE -each_with_index_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memo)) +each_with_index_i(RB_BLOCK_CALL_FUNC_ARGLIST(_, index)) { - struct MEMO *m = MEMO_CAST(memo); - VALUE n = imemo_count_value(m); + struct vm_ifunc *ifunc = rb_current_ifunc(); + ifunc->data = (const void *)rb_int_succ(index); - imemo_count_up(m); - return rb_yield_values(2, rb_enum_values_pack(argc, argv), n); + return rb_yield_values(2, rb_enum_values_pack(argc, argv), index); } /* * call-seq: - * enum.each_with_index(*args) { |obj, i| block } -> enum - * enum.each_with_index(*args) -> an_enumerator + * each_with_index(*args) {|element, i| ..... } -> self + * each_with_index(*args) -> enumerator * - * Calls <em>block</em> with two arguments, the item and its index, - * for each item in <i>enum</i>. Given arguments are passed through - * to #each(). + * Invoke <tt>self.each</tt> with <tt>*args</tt>. + * With a block given, the block receives each element and its index; + * returns +self+: * - * If no block is given, an enumerator is returned instead. + * h = {} + * (1..4).each_with_index {|element, i| h[element] = i } # => 1..4 + * h # => {1=>0, 2=>1, 3=>2, 4=>3} * - * hash = Hash.new - * %w(cat dog wombat).each_with_index { |item, index| - * hash[item] = index - * } - * hash #=> {"cat"=>0, "dog"=>1, "wombat"=>2} + * h = {} + * %w[a b c d].each_with_index {|element, i| h[element] = i } + * # => ["a", "b", "c", "d"] + * h # => {"a"=>0, "b"=>1, "c"=>2, "d"=>3} + * + * a = [] + * h = {foo: 0, bar: 1, baz: 2} + * h.each_with_index {|element, i| a.push([i, element]) } + * # => {:foo=>0, :bar=>1, :baz=>2} + * a # => [[0, [:foo, 0]], [1, [:bar, 1]], [2, [:baz, 2]]] + * + * With no block given, returns an Enumerator. * */ static VALUE enum_each_with_index(int argc, VALUE *argv, VALUE obj) { - struct MEMO *memo; - RETURN_SIZED_ENUMERATOR(obj, argc, argv, enum_size); - memo = MEMO_NEW(0, 0, 0); - rb_block_call(obj, id_each, argc, argv, each_with_index_i, (VALUE)memo); + rb_block_call(obj, id_each, argc, argv, each_with_index_i, INT2FIX(0)); return obj; } /* * call-seq: - * enum.reverse_each(*args) { |item| block } -> enum - * enum.reverse_each(*args) -> an_enumerator + * reverse_each(*args) {|element| ... } -> self + * reverse_each(*args) -> enumerator + * + * With a block given, calls the block with each element, + * but in reverse order; returns +self+: * - * Builds a temporary array and traverses that array in reverse order. + * a = [] + * (1..4).reverse_each {|element| a.push(-element) } # => 1..4 + * a # => [-4, -3, -2, -1] * - * If no block is given, an enumerator is returned instead. + * a = [] + * %w[a b c d].reverse_each {|element| a.push(element) } + * # => ["a", "b", "c", "d"] + * a # => ["d", "c", "b", "a"] * - * (1..3).reverse_each { |v| p v } + * a = [] + * h.reverse_each {|element| a.push(element) } + * # => {:foo=>0, :bar=>1, :baz=>2} + * a # => [[:baz, 2], [:bar, 1], [:foo, 0]] * - * produces: + * With no block given, returns an Enumerator. * - * 3 - * 2 - * 1 */ static VALUE enum_reverse_each(int argc, VALUE *argv, VALUE obj) { VALUE ary; - long i; + long len; RETURN_SIZED_ENUMERATOR(obj, argc, argv, enum_size); ary = enum_to_a(argc, argv, obj); - for (i = RARRAY_LEN(ary); --i >= 0; ) { - rb_yield(RARRAY_AREF(ary, i)); + len = RARRAY_LEN(ary); + while (len--) { + long nlen; + rb_yield(RARRAY_AREF(ary, len)); + nlen = RARRAY_LEN(ary); + if (nlen < len) { + len = nlen; + } } return obj; @@ -2448,30 +3092,39 @@ each_val_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, p)) /* * call-seq: - * enum.each_entry { |obj| block } -> enum - * enum.each_entry -> an_enumerator - * - * Calls <i>block</i> once for each element in +self+, passing that - * element as a parameter, converting multiple values from yield to an - * array. - * - * If no block is given, an enumerator is returned instead. - * - * class Foo - * include Enumerable - * def each - * yield 1 - * yield 1, 2 - * yield - * end - * end - * Foo.new.each_entry{ |o| p o } + * each_entry(*args) {|element| ... } -> self + * each_entry(*args) -> enumerator + * + * Calls the given block with each element, + * converting multiple values from yield to an array; returns +self+: + * + * a = [] + * (1..4).each_entry {|element| a.push(element) } # => 1..4 + * a # => [1, 2, 3, 4] + * + * a = [] + * h = {foo: 0, bar: 1, baz:2} + * h.each_entry {|element| a.push(element) } + * # => {:foo=>0, :bar=>1, :baz=>2} + * a # => [[:foo, 0], [:bar, 1], [:baz, 2]] + * + * class Foo + * include Enumerable + * def each + * yield 1 + * yield 1, 2 + * yield + * end + * end + * Foo.new.each_entry {|yielded| p yielded } + * + * Output: * - * produces: + * 1 + * [1, 2] + * nil * - * 1 - * [1, 2] - * nil + * With no block given, returns an Enumerator. * */ @@ -2513,14 +3166,14 @@ each_slice_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, m)) rb_ary_push(ary, i); if (RARRAY_LEN(ary) == size) { - v = rb_yield(ary); + v = rb_yield(ary); - if (memo->v2) { - MEMO_V1_SET(memo, rb_ary_new2(size)); - } - else { - rb_ary_clear(ary); - } + if (memo->v2) { + MEMO_V1_SET(memo, rb_ary_new2(size)); + } + else { + rb_ary_clear(ary); + } } return v; @@ -2536,7 +3189,7 @@ enum_each_slice_size(VALUE obj, VALUE args, VALUE eobj) if (slice_size <= 0) rb_raise(rb_eArgError, "invalid slice size"); size = enum_size(obj, 0, 0); - if (size == Qnil) return Qnil; + if (NIL_P(size)) return Qnil; if (RB_FLOAT_TYPE_P(size) && RTEST(rb_funcall(size, infinite_p, 0))) { return size; } @@ -2547,18 +3200,22 @@ enum_each_slice_size(VALUE obj, VALUE args, VALUE eobj) /* * call-seq: - * enum.each_slice(n) { ... } -> nil - * enum.each_slice(n) -> an_enumerator + * each_slice(n) { ... } -> self + * each_slice(n) -> enumerator + * + * Calls the block with each successive disjoint +n+-tuple of elements; + * returns +self+: * - * Iterates the given block for each slice of <n> elements. If no - * block is given, returns an enumerator. + * a = [] + * (1..10).each_slice(3) {|tuple| a.push(tuple) } + * a # => [[1, 2, 3], [4, 5, 6], [7, 8, 9], [10]] * - * (1..10).each_slice(3) { |a| p a } - * # outputs below - * [1, 2, 3] - * [4, 5, 6] - * [7, 8, 9] - * [10] + * a = [] + * h = {foo: 0, bar: 1, baz: 2, bat: 3, bam: 4} + * h.each_slice(2) {|tuple| a.push(tuple) } + * a # => [[[:foo, 0], [:bar, 1]], [[:baz, 2], [:bat, 3]], [[:bam, 4]]] + * + * With no block given, returns an Enumerator. * */ static VALUE @@ -2574,12 +3231,12 @@ enum_each_slice(VALUE obj, VALUE n) size = limit_by_enum_size(obj, size); ary = rb_ary_new2(size); arity = rb_block_arity(); - memo = MEMO_NEW(ary, dont_recycle_block_arg(arity), size); + memo = rb_imemo_memo_new(ary, dont_recycle_block_arg(arity), size); rb_block_call(obj, id_each, 0, 0, each_slice_i, (VALUE)memo); ary = memo->v1; if (RARRAY_LEN(ary) > 0) rb_yield(ary); - return Qnil; + return obj; } static VALUE @@ -2592,14 +3249,14 @@ each_cons_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); if (RARRAY_LEN(ary) == size) { - rb_ary_shift(ary); + rb_ary_shift(ary); } rb_ary_push(ary, i); if (RARRAY_LEN(ary) == size) { - if (memo->v2) { - ary = rb_ary_dup(ary); - } - v = rb_yield(ary); + if (memo->v2) { + ary = rb_ary_dup(ary); + } + v = rb_yield(ary); } return v; } @@ -2607,38 +3264,36 @@ each_cons_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) static VALUE enum_each_cons_size(VALUE obj, VALUE args, VALUE eobj) { - struct cmp_opt_data cmp_opt = { 0, 0 }; const VALUE zero = LONG2FIX(0); VALUE n, size; long cons_size = NUM2LONG(RARRAY_AREF(args, 0)); if (cons_size <= 0) rb_raise(rb_eArgError, "invalid size"); size = enum_size(obj, 0, 0); - if (size == Qnil) return Qnil; + if (NIL_P(size)) return Qnil; n = add_int(size, 1 - cons_size); - return (OPTIMIZED_CMP(n, zero, cmp_opt) == -1) ? zero : n; + return (OPTIMIZED_CMP(n, zero) == -1) ? zero : n; } /* * call-seq: - * enum.each_cons(n) { ... } -> nil - * enum.each_cons(n) -> an_enumerator - * - * Iterates the given block for each array of consecutive <n> - * elements. If no block is given, returns an enumerator. - * - * e.g.: - * (1..10).each_cons(3) { |a| p a } - * # outputs below - * [1, 2, 3] - * [2, 3, 4] - * [3, 4, 5] - * [4, 5, 6] - * [5, 6, 7] - * [6, 7, 8] - * [7, 8, 9] - * [8, 9, 10] + * each_cons(n) { ... } -> self + * each_cons(n) -> enumerator + * + * Calls the block with each successive overlapped +n+-tuple of elements; + * returns +self+: + * + * a = [] + * (1..5).each_cons(3) {|element| a.push(element) } + * a # => [[1, 2, 3], [2, 3, 4], [3, 4, 5]] + * + * a = [] + * h = {foo: 0, bar: 1, baz: 2, bam: 3} + * h.each_cons(2) {|element| a.push(element) } + * a # => [[[:foo, 0], [:bar, 1]], [[:bar, 1], [:baz, 2]], [[:baz, 2], [:bam, 3]]] + * + * With no block given, returns an Enumerator. * */ static VALUE @@ -2651,11 +3306,11 @@ enum_each_cons(VALUE obj, VALUE n) if (size <= 0) rb_raise(rb_eArgError, "invalid size"); RETURN_SIZED_ENUMERATOR(obj, 1, &n, enum_each_cons_size); arity = rb_block_arity(); - if (enum_size_over_p(obj, size)) return Qnil; - memo = MEMO_NEW(rb_ary_new2(size), dont_recycle_block_arg(arity), size); + if (enum_size_over_p(obj, size)) return obj; + memo = rb_imemo_memo_new(rb_ary_new2(size), dont_recycle_block_arg(arity), size); rb_block_call(obj, id_each, 0, 0, each_cons_i, (VALUE)memo); - return Qnil; + return obj; } static VALUE @@ -2667,16 +3322,19 @@ each_with_object_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, memo)) /* * call-seq: - * enum.each_with_object(obj) { |(*args), memo_obj| ... } -> obj - * enum.each_with_object(obj) -> an_enumerator + * each_with_object(object) { |(*args), memo_object| ... } -> object + * each_with_object(object) -> enumerator + * + * Calls the block once for each element, passing both the element + * and the given object: * - * Iterates the given block for each element with an arbitrary - * object given, and returns the initially given object. + * (1..4).each_with_object([]) {|i, a| a.push(i**2) } + * # => [1, 4, 9, 16] * - * If no block is given, returns an enumerator. + * {foo: 0, bar: 1, baz: 2}.each_with_object({}) {|(k, v), h| h[v] = k } + * # => {0=>:foo, 1=>:bar, 2=>:baz} * - * evens = (1..10).each_with_object([]) { |i, a| a << i*2 } - * #=> [2, 4, 6, 8, 10, 12, 14, 16, 18, 20] + * With no block given, returns an Enumerator. * */ static VALUE @@ -2702,20 +3360,20 @@ zip_ary(RB_BLOCK_CALL_FUNC_ARGLIST(val, memoval)) tmp = rb_ary_new2(RARRAY_LEN(args) + 1); rb_ary_store(tmp, 0, rb_enum_values_pack(argc, argv)); for (i=0; i<RARRAY_LEN(args); i++) { - VALUE e = RARRAY_AREF(args, i); + VALUE e = RARRAY_AREF(args, i); - if (RARRAY_LEN(e) <= n) { - rb_ary_push(tmp, Qnil); - } - else { - rb_ary_push(tmp, RARRAY_AREF(e, n)); - } + if (RARRAY_LEN(e) <= n) { + rb_ary_push(tmp, Qnil); + } + else { + rb_ary_push(tmp, RARRAY_AREF(e, n)); + } } if (NIL_P(result)) { - enum_yield_array(tmp); + enum_yield_array(tmp); } else { - rb_ary_push(result, tmp); + rb_ary_push(result, tmp); } RB_GC_GUARD(args); @@ -2724,14 +3382,16 @@ zip_ary(RB_BLOCK_CALL_FUNC_ARGLIST(val, memoval)) } static VALUE -call_next(VALUE *v) +call_next(VALUE w) { + VALUE *v = (VALUE *)w; return v[0] = rb_funcallv(v[1], id_next, 0, 0); } static VALUE -call_stop(VALUE *v) +call_stop(VALUE w, VALUE _) { + VALUE *v = (VALUE *)w; return v[0] = Qundef; } @@ -2747,26 +3407,26 @@ zip_i(RB_BLOCK_CALL_FUNC_ARGLIST(val, memoval)) tmp = rb_ary_new2(RARRAY_LEN(args) + 1); rb_ary_store(tmp, 0, rb_enum_values_pack(argc, argv)); for (i=0; i<RARRAY_LEN(args); i++) { - if (NIL_P(RARRAY_AREF(args, i))) { - rb_ary_push(tmp, Qnil); - } - else { - VALUE v[2]; - - v[1] = RARRAY_AREF(args, i); - rb_rescue2(call_next, (VALUE)v, call_stop, (VALUE)v, rb_eStopIteration, (VALUE)0); - if (v[0] == Qundef) { - RARRAY_ASET(args, i, Qnil); - v[0] = Qnil; - } - rb_ary_push(tmp, v[0]); - } + if (NIL_P(RARRAY_AREF(args, i))) { + rb_ary_push(tmp, Qnil); + } + else { + VALUE v[2]; + + v[1] = RARRAY_AREF(args, i); + rb_rescue2(call_next, (VALUE)v, call_stop, (VALUE)v, rb_eStopIteration, (VALUE)0); + if (UNDEF_P(v[0])) { + RARRAY_ASET(args, i, Qnil); + v[0] = Qnil; + } + rb_ary_push(tmp, v[0]); + } } if (NIL_P(result)) { - enum_yield_array(tmp); + enum_yield_array(tmp); } else { - rb_ary_push(result, tmp); + rb_ary_push(result, tmp); } RB_GC_GUARD(args); @@ -2776,29 +3436,68 @@ zip_i(RB_BLOCK_CALL_FUNC_ARGLIST(val, memoval)) /* * call-seq: - * enum.zip(arg, ...) -> an_array_of_array - * enum.zip(arg, ...) { |arr| block } -> nil - * - * Takes one element from <i>enum</i> and merges corresponding - * elements from each <i>args</i>. This generates a sequence of - * <em>n</em>-element arrays, where <em>n</em> is one more than the - * count of arguments. The length of the resulting sequence will be - * <code>enum#size</code>. If the size of any argument is less than - * <code>enum#size</code>, <code>nil</code> values are supplied. If - * a block is given, it is invoked for each output array, otherwise - * an array of arrays is returned. - * - * a = [ 4, 5, 6 ] - * b = [ 7, 8, 9 ] - * - * a.zip(b) #=> [[4, 7], [5, 8], [6, 9]] - * [1, 2, 3].zip(a, b) #=> [[1, 4, 7], [2, 5, 8], [3, 6, 9]] - * [1, 2].zip(a, b) #=> [[1, 4, 7], [2, 5, 8]] - * a.zip([1, 2], [8]) #=> [[4, 1, 8], [5, 2, nil], [6, nil, nil]] - * - * c = [] - * a.zip(b) { |x, y| c << x + y } #=> nil - * c #=> [11, 13, 15] + * zip(*other_enums) -> array + * zip(*other_enums) {|array| ... } -> nil + * + * With no block given, returns a new array +new_array+ of size self.size + * whose elements are arrays. + * Each nested array <tt>new_array[n]</tt> + * is of size <tt>other_enums.size+1</tt>, and contains: + * + * - The +n+-th element of self. + * - The +n+-th element of each of the +other_enums+. + * + * If all +other_enums+ and self are the same size, + * all elements are included in the result, and there is no +nil+-filling: + * + * a = [:a0, :a1, :a2, :a3] + * b = [:b0, :b1, :b2, :b3] + * c = [:c0, :c1, :c2, :c3] + * d = a.zip(b, c) + * d # => [[:a0, :b0, :c0], [:a1, :b1, :c1], [:a2, :b2, :c2], [:a3, :b3, :c3]] + * + * f = {foo: 0, bar: 1, baz: 2} + * g = {goo: 3, gar: 4, gaz: 5} + * h = {hoo: 6, har: 7, haz: 8} + * d = f.zip(g, h) + * d # => [ + * # [[:foo, 0], [:goo, 3], [:hoo, 6]], + * # [[:bar, 1], [:gar, 4], [:har, 7]], + * # [[:baz, 2], [:gaz, 5], [:haz, 8]] + * # ] + * + * If any enumerable in other_enums is smaller than self, + * fills to <tt>self.size</tt> with +nil+: + * + * a = [:a0, :a1, :a2, :a3] + * b = [:b0, :b1, :b2] + * c = [:c0, :c1] + * d = a.zip(b, c) + * d # => [[:a0, :b0, :c0], [:a1, :b1, :c1], [:a2, :b2, nil], [:a3, nil, nil]] + * + * If any enumerable in other_enums is larger than self, + * its trailing elements are ignored: + * + * a = [:a0, :a1, :a2, :a3] + * b = [:b0, :b1, :b2, :b3, :b4] + * c = [:c0, :c1, :c2, :c3, :c4, :c5] + * d = a.zip(b, c) + * d # => [[:a0, :b0, :c0], [:a1, :b1, :c1], [:a2, :b2, :c2], [:a3, :b3, :c3]] + * + * When a block is given, calls the block with each of the sub-arrays + * (formed as above); returns nil: + * + * a = [:a0, :a1, :a2, :a3] + * b = [:b0, :b1, :b2, :b3] + * c = [:c0, :c1, :c2, :c3] + * a.zip(b, c) {|sub_array| p sub_array} # => nil + * + * Output: + * + * [:a0, :b0, :c0] + * [:a1, :b1, :c1] + * [:a2, :b2, :c2] + * [:a3, :b3, :c3] * */ @@ -2814,30 +3513,30 @@ enum_zip(int argc, VALUE *argv, VALUE obj) argv = RARRAY_PTR(args); for (i=0; i<argc; i++) { - VALUE ary = rb_check_array_type(argv[i]); - if (NIL_P(ary)) { - allary = FALSE; - break; - } - argv[i] = ary; + VALUE ary = rb_check_array_type(argv[i]); + if (NIL_P(ary)) { + allary = FALSE; + break; + } + argv[i] = ary; } if (!allary) { - static const VALUE sym_each = STATIC_ID2SYM(id_each); - CONST_ID(conv, "to_enum"); - for (i=0; i<argc; i++) { - if (!rb_respond_to(argv[i], id_each)) { - rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (must respond to :each)", - rb_obj_class(argv[i])); + static const VALUE sym_each = STATIC_ID2SYM(id_each); + CONST_ID(conv, "to_enum"); + for (i=0; i<argc; i++) { + if (!rb_respond_to(argv[i], id_each)) { + rb_raise(rb_eTypeError, "wrong argument type %"PRIsVALUE" (must respond to :each)", + rb_obj_class(argv[i])); } - argv[i] = rb_funcallv(argv[i], conv, 1, &sym_each); - } + argv[i] = rb_funcallv(argv[i], conv, 1, &sym_each); + } } if (!rb_block_given_p()) { - result = rb_ary_new(); + result = rb_ary_new(); } /* TODO: use NODE_DOT2 as memo(v, v, -) */ - memo = MEMO_NEW(result, args, 0); + memo = rb_imemo_memo_new(result, args, 0); rb_block_call(obj, id_each, 0, 0, allary ? zip_ary : zip_i, (VALUE)memo); return result; @@ -2854,13 +3553,16 @@ take_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) /* * call-seq: - * enum.take(n) -> array + * take(n) -> array * - * Returns first n elements from <i>enum</i>. + * For non-negative integer +n+, returns the first +n+ elements: * - * a = [1, 2, 3, 4, 5, 0] - * a.take(3) #=> [1, 2, 3] - * a.take(30) #=> [1, 2, 3, 4, 5, 0] + * r = (1..4) + * r.take(2) # => [1, 2] + * r.take(0) # => [] + * + * h = {foo: 0, bar: 1, baz: 2, bat: 3} + * h.take(2) # => [[:foo, 0], [:bar, 1]] * */ @@ -2872,12 +3574,12 @@ enum_take(VALUE obj, VALUE n) long len = NUM2LONG(n); if (len < 0) { - rb_raise(rb_eArgError, "attempt to take negative size"); + rb_raise(rb_eArgError, "attempt to take negative size"); } if (len == 0) return rb_ary_new2(0); result = rb_ary_new2(len); - memo = MEMO_NEW(result, 0, len); + memo = rb_imemo_memo_new(result, 0, len); rb_block_call(obj, id_each, 0, 0, take_i, (VALUE)memo); return result; } @@ -2893,16 +3595,20 @@ take_while_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) /* * call-seq: - * enum.take_while { |obj| block } -> array - * enum.take_while -> an_enumerator + * take_while {|element| ... } -> array + * take_while -> enumerator + * + * Calls the block with successive elements as long as the block + * returns a truthy value; + * returns an array of all elements up to that point: * - * Passes elements to the block until the block returns +nil+ or +false+, - * then stops iterating and returns an array of all prior elements. * - * If no block is given, an enumerator is returned instead. + * (1..4).take_while{|i| i < 3 } # => [1, 2] + * h = {foo: 0, bar: 1, baz: 2} + * h.take_while{|element| key, value = *element; value < 2 } + * # => [[:foo, 0], [:bar, 1]] * - * a = [1, 2, 3, 4, 5, 0] - * a.take_while { |i| i < 3 } #=> [1, 2] + * With no block given, returns an Enumerator. * */ @@ -2922,23 +3628,30 @@ drop_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) { struct MEMO *memo = MEMO_CAST(args); if (memo->u3.cnt == 0) { - rb_ary_push(memo->v1, rb_enum_values_pack(argc, argv)); + rb_ary_push(memo->v1, rb_enum_values_pack(argc, argv)); } else { - memo->u3.cnt--; + memo->u3.cnt--; } return Qnil; } /* * call-seq: - * enum.drop(n) -> array + * drop(n) -> array * - * Drops first n elements from <i>enum</i>, and returns rest elements - * in an array. + * For positive integer +n+, returns an array containing + * all but the first +n+ elements: * - * a = [1, 2, 3, 4, 5, 0] - * a.drop(3) #=> [4, 5, 0] + * r = (1..4) + * r.drop(3) # => [4] + * r.drop(2) # => [3, 4] + * r.drop(1) # => [2, 3, 4] + * r.drop(0) # => [1, 2, 3, 4] + * r.drop(50) # => [] + * + * h = {foo: 0, bar: 1, baz: 2, bat: 3} + * h.drop(2) # => [[:baz, 2], [:bat, 3]] * */ @@ -2950,11 +3663,11 @@ enum_drop(VALUE obj, VALUE n) long len = NUM2LONG(n); if (len < 0) { - rb_raise(rb_eArgError, "attempt to drop negative size"); + rb_raise(rb_eArgError, "attempt to drop negative size"); } result = rb_ary_new(); - memo = MEMO_NEW(result, 0, len); + memo = rb_imemo_memo_new(result, 0, len); rb_block_call(obj, id_each, 0, 0, drop_i, (VALUE)memo); return result; } @@ -2967,27 +3680,41 @@ drop_while_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, args)) ENUM_WANT_SVALUE(); if (!memo->u3.state && !RTEST(enum_yield(argc, i))) { - memo->u3.state = TRUE; + memo->u3.state = TRUE; } if (memo->u3.state) { - rb_ary_push(memo->v1, i); + rb_ary_push(memo->v1, i); } return Qnil; } /* * call-seq: - * enum.drop_while { |obj| block } -> array - * enum.drop_while -> an_enumerator - * - * Drops elements up to, but not including, the first element for - * which the block returns +nil+ or +false+ and returns an array - * containing the remaining elements. - * - * If no block is given, an enumerator is returned instead. - * - * a = [1, 2, 3, 4, 5, 0] - * a.drop_while { |i| i < 3 } #=> [3, 4, 5, 0] + * drop_while {|element| ... } -> array + * drop_while -> enumerator + * + * Calls the block with successive elements as long as the block + * returns a truthy value; + * returns an array of all elements after that point: + * + * + * (1..4).drop_while{|i| i < 3 } # => [3, 4] + * h = {foo: 0, bar: 1, baz: 2} + * a = h.drop_while{|element| key, value = *element; value < 2 } + * a # => [[:baz, 2]] + * + * With no block given, returns an Enumerator. + * + * e = (1..4).drop_while + * p e #=> #<Enumerator: 1..4:drop_while> + * i = e.next; p i; e.feed(i < 3) #=> 1 + * i = e.next; p i; e.feed(i < 3) #=> 2 + * i = e.next; p i; e.feed(i < 3) #=> 3 + * begin + * e.next + * rescue StopIteration + * p $!.result #=> [3, 4] + * end * */ @@ -2999,7 +3726,7 @@ enum_drop_while(VALUE obj) RETURN_ENUMERATOR(obj, 0, 0); result = rb_ary_new(); - memo = MEMO_NEW(result, 0, FALSE); + memo = rb_imemo_memo_new(result, 0, FALSE); rb_block_call(obj, id_each, 0, 0, drop_while_i, (VALUE)memo); return result; } @@ -3022,8 +3749,8 @@ enum_cycle_size(VALUE self, VALUE args, VALUE eobj) VALUE size; if (args && (RARRAY_LEN(args) > 0)) { - n = RARRAY_AREF(args, 0); - if (!NIL_P(n)) mul = NUM2LONG(n); + n = RARRAY_AREF(args, 0); + if (!NIL_P(n)) mul = NUM2LONG(n); } size = enum_size(self, args, 0); @@ -3037,22 +3764,28 @@ enum_cycle_size(VALUE self, VALUE args, VALUE eobj) /* * call-seq: - * enum.cycle(n=nil) { |obj| block } -> nil - * enum.cycle(n=nil) -> an_enumerator + * cycle(n = nil) {|element| ...} -> nil + * cycle(n = nil) -> enumerator + * + * When called with positive integer argument +n+ and a block, + * calls the block with each element, then does so again, + * until it has done so +n+ times; returns +nil+: * - * Calls <i>block</i> for each element of <i>enum</i> repeatedly _n_ - * times or forever if none or +nil+ is given. If a non-positive - * number is given or the collection is empty, does nothing. Returns - * +nil+ if the loop has finished without getting interrupted. + * a = [] + * (1..4).cycle(3) {|element| a.push(element) } # => nil + * a # => [1, 2, 3, 4, 1, 2, 3, 4, 1, 2, 3, 4] + * a = [] + * ('a'..'d').cycle(2) {|element| a.push(element) } + * a # => ["a", "b", "c", "d", "a", "b", "c", "d"] + * a = [] + * {foo: 0, bar: 1, baz: 2}.cycle(2) {|element| a.push(element) } + * a # => [[:foo, 0], [:bar, 1], [:baz, 2], [:foo, 0], [:bar, 1], [:baz, 2]] * - * Enumerable#cycle saves elements in an internal array so changes - * to <i>enum</i> after the first pass have no effect. + * If count is zero or negative, does not call the block. * - * If no block is given, an enumerator is returned instead. + * When called with a block and +n+ is +nil+, cycles forever. * - * a = ["a", "b", "c"] - * a.cycle { |x| puts x } # print, a, b, c, a, b, c,.. forever. - * a.cycle(2) { |x| puts x } # print, a, b, c, a, b, c. + * When no block is given, returns an Enumerator. * */ @@ -3080,7 +3813,7 @@ enum_cycle(int argc, VALUE *argv, VALUE obj) if (len == 0) return Qnil; while (n < 0 || 0 < --n) { for (i=0; i<len; i++) { - enum_yield_array(RARRAY_AREF(ary, i)); + enum_yield_array(RARRAY_AREF(ary, i)); } } return Qnil; @@ -3098,8 +3831,8 @@ chunk_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _argp)) { struct chunk_arg *argp = MEMO_FOR(struct chunk_arg, _argp); VALUE v, s; - VALUE alone = ID2SYM(rb_intern("_alone")); - VALUE separator = ID2SYM(rb_intern("_separator")); + VALUE alone = ID2SYM(id__alone); + VALUE separator = ID2SYM(id__separator); ENUM_WANT_SVALUE(); @@ -3107,22 +3840,22 @@ chunk_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _argp)) if (v == alone) { if (!NIL_P(argp->prev_value)) { - s = rb_assoc_new(argp->prev_value, argp->prev_elts); + s = rb_assoc_new(argp->prev_value, argp->prev_elts); rb_funcallv(argp->yielder, id_lshift, 1, &s); argp->prev_value = argp->prev_elts = Qnil; } - v = rb_assoc_new(v, rb_ary_new3(1, i)); + v = rb_assoc_new(v, rb_ary_new3(1, i)); rb_funcallv(argp->yielder, id_lshift, 1, &v); } else if (NIL_P(v) || v == separator) { if (!NIL_P(argp->prev_value)) { - v = rb_assoc_new(argp->prev_value, argp->prev_elts); + v = rb_assoc_new(argp->prev_value, argp->prev_elts); rb_funcallv(argp->yielder, id_lshift, 1, &v); argp->prev_value = argp->prev_elts = Qnil; } } else if (SYMBOL_P(v) && (s = rb_sym2str(v), RSTRING_PTR(s)[0] == '_')) { - rb_raise(rb_eRuntimeError, "symbols beginning with an underscore are reserved"); + rb_raise(rb_eRuntimeError, "symbols beginning with an underscore are reserved"); } else { if (NIL_P(argp->prev_value)) { @@ -3134,7 +3867,7 @@ chunk_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _argp)) rb_ary_push(argp->prev_elts, i); } else { - s = rb_assoc_new(argp->prev_value, argp->prev_elts); + s = rb_assoc_new(argp->prev_value, argp->prev_elts); rb_funcallv(argp->yielder, id_lshift, 1, &s); argp->prev_value = v; argp->prev_elts = rb_ary_new3(1, i); @@ -3151,8 +3884,8 @@ chunk_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) VALUE arg; struct chunk_arg *memo = NEW_MEMO_FOR(struct chunk_arg, arg); - enumerable = rb_ivar_get(enumerator, rb_intern("chunk_enumerable")); - memo->categorize = rb_ivar_get(enumerator, rb_intern("chunk_categorize")); + enumerable = rb_ivar_get(enumerator, id_chunk_enumerable); + memo->categorize = rb_ivar_get(enumerator, id_chunk_categorize); memo->prev_value = Qnil; memo->prev_elts = Qnil; memo->yielder = yielder; @@ -3160,58 +3893,86 @@ chunk_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) rb_block_call(enumerable, id_each, 0, 0, chunk_ii, arg); memo = MEMO_FOR(struct chunk_arg, arg); if (!NIL_P(memo->prev_elts)) { - arg = rb_assoc_new(memo->prev_value, memo->prev_elts); - rb_funcallv(memo->yielder, id_lshift, 1, &arg); + arg = rb_assoc_new(memo->prev_value, memo->prev_elts); + rb_funcallv(memo->yielder, id_lshift, 1, &arg); } return Qnil; } /* * call-seq: - * enum.chunk { |elt| ... } -> an_enumerator + * chunk {|array| ... } -> enumerator * - * Enumerates over the items, chunking them together based on the return - * value of the block. + * Each element in the returned enumerator is a 2-element array consisting of: * - * Consecutive elements which return the same block value are chunked together. + * - A value returned by the block. + * - An array ("chunk") containing the element for which that value was returned, + * and all following elements for which the block returned the same value: * - * For example, consecutive even numbers and odd numbers can be - * chunked as follows. + * So that: * - * [3, 1, 4, 1, 5, 9, 2, 6, 5, 3, 5].chunk { |n| - * n.even? - * }.each { |even, ary| - * p [even, ary] - * } - * #=> [false, [3, 1]] - * # [true, [4]] - * # [false, [1, 5, 9]] - * # [true, [2, 6]] - * # [false, [5, 3, 5]] + * - Each block return value that is different from its predecessor + * begins a new chunk. + * - Each block return value that is the same as its predecessor + * continues the same chunk. + * + * Example: + * + * e = (0..10).chunk {|i| (i / 3).floor } # => #<Enumerator: ...> + * # The enumerator elements. + * e.next # => [0, [0, 1, 2]] + * e.next # => [1, [3, 4, 5]] + * e.next # => [2, [6, 7, 8]] + * e.next # => [3, [9, 10]] + * + * Method +chunk+ is especially useful for an enumerable that is already sorted. + * This example counts words for each initial letter in a large array of words: + * + * # Get sorted words from a web page. + * url = 'https://raw.githubusercontent.com/eneko/data-repository/master/data/words.txt' + * words = URI::open(url).readlines + * # Make chunks, one for each letter. + * e = words.chunk {|word| word.upcase[0] } # => #<Enumerator: ...> + * # Display 'A' through 'F'. + * e.each {|c, words| p [c, words.length]; break if c == 'F' } + * + * Output: + * + * ["A", 17096] + * ["B", 11070] + * ["C", 19901] + * ["D", 10896] + * ["E", 8736] + * ["F", 6860] + * + * You can use the special symbol <tt>:_alone</tt> to force an element + * into its own separate chunk: * - * This method is especially useful for sorted series of elements. - * The following example counts words for each initial letter. + * a = [0, 0, 1, 1] + * e = a.chunk{|i| i.even? ? :_alone : true } + * e.to_a # => [[:_alone, [0]], [:_alone, [0]], [true, [1, 1]]] * - * open("/usr/share/dict/words", "r:iso-8859-1") { |f| - * f.chunk { |line| line.ord }.each { |ch, lines| p [ch.chr, lines.length] } + * For example, you can put each line that contains a URL into its own chunk: + * + * pattern = /http/ + * open(filename) { |f| + * f.chunk { |line| line =~ pattern ? :_alone : true }.each { |key, lines| + * pp lines + * } * } - * #=> ["\n", 1] - * # ["A", 1327] - * # ["B", 1372] - * # ["C", 1507] - * # ["D", 791] - * # ... * - * The following key values have special meaning: - * - +nil+ and +:_separator+ specifies that the elements should be dropped. - * - +:_alone+ specifies that the element should be chunked by itself. + * You can use the special symbol <tt>:_separator</tt> or +nil+ + * to force an element to be ignored (not included in any chunk): * - * Any other symbols that begin with an underscore will raise an error: + * a = [0, 0, -1, 1, 1] + * e = a.chunk{|i| i < 0 ? :_separator : true } + * e.to_a # => [[true, [0, 0]], [true, [1, 1]]] * - * items.chunk { |item| :_underscore } - * #=> RuntimeError: symbols beginning with an underscore are reserved + * Note that the separator does end the chunk: * - * +nil+ and +:_separator+ can be used to ignore some elements. + * a = [0, 0, -1, 1, -1, 1] + * e = a.chunk{|i| i < 0 ? :_separator : true } + * e.to_a # => [[true, [0, 0]], [true, [1]], [true, [1]]] * * For example, the sequence of hyphens in svn log can be eliminated as follows: * @@ -3241,18 +4002,6 @@ chunk_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) * pp lines * } * - * +:_alone+ can be used to force items into their own chunk. - * For example, you can put lines that contain a URL by themselves, - * and chunk the rest of the lines together, like this: - * - * pattern = /http/ - * open(filename) { |f| - * f.chunk { |line| line =~ pattern ? :_alone : true }.each { |key, lines| - * pp lines - * } - * } - * - * If no block is given, an enumerator to `chunk` is returned instead. */ static VALUE enum_chunk(VALUE enumerable) @@ -3262,8 +4011,8 @@ enum_chunk(VALUE enumerable) RETURN_SIZED_ENUMERATOR(enumerable, 0, 