line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package Data::Enumerable::Lazy; |
2
|
|
|
|
|
|
|
|
3
|
19
|
|
|
19
|
|
1228206
|
use 5.18.2; |
|
19
|
|
|
|
|
177
|
|
4
|
|
|
|
|
|
|
|
5
|
19
|
|
|
19
|
|
97
|
use strict; |
|
19
|
|
|
|
|
28
|
|
|
19
|
|
|
|
|
451
|
|
6
|
19
|
|
|
19
|
|
87
|
use warnings; |
|
19
|
|
|
|
|
48
|
|
|
19
|
|
|
|
|
2031
|
|
7
|
|
|
|
|
|
|
|
8
|
|
|
|
|
|
|
our $VERSION = '0.031'; |
9
|
|
|
|
|
|
|
|
10
|
|
|
|
|
|
|
=pod |
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
=head1 NAME |
13
|
|
|
|
|
|
|
|
14
|
|
|
|
|
|
|
Data::Enumerable::Lazy - Lazy generator + enumerable for Perl5. |
15
|
|
|
|
|
|
|
|
16
|
|
|
|
|
|
|
=head1 SYNOPSIS |
17
|
|
|
|
|
|
|
|
18
|
|
|
|
|
|
|
A basic lazy range implementation picking even numbers only: |
19
|
|
|
|
|
|
|
|
20
|
|
|
|
|
|
|
my ($from, $to) = (0, 10); |
21
|
|
|
|
|
|
|
my $current = $from; |
22
|
|
|
|
|
|
|
my $tream = Data::Enumerable::Lazy->new({ |
23
|
|
|
|
|
|
|
on_has_next => sub { $current <= $to }, |
24
|
|
|
|
|
|
|
on_next => sub { shift->yield($current++) }, |
25
|
|
|
|
|
|
|
})->grep(sub{ shift % 2 == 0 }); |
26
|
|
|
|
|
|
|
$tream->to_list(); # generates: [0, 2, 4, 6, 8, 10] |
27
|
|
|
|
|
|
|
|
28
|
|
|
|
|
|
|
=head2 DESCRIPTION |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
This library is another one implementation of a lazy generator + enumerable |
31
|
|
|
|
|
|
|
for Perl5. It might be handy if the elements of the collection are resolved on |
32
|
|
|
|
|
|
|
the flight and the iteration itself should be hidden from the end users. |
33
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
The enumerables are single-pass composable calculation units. What it means: |
35
|
|
|
|
|
|
|
An enumerable is stateful, once it reached the end of the sequence, it will |
36
|
|
|
|
|
|
|
not rewind to the beginning unless explicitly forced to. |
37
|
|
|
|
|
|
|
Enumerables are composable: one enumerable might be an extension of another by |
38
|
|
|
|
|
|
|
applying some additional logic. Enumerables resolve steps on demand, one by one. |
39
|
|
|
|
|
|
|
A single step might return another enumerable (micro batches). The library |
40
|
|
|
|
|
|
|
flattens these enumerables, so for the end user this looks like a single |
41
|
|
|
|
|
|
|
continuous sequence of elements. |
42
|
|
|
|
|
|
|
|
43
|
|
|
|
|
|
|
|
44
|
|
|
|
|
|
|
[enumerable.has_next] -> [_buffer.has_next] -> yes -> return true |
45
|
|
|
|
|
|
|
-> no -> result = [enumerable.on_has_next] -> return result |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
[enumerable.next] -> [_buffer.has_next] -> yes -> return [_buffer.next] |
48
|
|
|
|
|
|
|
-> no -> result = [enumerable.next] -> [enumerable.set_buffer(result)] -> return result |
49
|
|
|
|
|
|
|
|
50
|
|
|
|
|
|
|
=head1 EXAMPLES |
51
|
|
|
|
|
|
|
|
52
|
|
|
|
|
|
|
=head2 A basic range |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
This example implements a range generator from $from until $to. In order to |
55
|
|
|
|
|
|
|
generate this range we define 2 callbacks: C and C. |
56
|
|
|
|
|
|
|
The first one is used as point of truth whether the sequence has any more |
57
|
|
|
|
|
|
|
non-iterated elements, and the 2nd one is here to return the next element in |
58
|
|
|
|
|
|
|
the sequence and the one that changes the state of the internal sequence |
59
|
|
|
|
|
|
|
iterator. |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
sub basic_range { |
62
|
|
|
|
|
|
|
my ($from, $to) = @_; |
63
|
|
|
|
|
|
|
$from <= $to or die '$from should be less or equal $to'; |
64
|
|
|
|
|
|
|
my $current = $from; |
65
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
66
|
|
|
|
|
|
|
on_has_next => sub { |
67
|
|
|
|
|
|
|
return $current <= $to; |
68
|
|
|
|
|
|
|
}, |
69
|
|
|
|
|
|
|
on_next => sub { |
70
|
|
|
|
|
|
|
my ($self) = @_; |
71
|
|
|
|
|
|
|
return $self->yield($current++); |
72
|
|
|
|
|
|
|
}, |
73
|
|
|
|
|
|
|
}); |
74
|
|
|
|
|
|
|
} |
75
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
on_has_next() makes sure the current value does not exceed $to value, and |
77
|
|
|
|
|
|
|
on_next() yields the next value of the sequence. Note the yield method. |
78
|
|
|
|
|
|
|
An enumerable developer is expected to use this method in order to return |
79
|
|
|
|
|
|
|
the next step value. This method does some internal bookkeeping and smart |
80
|
|
|
|
|
|
|
caching. |
81
|
|
|
|
|
|
|
|
82
|
|
|
|
|
|
|
Usage: |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
# We initialize a new range generator from 0 to 10 including. |
85
|
|
|
|
|
|
|
my $range = basic_range(0, 10); |
86
|
|
|
|
|
|
|
# We check if the sequence has elements in it's tail. |
87
|
|
|
|
|
|
|
while ($range->has_next) { |
88
|
|
|
|
|
|
|
# In this very line the state of $range is being changed |
89
|
|
|
|
|
|
|
say $range->next; |
90
|
|
|
|
|
|
|
} |
91
|
|
|
|
|
|
|
|
92
|
|
|
|
|
|
|
is $range->has_next, 0, '$range has been iterated completely' |
93
|
|
|
|
|
|
|
is $range->next, undef, 'A fully iterated sequence returns undef on next()' |
94
|
|
|
|
|
|
|
|
95
|
|
|
|
|
|
|
=head2 Prime numbers |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
Prime numbers is an infinite sequence of natural numbers. This example |
98
|
|
|
|
|
|
|
implements a very naive suboptimal prime number generator. |
99
|
|
|
|
|
|
|
|
100
|
|
|
|
|
|
|
my $prime_num_stream = Data::Enumerable::Lazy->new({ |
101
|
|
|
|
|
|
|
# This is an infinite sequence |
102
|
|
|
|
|
|
|
on_has_next => sub { 1 }, |
103
|
|
|
|
|
|
|
on_next => sub { |
104
|
|
|
|
|
|
|
my $self = shift; |
105
|
|
|
|
|
|
|
# We save the result of the previous step |
106
|
|
|
|
|
|
|
my $next = $self->{_prev_} // 1; |
107
|
|
|
|
|
|
|
LOOKUP: while (1) { |
108
|
|
|
|
|
|
|
$next++; |
109
|
|
|
|
|
|
|
# Check all numbers from 2 to sqrt(N) |
110
|
|
|
|
|
|
|
foreach (2..floor(sqrt($next))) { |
111
|
|
|
|
|
|
|
($next % $_ == 0) and next LOOKUP; |
112
|
|
|
|
|
|
|
} |
113
|
|
|
|
|
|
|
last LOOKUP; |
114
|
|
|
|
|
|
|
} |
115
|
|
|
|
|
|
|
