litestack 0.3.0 → 0.4.2
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/.standard.yml +3 -0
- data/BENCHMARKS.md +34 -7
- data/CHANGELOG.md +21 -0
- data/Gemfile +1 -5
- data/Gemfile.lock +92 -0
- data/README.md +120 -6
- data/ROADMAP.md +45 -0
- data/Rakefile +3 -1
- data/WHYLITESTACK.md +1 -1
- data/assets/litecache_metrics.png +0 -0
- data/assets/litedb_metrics.png +0 -0
- data/assets/litemetric_logo_teal.png +0 -0
- data/assets/litesearch_logo_teal.png +0 -0
- data/bench/bench.rb +17 -10
- data/bench/bench_cache_rails.rb +10 -13
- data/bench/bench_cache_raw.rb +17 -22
- data/bench/bench_jobs_rails.rb +19 -13
- data/bench/bench_jobs_raw.rb +17 -10
- data/bench/bench_queue.rb +4 -6
- data/bench/rails_job.rb +5 -7
- data/bench/skjob.rb +4 -4
- data/bench/uljob.rb +6 -6
- data/lib/action_cable/subscription_adapter/litecable.rb +5 -8
- data/lib/active_job/queue_adapters/litejob_adapter.rb +6 -8
- data/lib/active_record/connection_adapters/litedb_adapter.rb +65 -75
- data/lib/active_support/cache/litecache.rb +38 -41
- data/lib/generators/litestack/install/install_generator.rb +3 -3
- data/lib/generators/litestack/install/templates/database.yml +7 -1
- data/lib/litestack/liteboard/liteboard.rb +269 -149
- data/lib/litestack/litecable.rb +44 -40
- data/lib/litestack/litecable.sql.yml +22 -11
- data/lib/litestack/litecache.rb +80 -89
- data/lib/litestack/litecache.sql.yml +81 -22
- data/lib/litestack/litecache.yml +1 -1
- data/lib/litestack/litedb.rb +39 -38
- data/lib/litestack/litejob.rb +31 -31
- data/lib/litestack/litejobqueue.rb +107 -106
- data/lib/litestack/litemetric.rb +83 -95
- data/lib/litestack/litemetric.sql.yml +244 -234
- data/lib/litestack/litemetric_collector.sql.yml +38 -41
- data/lib/litestack/litequeue.rb +39 -41
- data/lib/litestack/litequeue.sql.yml +39 -31
- data/lib/litestack/litescheduler.rb +84 -0
- data/lib/litestack/litesearch/index.rb +260 -0
- data/lib/litestack/litesearch/model.rb +179 -0
- data/lib/litestack/litesearch/schema.rb +190 -0
- data/lib/litestack/litesearch/schema_adapters/backed_adapter.rb +143 -0
- data/lib/litestack/litesearch/schema_adapters/basic_adapter.rb +137 -0
- data/lib/litestack/litesearch/schema_adapters/contentless_adapter.rb +14 -0
- data/lib/litestack/litesearch/schema_adapters/standalone_adapter.rb +31 -0
- data/lib/litestack/litesearch/schema_adapters.rb +4 -0
- data/lib/litestack/litesearch.rb +34 -0
- data/lib/litestack/litesupport.rb +85 -186
- data/lib/litestack/railtie.rb +1 -1
- data/lib/litestack/version.rb +2 -2
- data/lib/litestack.rb +7 -4
- data/lib/railties/rails/commands/dbconsole.rb +11 -15
- data/lib/sequel/adapters/litedb.rb +18 -22
- data/lib/sequel/adapters/shared/litedb.rb +168 -168
- data/scripts/build_metrics.rb +91 -0
- data/scripts/test_cable.rb +30 -0
- data/scripts/test_job_retry.rb +33 -0
- data/scripts/test_metrics.rb +60 -0
- data/template.rb +2 -2
- metadata +112 -7
@@ -1,56 +1,53 @@
|
|
1
1
|
schema:
|
2
2
|
1:
|
3
|
-
|
4
3
|
create_events: >
|
5
4
|
CREATE TABLE IF NOT EXISTS local_events(
|
6
5
|
topic TEXT NOT NULL,
|
7
|
-
name TEXT DEFAULT('___') NOT NULL ON CONFLICT REPLACE,
|
8
|
-
key TEXT DEFAULT('___') NOT NULL ON CONFLICT REPLACE,
|
9
|
-
count INTEGER DEFAULT(0) NOT NULL ON CONFLICT REPLACE,
|
10
|
-
value REAL,
|
11
|
-
minimum REAL,
|
12
|
-
maximum REAL,
|
13
|
-
created_at INTEGER DEFAULT((unixepoch()/300*300)) NOT NULL ON CONFLICT REPLACE,
|
14
|
-
resolution TEXT DEFAULT('minute') NOT NULL,
|
6
|
+
name TEXT DEFAULT('___') NOT NULL ON CONFLICT REPLACE,
|
7
|
+
key TEXT DEFAULT('___') NOT NULL ON CONFLICT REPLACE,
|
8
|
+
