sqlite3 2.0.0-x86-linux-musl
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.gemtest +0 -0
- data/CHANGELOG.md +800 -0
- data/CONTRIBUTING.md +56 -0
- data/FAQ.md +388 -0
- data/INSTALLATION.md +267 -0
- data/LICENSE +23 -0
- data/README.md +181 -0
- data/dependencies.yml +13 -0
- data/ext/sqlite3/aggregator.c +270 -0
- data/ext/sqlite3/aggregator.h +10 -0
- data/ext/sqlite3/backup.c +190 -0
- data/ext/sqlite3/backup.h +15 -0
- data/ext/sqlite3/database.c +931 -0
- data/ext/sqlite3/database.h +22 -0
- data/ext/sqlite3/exception.c +117 -0
- data/ext/sqlite3/exception.h +10 -0
- data/ext/sqlite3/extconf.rb +284 -0
- data/ext/sqlite3/sqlite3.c +208 -0
- data/ext/sqlite3/sqlite3_ruby.h +48 -0
- data/ext/sqlite3/statement.c +667 -0
- data/ext/sqlite3/statement.h +16 -0
- data/ext/sqlite3/timespec.h +20 -0
- data/lib/sqlite3/3.0/sqlite3_native.so +0 -0
- data/lib/sqlite3/3.1/sqlite3_native.so +0 -0
- data/lib/sqlite3/3.2/sqlite3_native.so +0 -0
- data/lib/sqlite3/3.3/sqlite3_native.so +0 -0
- data/lib/sqlite3/constants.rb +174 -0
- data/lib/sqlite3/database.rb +701 -0
- data/lib/sqlite3/errors.rb +60 -0
- data/lib/sqlite3/pragmas.rb +585 -0
- data/lib/sqlite3/resultset.rb +96 -0
- data/lib/sqlite3/statement.rb +190 -0
- data/lib/sqlite3/value.rb +54 -0
- data/lib/sqlite3/version.rb +3 -0
- data/lib/sqlite3.rb +17 -0
- metadata +101 -0
data/CONTRIBUTING.md
ADDED
@@ -0,0 +1,56 @@
|
|
1
|
+
# Contributing to sqlite3-ruby
|
2
|
+
|
3
|
+
**This document is a work-in-progress.**
|
4
|
+
|
5
|
+
This doc is a short introduction on how to modify and maintain the sqlite3-ruby gem.
|
6
|
+
|
7
|
+
|
8
|
+
## Architecture notes
|
9
|
+
|
10
|
+
### Garbage collection
|
11
|
+
|
12
|
+
All statements keep pointers back to their respective database connections.
|
13
|
+
The `@connection` instance variable on the `Statement` handle keeps the database
|
14
|
+
connection alive. Memory allocated for a statement handler will be freed in
|
15
|
+
two cases:
|
16
|
+
|
17
|
+
1. `#close` is called on the statement
|
18
|
+
2. The `SQLite3::Database` object gets garbage collected
|
19
|
+
|
20
|
+
We can't free the memory for the statement in the garbage collection function
|
21
|
+
for the statement handler. The reason is because there exists a race
|
22
|
+
condition. We cannot guarantee the order in which objects will be garbage
|
23
|
+
collected. So, it is possible that a connection and a statement are up for
|
24
|
+
garbage collection. If the database connection were to be free'd before the
|
25
|
+
statement, then boom. Instead we'll be conservative and free unclosed
|
26
|
+
statements when the connection is terminated.
|
27
|
+
|
28
|
+
|
29
|
+
|
30
|
+
## Building gems
|
31
|
+
|
32
|
+
As a prerequisite please make sure you have `docker` correctly installed, so that you're able to cross-compile the native gems.
|
33
|
+
|
34
|
+
Run `bin/build-gems` which will package gems for all supported platforms, and run some basic sanity tests on those packages using `bin/test-gem-set` and `bin/test-gem-file-contents`.
|
35
|
+
|
36
|
+
|
37
|
+
## Updating the version of libsqlite3
|
38
|
+
|
39
|
+
Update `/dependencies.yml` to reflect:
|
40
|
+
|
41
|
+
- the version of libsqlite3
|
42
|
+
- the URL from which to download
|
43
|
+
- the checksum of the file, which will need to be verified manually (see comments in that file)
|
44
|
+
|
45
|
+
|
46
|
+
## Making a release
|
47
|
+
|
48
|
+
A quick checklist:
|
49
|
+
|
50
|
+
- [ ] make sure CI is green!