# Save the result in order to use it in the next step |
116
|
|
|
|
|
|
|
$self->{_prev_} = $next; |
117
|
|
|
|
|
|
|
# Return the result |
118
|
|
|
|
|
|
|
$self->yield($next); |
119
|
|
|
|
|
|
|
}, |
120
|
|
|
|
|
|
|
}); |
121
|
|
|
|
|
|
|
|
122
|
|
|
|
|
|
|
What's remarkable regarding this specific example is that one can not simply |
123
|
|
|
|
|
|
|
call C in order to get all elements of the sequence. The enumerable |
124
|
|
|
|
|
|
|
will throw an exception claiming it's an infinitive sequence. Therefore, we |
125
|
|
|
|
|
|
|
should use C in order to get elements one by one or use another handy |
126
|
|
|
|
|
|
|
method C which returns first N results. |
127
|
|
|
|
|
|
|
|
128
|
|
|
|
|
|
|
=head2 Flat enumeration (Nested generators) |
129
|
|
|
|
|
|
|
|
130
|
|
|
|
|
|
|
In this example we will output a numbers of a multiplication table 10x10. |
131
|
|
|
|
|
|
|
What's interesting in this example is that there are 2 levels of sequences: |
132
|
|
|
|
|
|
|
primary and secondary. Primary C returns secondary sequence, which |
133
|
|
|
|
|
|
|
multiplicates 2 numbers.. |
134
|
|
|
|
|
|
|
|
135
|
|
|
|
|
|
|
# A new stream based on a range from 1 to 10 |
136
|
|
|
|
|
|
|
my $mult_table = Data::Enumerable::Lazy->from_list(1..10)->continue({ |
137
|
|
|
|
|
|
|
on_has_next => sub { |
138
|
|
|
|
|
|
|
my ($self, $i) = @_; |
139
|
|
|
|
|
|
|
# The primary stream returns another sequence, based on range |
140
|
|
|
|
|
|
|
$self->yield(Data::Enumerable::Lazy->from_list(1..10)->continue({ |
141
|
|
|
|
|
|
|
on_next => sub { |
142
|
|
|
|
|
|
|
# $_[0] is a substream self |
143
|
|
|
|
|
|
|
# $_[1] is a next substream sequence element |
144
|
|
|
|
|
|
|
$_[0]->yield( $_[1] * $i ) |
145
|
|
|
|
|
|
|
}, |
146
|
|
|
|
|
|
|
})); |
147
|
|
|
|
|
|
|
}, |
148
|
|
|
|
|
|
|
}); |
149
|
|
|
|
|
|
|
|
150
|
|
|
|
|
|
|
Another feature which is represented here is the nested result generation. |
151
|
|
|
|
|
|
|
Let's walk trough the sequence generation step by step and see what happens. |
152
|
|
|
|
|
|
|
|
153
|
|
|
|
|
|
|
$mult_table->has_next; # returns true based on the primary range, _buffer is |
154
|
|
|
|
|
|
|
# empty |
155
|
|
|
|
|
|
|
$mult_table->next; # returns 1, the secondary sequence is now stored as |
156
|
|
|
|
|
|
|
# the primary enumerable buffer and 1 is being served |
157
|
|
|
|
|
|
|
# from this buffer |
158
|
|
|
|
|
|
|
$mult_table->has_next; # returns true, resolved by the state of the buffer |
159
|
|
|
|
|
|
|
$mult_table->next; # returns 2, moves buffer iterator forward, the |
160
|
|
|
|
|
|
|
# primary sequence on_next() is _not_ being called |
161
|
|
|
|
|
|
|
# this time |
162
|
|
|
|
|
|
|
$mult_table->next for (3..10); # The last iteration completes the buffer |
163
|
|
|
|
|
|
|
# iteration cycle |
164
|
|
|
|
|
|
|
$mult_table->has_next; # returns true, but now it calls the primary |
165
|
|
|
|
|
|
|
# on_has_next() |
166
|
|
|
|
|
|
|
$mult_table->next; # returns 2 as the first element in the next |
167
|
|
|
|
|
|
|
# secondary sequence (which is 1 again) multiplied by |
168
|
|
|
|
|
|
|
# the 2nd element of the primary sequence (which is 2) |
169
|
|
|
|
|
|
|
$mult_table->to_list; # Generates the tail of the sesquence: |
170
|
|
|
|
|
|
|
# [4, 6, ..., 80, 90, 100] |
171
|
|
|
|
|
|
|
$mult_table->has_next; # returns false as the buffer is empty now and the |
172
|
|
|
|
|
|
|
# primary sequence on_has_next() says there is nothing |
173
|
|
|
|
|
|
|
# more to iterate over. |
174
|
|
|
|
|
|
|
|
175
|
|
|
|
|
|
|
=head2 DBI paginator example |
176
|
|
|
|
|
|
|
|
177
|
|
|
|
|
|
|
As mentioned earlier, lazy enumerables are useful when the number of the |
178
|
|
|
|
|
|
|
elements in the sequence is not known in advance. So far, we were looking at |
179
|
|
|
|
|
|
|
some synthetic examples, but the majority of us are not being paid for prime |
180
|
|
|
|
|
|
|
number generators. Hands on some real life example. Say, we have a table and |
181
|
|
|
|
|
|
|
we want to iterate over all entries in the table, and we want the data to be |
182
|
|
|
|
|
|
|
retrieved in batches by 10 elements in order to reduce the number of queries. |
183
|
|
|
|
|
|
|
We don't want to compute the number of steps in advance, as the number might |
184
|
|
|
|
|
|
|
be inaccurate: let's assume we're paginating over some new tweets and the new |
185
|
|
|
|
|
|
|
entries might be created on the flight. |
186
|
|
|
|
|
|
|
|
187
|
|
|
|
|
|
|
use DBI; |
188
|
|
|
|
|
|
|
my $dbh = setup_dbh(); # Some config |
189
|
|
|
|
|
|
|
|
190
|
|
|
|
|
|
|
my $last_id = -1; |
191
|
|
|
|
|
|
|
my $limit = 10; |
192
|
|
|
|
|
|
|
my $offset = 0; |
193
|
|
|
|
|
|
|
my $tweet_enum = Data::Enumerable::Lazy->new({ |
194
|
|
|
|
|
|
|
on_has_next => sub { |
195
|
|
|
|
|
|
|
my $sth = $dbh->prepare('SELECT count(1) from Tweets where id > ?'); |
196
|
|
|
|
|
|
|
$sth->execute($last_id); |
197
|
|
|
|
|
|
|
my ($cnt) = $sth->fetchrow_array; |
198
|
|
|
|
|
|
|
return int($cnt) > 0; |
199
|
|
|
|
|
|
|
}, |
200
|
|
|
|
|
|
|
on_next => sub { |
201
|
|
|
|
|
|
|
my ($self) = @_; |
202
|
|
|
|
|
|
|
my $sth = $dbh->prepare('SELECT * from Tweets ORDER BY id LIMIT ? OFFSET ?'); |
203
|
|
|
|
|
|
|
$sth->execute($lmit, $offset); |
204
|
|
|
|
|
|
|
$offset += $limit; |
205
|
|
|
|
|
|
|
my @tweets = $sth->fetchrow_array; |
206
|
|
|
|
|
|
|
$last_id = $tweets[-1]->{id}; |
207
|
|
|
|
|
|
|
$self->yield(Data::Enumerable::Lazy->from_list(@tweets)); |
208
|
|
|
|
|
|
|
}, |
209
|
|
|
|
|
|
|
is_finite => 1, |
210
|
|
|
|
|
|
|
}); |
211
|
|
|
|
|
|
|
|
212
|
|
|
|
|
|
|
while ($tweet_enum->has_next) { |
213
|
|
|
|
|
|
|
my $tweet = $tweet_enum->next; |
214
|
|
|
|
|
|
|
# do something with this tweet |
215
|
|
|
|
|
|
|
} |
216
|
|
|
|
|
|
|
|
217
|
|
|
|
|
|
|
In this example a tweet consumer is abstracted from any DBI bookkeeping and |
218
|
|
|
|
|
|
|
consumes tweet entries one by one without any prior knowledge about the table |
219
|
|
|
|
|
|
|
size and might work on a rapidly growing dataset. |
220
|
|
|
|
|
|
|
|
221
|
|
|
|
|
|
|