count INTEGER DEFAULT(0) NOT NULL ON CONFLICT REPLACE,
|
9
|
+
value REAL,
|
10
|
+
minimum REAL,
|
11
|
+
maximum REAL,
|
12
|
+
created_at INTEGER DEFAULT((unixepoch()/300*300)) NOT NULL ON CONFLICT REPLACE,
|
13
|
+
resolution TEXT DEFAULT('minute') NOT NULL,
|
15
14
|
PRIMARY KEY(resolution, created_at, topic, name, key)
|
16
|
-
) STRICT;
|
15
|
+
) STRICT;
|
17
16
|
|
18
17
|
stmts:
|
19
|
-
|
20
18
|
capture_event: >
|
21
|
-
INSERT INTO local_events(topic, name, key, created_at, count, value, minimum, maximum)
|
22
|
-
VALUES
|
23
|
-
(?1, ?2, ?3, ?4, ?5, ?6, ?6, ?6),
|
19
|
+
INSERT INTO local_events(topic, name, key, created_at, count, value, minimum, maximum)
|
20
|
+
VALUES
|
21
|
+
(?1, ?2, ?3, ?4, ?5, ?6, ?6, ?6),
|
24
22
|
(?1, ?2, '___', ?4, ?5, ?6, ?6, ?6),
|
25
|
-
(?1, '___', '___', ?4, ?5, ?6, ?6, ?6)
|
26
|
-
ON CONFLICT DO UPDATE
|
27
|
-
SET
|
28
|
-
count = count + EXCLUDED.count,
|
29
|
-
value = value + EXCLUDED.value,
|
30
|
-
minimum = min(minimum, EXCLUDED.minimum),
|
31
|
-
maximum = max(maximum, EXCLUDED.maximum)
|
32
|
-
|
33
|
-
migrate_events:
|
34
|
-
INSERT INTO m.events(topic, name, key, created_at, count, value, minimum, maximum)
|
35
|
-
SELECT topic, name, key, created_at, count, value, minimum, maximum
|
36
|
-
FROM local_events
|
37
|
-
ORDER BY resolution, created_at ASC, topic, name, key
|
23
|
+
(?1, '___', '___', ?4, ?5, ?6, ?6, ?6)
|
24
|
+
ON CONFLICT DO UPDATE
|
25
|
+
SET
|
26
|
+
count = count + EXCLUDED.count,
|
27
|
+
value = value + EXCLUDED.value,
|
28
|
+
minimum = min(minimum, EXCLUDED.minimum),
|
29
|
+
maximum = max(maximum, EXCLUDED.maximum);
|
30
|
+
|
31
|
+
migrate_events: >
|
32
|
+
INSERT INTO m.events(topic, name, key, created_at, count, value, minimum, maximum)
|
33
|
+
SELECT topic, name, key, created_at, count, value, minimum, maximum
|
34
|
+
FROM local_events
|
35
|
+
ORDER BY resolution, created_at ASC, topic, name, key
|
38
36
|
LIMIT ?
|
39
|
-
ON CONFLICT DO UPDATE
|
40
|
-
SET
|
41
|
-
count = count + EXCLUDED.count,
|
42
|
-
value = value + EXCLUDED.value,
|
43
|
-
minimum = min(minimum, EXCLUDED.minimum),
|
44
|
-
maximum = max(maximum, EXCLUDED.maximum)
|
37
|
+
ON CONFLICT DO UPDATE
|
38
|
+
SET
|
39
|
+
count = count + EXCLUDED.count,
|
40
|
+
value = value + EXCLUDED.value,
|
41
|
+
minimum = min(minimum, EXCLUDED.minimum),
|
42
|
+
maximum = max(maximum, EXCLUDED.maximum);
|
45
43
|
|
46
|
-
delete_migrated_events:
|
44
|
+
delete_migrated_events: >
|
47
45
|
DELETE FROM local_events WHERE rowid IN (
|
48
|
-
SELECT rowid
|
49
|
-
FROM local_events
|
50
|
-
ORDER BY resolution, created_at ASC, topic, name, key
|
46
|
+
SELECT rowid
|
47
|
+
FROM local_events
|
48
|
+
ORDER BY resolution, created_at ASC, topic, name, key
|
51
49
|
LIMIT ?
|
52
|
-
)
|
53
|
-
|
54
|
-
event_count:
|
50
|
+
);
|
51
|
+
|
52
|
+
event_count: >
|
55
53
|
SELECT count(*) FROM local_events;
|
56
|
-
|
data/lib/litestack/litequeue.rb
CHANGED
@@ -1,27 +1,26 @@
|
|
1
1
|
# frozen_stringe_literal: true
|
2
2
|
|
3
3
|
# all components should require the support module
|
4
|
-
require_relative
|
4
|
+
require_relative "litesupport"
|
5
5
|
|
6
|
-
#require 'securerandom'
|
6
|
+
# require 'securerandom'
|
7
7
|
|
8
8
|
##
|
9
|
-
#Litequeue is a simple queueing system for Ruby applications that allows you to push and pop values from a queue. It provides a straightforward API for creating and managing named queues, and for adding and removing values from those queues. Additionally, it offers options for scheduling pops at a certain time in the future, which can be useful for delaying processing until a later time.