0, enum_size); enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("chunk_enumerable"), enumerable); - rb_ivar_set(enumerator, rb_intern("chunk_categorize"), rb_block_proc()); + rb_ivar_set(enumerator, id_chunk_enumerable, enumerable); + rb_ivar_set(enumerator, id_chunk_categorize, rb_block_proc()); rb_block_call(enumerator, idInitialize, 0, 0, chunk_i, enumerator); return enumerator; } @@ -3310,9 +4059,9 @@ slicebefore_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) VALUE arg; struct slicebefore_arg *memo = NEW_MEMO_FOR(struct slicebefore_arg, arg); - enumerable = rb_ivar_get(enumerator, rb_intern("slicebefore_enumerable")); - memo->sep_pred = rb_attr_get(enumerator, rb_intern("slicebefore_sep_pred")); - memo->sep_pat = NIL_P(memo->sep_pred) ? rb_ivar_get(enumerator, rb_intern("slicebefore_sep_pat")) : Qnil; + enumerable = rb_ivar_get(enumerator, id_slicebefore_enumerable); + memo->sep_pred = rb_attr_get(enumerator, id_slicebefore_sep_pred); + memo->sep_pat = NIL_P(memo->sep_pred) ? rb_ivar_get(enumerator, id_slicebefore_sep_pat) : Qnil; memo->prev_elts = Qnil; memo->yielder = yielder; @@ -3325,24 +4074,41 @@ slicebefore_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) /* * call-seq: - * enum.slice_before(pattern) -> an_enumerator - * enum.slice_before { |elt| bool } -> an_enumerator + * slice_before(pattern) -> enumerator + * slice_before {|elt| ... } -> enumerator * - * Creates an enumerator for each chunked elements. - * The beginnings of chunks are defined by _pattern_ and the block. - - * If <code>_pattern_ === _elt_</code> returns <code>true</code> or the block - * returns <code>true</code> for the element, the element is beginning of a - * chunk. - - * The <code>===</code> and _block_ is called from the first element to the last - * element of _enum_. The result for the first element is ignored. - - * The result enumerator yields the chunked elements as an array. - * So +each+ method can be called as follows: + * With argument +pattern+, returns an enumerator that uses the pattern + * to partition elements into arrays ("slices"). + * An element begins a new slice if <tt>element === pattern</tt> + * (or if it is the first element). + * + * a = %w[foo bar fop for baz fob fog bam foy] + * e = a.slice_before(/ba/) # => #<Enumerator: ...> + * e.each {|array| p array } + * + * Output: * - * enum.slice_before(pattern).each { |ary| ... } - * enum.slice_before { |elt| bool }.each { |ary| ... } + * ["foo"] + * ["bar", "fop", "for"] + * ["baz", "fob", "fog"] + * ["bam", "foy"] + * + * With a block, returns an enumerator that uses the block + * to partition elements into arrays. + * An element begins a new slice if its block return is a truthy value + * (or if it is the first element): + * + * e = (1..20).slice_before {|i| i % 4 == 2 } # => #<Enumerator: ...> + * e.each {|array| p array } + * + * Output: + * + * [1] + * [2, 3, 4, 5] + * [6, 7, 8, 9] + * [10, 11, 12, 13] + * [14, 15, 16, 17] + * [18, 19, 20] * * Other methods of the Enumerator class and Enumerable module, * such as +to_a+, +map+, etc., are also usable. @@ -3360,7 +4126,6 @@ slicebefore_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) * f.slice_before { |line| /\A\S/ === line }.each { |e| pp e } * } * - * * "svn proplist -R" produces multiline output for each file. * They can be chunked as follows: * @@ -3476,15 +4241,15 @@ enum_slice_before(int argc, VALUE *argv, VALUE enumerable) if (argc != 0) rb_error_arity(argc, 0, 0); enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("slicebefore_sep_pred"), rb_block_proc()); + rb_ivar_set(enumerator, id_slicebefore_sep_pred, rb_block_proc()); } else { VALUE sep_pat; rb_scan_args(argc, argv, "1", &sep_pat); enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("slicebefore_sep_pat"), sep_pat); + rb_ivar_set(enumerator, id_slicebefore_sep_pat, sep_pat); } - rb_ivar_set(enumerator, rb_intern("slicebefore_enumerable"), enumerable); + rb_ivar_set(enumerator, id_slicebefore_enumerable, enumerable); rb_block_call(enumerator, idInitialize, 0, 0, slicebefore_i, enumerator); return enumerator; } @@ -3540,9 +4305,9 @@ sliceafter_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) VALUE arg; struct sliceafter_arg *memo = NEW_MEMO_FOR(struct sliceafter_arg, arg); - enumerable = rb_ivar_get(enumerator, rb_intern("sliceafter_enum")); - memo->pat = rb_ivar_get(enumerator, rb_intern("sliceafter_pat")); - memo->pred = rb_attr_get(enumerator, rb_intern("sliceafter_pred")); + enumerable = rb_ivar_get(enumerator, id_sliceafter_enum); + memo->pat = rb_ivar_get(enumerator, id_sliceafter_pat); + memo->pred = rb_attr_get(enumerator, id_sliceafter_pred); memo->prev_elts = Qnil; memo->yielder = yielder; @@ -3605,9 +4370,9 @@ enum_slice_after(int argc, VALUE *argv, VALUE enumerable) } enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("sliceafter_enum"), enumerable); - rb_ivar_set(enumerator, rb_intern("sliceafter_pat"), pat); - rb_ivar_set(enumerator, rb_intern("sliceafter_pred"), pred); + rb_ivar_set(enumerator, id_sliceafter_enum, enumerable); + rb_ivar_set(enumerator, id_sliceafter_pat, pat); + rb_ivar_set(enumerator, id_sliceafter_pred, pred); rb_block_call(enumerator, idInitialize, 0, 0, sliceafter_i, enumerator); return enumerator; @@ -3631,15 +4396,15 @@ slicewhen_ii(RB_BLOCK_CALL_FUNC_ARGLIST(i, _memo)) ENUM_WANT_SVALUE(); - if (memo->prev_elt == Qundef) { + if (UNDEF_P(memo->prev_elt)) { /* The first element */ memo->prev_elt = i; memo->prev_elts = rb_ary_new3(1, i); } else { - VALUE args[2]; - args[0] = memo->prev_elt; - args[1] = i; + VALUE args[2]; + args[0] = memo->prev_elt; + args[1] = i; split_p = RTEST(rb_funcallv(memo->pred, id_call, 2, args)); UPDATE_MEMO; @@ -3668,14 +4433,14 @@ slicewhen_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) VALUE enumerable; VALUE arg; struct slicewhen_arg *memo = - NEW_PARTIAL_MEMO_FOR(struct slicewhen_arg, arg, inverted); + NEW_PARTIAL_MEMO_FOR(struct slicewhen_arg, arg, inverted); - enumerable = rb_ivar_get(enumerator, rb_intern("slicewhen_enum")); - memo->pred = rb_attr_get(enumerator, rb_intern("slicewhen_pred")); + enumerable = rb_ivar_get(enumerator, id_slicewhen_enum); + memo->pred = rb_attr_get(enumerator, id_slicewhen_pred); memo->prev_elt = Qundef; memo->prev_elts = Qnil; memo->yielder = yielder; - memo->inverted = RTEST(rb_attr_get(enumerator, rb_intern("slicewhen_inverted"))); + memo->inverted = RTEST(rb_attr_get(enumerator, id_slicewhen_inverted)); rb_block_call(enumerable, id_each, 0, 0, slicewhen_ii, arg); memo = MEMO_FOR(struct slicewhen_arg, arg); @@ -3691,7 +4456,7 @@ slicewhen_i(RB_BLOCK_CALL_FUNC_ARGLIST(yielder, enumerator)) * Creates an enumerator for each chunked elements. * The beginnings of chunks are defined by the block. * - * This method split each chunk using adjacent elements, + * This method splits each chunk using adjacent elements, * _elt_before_ and _elt_after_, * in the receiver enumerator. * This method split chunks between _elt_before_ and _elt_after_ where @@ -3755,9 +4520,9 @@ enum_slice_when(VALUE enumerable) pred = rb_block_proc(); enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("slicewhen_enum"), enumerable); - rb_ivar_set(enumerator, rb_intern("slicewhen_pred"), pred); - rb_ivar_set(enumerator, rb_intern("slicewhen_inverted"), Qfalse); + rb_ivar_set(enumerator, id_slicewhen_enum, enumerable); + rb_ivar_set(enumerator, id_slicewhen_pred, pred); + rb_ivar_set(enumerator, id_slicewhen_inverted, Qfalse); rb_block_call(enumerator, idInitialize, 0, 0, slicewhen_i, enumerator); return enumerator; @@ -3770,7 +4535,7 @@ enum_slice_when(VALUE enumerable) * Creates an enumerator for each chunked elements. * The beginnings of chunks are defined by the block. * - * This method split each chunk using adjacent elements, + * This method splits each chunk using adjacent