In order to reduce the number of queries, we query the data in batches by 10 |
222
|
|
|
|
|
|
|
elements max. |
223
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
=head2 Redis queue consumer |
225
|
|
|
|
|
|
|
|
226
|
|
|
|
|
|
|
use Redis; |
227
|
|
|
|
|
|
|
|
228
|
|
|
|
|
|
|
my $redis = Redis->new; |
229
|
|
|
|
|
|
|
my $queue_enum = Data::Enumerable::Lazy->new({ |
230
|
|
|
|
|
|
|
on_has_next => sub { 1 }, |
231
|
|
|
|
|
|
|
on_next => sub { |
232
|
|
|
|
|
|
|
# Blocking right POP |
233
|
|
|
|
|
|
|
$redis->brpop(); |
234
|
|
|
|
|
|
|
}, |
235
|
|
|
|
|
|
|
}); |
236
|
|
|
|
|
|
|
|
237
|
|
|
|
|
|
|
while (my $queue_item = $queue_enum->next) { |
238
|
|
|
|
|
|
|
# do something with the queue item |
239
|
|
|
|
|
|
|
} |
240
|
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
In this example the client is blocked until there is an element available in |
242
|
|
|
|
|
|
|
the queue, but it's hidden away from the clients who consume the data item by |
243
|
|
|
|
|
|
|
item. |
244
|
|
|
|
|
|
|
|
245
|
|
|
|
|
|
|
=head2 Kafka example |
246
|
|
|
|
|
|
|
|
247
|
|
|
|
|
|
|
Kafka consumer wrapper is another example of a lazy calculation application. |
248
|
|
|
|
|
|
|
Lazy enumerables are very naturally co-operated with streaming data, like |
249
|
|
|
|
|
|
|
Kafka. In this example we're fetching batches of messages from Kafka topic, |
250
|
|
|
|
|
|
|
grep out corrupted ones and proceed with the mesages. |
251
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
use Kafka qw($DEFAULT_MAX_BYTES); |
253
|
|
|
|
|
|
|
use Kafka::Connection; |
254
|
|
|
|
|
|
|
use Kafka::Consumer; |
255
|
|
|
|
|
|
|
|
256
|
|
|
|
|
|
|
my $kafka_consumer = Kafka::Consumer->new( |
257
|
|
|
|
|
|
|
Connection => Kafka::Connection->new( host => 'localhost', ), |
258
|
|
|
|
|
|
|
); |
259
|
|
|
|
|
|
|
|
260
|
|
|
|
|
|
|
my $partition = 0; |
261
|
|
|
|
|
|
|
my $offset = 0; |
262
|
|
|
|
|
|
|
my $kafka_enum = Data::Enumerable::Lazy->new({ |
263
|
|
|
|
|
|
|
on_has_next => sub { 1 }, |
264
|
|
|
|
|
|
|
on_next => sub { |
265
|
|
|
|
|
|
|
my ($self) = @_; |
266
|
|
|
|
|
|
|
# Fetch messages in batch |
267
|
|
|
|
|
|
|
my $messages = $kafka_consumer->fetch({ |
268
|
|
|
|
|
|
|
'topic', |
269
|
|
|
|
|
|
|
$partition, |
270
|
|
|
|
|
|
|
$offset, |
271
|
|
|
|
|
|
|
$DEFAULT_MAX_BYTES |
272
|
|
|
|
|
|
|
}); |
273
|
|
|
|
|
|
|
if ($messages) { |
274
|
|
|
|
|
|
|
# Note the grep function applied: we're filtering away corrupted messages |
275
|
|
|
|
|
|
|
$self->yield(Data::Enumerable::Lazy->from_list(@$messages))->grep(sub { $_[0]->valid }); |
276
|
|
|
|
|
|
|
} else { |
277
|
|
|
|
|
|
|
# If there are no more messages, we return an empty enum, this is |
278
|
|
|
|
|
|
|
# another handy use-case for nested enums. |
279
|
|
|
|
|
|
|
$self->yield(Data::Enumerable::Lazy->empty); |
280
|
|
|
|
|
|
|
} |
281
|
|
|
|
|
|
|
}, |
282
|
|
|
|
|
|
|
}); |
283
|
|
|
|
|
|
|
|
284
|
|
|
|
|
|
|
while (my $message = $kafka_enum->next) { |
285
|
|
|
|
|
|
|
# handle the message |
286
|
|
|
|
|
|
|
} |
287
|
|
|
|
|
|
|
|
288
|
|
|
|
|
|
|
=cut |
289
|
|
|
|
|
|
|
|
290
|
|
|
|
|
|
|
=head1 INSTALLATION |
291
|
|
|
|
|
|
|
|
292
|
|
|
|
|
|
|
To install this module type the following: |
293
|
|
|
|
|
|
|
perl Makefile.PL |
294
|
|
|
|
|
|
|
make |
295
|
|
|
|
|
|
|
make test |
296
|
|
|
|
|
|
|
make install |
297
|
|
|
|
|
|
|
|
298
|
|
|
|
|
|
|
=cut |
299
|
|
|
|
|
|
|
|
300
|
19
|
|
|
19
|
|
121
|
use Carp; |
|
19
|
|
|
|
|
41
|
|
|
19
|
|
|
|
|
1206
|
|
301
|
19
|
|
|
19
|
|
130
|
use List::Util; |
|
19
|
|
|
|
|
49
|
|
|
19
|
|
|
|
|
48795
|
|
302
|
|
|
|
|
|
|
|
303
|
|
|
|
|
|
|
sub new { |
304
|
1563
|
|
|
1563
|
0
|
6676
|
my ($class, $opts) = @_; |
305
|
1563
|
|
|
|
|
7302
|
return bless({ _opts => $opts, _buff => undef }, $class); |
306
|
|
|
|
|
|
|
} |
307
|
|
|
|
|
|
|
|
308
|
|
|
|
|
|
|
=head1 OPTIONS |
309
|
|
|
|
|
|
|
|
310
|
|
|
|
|
|
|
=head2 on_next($self, $element) :: CodeRef -> Data::Enumerable::Lazy | Any |
311
|
|
|
|
|
|
|
|
312
|
|
|
|
|
|
|
C is a code ref, a callback which is being called every time the |
313
|
|
|
|
|
|
|
generator is in demand for a new bit of data. Enumerable buffers up the result |
314
|
|
|
|
|
|
|
of the previous calculation and if there are no more elements left in the |
315
|
|
|
|
|
|
|
buffer, C would be called. |
316
|
|
|
|
|
|
|
|
317
|
|
|
|
|
|
|
C<$element> is defined when the current collection is a contuniation of another |
318
|
|
|
|
|
|
|
enumerable. I.e.: |
319
|
|
|
|
|
|
|
my $enum = Data::Enumerable::Lazy->from_list(1, 2, 3); |
320
|
|
|
|
|
|
|
my $enum2 = $enum->continue({ |
321
|
|
|
|
|
|
|
on_next => sub { my ($self, $i) = @_; $self->yield($i * $i) } |
322
|
|
|
|
|
|
|
}); |
323
|
|
|
|
|
|
|
$enum2->to_list; # generates 1, 4, 9 |
324
|
|
|
|
|
|
|
In this case $i would be defined and it comes from the original enumerable. |
325
|
|
|
|
|
|
|
|
326
|
|
|
|
|
|
|
The function is supposed to return an enumerable, in this case it would be |
327
|
|
|
|
|
|
|
kept as the buffer object. If this function method returns any other value, |
328
|
|
|
|
|
|
|
it would be wrapped in a Csingular()>. There is a |
329
|
|
|
|
|
|
|
way to prevent an enumerable from wrapping your return value in an enum and |
330
|
|
|
|
|
|
|
keeping it in a raw state by providing C. |
331
|
|
|
|
|
|
|
|
332
|
|
|
|
|
|
|
=cut |
333
|
|
|
|
|
|
|
|
334
|
3229
|
|
100
|
3
|
1
|
7444
|
sub on_next { $_[0]->{_opts}->{on_next} // sub {} } |
|
|
|
|
3229
|
|
|
|
335
|
|
|
|
|
|
|
|
336
|
|
|
|
|
|
|
=head2 on_has_next($self) :: CodeRef -> Bool |
337
|
|
|
|
|
|
|
|
338
|
|
|
|
|
|
|
C is a code ref, a callback to be called whenever the enumerable |
339
|
|
|
|
|
|
|
is about to resolve C method call. Similar to C call, |
340
|
|
|
|
|
|
|
this one is also triggered whenever an enumerable runs out of buffered |
341
|
|
|
|
|
|
|
elements. The function shoiuld return boolean. |
342
|
|
|
|
|
|
|
|
343
|
|
|
|
|
|
|
A method that returns 1 all the time is the way to initialize an infinite |
344
|
|
|
|
|
|
|