|
9
|
+
# Litequeue is a simple queueing system for Ruby applications that allows you to push and pop values from a queue. It provides a straightforward API for creating and managing named queues, and for adding and removing values from those queues. Additionally, it offers options for scheduling pops at a certain time in the future, which can be useful for delaying processing until a later time.
|
10
10
|
#
|
11
|
-
#Litequeue is built on top of SQLite, which makes it very fast and efficient, even when handling large volumes of data. This lightweight and easy-to-use queueing system serves as a good foundation for building more advanced job processing frameworks that require basic queuing capabilities.
|
11
|
+
# Litequeue is built on top of SQLite, which makes it very fast and efficient, even when handling large volumes of data. This lightweight and easy-to-use queueing system serves as a good foundation for building more advanced job processing frameworks that require basic queuing capabilities.
|
12
12
|
#
|
13
|
-
|
14
|
-
class Litequeue
|
15
13
|
|
14
|
+
class Litequeue
|
16
15
|
# the default options for the queue
|
17
|
-
# can be overriden by passing new options in a hash
|
16
|
+
# can be overriden by passing new options in a hash
|
18
17
|
# to Litequeue.new
|
19
18
|
# path: "./queue.db"
|
20
19
|
# mmap_size: 128 * 1024 * 1024 -> 128MB to be held in memory
|
21
20
|
# sync: 1 -> sync only when checkpointing
|
22
|
-
|
21
|
+
|
23
22
|
include Litesupport::Liteconnection
|
24
|
-
|
23
|
+
|
25
24
|
DEFAULT_OPTIONS = {
|
26
25
|
path: Litesupport.root.join("queue.sqlite3"),
|
27
26
|
mmap_size: 32 * 1024,
|
@@ -35,70 +34,70 @@ class Litequeue
|
|
35
34
|
# queue.pop # => nil
|
36
35
|
# sleep 2
|
37
36
|
# queue.pop # => "somevalue"
|
38
|
-
|
37
|
+
|
39
38
|
def initialize(options = {})
|
40
39
|
init(options)
|
41
40
|
end
|
42
|
-
|
41
|
+
|
43
42
|
# push an item to the queue, optionally specifying the queue name (defaults to default) and after how many seconds it should be ready to pop (defaults to zero)
|
44
43
|
# a unique job id is returned from this method, can be used later to delete it before it fires. You can push string, integer, float, true, false or nil values
|
45
44
|
#
|
46
|
-
def push(value, delay=0, queue=
|
45
|
+
def push(value, delay = 0, queue = "default")
|
47
46
|
# @todo - check if queue is busy, back off if it is
|
48
47
|
# also bring back the synchronize block, to prevent
|
49
48
|
# a race condition if a thread hits the busy handler
|
50
49
|
# before the current thread proceeds after a backoff
|
51
|
-
#id = SecureRandom.uuid # this is somehow expensive, can we improve?
|
50
|
+
# id = SecureRandom.uuid # this is somehow expensive, can we improve?
|
52
51
|
run_stmt(:push, queue, delay, value)[0]
|
53
52
|
end
|
54
|
-
|
55
|
-
def repush(id, value, delay=0, queue=
|
53
|
+
|
54
|
+
def repush(id, value, delay = 0, queue = "default")
|
56
55
|
run_stmt(:repush, id, queue, delay, value)[0]
|
57
56
|
end
|
58
|
-
|
59
|
-
alias_method
|
57
|
+
|
58
|
+
alias_method :<<, :push
|
60
59
|
alias_method :"<<<", :repush
|
61
|
-
|
60
|
+
|
62
61
|
# pop an item from the queue, optionally with a specific queue name (default queue name is 'default')
|
63
|
-
def pop(queue=
|
64
|
-
|
65
|
-
|
66
|
-
|
67
|
-
|
62
|
+
def pop(queue = "default", limit = 1)
|
63
|
+
res = run_stmt(:pop, queue, limit)
|
64
|
+
return res[0] if res.length == 1
|
65
|
+
return nil if res.empty?