elements, * _elt_before_ and _elt_after_, * in the receiver enumerator. * This method split chunks between _elt_before_ and _elt_after_ where @@ -3821,9 +4586,9 @@ enum_chunk_while(VALUE enumerable) pred = rb_block_proc(); enumerator = rb_obj_alloc(rb_cEnumerator); - rb_ivar_set(enumerator, rb_intern("slicewhen_enum"), enumerable); - rb_ivar_set(enumerator, rb_intern("slicewhen_pred"), pred); - rb_ivar_set(enumerator, rb_intern("slicewhen_inverted"), Qtrue); + rb_ivar_set(enumerator, id_slicewhen_enum, enumerable); + rb_ivar_set(enumerator, id_slicewhen_pred, pred); + rb_ivar_set(enumerator, id_slicewhen_inverted, Qtrue); rb_block_call(enumerator, idInitialize, 0, 0, slicewhen_i, enumerator); return enumerator; @@ -3838,122 +4603,141 @@ struct enum_sum_memo { }; static void -sum_iter(VALUE i, struct enum_sum_memo *memo) +sum_iter_normalize_memo(struct enum_sum_memo *memo) { - const int unused = (assert(memo != NULL), 0); + RUBY_ASSERT(FIXABLE(memo->n)); + memo->v = rb_fix_plus(LONG2FIX(memo->n), memo->v); + memo->n = 0; - long n = memo->n; - VALUE v = memo->v; - VALUE r = memo->r; - double f = memo->f; - double c = memo->c; + switch (TYPE(memo->r)) { + case T_RATIONAL: memo->v = rb_rational_plus(memo->r, memo->v); break; + case T_UNDEF: break; + default: UNREACHABLE; /* or ...? */ + } + memo->r = Qundef; +} - if (memo->block_given) - i = rb_yield(i); +static void +sum_iter_fixnum(VALUE i, struct enum_sum_memo *memo) +{ + memo->n += FIX2LONG(i); /* should not overflow long type */ + if (! FIXABLE(memo->n)) { + memo->v = rb_big_plus(LONG2NUM(memo->n), memo->v); + memo->n = 0; + } +} - if (memo->float_value) - goto float_value; +static void +sum_iter_bignum(VALUE i, struct enum_sum_memo *memo) +{ + memo->v = rb_big_plus(i, memo->v); +} - if (FIXNUM_P(v) || RB_TYPE_P(v, T_BIGNUM) || RB_TYPE_P(v, T_RATIONAL)) { - if (FIXNUM_P(i)) { - n += FIX2LONG(i); /* should not overflow long type */ - if (!FIXABLE(n)) { - v = rb_big_plus(LONG2NUM(n), v); - n = 0; - } - } - else if (RB_TYPE_P(i, T_BIGNUM)) - v = rb_big_plus(i, v); - else if (RB_TYPE_P(i, T_RATIONAL)) { - if (r == Qundef) - r = i; - else - r = rb_rational_plus(r, i); - } - else { - if (n != 0) { - v = rb_fix_plus(LONG2FIX(n), v); - n = 0; - } - if (r != Qundef) { - /* r can be an Integer when mathn is loaded */ - if (FIXNUM_P(r)) - v = rb_fix_plus(r, v); - else if (RB_TYPE_P(r, T_BIGNUM)) - v = rb_big_plus(r, v); - else - v = rb_rational_plus(r, v); - r = Qundef; - } - if (RB_FLOAT_TYPE_P(i)) { - f = NUM2DBL(v); - c = 0.0; - memo->float_value = 1; - goto float_value; - } - else - goto some_value; - } +static void +sum_iter_rational(VALUE i, struct enum_sum_memo *memo) +{ + if (UNDEF_P(memo->r)) { + memo->r = i; } - else if (RB_FLOAT_TYPE_P(v)) { - /* - * Kahan-Babuska balancing compensated summation algorithm - * See http://link.springer.com/article/10.1007/s00607-005-0139-x - */ - double x, t; - - float_value: - if (RB_FLOAT_TYPE_P(i)) - x = RFLOAT_VALUE(i); - else if (FIXNUM_P(i)) - x = FIX2LONG(i); - else if (RB_TYPE_P(i, T_BIGNUM)) - x = rb_big2dbl(i); - else if (RB_TYPE_P(i, T_RATIONAL)) - x = rb_num2dbl(i); - else { - v = DBL2NUM(f); - memo->float_value = 0; - goto some_value; - } + else { + memo->r = rb_rational_plus(memo->r, i); + } +} - if (isnan(f)) return; - if (isnan(x)) { - memo->v = i; - memo->f = x; - return; - } - if (isinf(x)) { - if (isinf(f) && signbit(x) != signbit(f)) { - memo->f = NAN; - memo->v = DBL2NUM(f); - } - else { - memo->f = x; - memo->v = i; - } - return; +static void +sum_iter_some_value(VALUE i, struct enum_sum_memo *memo) +{ + memo->v = rb_funcallv(memo->v, idPLUS, 1, &i); +} + +static void +sum_iter_Kahan_Babuska(VALUE i, struct enum_sum_memo *memo) +{ + /* + * Kahan-Babuska balancing compensated summation algorithm + * See https://link.springer.com/article/10.1007/s00607-005-0139-x + */ + double x; + + switch (TYPE(i)) { + case T_FLOAT: x = RFLOAT_VALUE(i); break; + case T_FIXNUM: x = FIX2LONG(i); break; + case T_BIGNUM: x = rb_big2dbl(i); break; + case T_RATIONAL: x = rb_num2dbl(i); break; + default: + memo->v = DBL2NUM(memo->f); + memo->float_value = 0; + sum_iter_some_value(i, memo); + return; + } + + double f = memo->f; + + if (isnan(f)) { + return; + } + else if (! isfinite(x)) { + if (isinf(x) && isinf(f) && signbit(x) != signbit(f)) { + i = DBL2NUM(f); + x = nan(""); } - if (isinf(f)) return; + memo->v = i; + memo->f = x; + return; + } + else if (isinf(f)) { + return; + } - t = f + x; - if (fabs(f) >= fabs(x)) - c += ((f - t) + x); - else - c += ((x - t) + f); - f = t; + double c = memo->c; + double t = f + x; + + if (fabs(f) >= fabs(x)) { + c += ((f - t) + x); } else { - some_value: - v = rb_funcallv(v, idPLUS, 1, &i); + c += ((x - t) + f); } + f = t; - memo->v = v; - memo->n = n; - memo->r = r; memo->f = f; memo->c = c; - (void)unused; +} + +static void +sum_iter(VALUE i, struct enum_sum_memo *memo) +{ + RUBY_ASSERT(memo != NULL); + if (memo->block_given) { + i = rb_yield(i); + } + + if (memo->float_value) { + sum_iter_Kahan_Babuska(i, memo); + } + else switch (TYPE(memo->v)) { + default: sum_iter_some_value(i, memo); return; + case T_FLOAT: + case T_FIXNUM: + case T_BIGNUM: + case T_RATIONAL: + switch (TYPE(i)) { + case T_FIXNUM: sum_iter_fixnum(i, memo); return; + case T_BIGNUM: sum_iter_bignum(i, memo); return; + case T_RATIONAL: sum_iter_rational(i, memo); return; + case T_FLOAT: + sum_iter_normalize_memo(memo); + memo->f = NUM2DBL(memo->v); + memo->c = 0.0; + memo->float_value = 1; + sum_iter_Kahan_Babuska(i, memo); + return; + default: + sum_iter_normalize_memo(memo); + sum_iter_some_value(i, memo); + return; + } + } } static VALUE @@ -3974,8 +4758,8 @@ hash_sum_i(VALUE key, VALUE value, VALUE arg) static void hash_sum(VALUE hash, struct enum_sum_memo *memo) { - assert(RB_TYPE_P(hash, T_HASH)); - assert(memo != NULL); + RUBY_ASSERT(RB_TYPE_P(hash, T_HASH)); + RUBY_ASSERT(memo != NULL); rb_hash_foreach(hash, hash_sum_i, (VALUE)memo); } @@ -4002,32 +4786,35 @@ int_range_sum(VALUE beg, VALUE end, int excl, VALUE init) } /* - * call-seq: - * enum.sum(init=0) -> number - * enum.sum(init=0) {|e| expr } -> number + * call-seq: + * sum(initial_value = 0) -> number + * sum(initial_value = 0) {|element| ... } -> object * - * Returns the sum of elements in an Enumerable. + * With no block given, + * returns the sum of +initial_value+ and the elements: * - * If a block is given, the block is applied to each element - * before addition. + * (1..100).sum # => 5050 + * (1..100).sum(1) # => 5051 + * ('a'..'d').sum('foo') # => "fooabcd" * - * If <i>enum</i> is empty, it returns <i>init</i>. + * Generally, the sum is computed using methods <tt>+</tt> and +each+; + * for performance optimizations, those methods may not be used, + * and so any redefinition of those methods may not have effect here. * - * For example: + * One such optimization: When possible, computes using Gauss's summation + * formula <em>n(n+1)/2</em>: * - * { 1 => 10, 2 => 20 }.sum {|k, v| k * v } #=> 50 - * (1..10).sum #=> 55 - * (1..10).sum {|v| v * 2 } #=> 110 - * [Object.new].each.sum #=> TypeError + * 100 * (100 + 1) / 2 # => 5050 * - * This method can be used for non-numeric objects by - * explicit <i>init</i> argument. + * With a block given, calls the block with each element; + * returns the sum of +initial_value+ and the block return values: * - * { 1 => 10, 2 => 20 }.sum([]) #=> [1, 10, 2, 20] - * "a\nb\nc".each_line.lazy.map(&:chomp).sum("") #=> "abc" + * (1..4).sum {|i| i*i } # => 30 + * (1..4).sum(100) {|i| i*i } # => 130 + * h = {a: 0, b: 1, c: 2, d: 3, e: 4, f: 5} + * h.sum {|key, value| value.odd? ? value : 0 } # => 9 + * ('a'..'f').sum('x') {|c| c < 'd' ? c : '' } # => "xabc" * - * Enumerable#sum