enumerable (see C). If it returns 0 no matter what, it would be |
345
|
|
|
|
|
|
|
an empty enumerable (see C). Normally you want to stay somewhere in |
346
|
|
|
|
|
|
|
the middle and implement some state check login in there. |
347
|
|
|
|
|
|
|
|
348
|
|
|
|
|
|
|
=cut |
349
|
|
|
|
|
|
|
|
350
|
46
|
|
100
|
46
|
1
|
134
|
sub on_has_next { $_[0]->{_opts}->{on_has_next} // sub {0} } |
|
5214
|
|
|
5214
|
|
12119
|
|
351
|
|
|
|
|
|
|
|
352
|
|
|
|
|
|
|
=head2 on_reset($self) :: CodeRef -> void |
353
|
|
|
|
|
|
|
|
354
|
|
|
|
|
|
|
This is a callback to be called in order to reset the state of the enumerable. |
355
|
|
|
|
|
|
|
This callback should be defined in the same scope as the enumerable itself. |
356
|
|
|
|
|
|
|
The library provides nothing magical but a callback and a handle to call it, |
357
|
|
|
|
|
|
|
so the state cleanup is completely on the developer's side. |
358
|
|
|
|
|
|
|
|
359
|
|
|
|
|
|
|
=cut |
360
|
|
|
|
|
|
|
|
361
|
0
|
|
0
|
0
|
1
|
0
|
sub on_reset { $_[0]->{_opts}->{on_reset} // sub {} } |
|
|
|
|
0
|
|
|
|
362
|
|
|
|
|
|
|
|
363
|
|
|
|
|
|
|
=head2 is_finite :: Bool |
364
|
|
|
|
|
|
|
|
365
|
|
|
|
|
|
|
A boolean flag indicating whether an enumerable is finite or not. By default |
366
|
|
|
|
|
|
|
enumerables are treated as infinite, which means some functions will throw |
367
|
|
|
|
|
|
|
an exception, like: C or C. |
368
|
|
|
|
|
|
|
|
369
|
|
|
|
|
|
|
Make sure to not mark an enumerable as finite and to call finite-size defined |
370
|
|
|
|
|
|
|
methods, in this case it will create an infinite loop on the resolution. |
371
|
|
|
|
|
|
|
|
372
|
|
|
|
|
|
|
=cut |
373
|
|
|
|
|
|
|
|
374
|
78
|
|
50
|
78
|
1
|
444
|
sub is_finite { $_[0]->{_opts}->{is_finite} // 0 } |
375
|
|
|
|
|
|
|
|
376
|
10021
|
|
100
|
10021
|
0
|
30780
|
sub no_wrap { $_[0]->{_opts}->{no_wrap} // 0 } |
377
|
|
|
|
|
|
|
|
378
|
|
|
|
|
|
|
=head1 INSTANCE METHODS |
379
|
|
|
|
|
|
|
|
380
|
|
|
|
|
|
|
=head2 next() |
381
|
|
|
|
|
|
|
|
382
|
|
|
|
|
|
|
Function C is the primary interface for accessing elements of an |
383
|
|
|
|
|
|
|
enumerable. It will do some internal checks and if there is no elements to be |
384
|
|
|
|
|
|
|
served from an intermediate buffer, it will resolve the next step by calling |
385
|
|
|
|
|
|
|
C callback. |
386
|
|
|
|
|
|
|
Enumerables are composable: one enumerable might be based on another |
387
|
|
|
|
|
|
|
enumeration. E.g.: a sequence of natural number squares is based on the |
388
|
|
|
|
|
|
|
sequence of natural numbers themselves. In other words, a sequence is defined |
389
|
|
|
|
|
|
|
as a tuple of another sequence and a function which would be lazily applied to |
390
|
|
|
|
|
|
|
every element of this sequence. |
391
|
|
|
|
|
|
|
|
392
|
|
|
|
|
|
|
C accepts 0 or more arguments, which would be passed to C |
393
|
|
|
|
|
|
|
callback. |
394
|
|
|
|
|
|
|
|
395
|
|
|
|
|
|
|
C is expected to do the heavy-lifting job in opposite to C, |
396
|
|
|
|
|
|
|
which is supposed to be cheap and fast. This statement flips upside down |
397
|
|
|
|
|
|
|
whenever C is applied to a stream. See C for more details. |
398
|
|
|
|
|
|
|
|
399
|
|
|
|
|
|
|
=cut |
400
|
|
|
|
|
|
|
|
401
|
|
|
|
|
|
|
sub next { |
402
|
3431
|
|
|
3431
|
1
|
5448
|
my $self = shift; |
403
|
3431
|
|
|
|
|
4276
|
my $res; |
404
|
3431
|
100
|
100
|
|
|
7315
|
unless ($self->{_buff} && $self->{_buff}->has_next()) { |
405
|
3229
|
|
|
|
|
5746
|
$res = $self->on_next()->($self, @_); |
406
|
3229
|
100
|
|
|
|
5831
|
$self->{_buff} = $res |
407
|
|
|
|
|
|
|
unless $self->no_wrap(); |
408
|
|
|
|
|
|
|
} |
409
|
3431
|
100
|
|
|
|
5810
|
my $return = $self->no_wrap() ? $res : $self->{_buff}->next(); |
410
|
3431
|
|
|
|
|
7090
|
return $return; |
411
|
|
|
|
|
|
|
} |
412
|
|
|
|
|
|
|
|
413
|
|
|
|
|
|
|
=head2 has_next() |
414
|
|
|
|
|
|
|
|
415
|
|
|
|
|
|
|
C is the primary entry point to get an information about the state |
416
|
|
|
|
|
|
|
of an enumerable. If the method returned false, there are no more elements to be |
417
|
|
|
|
|
|
|
consumed. I.e. the sequence has been iterated completely. Normally it means |
418
|
|
|
|
|
|
|
the end of an iteration cycle. |
419
|
|
|
|
|
|
|
|
420
|
|
|
|
|
|
|
Enumerables use internal buffers in order to support batched C |
421
|
|
|
|
|
|
|
resolutions. If there are some elements left in the buffer, C |
422
|
|
|
|
|
|
|
won't call C callback immediately. If the buffer has been |
423
|
|
|
|
|
|
|
iterated completely, C would be called. |
424
|
|
|
|
|
|
|
|
425
|
|
|
|
|
|
|
C should be fast on resolving the state of an enumerable as it's going |
426
|
|
|
|
|
|
|
to be used for a condition state check. |
427
|
|
|
|
|
|
|
|
428
|
|
|
|
|
|
|
=cut |
429
|
|
|
|
|
|
|
|
430
|
|
|
|
|
|
|
sub has_next { |
431
|
5399
|
|
|
5399
|
1
|
12201
|
my $self = shift; |
432
|
5399
|
|
|
|
|
6639
|
my $res; |
433
|
|
|
|
|
|
|
eval { |
434
|
5399
|
|
100
|
|
|
8423
|
$res = $self->_has_next_in_buffer() || |
435
|
|
|
|
|
|
|
$self->_has_next_in_generator(); |
436
|
5399
|
|
|
|
|
13506
|
1; |
437
|
5399
|
50
|
|
|
|
6864
|
} or do { |
438
|
0
|
|
0
|
|
|
0
|
croak sprintf('Problem calling on_has_next(): %s', $@ // 'zombie error'); |
439
|
|
|
|
|
|
|
}; |
440
|
5399
|
|
|
|
|
13989
|
return int $res; |
441
|
|
|
|
|
|
|
} |
442
|
|
|
|
|
|
|
|
443
|
|
|
|
|
|
|
=head2 reset() |
444
|
|
|
|
|
|
|
|
445
|
|
|
|
|
|
|
This method is a generic entry point for a enum reset. In fact, it is basically |
446
|
|
|
|
|
|
|
a wrapper around user-defined C. Use with caution: if C |
447
|
|
|
|
|
|
|
was not defined, it will reset the buffer and might cause a partial calculation |
448
|
|
|
|
|
|
|
skip (reset implicitly clears the internal buffer) if the buffer was not fully |
449
|
|
|
|
|
|
|
iterated yet. |
450
|
|
|
|
|
|
|
|
451
|
|
|
|
|
|
|
=cut |
452
|
|
|
|
|
|
|
|
453
|
|
|
|
|
|
|
sub reset { |
454
|
0
|
|
|
0
|
1
|
0
|
my $self = shift; |
455
|
0
|
|
|
|
|
0
|
$self->{_buff} = undef; |
456
|
0
|
0
|
|
|
|
0
|
eval { $self->on_reset(); 1 } or do { |
|
0
|
|
|
|
|
0