|
66
|
+
res
|
68
67
|
end
|
69
|
-
|
68
|
+
|
70
69
|
# delete an item from the queue
|
71
70
|
# queue = Litequeue.new
|
72
71
|
# id = queue.push("somevalue")
|
73
72
|
# queue.delete(id) # => "somevalue"
|
74
73
|
# queue.pop # => nil
|
75
74
|
def delete(id)
|
76
|
-
|
75
|
+
run_stmt(:delete, id)[0]
|
77
76
|
end
|
78
|
-
|
77
|
+
|
79
78
|
# deletes all the entries in all queues, or if a queue name is given, deletes all entries in that specific queue
|
80
|
-
def clear(queue=nil)
|
79
|
+
def clear(queue = nil)
|
81
80
|
run_sql("DELETE FROM queue WHERE iif(?1 IS NOT NULL, name = ?1, TRUE)", queue)
|
82
81
|
end
|
83
82
|
|
84
83
|
# returns a count of entries in all queues, or if a queue name is given, reutrns the count of entries in that queue
|
85
|
-
def count(queue=nil)
|
84
|
+
def count(queue = nil)
|
86
85
|
run_sql("SELECT count(*) FROM queue WHERE iif(?1 IS NOT NULL, name = ?1, TRUE)", queue)[0][0]
|
87
86
|
end
|
88
|
-
|
87
|
+
|
89
88
|
# return the size of the queue file on disk
|
90
|
-
#def size
|
91
|
-
# run_sql("SELECT size.page_size * count.page_count FROM pragma_page_size() AS size, pragma_page_count() AS count")[0][0]
|
92
|
-
#end
|
93
|
-
|
89
|
+
# def size
|
90
|
+
# run_sql("SELECT size.page_size * count.page_count FROM pragma_page_size() AS size, pragma_page_count() AS count")[0][0]
|
91
|
+
# end
|
92
|
+
|
94
93
|
def queues_info
|
95
94
|
run_stmt(:info)
|
96
95
|
end
|
97
|
-
|
96
|
+
|
98
97
|
def snapshot
|
99
98
|
queues = {}
|
100
99
|
queues_info.each do |qc|
|
101
|
-
#queues[qc[0]] = {count: qc[1], time_in_queue: {avg: qc[2], min: qc[3], max: qc[4]}}
|
100
|
+
# queues[qc[0]] = {count: qc[1], time_in_queue: {avg: qc[2], min: qc[3], max: qc[4]}}
|
102
101
|
queues[qc[0]] = qc[1]
|
103
102
|
end
|
104
103
|
{
|
@@ -109,15 +108,15 @@ class Litequeue
|
|
109
108
|
size: size,
|
110
109
|
jobs: count
|
111
110
|
},
|
112
|
-
queues: queues
|
111
|
+
queues: queues
|
113
112
|
}
|
114
113
|
end
|
115
114
|
|
116
|
-
private
|
117
|
-
|
115
|
+
private
|
116
|
+
|
118
117
|
def create_connection
|
119
118
|
super("#{__dir__}/litequeue.sql.yml") do |conn|
|
120
|
-
conn.wal_autocheckpoint = 10000
|
119
|
+
conn.wal_autocheckpoint = 10000
|
121
120
|
# check if there is an old database and convert entries to the new format
|
122
121
|
if conn.get_first_value("select count(*) from sqlite_master where name = '_ul_queue_'") == 1
|
123
122
|
conn.transaction(:immediate) do
|
@@ -127,5 +126,4 @@ class Litequeue
|
|
127
126
|
end
|
128
127
|
end
|
129
128
|
end
|
130
|
-
|
131
|
-
end
|
129
|
+
end
|
@@ -1,45 +1,53 @@
|
|
1
1
|
schema:
|
2
|
-
1:
|
3
|
-
create_table_queue: >
|
2
|
+
1:
|
3
|
+
create_table_queue: >
|
4
4
|
CREATE TABLE IF NOT EXISTS queue(
|
5
|
-
id TEXT PRIMARY KEY DEFAULT(hex(randomblob(32))) NOT NULL ON CONFLICT REPLACE,
|
6
|
-
name TEXT DEFAULT('default') NOT NULL ON CONFLICT REPLACE,
|
5
|
+
id TEXT PRIMARY KEY DEFAULT(hex(randomblob(32))) NOT NULL ON CONFLICT REPLACE,
|
6
|
+
name TEXT DEFAULT('default') NOT NULL ON CONFLICT REPLACE,
|
7
7
|
fire_at INTEGER DEFAULT(unixepoch()) NOT NULL ON CONFLICT REPLACE,
|
8
8
|
value TEXT,
|
9
9
|
created_at INTEGER DEFAULT(unixepoch()) NOT NULL ON CONFLICT REPLACE
|
10
|
-
) WITHOUT ROWID
|
11
|
-
|
10
|
+
) WITHOUT ROWID;
|
11
|
+
|
12
12
|
create_index_queue_by_name: >
|
13
|
-
CREATE INDEX IF NOT EXISTS idx_queue_by_name ON queue(name, fire_at ASC)
|
13
|
+
CREATE INDEX IF NOT EXISTS idx_queue_by_name ON queue(name, fire_at ASC);
|
14
14
|
|
15
15
|
stmts:
|
16
16
|
|
17
|
-
push:
|
18
|
-
|
19
|
-
|
17
|
+
push: >
|
18
|
+
INSERT INTO queue(id, name, fire_at, value)
|
19
|
+
VALUES (hex(randomblob(32)), $1, (unixepoch() + $2), $3)
|
20
|
+
RETURNING id, name;
|
21
|
+
|
22
|
+
repush: >
|
23
|
+
INSERT INTO queue(id, name, fire_at, value)
|
24
|
+
VALUES (?, ?, (unixepoch() + ?), ?)