method may not respect method redefinition of "+" - * methods such as Integer#+. */ static VALUE enum_sum(int argc, VALUE* argv, VALUE obj) @@ -4052,8 +4839,8 @@ enum_sum(int argc, VALUE* argv, VALUE obj) if (RTEST(rb_range_values(obj, &beg, &end, &excl))) { if (!memo.block_given && !memo.float_value && - (FIXNUM_P(beg) || RB_TYPE_P(beg, T_BIGNUM)) && - (FIXNUM_P(end) || RB_TYPE_P(end, T_BIGNUM))) { + (FIXNUM_P(beg) || RB_BIGNUM_TYPE_P(beg)) && + (FIXNUM_P(end) || RB_BIGNUM_TYPE_P(end))) { return int_range_sum(beg, end, excl, memo.v); } } @@ -4070,14 +4857,8 @@ enum_sum(int argc, VALUE* argv, VALUE obj) else { if (memo.n != 0) memo.v = rb_fix_plus(LONG2FIX(memo.n), memo.v); - if (memo.r != Qundef) { - /* r can be an Integer when mathn is loaded */ - if (FIXNUM_P(memo.r)) - memo.v = rb_fix_plus(memo.r, memo.v); - else if (RB_TYPE_P(memo.r, T_BIGNUM)) - memo.v = rb_big_plus(memo.r, memo.v); - else - memo.v = rb_rational_plus(memo.r, memo.v); + if (!UNDEF_P(memo.r)) { + memo.v = rb_rational_plus(memo.r, memo.v); } return memo.v; } @@ -4101,12 +4882,23 @@ uniq_iter(RB_BLOCK_CALL_FUNC_ARGLIST(i, hash)) /* * call-seq: - * enum.uniq -> new_ary - * enum.uniq { |item| ... } -> new_ary + * uniq -> array + * uniq {|element| ... } -> array + * + * With no block, returns a new array containing only unique elements; + * the array has no two elements +e0+ and +e1+ such that <tt>e0.eql?(e1)</tt>: * - * Returns a new array by removing duplicate values in +self+. + * %w[a b c c b a a b c].uniq # => ["a", "b", "c"] + * [0, 1, 2, 2, 1, 0, 0, 1, 2].uniq # => [0, 1, 2] + * + * With a block, returns a new array containing elements only for which the block + * returns a unique value: + * + * a = [0, 1, 2, 3, 4, 5, 5, 4, 3, 2, 1] + * a.uniq {|i| i.even? ? i : 0 } # => [0, 2, 4] + * a = %w[a b c d e e d c b a a b c d e] + * a.uniq {|c| c < 'c' } # => ["a", "c"] * - * See also Array#uniq. */ static VALUE @@ -4114,7 +4906,7 @@ enum_uniq(VALUE obj) { VALUE hash, ret; rb_block_call_func *const func = - rb_block_given_p() ? uniq_iter : uniq_func; + rb_block_given_p() ? uniq_iter : uniq_func; hash = rb_obj_hide(rb_hash_new()); rb_block_call(obj, id_each, 0, 0, func, hash); @@ -4123,22 +4915,292 @@ enum_uniq(VALUE obj) return ret; } +static VALUE +compact_i(RB_BLOCK_CALL_FUNC_ARGLIST(i, ary)) +{ + ENUM_WANT_SVALUE(); + + if (!NIL_P(i)) { + rb_ary_push(ary, i); + } + return Qnil; +} + /* - * The Enumerable mixin provides collection classes with several - * traversal and searching methods, and with the ability to sort. The - * class must provide a method #each, which yields - * successive members of the collection. If Enumerable#max, #min, or - * #sort is used, the objects in the collection must also implement a - * meaningful <code><=></code> operator, as these methods rely on an - * ordering between members of the collection. + * call-seq: + * compact -> array + * + * Returns an array of all non-+nil+ elements: + * + * a = [nil, 0, nil, 'a', false, nil, false, nil, 'a', nil, 0, nil] + * a.compact # => [0, "a", false, false, "a", 0] + * + */ + +static VALUE +enum_compact(VALUE obj) +{ + VALUE ary; + + ary = rb_ary_new(); + rb_block_call(obj, id_each, 0, 0, compact_i, ary); + + return ary; +} + + +/* + * == What's Here + * + * Module \Enumerable provides methods that are useful to a collection class for: + * + * - {Querying}[rdoc-ref:Enumerable@Methods+for+Querying] + * - {Fetching}[rdoc-ref:Enumerable@Methods+for+Fetching] + * - {Searching and Filtering}[rdoc-ref:Enumerable@Methods+for+Searching+and+Filtering] + * - {Sorting}[rdoc-ref:Enumerable@Methods+for+Sorting] + * - {Iterating}[rdoc-ref:Enumerable@Methods+for+Iterating] + * - {And more....}[rdoc-ref:Enumerable@Other+Methods] + * + * === Methods for Querying + * + * These methods return information about the \Enumerable other than the elements themselves: + * + * - #member? (aliased as #include?): Returns +true+ if <tt>self == object</tt>, +false+ otherwise. + * - #all?: Returns +true+ if all elements meet a specified criterion; +false+ otherwise. + * - #any?: Returns +true+ if any element meets a specified criterion; +false+ otherwise. + * - #none?: Returns +true+ if no element meets a specified criterion; +false+ otherwise. + * - #one?: Returns +true+ if exactly one element meets a specified criterion; +false+ otherwise. + * - #count: Returns the count of elements, + * based on an argument or block criterion, if given. + * - #tally: Returns a new Hash containing the counts of occurrences of each element. + * + * === Methods for Fetching + * + * These methods return entries from the \Enumerable, without modifying it: + * + * <i>Leading, trailing, or all elements</i>: + * + * - #to_a (aliased as #entries): Returns all elements. + * - #first: Returns the first element or leading elements. + * - #take: Returns a specified number of leading elements. + * - #drop: Returns a specified number of trailing elements. + * - #take_while: Returns leading elements as specified by the given block. + * - #drop_while: Returns trailing elements as specified by the given block. + * + * <i>Minimum and maximum value elements</i>: + * + * - #min: Returns the elements whose values are smallest among the elements, + * as determined by <tt>#<=></tt> or a given block. + * - #max: Returns the elements whose values are largest among the elements, + * as determined by <tt>#<=></tt> or a given block. + * - #minmax: Returns a 2-element Array containing the smallest and largest elements. + * - #min_by: Returns the smallest element, as determined by the given block. + * - #max_by: Returns the largest element, as determined by the given block. + * - #minmax_by: Returns the smallest and largest elements, as determined by the given block. + * + * <i>Groups, slices, and partitions</i>: + * + * - #group_by: Returns a Hash that partitions the elements into groups. + * - #partition: Returns elements partitioned into two new Arrays, as determined by the given block. + * - #slice_after: Returns a new Enumerator whose entries are a partition of +self+, + * based either on a given +object+ or a given block. + * - #slice_before: Returns a new Enumerator whose entries are a partition of +self+, + * based either on a given +object+ or a given block. + * - #slice_when: Returns a new Enumerator whose entries are a partition of +self+ + * based on the given block. + * - #chunk: Returns elements organized into chunks as specified by the given block. + * - #chunk_while: Returns elements organized into chunks as specified by the given block. + * + * === Methods for Searching and Filtering + * + * These methods return elements that meet a specified criterion: + * + * - #find (aliased as #detect): Returns an element selected by the block. + * - #find_all (aliased as #filter, #select): Returns elements selected by the block. + * - #find_index: Returns the index of an element selected by a given object or block. + * - #reject: Returns elements not rejected by the block. + * - #uniq: Returns elements that are not duplicates. + * + * === Methods for Sorting + * + * These methods return elements in sorted order: + * + * - #sort: Returns the elements, sorted by <tt>#<=></tt> or the given block. + * - #sort_by: Returns the elements, sorted by the given block. + * + * === Methods for Iterating + * + * - #each_entry: Calls the block with each successive element + * (slightly different from #each). + * - #each_with_index: Calls the block with each successive element and its index. + * - #each_with_object: Calls the block with each successive element and a given object. + * - #each_slice: Calls the block with successive non-overlapping slices. + * - #each_cons: Calls the block with successive overlapping slices. + * (different from #each_slice). + * - #reverse_each: Calls the block with each successive element, in reverse order. + * + * === Other Methods + * + * - #collect (aliased as #map): Returns objects returned by the block. + * - #filter_map: Returns truthy objects returned by the block. + * - #flat_map (aliased as #collect_concat): Returns flattened objects returned by the block. + * - #grep: Returns elements selected by a given object + * or objects returned by a given block. + * - #grep_v: Returns elements not selected by a given object + * or objects returned by a given block. + * - #inject (aliased as #reduce): Returns the object formed by combining all elements. + * - #sum: Returns the sum of the elements, using method <tt>+</tt>. + * - #zip: Combines each element with elements from other enumerables; + * returns the n-tuples or calls the block with each. + * - #cycle: Calls the block with each element, cycling repeatedly. + * + * == Usage + * + * To use module \Enumerable in a collection class: + * + * - Include it: + * + * include Enumerable + * + * - Implement method <tt>#each</tt> + * which must yield successive elements of the collection. + * The method will be called by almost any \Enumerable method. + * + * Example: + * + * class Foo + * include Enumerable + * def each + * yield 1 + * yield 1, 2 + * yield + * end + * end + * Foo.new.each_entry{ |element| p element } + * + * Output: + * + * 1 + * [1, 2] + * nil + * + * == \Enumerable in Ruby Classes + * + * These Ruby core classes include (or extend) \Enumerable: + * + * - ARGF + * - Array + * - Dir + * - Enumerator + * - ENV (extends) + * - Hash + * - IO + * - Range + * - Struct + * + * These Ruby standard library classes include \Enumerable: + * + * - CSV + * - CSV::Table + * - CSV::Row + * - Set + * + * Virtually all methods in \Enumerable call method +#each+ in the including class: + * + * - <tt>Hash#each</tt> yields the next key-value pair as a 2-element Array. + * - <tt>Struct#each</tt> yields the next name-value pair as a 2-element Array. + * - For the other classes above, +#each+ yields the next object from the collection. + * + * == About the Examples + * + * The example code snippets for the \Enumerable methods: + * + * - Always show the use of one or more Array-like classes (often Array itself). + * - Sometimes show the use of a Hash-like class. + * For some methods, though, the usage would not make sense, + * and so it is not shown. Example: #tally would find exactly one of each Hash entry. + * + * == Extended Methods + * + * A Enumerable class may define extended methods. This section describes the standard + * behavior of extension methods for reference purposes. + * + * === #size + * + * \Enumerator has a #size method. + * It uses the size function argument passed to +Enumerator.new+. + * + * e = Enumerator.new(-> { 3 }) {|y| p y; y.yield :a; y.yield :b; y.yield :c; :z } + * p e.size #=> 3 + * p e.next #=> :a + * p e.next #=> :b + * p e.next #=> :c + * begin + * e.next + * rescue StopIteration + * p $!.result #=> :z + * end + * + * The result of the size function should represent the number of iterations + * (i.e., the number of times Enumerator::Yielder#yield is called). + * In the above example, the block calls #yield three times, and + * the size function, +-> { 3 }+, returns 3 accordingly. + * The result of the size function can be an integer, +Float::INFINITY+, + * or +nil+. + * An integer means the exact number of times #yield will be called, + * as shown above. + * +Float::INFINITY+ indicates an infinite number of #yield calls. + * +nil+ means the number of #yield calls is difficult or impossible to + * determine. + * + * Many iteration methods return an \Enumerator object with an + * appropriate size function if no block is given. + * + * Examples: + * + * ["a", "b", "c"].each.size #=> 3 + * {a: "x", b: "y", c: "z"}.each.size #=> 3 + * (0..20).to_a.permutation.size #=> 51090942171709440000 + * loop.size #=> Float::INFINITY + * (1..100).drop_while.size #=> nil # size depends on the block's behavior + * STDIN.each.size #=> nil # cannot be computed without consuming input + * File.open("/etc/resolv.conf").each.size #=> nil # cannot be computed without reading the file + * + * The behavior of #size for Range-based enumerators depends on the #begin element: + * + * - If the #begin element is an Integer, the #size method returns an Integer or +Float::INFINITY+. + * - If the #begin element is an object with a #succ method (other than Integer), #size returns +nil+. + * (Computing the size would require repeatedly calling #succ, which may be too slow.) + * - If the #begin element does not have a #succ method, #size raises a TypeError. + * + * Examples: + * + * (10..42).each.size #=> 33 + * (10..42.9).each.size #=> 33 (the #end element may be a non-integer numeric) + * (10..).each.size #=> Float::INFINITY + * ("a".."z").each.size #=> nil + * ("a"..).each.size #=> nil + * (1.0..9.0).each.size # raises TypeError (Float does not have #succ) + * (..10).each.size # raises TypeError (beginless range has nil as its #begin) + * + * The \Enumerable module itself does not define a #size method. + * A class that includes \Enumerable may define its own #size method. + * It is recommended that such a #size method be consistent with + * Enumerator#size. + * + * Array and Hash implement #size and return values consistent with + * Enumerator#size. + * IO and Dir do not define #size, which is also consistent because the + * corresponding enumerator's size function returns +nil+. + * + * However, it is not strictly required for a class's #size method to match Enumerator#size. + * For example, File#size returns the number of bytes in the file, not the number of lines. + * */ void Init_Enumerable(void) { -#undef rb_intern -#define rb_intern(str) rb_intern_const(str) - rb_mEnumerable = rb_define_module("Enumerable"); rb_define_method(rb_mEnumerable, "to_a", enum_to_a, -1); @@ -4166,7 +5228,7 @@ Init_Enumerable(void) rb_define_method(rb_mEnumerable, "reduce", enum_inject, -1); rb_define_method(rb_mEnumerable, "partition", enum_partition, 0); rb_define_method(rb_mEnumerable, "group_by", enum_group_by, 0); - rb_define_method(rb_mEnumerable, "tally", enum_tally, 0); + rb_define_method(rb_mEnumerable, "tally", enum_tally, -1); rb_define_method(rb_mEnumerable, "first", enum_first, -1); rb_define_method(rb_mEnumerable, "all?", enum_all, -1); rb_define_method(rb_mEnumerable, "any?", enum_any, -1); @@ -4199,7 +5261,20 @@ Init_Enumerable(void) rb_define_method(rb_mEnumerable, "chunk_while", enum_chunk_while, 0); rb_define_method(rb_mEnumerable, "sum", enum_sum, -1); rb_define_method(rb_mEnumerable, "uniq", enum_uniq, 0); - - id_next = rb_intern("next"); - id_div = rb_intern("div"); + rb_define_method(rb_mEnumerable, "compact", enum_compact, 0); + + id__alone = rb_intern_const("_alone"); + id__separator = rb_intern_const("_separator"); + id_chunk_categorize = rb_intern_const("chunk_categorize"); + id_chunk_enumerable = rb_intern_const("chunk_enumerable"); + id_next = rb_intern_const("next"); + id_sliceafter_enum = rb_intern_const("sliceafter_enum"); + id_sliceafter_pat = rb_intern_const("sliceafter_pat"); + id_sliceafter_pred = rb_intern_const("sliceafter_pred"); + id_slicebefore_enumerable = rb_intern_const("slicebefore_enumerable"); + id_slicebefore_sep_pat = rb_intern_const("slicebefore_sep_pat"); + id_slicebefore_sep_pred = rb_intern_const("slicebefore_sep_pred"); + id_slicewhen_enum = rb_intern_const("slicewhen_enum"); + id_slicewhen_inverted = rb_intern_const("slicewhen_inverted"); + id_slicewhen_pred = rb_intern_const("slicewhen_pred"); } |