|
|
|
0
|
|
|
|
|
0
|
|
457
|
0
|
|
0
|
|
|
0
|
croak sprintf('Problem calling on_reset(): %s', $@ // 'zombie error'); |
458
|
|
|
|
|
|
|
}; |
459
|
|
|
|
|
|
|
} |
460
|
|
|
|
|
|
|
|
461
|
|
|
|
|
|
|
=head2 to_list() |
462
|
|
|
|
|
|
|
|
463
|
|
|
|
|
|
|
This function transforms a lazy enumerable to a list. Only finite enumerables |
464
|
|
|
|
|
|
|
can be transformed to a list, so the method checks if an enumerable is created |
465
|
|
|
|
|
|
|
with C flag. An exception would be thrown otherwise. |
466
|
|
|
|
|
|
|
|
467
|
|
|
|
|
|
|
=cut |
468
|
|
|
|
|
|
|
|
469
|
|
|
|
|
|
|
sub to_list { |
470
|
24
|
|
|
24
|
1
|
110
|
my ($self) = @_; |
471
|
24
|
50
|
|
|
|
52
|
croak 'Only finite enumerables might be converted to list. Use is_finite=1' |
472
|
|
|
|
|
|
|
unless $self->is_finite(); |
473
|
24
|
|
|
|
|
50
|
my @acc; |
474
|
24
|
|
|
|
|
59
|
push @acc, $self->next() while $self->has_next(); |
475
|
24
|
|
|
|
|
140
|
return \@acc; |
476
|
|
|
|
|
|
|
} |
477
|
|
|
|
|
|
|
|
478
|
|
|
|
|
|
|
=head2 map($callback) |
479
|
|
|
|
|
|
|
|
480
|
|
|
|
|
|
|
Creates a new enumerable by applying a user-defined function to the original |
481
|
|
|
|
|
|
|
enumerable. Works the same way as perl map {} function but it's lazy. |
482
|
|
|
|
|
|
|
|
483
|
|
|
|
|
|
|
Example |
484
|
|
|
|
|
|
|
Data::Enumerable::Lazy |
485
|
|
|
|
|
|
|
->from_array(1, 2, 3) |
486
|
|
|
|
|
|
|
->map(sub { |
487
|
|
|
|
|
|
|
my ($number) = @_; |
488
|
|
|
|
|
|
|
return $number * $number |
489
|
|
|
|
|
|
|
}); |
490
|
|
|
|
|
|
|
|
491
|
|
|
|
|
|
|
=cut |
492
|
|
|
|
|
|
|
|
493
|
|
|
|
|
|
|
sub map { |
494
|
1
|
|
|
1
|
1
|
2
|
my ($self, $callback) = @_; |
495
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
496
|
|
|
|
|
|
|
on_has_next => $self->on_has_next(), |
497
|
10
|
|
|
10
|
|
20
|
on_next => sub { shift->yield($callback->($self->next())) }, |
498
|
1
|
|
|
|
|
2
|
is_finite => $self->is_finite(), |
499
|
|
|
|
|
|
|
no_wrap => $self->no_wrap(), |
500
|
|
|
|
|
|
|
}); |
501
|
|
|
|
|
|
|
} |
502
|
|
|
|
|
|
|
|
503
|
|
|
|
|
|
|
=head2 reduce($acc, $callback) |
504
|
|
|
|
|
|
|
|
505
|
|
|
|
|
|
|
Resolves the enumerable and returns the resulting state of the accumulator $acc |
506
|
|
|
|
|
|
|
provided as the 1st argument. C<$callback> should always return the new state of |
507
|
|
|
|
|
|
|
C<$acc>. |
508
|
|
|
|
|
|
|
|
509
|
|
|
|
|
|
|
C is defined for finite enumerables only. |
510
|
|
|
|
|
|
|
|
511
|
|
|
|
|
|
|
Example |
512
|
|
|
|
|
|
|
Data::Enumerable::Lazy |
513
|
|
|
|
|
|
|
->from_array(1, 2, 3) |
514
|
|
|
|
|
|
|
->reduce(1, sub { |
515
|
|
|
|
|
|
|
my ($acc, $number) = @_; |
516
|
|
|
|
|
|
|
return $acc *= $number |
517
|
|
|
|
|
|
|
}); |
518
|
|
|
|
|
|
|
|
519
|
|
|
|
|
|
|
=cut |
520
|
|
|
|
|
|
|
|
521
|
|
|
|
|
|
|
sub reduce { |
522
|
2
|
|
|
2
|
1
|
5
|
my ($self, $acc, $callback) = @_; |
523
|
2
|
50
|
|
|
|
5
|
croak 'Only finite enumerables might be reduced. Use is_finite=1' |
524
|
|
|
|
|
|
|
unless $self->is_finite(); |
525
|
2
|
|
|
|
|
7
|
($acc = $callback->($acc, $self->next())) while $self->has_next(); |
526
|
2
|
|
|
|
|
4
|
return $acc; |
527
|
|
|
|
|
|
|
} |
528
|
|
|
|
|
|
|
|
529
|
|
|
|
|
|
|
=head2 grep($callback, $max_lookahead) |
530
|
|
|
|
|
|
|
|
531
|
|
|
|
|
|
|
C is a function which returns a new enumerable by applying a |
532
|
|
|
|
|
|
|
user-defined filter function. |
533
|
|
|
|
|
|
|
|
534
|
|
|
|
|
|
|
C might be applied to both finite and infinite enumerables. In case of |
535
|
|
|
|
|
|
|
an infinitive enumerable there is an additional argument specifying max number |
536
|
|
|
|
|
|
|
of lookahead steps. If an element satisfying the condition could not be found in |
537
|
|
|
|
|
|
|
C steps, an enumerable is considered to be completely iterated |
538
|
|
|
|
|
|
|
and C will return false. |
539
|
|
|
|
|
|
|
|
540
|
|
|
|
|
|
|
C returns a new enumerable with quite special properties: C |
541
|
|
|
|
|
|
|
will perform a look ahead and call the original enumerable C method |
542
|
|
|
|
|
|
|
in order to find an element for which the user-defined function will return |
543
|
|
|
|
|
|
|
true. C, on the other side, returns the value that was pre-fetched |
544
|
|
|
|
|
|
|
by C. |
545
|
|
|
|
|
|
|
|
546
|
|
|
|
|
|
|
Example |
547
|
|
|
|
|
|
|
Data::Enumerable::Lazy |
548
|
|
|
|
|
|
|
->from_list(1, 2, 3) |
549
|
|
|
|
|
|
|
->grep(sub { |
550
|
|
|
|
|
|
|
my ($number) = @_; |
551
|
|
|
|
|
|
|
return $number % 2 |
552
|
|
|
|
|
|
|
}); |
553
|
|
|
|
|
|
|
|
554
|
|
|
|
|
|
|
=cut |
555
|
|
|
|
|
|
|
|
556
|
|
|
|
|
|
|
sub grep { |
557
|
9
|
|
|
9
|
1
|
18
|
my ($self, $callback, $max_lookahead) = @_; |
558
|
9
|
|
|
|
|
12
|
my $next; |
559
|
9
|
|
|
|
|
12
|
my $initialized = 0; |
560
|
9
|
|
50
|
|
|
96
|
$max_lookahead //= 0; |
561
|
9
|
100
|
|
|
|
17
|
$max_lookahead = 0 |
562
|
|
|
|
|
|
|
if $self->is_finite; |
563
|
9
|
|
|
|
|
14
|
my $prev_has_next; |
564
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
565
|
|
|
|
|
|
|
on_has_next => sub { |
566
|
65
|
50
|
|
65
|
|
107
|
defined $prev_has_next |
567
|
|
|
|
|
|
|
and return $prev_has_next; |
568
|
65
|
|
|
|
|
77
|
my $ix = 0; |
569
|
65
|
|
|
|
|
84
|
$initialized = 1; |
570
|
65
|
|
|
|
|
99
|
undef $next; |
571
|
65
|
|
|
|
|
98
|
while ($self->has_next()) { |
572
|
84
|
50
|
|
|
|
133
|
if ($max_lookahead > 0) { |
573
|
|
|
|
|
|
|
$ix > $max_lookahead |
574
|
0
|
0
|
|
|
|
0
|
and do { |
575
|
0
|
|
|
|
|
0
|
carp sprintf 'Max lookahead steps cnt reached. Bailing out'; |
576
|
0
|
|
|
|
|
0
|
return $prev_has_next = 0; |
577
|
|
|
|
|
|
|
}; |
578
|
|
|
|
|
|
|
} |
579
|
84
|
|
|
|
|
139
|
$next = $self->next(); |
580
|
84
|
100
|
|
|
|
138
|
$callback->($next) and last; |
581
|
27
|
|
|
|
|
61
|
undef $next; |
582
|
27
|
|
|
|
|
43
|
$ix++; |
583
|
|
|
|
|
|
|
} |
584
|
65
|
|
|
|
|
154
|
return $prev_has_next = (defined $next); |
585
|
|
|
|
|
|
|
}, |
586
|
|
|
|
|
|
|
on_next => sub { |
587
|
56
|
|
|
56
|
|
66
|
my $self = shift; |
588
|
56
|
50
|
|
|
|
97
|
$initialized or $self->has_next(); |