|
25
|
+
RETURNING name;
|
20
26
|
|
21
27
|
pop: >
|
22
|
-
DELETE FROM queue
|
23
|
-
WHERE (name, fire_at, id)
|
28
|
+
DELETE FROM queue
|
29
|
+
WHERE (name, fire_at, id)
|
24
30
|
IN (
|
25
|
-
|
26
|
-
|
27
|
-
AND fire_at <= (unixepoch())
|
28
|
-
|
29
|
-
|
30
|
-
)
|
31
|
-
RETURNING id, value
|
32
|
-
|
33
|
-
delete:
|
34
|
-
|
31
|
+
SELECT name, fire_at, id FROM queue
|
32
|
+
WHERE name = ifnull($1, 'default')
|
33
|
+
AND fire_at <= (unixepoch())
|
34
|
+
ORDER BY fire_at ASC
|
35
|
+
LIMIT ifnull($2, 1)
|
36
|
+
)
|
37
|
+
RETURNING id, value;
|
38
|
+
|
39
|
+
delete: >
|
40
|
+
DELETE FROM queue
|
41
|
+
WHERE id = $1
|
42
|
+
RETURNING value;
|
43
|
+
|
35
44
|
info: >
|
36
|
-
SELECT
|
37
|
-
name,
|
38
|
-
count(*) AS count,
|
39
|
-
avg(unixepoch() - created_at) AS avg,
|
40
|
-
min(unixepoch() - created_at) AS min,
|
45
|
+
SELECT
|
46
|
+
name,
|
47
|
+
count(*) AS count,
|
48
|
+
avg(unixepoch() - created_at) AS avg,
|
49
|
+
min(unixepoch() - created_at) AS min,
|
41
50
|
max(unixepoch() - created_at) AS max
|
42
|
-
FROM queue
|
43
|
-
GROUP BY name
|
44
|
-
ORDER BY count DESC
|
45
|
-
|
51
|
+
FROM queue
|
52
|
+
GROUP BY name
|
53
|
+
ORDER BY count DESC;
|
@@ -0,0 +1,84 @@
|
|
1
|
+
# frozen_stringe_literal: true
|
2
|
+
|
3
|
+
module Litescheduler
|
4
|
+
# cache the scheduler we are running in
|
5
|
+
# it is an error to change the scheduler for a process
|
6
|
+
# or for a child forked from that process
|
7
|
+
def self.backend
|
8
|
+
@backend ||= if Fiber.scheduler
|
9
|
+
:fiber
|
10
|
+
elsif defined? Polyphony
|
11
|
+
:polyphony
|
12
|
+
elsif defined? Iodine
|
13
|
+
:iodine
|
14
|
+
else
|
15
|
+
:threaded
|
16
|
+
end
|
17
|
+
end
|
18
|
+
|
19
|
+
# spawn a new execution context
|
20
|
+
def self.spawn(&block)
|
21
|
+
if backend == :fiber
|
22
|
+
Fiber.schedule(&block)
|
23
|
+
elsif backend == :polyphony
|
24
|
+
spin(&block)
|
25
|
+
elsif (backend == :threaded) || (backend == :iodine)
|
26
|
+
Thread.new(&block)
|
27
|
+
end
|
28
|
+
# we should never reach here
|
29
|
+
end
|
30
|
+
|
31
|
+
def self.storage
|
32
|
+
if backend == :fiber || backend == :poylphony
|
33
|
+
Fiber.current.storage
|
34
|
+
else
|
35
|
+
Thread.current
|
36
|
+
end
|
37
|
+
end
|
38
|
+
|
39
|
+
def self.current
|
40
|
+
if backend == :fiber || backend == :poylphony
|
41
|
+
Fiber.current
|
42
|
+
else
|
43
|
+
Thread.current
|
44
|
+
end
|
45
|
+
end
|
46
|
+
|
47
|
+
# switch the execution context to allow others to run
|
48
|
+
def self.switch
|
49
|
+
if backend == :fiber
|
50
|
+
Fiber.scheduler.yield
|
51
|
+
true
|
52
|
+
elsif backend == :polyphony
|
53
|
+
Fiber.current.schedule
|
54
|
+
Thread.current.switch_fiber
|
55
|
+
true
|
56
|
+
else
|
57
|
+
# Thread.pass
|
58
|
+
false
|
59
|
+
end
|
60
|
+
end
|
61
|
+
|
62
|
+
# bold assumption, we will only synchronize threaded code!
|
63
|
+
# If some code explicitly wants to synchronize a fiber
|
64
|
+
# they must send (true) as a parameter to this method
|
65
|
+
# else it is a no-op for fibers
|
66
|
+
def self.synchronize(fiber_sync = false, &block)
|
67
|
+
if (backend == :fiber) || (backend == :polyphony)
|
68
|
+
yield # do nothing, just run the block as is
|
69
|
+
else
|
70
|
+
mutex.synchronize(&block)
|
71
|
+
end
|
72
|
+
end
|
73
|
+
|
74
|
+
def self.max_contexts
|
75
|
+
return 50 if backend == :fiber || backend == :polyphony
|
76
|
+
5
|
77
|
+
end
|
78
|
+
|
79
|
+
# mutex initialization
|
80
|
+
def self.mutex
|
81
|
+
# a single mutex per process (is that ok?)
|
82
|
+
@@mutex ||= Mutex.new
|
83
|
+
end
|
84
|
+
end
|
@@ -0,0 +1,260 @@
|
|
1
|
+
require "oj"
|
2
|
+
require_relative "./schema"
|
3
|
+
|
4
|
+
class Litesearch::Index
|
5
|
+
DEFAULT_SEARCH_OPTIONS = {limit: 25, offset: 0}
|
6
|
+
|
7
|
+
def initialize(db, name)
|
8
|
+
@db = db # this index instance will always belong to this db instance
|
9
|
+
@stmts = {}
|
10
|
+
name = name.to_s.downcase.to_sym
|
11
|
+
# if in the db then put in cache and return if no schema is given
|
12
|
+
# if a schema is given then compare the new and the existing schema
|
13
|
+
# if they are the same put in cache and return
|
14
|
+
# if they differ only in weights then set the new weights, update the schema, put in cache and return
|
15
|
+
# if they differ in fields (added/removed/renamed) then update the structure, then rebuild if auto-rebuild is on
|
16
|
+
# if they differ in tokenizer then rebuild if auto-rebuild is on (error otherwise)
|
17
|
+
# if they differ in both then update the structure and rebuild if auto-rebuild is on (error otherwise)
|
18
|
+
load_index(name) if exists?(name)
|
19
|
+
|
20
|
+
if block_given?
|
21
|
+
schema = Litesearch::Schema.new
|
22
|
+
schema.schema[:name] = name
|
23
|
+
yield schema
|
24
|
+
schema.post_init
|
25
|
+
# now that we have a schema object we need to check if we need to create or modify and existing index
|
26
|
+
if @db.transaction_active?
|
27
|
+
if exists?(name)
|
28
|
+
load_index(name)
|
29
|
+
do_modify(schema)
|
30
|
+
else
|
31
|
+
do_create(schema)
|
32
|
+
end
|
33
|
+
prepare_statements
|
34
|
+
else
|
35
|
+
@db.transaction(:immediate) do
|
36
|
+
if exists?(name)
|
37
|
+
load_index(name)
|
38
|
+
do_modify(schema)
|
39
|
+
else
|
40
|
+
do_create(schema)
|
41
|
+
end
|
42
|
+
prepare_statements
|
43
|
+
end
|
44
|
+
end
|
45
|
+
elsif exists?(name)
|
46
|
+
load_index(name)
|
47
|
+
prepare_statements
|
48
|
+
# an index already exists, load it from the database and return the index instance to the caller
|
49
|
+
else
|
50
|
+
raise "index does not exist and no schema was supplied"
|
51
|
+
end
|
52
|
+
end
|
53
|
+
|
54
|
+
def load_index(name)
|
55
|
+
# we cannot use get_config_value here since the schema object is not created yet, should we allow something here?
|
56
|
+
@schema = begin
|
57
|
+
Litesearch::Schema.new(Oj.load(@db.get_first_value("SELECT v from #{name}_config where k = ?", :litesearch_schema.to_s)))
|
58
|
+
rescue
|
59
|
+
nil
|
60
|
+
end
|
61
|
+
raise "index configuration not found, either corrupted or not a litesearch index!" if @schema.nil?
|
62
|
+
self
|
63
|
+
end
|
64
|
+
|
65
|
+
def modify
|
66
|
+
schema = Litesearch::Schema.new
|
67
|
+
yield schema
|
68
|
+
schema.schema[:name] = @schema.schema[:name]
|
69
|
+
do_modify(schema)
|
70
|
+
end
|
71
|
+
|
72
|
+
def rebuild!
|
73
|
+
if @db.transaction_active?