589
|
56
|
|
|
|
|
75
|
undef $prev_has_next; |
590
|
56
|
|
|
|
|
86
|
$self->yield($next); |
591
|
|
|
|
|
|
|
}, |
592
|
9
|
|
|
|
|
68
|
is_finite => $self->is_finite(), |
593
|
|
|
|
|
|
|
no_wrap => $self->no_wrap(), |
594
|
|
|
|
|
|
|
}); |
595
|
|
|
|
|
|
|
} |
596
|
|
|
|
|
|
|
|
597
|
|
|
|
|
|
|
=head2 resolve() |
598
|
|
|
|
|
|
|
|
599
|
|
|
|
|
|
|
Resolves an enumerable completely. Applicable for finite enumerables only. |
600
|
|
|
|
|
|
|
The method returns nothing. |
601
|
|
|
|
|
|
|
|
602
|
|
|
|
|
|
|
=cut |
603
|
|
|
|
|
|
|
|
604
|
|
|
|
|
|
|
sub resolve { |
605
|
1
|
|
|
1
|
1
|
5
|
my ($self) = @_; |
606
|
1
|
50
|
|
|
|
2
|
croak 'Only finite enumerables might be resolved. Use is_finite=1' |
607
|
|
|
|
|
|
|
unless $self->is_finite(); |
608
|
1
|
|
|
|
|
3
|
$self->next() while $self->has_next(); |
609
|
|
|
|
|
|
|
} |
610
|
|
|
|
|
|
|
|
611
|
|
|
|
|
|
|
=head2 take($N_elements) |
612
|
|
|
|
|
|
|
|
613
|
|
|
|
|
|
|
Resolves first $N_elements and returns the resulting list. If there are |
614
|
|
|
|
|
|
|
fewer than N elements in the enumerable, the entire enumerable would be |
615
|
|
|
|
|
|
|
returned as a list. |
616
|
|
|
|
|
|
|
|
617
|
|
|
|
|
|
|
=cut |
618
|
|
|
|
|
|
|
|
619
|
|
|
|
|
|
|
sub take { |
620
|
10
|
|
|
10
|
1
|
44
|
my ($self, $slice_size) = @_; |
621
|
10
|
|
|
|
|
14
|
my $ix = 0; |
622
|
10
|
|
|
|
|
12
|
my @acc; |
623
|
10
|
|
100
|
|
|
22
|
push @acc, $self->next() while ($self->has_next() && $ix++ < $slice_size); |
624
|
10
|
|
|
|
|
39
|
return \@acc; |
625
|
|
|
|
|
|
|
} |
626
|
|
|
|
|
|
|
|
627
|
|
|
|
|
|
|
=head2 take_while($callback) |
628
|
|
|
|
|
|
|
|
629
|
|
|
|
|
|
|
This function takes elements until it meets the first one that does not |
630
|
|
|
|
|
|
|
satisfy the conditional callback. |
631
|
|
|
|
|
|
|
The callback takes only 1 argument: an element. It should return true if |
632
|
|
|
|
|
|
|
the element should be taken. Once it returned false, the stream is over. |
633
|
|
|
|
|
|
|
|
634
|
|
|
|
|
|
|
=cut |
635
|
|
|
|
|
|
|
|
636
|
|
|
|
|
|
|
sub take_while { |
637
|
7
|
|
|
7
|
1
|
11
|
my ($self, $callback) = @_; |
638
|
7
|
|
|
|
|
9
|
my $next_el; |
639
|
|
|
|
|
|
|
my $prev_has_next; |
640
|
7
|
|
|
|
|
7
|
my $initialized = 0; |
641
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
642
|
|
|
|
|
|
|
on_has_next => sub { |
643
|
53
|
|
|
53
|
|
63
|
$initialized = 1; |
644
|
53
|
50
|
|
|
|
71
|
defined $prev_has_next |
645
|
|
|
|
|
|
|
and return $prev_has_next; |
646
|
53
|
|
|
|
|
49
|
$prev_has_next = 0; |
647
|
53
|
100
|
|
|
|
61
|
if ($self->has_next()) { |
648
|
52
|
|
|
|
|
62
|
$next_el = $self->next(); |
649
|
52
|
100
|
|
|
|
65
|
if ($callback->($next_el)) { |
650
|
47
|
|
|
|
|
76
|
$prev_has_next = 1; |
651
|
|
|
|
|
|
|
} |
652
|
|
|
|
|
|
|
} |
653
|
53
|
|
|
|
|
88
|
return $prev_has_next; |
654
|
|
|
|
|
|
|
}, |
655
|
|
|
|
|
|
|
on_next => sub { |
656
|
47
|
|
|
47
|
|
59
|
my ($new_self) = @_; |
657
|
47
|
50
|
|
|
|
58
|
$initialized or $new_self->has_next(); |
658
|
47
|
50
|
|
|
|
55
|
$prev_has_next |
659
|
|
|
|
|
|
|
or return $new_self->yield(Data::Enumerable::Lazy->empty()); |
660
|
47
|
|
|
|
|
54
|
undef $prev_has_next; |
661
|
47
|
|
|
|
|
61
|
$new_self->yield($next_el); |
662
|
|
|
|
|
|
|
}, |
663
|
7
|
|
|
|
|
22
|
is_finite => $self->is_finite(), |
664
|
|
|
|
|
|
|
}); |
665
|
|
|
|
|
|
|
} |
666
|
|
|
|
|
|
|
|
667
|
|
|
|
|
|
|
=head2 continue($ext = %{ on_next => sub {}, ... }) |
668
|
|
|
|
|
|
|
|
669
|
|
|
|
|
|
|
Creates a new enumerable by extending the existing one. on_next is |
670
|
|
|
|
|
|
|
the only manfatory argument. on_has_next might be overriden if some |
671
|
|
|
|
|
|
|
custom logic comes into play. |
672
|
|
|
|
|
|
|
|
673
|
|
|
|
|
|
|
is_finite is inherited from the parent enumerable by default. All additional |
674
|
|
|
|
|
|
|
attributes would be transparently passed to the constuctor. |
675
|
|
|
|
|
|
|
|
676
|
|
|
|
|
|
|
=cut |
677
|
|
|
|
|
|
|
|
678
|
|
|
|
|
|
|
sub continue { |
679
|
16
|
|
|
16
|
1
|
34
|
my ($this, $ext) = @_; |
680
|
16
|
|
|
|
|
89
|
my %ext = %$ext; |
681
|
|
|
|
|
|
|
my $on_next = delete $ext{on_next} |
682
|
16
|
50
|
|
|
|
54
|
or croak '`on_next` should be defined on stream continuation'; |
683
|
16
|
50
|
|
|
|
44
|
ref($on_next) eq 'CODE' |
684
|
|
|
|
|
|
|
or croak '`on_next` should be a function'; |
685
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
686
|
|
|
|
|
|
|
on_next => sub { |
687
|
138
|
|
|
138
|
|
202
|
my $self = shift; |
688
|
138
|
100
|
|
|
|
596
|
$self->yield( |
689
|
|
|
|
|
|
|
$this->has_next() ? |
690
|
|
|
|
|
|
|
$on_next->($self, $this->next()) : |
691
|
|
|
|
|
|
|
Data::Enumerable::Lazy->empty |
692
|
|
|
|
|
|
|
); |
693
|
|
|
|
|
|
|
}, |
694
|
|
|
|
|
|
|
on_has_next => delete $ext->{on_has_next} // $this->on_has_next(), |
695
|
|
|
|
|
|
|
is_finite => delete $ext->{is_finite} // $this->is_finite(), |
696
|
16
|
|
33
|
|
|
93
|
no_wrap => delete $ext->{no_wrap} // 0, |
|
|
|
66
|
|
|
|
|
|
|
|
50
|
|
|
|
|
697
|
|
|
|
|
|
|
%ext, |
698
|
|
|
|
|
|
|
}); |
699
|
|
|
|
|
|
|
} |
700
|
|
|
|
|
|
|
|
701
|
|
|
|
|
|
|
=head2 count() |
702
|
|
|
|
|
|
|
|
703
|
|
|
|
|
|
|
Counts the number of the elements in the stream. This method iterates through |
704
|
|
|
|
|
|
|
the stream so it makes it exhausted by the end of the computatuion. |
705
|
|
|
|
|
|
|
|
706
|
|
|
|
|
|
|
=cut |
707
|
|
|
|
|
|
|
|
708
|
|
|
|
|
|
|
sub count { |
709
|
3
|
|
|
3
|
1
|
57
|
my ($self) = @_; |
710
|
3
|
100
|
|
|
|
8
|
croak 'Only finite enumerables might be counted. Use is_finite=1' |
711
|
|
|
|
|
|
|
unless $self->is_finite(); |
712
|
2
|
|
|
|
|
5
|
my $cnt = 0; |
713
|
2
|
|
|
|
|
6
|
for (; $self->has_next(); $self->next()) { |
714
|
10
|
|
|
|
|
20
|
$cnt++; |
715
|
|
|
|
|
|
|
} |
716
|
2
|
|
|
|
|
27
|
return $cnt; |
717
|
|
|
|
|
|
|
} |
718
|
|
|
|
|
|
|
|
719
|
|
|
|
|
|
|
=head2 yield($result) |
720
|
|
|
|
|
|
|
|
721
|
|
|
|
|
|
|
This method is supposed to be called from C callback only. This is |