|
74
|
+
do_rebuild
|
75
|
+
else
|
76
|
+
@db.transaction(:immediate) { do_rebuild }
|
77
|
+
end
|
78
|
+
end
|
79
|
+
|
80
|
+
def add(document)
|
81
|
+
@stmts[:insert].execute!(document)
|
82
|
+
@db.last_insert_row_id
|
83
|
+
end
|
84
|
+
|
85
|
+
def remove(id)
|
86
|
+
@stmts[:delete].execute!(id)
|
87
|
+
end
|
88
|
+
|
89
|
+
def count(term = nil)
|
90
|
+
if term
|
91
|
+
@stmts[:count].execute!(term)[0][0]
|
92
|
+
else
|
93
|
+
@stmts[:count_all].execute![0][0]
|
94
|
+
end
|
95
|
+
end
|
96
|
+
|
97
|
+
# search options include
|
98
|
+
# limit: how many records to return
|
99
|
+
# offset: start from which record
|
100
|
+
def search(term, options = {})
|
101
|
+
options = DEFAULT_SEARCH_OPTIONS.merge(options)
|
102
|
+
rs = @stmts[:search].execute(term, options[:limit], options[:offset])
|
103
|
+
generate_results(rs)
|
104
|
+
end
|
105
|
+
|
106
|
+
def similar(id, limit=10)
|
107
|
+
# pp term = @db.execute(@schema.sql_for(:similarity_query), id)
|
108
|
+
if @schema.schema[:tokenizer] == :trigram
|
109
|
+
# just use the normal similarity approach for now
|
110
|
+
# need to recondisder that for trigram indexes later
|
111
|
+
rs = @stmts[:similar].execute(id, limit)
|
112
|
+
else
|
113
|
+
rs = @stmts[:similar].execute(id, limit)
|
114
|
+
end
|
115
|
+
generate_results(rs)
|
116
|
+
end
|
117
|
+
|
118
|
+
def clear!
|
119
|
+
@stmts[:delete_all].execute!(id)
|
120
|
+
end
|
121
|
+
|
122
|
+
def drop!
|
123
|
+
if @schema.get(:type) == :backed
|
124
|
+
@db.execute_batch(@schema.sql_for(:drop_primary_triggers))
|
125
|
+
if @schema.sql_for(:create_secondary_triggers)
|
126
|
+
@db.execute_batch(@schema.sql_for(:drop_secondary_triggers))
|
127
|
+
end
|
128
|
+
end
|
129
|
+
@db.execute(@schema.sql_for(:drop))
|
130
|
+
end
|
131
|
+
|
132
|
+
private
|
133
|
+
|
134
|
+
def generate_results(rs)
|
135
|
+
result = []
|
136
|
+
if @db.results_as_hash
|
137
|
+
rs.each_hash do |hash|
|
138
|
+
result << hash
|
139
|
+
end
|
140
|
+
else
|
141
|
+
result = rs.to_a
|
142
|
+
end
|
143
|
+
result
|
144
|
+
end
|
145
|
+
|
146
|
+
def exists?(name)
|
147
|
+
@db.get_first_value("SELECT count(*) FROM SQLITE_MASTER WHERE name = ? AND type = 'table' AND (sql like '%fts5%' OR sql like '%FTS5%')", name.to_s) == 1
|
148
|
+
end
|
149
|
+
|
150
|
+
def prepare_statements
|
151
|
+
stmt_names = [:insert, :delete, :delete_all, :drop, :count, :count_all, :search, :similar]
|
152
|
+
stmt_names.each do |stmt_name|
|
153
|
+
@stmts[stmt_name] = @db.prepare(@schema.sql_for(stmt_name))
|
154
|
+
end
|
155
|
+
end
|
156
|
+
|
157
|
+
def do_create(schema)
|
158
|
+
@schema = schema
|
159
|
+
@schema.clean
|
160
|
+
# create index
|
161
|
+
@db.execute(schema.sql_for(:create_index, true))
|
162
|
+
@db.execute_batch(schema.sql_for(:create_vocab_tables))
|
163
|
+
# adjust ranking function
|
164
|
+
@db.execute(schema.sql_for(:ranks, true))
|
165
|
+
# create triggers (if any)
|
166
|
+
if @schema.get(:type) == :backed
|
167
|
+
@db.execute_batch(@schema.sql_for(:create_primary_triggers))
|
168
|
+
if (secondary_triggers_sql = @schema.sql_for(:create_secondary_triggers))
|
169
|
+
@db.execute_batch(secondary_triggers_sql)
|
170
|
+
end
|
171
|
+
@db.execute(@schema.sql_for(:rebuild)) if @schema.get(:rebuild_on_create)
|
172
|
+
end
|
173
|
+
set_config_value(:litesearch_schema, @schema.schema)
|
174
|
+
end
|
175
|
+
|
176
|
+
def do_modify(new_schema)
|
177
|
+
changes = @schema.compare(new_schema)
|
178
|
+
# ensure the new schema maintains feild order
|
179
|
+
new_schema.order_fields(@schema)
|
180
|
+
# with the changes object decide what needs to be done to the schema
|
181
|
+
requires_schema_change = false
|
182
|
+
requires_trigger_change = false
|
183
|
+
requires_rebuild = false
|
184
|
+
if changes[:fields] || changes[:table] || changes[:tokenizer] || changes[:filter_column] || changes[:removed_fields_count] > 0 # any change here will require a schema change
|
185
|