722
|
|
|
|
|
|
|
the only valid result for an Enumerable to return the next step result. |
723
|
|
|
|
|
|
|
Effectively, it ensures the returned result conforms to the required interface |
724
|
|
|
|
|
|
|
and is wrapped in a lazy wrapper if needed. |
725
|
|
|
|
|
|
|
|
726
|
|
|
|
|
|
|
=cut |
727
|
|
|
|
|
|
|
|
728
|
|
|
|
|
|
|
sub yield { |
729
|
3351
|
|
|
3351
|
1
|
116284
|
my $self = shift; |
730
|
3351
|
|
|
|
|
4218
|
my $val = shift; |
731
|
3351
|
|
66
|
|
|
10404
|
my $val_is_stream = $val && ref($val) eq 'Data::Enumerable::Lazy' && |
732
|
|
|
|
|
|
|
$val->isa('Data::Enumerable::Lazy'); |
733
|
3351
|
100
|
100
|
|
|
5618
|
if ($self->no_wrap() || $val_is_stream) { |
734
|
1917
|
|
|
|
|
3612
|
return $val; |
735
|
|
|
|
|
|
|
} else { |
736
|
1434
|
|
|
|
|
2638
|
return Data::Enumerable::Lazy->singular($val); |
737
|
|
|
|
|
|
|
} |
738
|
|
|
|
|
|
|
} |
739
|
|
|
|
|
|
|
|
740
|
|
|
|
|
|
|
# Private methods |
741
|
|
|
|
|
|
|
|
742
|
|
|
|
|
|
|
sub _has_next_in_buffer { |
743
|
5399
|
|
|
5399
|
|
6997
|
my $self = shift; |
744
|
5399
|
100
|
|
|
|
15826
|
defined($self->{_buff}) && $self->{_buff}->has_next(); |
745
|
|
|
|
|
|
|
} |
746
|
|
|
|
|
|
|
|
747
|
|
|
|
|
|
|
sub _has_next_in_generator { |
748
|
5197
|
|
|
5197
|
|
7187
|
my $self = shift; |
749
|
5197
|
|
|
|
|
8554
|
$self->on_has_next()->($self, @_); |
750
|
|
|
|
|
|
|
} |
751
|
|
|
|
|
|
|
|
752
|
|
|
|
|
|
|
=head1 CLASS METHODS |
753
|
|
|
|
|
|
|
|
754
|
|
|
|
|
|
|
=head2 empty() |
755
|
|
|
|
|
|
|
|
756
|
|
|
|
|
|
|
Returns an empty enumerable. Effectively it means an equivalent of an empty |
757
|
|
|
|
|
|
|
array. C will return false and C will return undef. Useful |
758
|
|
|
|
|
|
|
whenever a C step wants to return an empty resultset. |
759
|
|
|
|
|
|
|
|
760
|
|
|
|
|
|
|
=cut |
761
|
|
|
|
|
|
|
|
762
|
|
|
|
|
|
|
sub empty { |
763
|
16
|
|
|
16
|
1
|
2331
|
Data::Enumerable::Lazy->new({ |
764
|
|
|
|
|
|
|
is_finite => 1, |
765
|
|
|
|
|
|
|
no_wrap => 1, |
766
|
|
|
|
|
|
|
}); |
767
|
|
|
|
|
|
|
} |
768
|
|
|
|
|
|
|
|
769
|
|
|
|
|
|
|
=head2 singular($val) |
770
|
|
|
|
|
|
|
|
771
|
|
|
|
|
|
|
Returns an enumerable with a single element $val. Actively used as an internal |
772
|
|
|
|
|
|
|
data container. |
773
|
|
|
|
|
|
|
|
774
|
|
|
|
|
|
|
=cut |
775
|
|
|
|
|
|
|
|
776
|
|
|
|
|
|
|
sub singular { |
777
|
1445
|
|
|
1445
|
1
|
5608
|
my ($class, $val) = @_; |
778
|
1445
|
|
|
|
|
1922
|
my $resolved = 0; |
779
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
780
|
3070
|
|
|
3070
|
|
7961
|
on_has_next => sub { not $resolved }, |
781
|
1445
|
|
|
1445
|
|
2119
|
on_next => sub { $resolved = 1; shift->yield($val) }, |
|
1445
|
|
|
|
|
2261
|
|
782
|
1445
|
|
|
|
|
7049
|
is_finite => 1, |
783
|
|
|
|
|
|
|
no_wrap => 1, |
784
|
|
|
|
|
|
|
}); |
785
|
|
|
|
|
|
|
} |
786
|
|
|
|
|
|
|
|
787
|
|
|
|
|
|
|
=head2 from_list(@list) |
788
|
|
|
|
|
|
|
|
789
|
|
|
|
|
|
|
Returns a new enumerable instantiated from a list. The easiest way to |
790
|
|
|
|
|
|
|
initialize an enumerable. In fact, all elements are already resolved |
791
|
|
|
|
|
|
|
so this method sets C by default. |
792
|
|
|
|
|
|
|
|
793
|
|
|
|
|
|
|
=cut |
794
|
|
|
|
|
|
|
|
795
|
|
|
|
|
|
|
sub from_list { |
796
|
32
|
|
|
32
|
1
|
8677
|
my $class = shift; |
797
|
32
|
|
|
|
|
89
|
my @list = @_; |
798
|
32
|
|
|
|
|
50
|
my $ix = 0; |
799
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
800
|
550
|
|
|
550
|
|
1482
|
on_has_next => sub { $ix < scalar(@list) }, |
801
|
249
|
|
|
249
|
|
480
|
on_next => sub { shift->yield($list[$ix++]) }, |
802
|
32
|
|
|
|
|
250
|
is_finite => 1, |
803
|
|
|
|
|
|
|
no_wrap => 1, |
804
|
|
|
|
|
|
|
}); |
805
|
|
|
|
|
|
|
} |
806
|
|
|
|
|
|
|
|
807
|
|
|
|
|
|
|
=head2 cycle() |
808
|
|
|
|
|
|
|
|
809
|
|
|
|
|
|
|
Creates an infinitive enumerable by cycling the original list. E.g. if the |
810
|
|
|
|
|
|
|
original list is [1, 2, 3], C will generate an infinitive sequences |
811
|
|
|
|
|
|
|
like: 1, 2, 3, 1, 2, 3, 1, ... |
812
|
|
|
|
|
|
|
|
813
|
|
|
|
|
|
|
=cut |
814
|
|
|
|
|
|
|
|
815
|
|
|
|
|
|
|
sub cycle { |
816
|
6
|
|
|
6
|
1
|
90
|
my $class = shift; |
817
|
6
|
|
|
|
|
12
|
my @list = @_; |
818
|
6
|
|
|
|
|
7
|
my $ix = 0; |
819
|
6
|
|
|
|
|
8
|
my $max_ix = scalar(@list) - 1; |
820
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
821
|
62
|
|
|
62
|
|
130
|
on_has_next => sub { 1 }, |
822
|
|
|
|
|
|
|
on_next => sub { |
823
|
49
|
100
|
|
49
|
|
77
|
$ix = $ix > $max_ix ? 0 : $ix; |
824
|
49
|
|
|
|
|
71
|
shift->yield($list[$ix++]) |
825
|
|
|
|
|
|
|
}, |
826
|
6
|
|
|
|
|
29
|
is_finite => 0, |
827
|
|
|
|
|
|
|
no_wrap => 1, |
828
|
|
|
|
|
|
|
}); |
829
|
|
|
|
|
|
|
} |
830
|
|
|
|
|
|
|
|
831
|
|
|
|
|
|
|
=head2 infinity() |
832
|
|
|
|
|
|
|
|
833
|
|
|
|
|
|
|
Returns a new infinite enumerable. C always returns true whereas |
834
|
|
|
|
|
|
|
C returns undef all the time. Useful as an extension basis for infinite |
835
|
|
|
|
|
|
|
sequences. |
836
|
|
|
|
|
|
|
|
837
|
|
|
|
|
|
|
=cut |
838
|
|
|
|
|
|
|
|
839
|
|
|
|
|
|
|
sub infinity { |
840
|
2
|
|
|
2
|
1
|
947
|
my $class = shift; |
841
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
842
|
11
|
|
|
11
|
|
20
|
on_has_next => sub { 1 }, |
843
|
|
|
|
10
|
|
|
on_next => sub {}, |
844
|
2
|
|
|
|
|
14
|
is_finite => 0, |
845
|
|
|
|
|
|
|
no_wrap => 1, |
846
|
|
|
|
|
|
|
}); |
847
|
|
|
|
|
|
|
} |
848
|
|
|
|
|
|
|
|
849
|
|
|
|
|
|
|
=head2 merge($tream1 [, $tream2 [, $tream3 [, ...]]]) |
850
|
|
|
|
|
|
|
|
851
|
|
|
|
|
|
|
This function merges one or more streams together by fan-outing C |
852
|
|
|
|
|
|
|
method call among the non-empty streams. |
853
|
|
|
|
|
|
|
Returns a new enumerable instance, which: |
854
|
|
|
|
|
|
|
* Has next elements as far as at least one of the streams does. |
855
|
|
|
|
|
|
|
* Returns next element py picking it one-by-one from the streams. |
856
|
|
|
|
|
|
|
* Is finite if and only if all the streams are finite. |
857
|
|
|
|
|
|
|