+
requires_schema_change = true
|
186
|
+
# only a change in tokenizer
|
187
|
+
requires_rebuild = changes[:tokenizer] || new_schema.get(:rebuild_on_modify)
|
188
|
+
requires_trigger_change = (changes[:table] || changes[:fields] || changes[:filter_column]) && @schema.get(:type) == :backed
|
189
|
+
end
|
190
|
+
if requires_schema_change
|
191
|
+
# 1. enable schema editing
|
192
|
+
@db.execute("PRAGMA WRITABLE_SCHEMA = TRUE")
|
193
|
+
# 2. update the index sql
|
194
|
+
@db.execute(new_schema.sql_for(:update_index), new_schema.sql_for(:create_index))
|
195
|
+
# 3. update the content table sql (if it exists)
|
196
|
+
@db.execute(new_schema.sql_for(:update_content_table), new_schema.sql_for(:create_content_table, new_schema.schema[:fields].count))
|
197
|
+
# adjust shadow tables
|
198
|
+
@db.execute(new_schema.sql_for(:expand_data), changes[:extra_fields_count])
|
199
|
+
@db.execute(new_schema.sql_for(:expand_docsize), changes[:extra_fields_count])
|
200
|
+
@db.execute("PRAGMA WRITABLE_SCHEMA = RESET")
|
201
|
+
# need to reprepare statements
|
202
|
+
end
|
203
|
+
if requires_trigger_change
|
204
|
+
@db.execute_batch(new_schema.sql_for(:drop_primary_triggers))
|
205
|
+
@db.execute_batch(new_schema.sql_for(:create_primary_triggers))
|
206
|
+
if (secondary_triggers_sql = new_schema.sql_for(:create_secondary_triggers))
|
207
|
+
@db.execute_batch(new_schema.sql_for(:drop_secondary_triggers))
|
208
|
+
@db.execute_batch(secondary_triggers_sql)
|
209
|
+
end
|
210
|
+
end
|
211
|
+
if changes[:fields] || changes[:table] || changes[:tokenizer] || changes[:weights] || changes[:filter_column]
|
212
|
+
@schema = new_schema
|
213
|
+
set_config_value(:litesearch_schema, @schema.schema)
|
214
|
+
prepare_statements
|
215
|
+
# save_schema
|
216
|
+
end
|
217
|
+
# update the weights if they changed
|
218
|
+
@db.execute(@schema.sql_for(:ranks, true)) if changes[:weights]
|
219
|
+
@db.execute_batch(@schema.sql_for(:create_vocab_tables))
|
220
|
+
do_rebuild if requires_rebuild
|
221
|
+
end
|
222
|
+
|
223
|
+
def do_rebuild
|
224
|
+
# remove any zero weight columns
|
225
|
+
if @schema.get(:type) == :backed
|
226
|
+
@db.execute_batch(@schema.sql_for(:drop_primary_triggers))
|
227
|
+
if (secondary_triggers_sql = @schema.sql_for(:create_secondary_triggers))
|
228
|
+
@db.execute_batch(@schema.sql_for(:drop_secondary_triggers))
|
229
|
+
end
|
230
|
+
@db.execute(@schema.sql_for(:drop))
|
231
|
+
@db.execute(@schema.sql_for(:create_index, true))
|
232
|
+
@db.execute_batch(@schema.sql_for(:create_primary_triggers))
|
233
|
+
@db.execute_batch(secondary_triggers_sql) if secondary_triggers_sql
|
234
|
+
@db.execute(@schema.sql_for(:rebuild))
|
235
|
+
elsif @schema.get(:type) == :standalone
|
236
|
+
removables = []
|
237
|
+
@schema.get(:fields).each_with_index { |f, i| removables << [f[0], i] if f[1][:weight] == 0 }
|
238
|
+
removables.each do |col|
|
239
|
+
@db.execute(@schema.sql_for(:drop_content_col, col[1]))
|
240
|
+
@schema.get(:fields).delete(col[0])
|
241
|
+
end
|
242
|
+
@db.execute("PRAGMA WRITABLE_SCHEMA = TRUE")
|
243
|
+
@db.execute(@schema.sql_for(:update_index), @schema.sql_for(:create_index, true))
|
244
|
+
@db.execute(@schema.sql_for(:update_content_table), @schema.sql_for(:create_content_table, @schema.schema[:fields].count))
|
245
|
+
@db.execute("PRAGMA WRITABLE_SCHEMA = RESET")
|
246
|
+
@db.execute(@schema.sql_for(:rebuild))
|
247
|
+
end
|
248
|
+
@db.execute_batch(@schema.sql_for(:create_vocab_tables))
|
249
|
+
set_config_value(:litesearch_schema, @schema.schema)
|
250
|
+
@db.execute(@schema.sql_for(:ranks, true))
|
251
|
+
end
|
252
|
+
|
253
|
+
def get_config_value(key)
|
254
|
+
Oj.load(@db.get_first_value(@schema.sql_for(:get_config_value), key.to_s)) # rescue nil
|
255
|
+
end
|
256
|
+
|
257
|
+
def set_config_value(key, value)
|
258
|
+
@db.execute(@schema.sql_for(:set_config_value), key.to_s, Oj.dump(value))
|
259
|
+
end
|
260
|
+
end
|