If one of the streams is over, it would be taken into account and |
858
|
|
|
|
|
|
|
C will continue choosing from non-empty ones. |
859
|
|
|
|
|
|
|
|
860
|
|
|
|
|
|
|
=cut |
861
|
|
|
|
|
|
|
|
862
|
|
|
|
|
|
|
sub merge { |
863
|
4
|
|
|
4
|
1
|
7
|
my $class = shift; |
864
|
4
|
|
|
|
|
7
|
my @streams = @_; |
865
|
4
|
50
|
|
|
|
7
|
scalar @streams == 0 |
866
|
|
|
|
|
|
|
and croak '`merge` function takes at least 1 stream'; |
867
|
4
|
50
|
|
|
|
7
|
scalar @streams == 1 |
868
|
|
|
|
|
|
|
and return shift; |
869
|
|
|
|
|
|
|
my $ixs = Data::Enumerable::Lazy->cycle(0..scalar(@streams) - 1) |
870
|
38
|
|
|
38
|
|
104
|
-> take_while(sub { List::Util::any { $_->has_next() } @streams }) |
|
91
|
|
|
|
|
120
|
|
871
|
4
|
|
|
35
|
|
8
|
-> grep(sub { $streams[ shift ]->has_next() }); |
|
35
|
|
|
|
|
46
|
|
872
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
873
|
24
|
|
|
24
|
|
31
|
on_has_next => sub { $ixs->has_next() }, |
874
|
|
|
|
|
|
|
on_next => sub { |
875
|
20
|
|
|
20
|
|
28
|
shift->yield($streams[ $ixs->next() ]->next()); |
876
|
|
|
|
|
|
|
}, |
877
|
4
|
100
|
|
17
|
|
37
|
is_finite => (List::Util::reduce { $a || $b->is_finite() } 0, @streams), |
|
17
|
|
|
|
|
39
|
|
878
|
|
|
|
|
|
|
}); |
879
|
|
|
|
|
|
|
} |
880
|
|
|
|
|
|
|
|
881
|
|
|
|
|
|
|
=head2 chain($tream1(, $tream2(, $tream3(, ...)))) |
882
|
|
|
|
|
|
|
|
883
|
|
|
|
|
|
|
Executes streams sequentually, one after another: the next stream starts once |
884
|
|
|
|
|
|
|
the previous is over. |
885
|
|
|
|
|
|
|
|
886
|
|
|
|
|
|
|
=cut |
887
|
|
|
|
|
|
|
|
888
|
|
|
|
|
|
|
sub chain { |
889
|
3
|
|
|
3
|
1
|
5
|
my $class = shift; |
890
|
3
|
|
|
|
|
6
|
my @streams = @_; |
891
|
3
|
100
|
|
|
|
10
|
scalar(@streams) < 2 |
892
|
|
|
|
|
|
|
and return $streams[0]; |
893
|
|
|
|
|
|
|
Data::Enumerable::Lazy->from_list(@streams) |
894
|
|
|
|
|
|
|
-> continue({ |
895
|
7
|
|
|
7
|
|
15
|
on_next => sub { $_[0]->yield($_[1]) } |
896
|
|
|
|
|
|
|
}) |
897
|
29
|
|
|
29
|
|
64
|
-> grep(sub { defined $_[0] }) |
898
|
2
|
|
|
|
|
5
|
} |
899
|
|
|
|
|
|
|
|
900
|
|
|
|
|
|
|
=head2 from_text_file($file_handle(, $options)) |
901
|
|
|
|
|
|
|
|
902
|
|
|
|
|
|
|
Method takes an open file handle and an optional hash of options and creates a |
903
|
|
|
|
|
|
|
stream of it. The file would be read as a text file, line by line. For |
904
|
|
|
|
|
|
|
additional options see C perl core function reference. |
905
|
|
|
|
|
|
|
Options is a basic hash, supported attributes are: |
906
|
|
|
|
|
|
|
* chomp :: Bool | Whether the lines should be chomped, 0 by default. |
907
|
|
|
|
|
|
|
* is_finite :: Bool | Forces the stream to be processed as finite, 0 by default. |
908
|
|
|
|
|
|
|
|
909
|
|
|
|
|
|
|
=cut |
910
|
|
|
|
|
|
|
|
911
|
|
|
|
|
|
|
sub from_text_file { |
912
|
2
|
|
|
2
|
1
|
5985
|
my ($class, $file_handle, $options) = @_; |
913
|
2
|
|
50
|
|
|
7
|
$options //= +{}; |
914
|
|
|
|
|
|
|
my $str = Data::Enumerable::Lazy->new({ |
915
|
22
|
|
|
22
|
|
127
|
on_has_next => sub { !eof($file_handle) }, |
916
|
|
|
|
|
|
|
on_next => sub { |
917
|
20
|
|
|
20
|
|
45
|
my $line = readline($file_handle); |
918
|
20
|
|
|
|
|
34
|
$_[0]->yield($line); |
919
|
|
|
|
|
|
|
}, |
920
|
2
|
|
50
|
|
|
18
|
is_finite => $options->{is_finite} // 0, |
921
|
|
|
|
|
|
|
}); |
922
|
2
|
100
|
|
|
|
7
|
if ($options->{chomp}) { |
923
|
1
|
|
|
10
|
|
5
|
$str = $str->map(sub { my $s = $_[0]; chomp $s; $s }); |
|
10
|
|
|
|
|
14
|
|
|
10
|
|
|
|
|
19
|
|
|
10
|
|
|
|
|
19
|
|
924
|
|
|
|
|
|
|
} |
925
|
2
|
|
|
|
|
5
|
return $str; |
926
|
|
|
|
|
|
|
} |
927
|
|
|
|
|
|
|
|
928
|
|
|
|
|
|
|
=head2 from_bin_file($file_handle(, $options)) |
929
|
|
|
|
|
|
|
|
930
|
|
|
|
|
|
|
Method similar to C but forces binary reading from file. |
931
|
|
|
|
|
|
|
Takes a file handle created by C function and an optional hash of |
932
|
|
|
|
|
|
|
options. Supported attributes are: |
933
|
|
|
|
|
|
|
* block_size :: Integer | The size of read block, 1024 bytes by default. |
934
|
|
|
|
|
|
|
* is_finite :: Bool | Forces the stream to be processed as finite, 0 by default. |
935
|
|
|
|
|
|
|
|
936
|
|
|
|
|
|
|
=cut |
937
|
|
|
|
|
|
|
|
938
|
|
|
|
|
|
|
sub from_bin_file { |
939
|
1
|
|
|
1
|
1
|
90
|
my ($class, $file_handle, $options) = @_; |
940
|
1
|
|
50
|
|
|
3
|
$options //= +{}; |
941
|
1
|
|
50
|
|
|
4
|
my $block_size = $options->{block_size} // 1024; |
942
|
|
|
|
|
|
|
Data::Enumerable::Lazy->new({ |
943
|
23
|
|
|
23
|
|
104
|
on_has_next => sub { !eof($file_handle) }, |
944
|
|
|
|
|
|
|
on_next => sub { |
945
|
22
|
|
|
22
|
|
20
|
my $buf; |
946
|
22
|
|
|
|
|
87
|
read($file_handle, $buf, $block_size); |
947
|
22
|
|
|
|
|
34
|
$_[0]->yield($buf); |
948
|
|
|
|
|
|
|
}, |
949
|
1
|
|
50
|
|
|
7
|
is_finite => $options->{is_finite} // 0, |
950
|
|
|
|
|
|
|
}) |
951
|
|
|
|
|
|
|
} |
952
|
|
|
|
|
|
|
|
953
|
|
|
|
|
|
|
=head1 AUTHOR |
954
|
|
|
|
|
|
|
|
955
|
|
|
|
|
|
|
Oleg S |
956
|
|
|
|
|
|
|
|
957
|
|
|
|
|
|
|
=cut |
958
|
|
|
|
|
|
|
|
959
|
|
|
|
|
|
|
=head1 SEE ALSO |
960
|
|
|
|
|
|
|
|
961
|
|
|
|
|
|
|
=head2 Lazy evaluation in a nutshell |
962
|
|
|
|
|
|
|
|
963
|
|
|
|
|
|
|
L |
964
|
|
|
|
|
|
|
|
965
|
|
|
|
|
|
|
=head2 Library GitHub page: |
966
|
|
|
|
|
|
|
|
967
|
|
|
|
|
|
|
L |
968
|
|
|
|
|
|
|
|
969
|
|
|
|
|
|
|
=head2 Alternative implementations: |
970
|
|
|
|
|
|
|
|
971
|
|
|
|
|
|
|
L |
972
|
|
|
|
|
|
|
L |
973
|
|
|
|
|
|
|
L |
974
|
|
|
|
|
|
|
|
975
|
|
|
|
|
|
|
=cut |
976
|
|
|
|
|
|
|
|
977
|
|
|
|
|
|
|
=head1 COPYRIGHT AND LICENSE |
978
|
|
|
|
|
|
|
|
979
|
|
|
|
|
|
|
Copyright 2017, 2018 Oleg S |
980
|
|
|
|
|
|
|
|
981
|
|
|
|
|
|
|
Copying and distribution of this file, with or without modification, are |
982
|
|
|
|
|
|
|
permitted in any medium without royalty provided the copyright notice and this |
983
|
|
|
|
|
|
|
notice are preserved. This file is offered as-is, without any warranty. |
984
|
|
|
|
|
|
|
|
985
|
|
|
|
|
|
|
=cut |
986
|
|
|
|
|
|
|
|
987
|
|
|
|
|
|
|
1; |
988
|
|
|
|
|
|
|
|
989
|
|
|
|
|
|
|
__END__ |