|
51
|
+
- [ ] update `CHANGELOG.md` and `lib/sqlite3/version.rb`
|
52
|
+
- [ ] run `bin/build-gems` and make sure it completes and all the tests pass
|
53
|
+
- [ ] create a git tag using a format that matches the pattern `v\d+\.\d+\.\d+`, e.g. `v1.3.13`
|
54
|
+
- [ ] `git push && git push --tags`
|
55
|
+
- [ ] `for g in gems/*.gem ; do gem push $g ; done`
|
56
|
+
- [ ] create a release at https://github.com/sparklemotion/sqlite3-ruby/releases and include sha2 checksums
|
data/FAQ.md
ADDED
@@ -0,0 +1,388 @@
|
|
1
|
+
|
2
|
+
## How do I do a database query?
|
3
|
+
### I just want an array of the rows...
|
4
|
+
|
5
|
+
Use the `Database#execute` method. If you don't give it a block, it will
|
6
|
+
return an array of all the rows:
|
7
|
+
|
8
|
+
```ruby
|
9
|
+
require 'sqlite3'
|
10
|
+
|
11
|
+
db = SQLite3::Database.new( "test.db" )
|
12
|
+
rows = db.execute( "select * from test" )
|
13
|
+
```
|
14
|
+
|
15
|
+
### I'd like to use a block to iterate through the rows...
|
16
|
+
|
17
|
+
Use the `Database#execute` method. If you give it a block, each row of the
|
18
|
+
result will be yielded to the block:
|
19
|
+
|
20
|
+
|
21
|
+
```ruby
|
22
|
+
require 'sqlite3'
|
23
|
+
|
24
|
+
db = SQLite3::Database.new( "test.db" )
|
25
|
+
db.execute( "select * from test" ) do |row|
|
26
|
+
...
|
27
|
+
end
|
28
|
+
```
|
29
|
+
|
30
|
+
### I need to get the column names as well as the rows...
|
31
|
+
|
32
|
+
Use the `Database#execute2` method. This works just like `Database#execute`;
|
33
|
+
if you don't give it a block, it returns an array of rows; otherwise, it
|
34
|
+
will yield each row to the block. _However_, the first row returned is
|
35
|
+
always an array of the column names from the query:
|
36
|
+
|
37
|
+
|
38
|
+
```ruby
|
39
|
+
require 'sqlite3'
|
40
|
+
|
41
|
+
db = SQLite3::Database.new( "test.db" )
|
42
|
+
columns, *rows = db.execute2( "select * from test" )
|
43
|
+
|
44
|
+
# or use a block:
|
45
|
+
|
46
|
+
columns = nil
|
47
|
+
db.execute2( "select * from test" ) do |row|
|
48
|
+
if columns.nil?
|
49
|
+
columns = row
|
50
|
+
else
|
51
|
+
# process row
|
52
|
+
end
|
53
|
+
end
|
54
|
+
```
|
55
|
+
|
56
|
+
### I just want the first row of the result set...
|
57
|
+
|
58
|
+
Easy. Just call `Database#get_first_row`:
|
59
|
+
|
60
|
+
|
61
|
+
```ruby
|
62
|
+
row = db.get_first_row( "select * from table" )
|
63
|
+
```
|
64
|
+
|
65
|
+
|
66
|
+
This also supports bind variables, just like `Database#execute`
|
67
|
+
and friends.
|
68
|
+
|
69
|
+
### I just want the first value of the first row of the result set...
|
70
|
+
|
71
|
+
Also easy. Just call `Database#get_first_value`:
|
72
|
+
|
73
|
+
|
74
|
+
```ruby
|
75
|
+
count = db.get_first_value( "select count(*) from table" )
|
76
|
+
```
|
77
|
+
|
78
|
+
|
79
|
+
This also supports bind variables, just like `Database#execute`
|
80
|
+
and friends.
|
81
|
+
|
82
|
+
## How do I prepare a statement for repeated execution?
|
83
|
+
|
84
|
+
If the same statement is going to be executed repeatedly, you can speed
|
85
|
+
things up a bit by _preparing_ the statement. You do this via the
|
86
|
+
`Database#prepare` method. It returns a `Statement` object, and you can
|
87
|
+
then invoke `#execute` on that to get the `ResultSet`:
|
88
|
+
|
89
|
+
|
90
|
+
```ruby
|
91
|
+
stmt = db.prepare( "select * from person" )
|
92
|
+
|
93
|
+
1000.times do
|
94
|
+
stmt.execute do |result|
|
95
|
+
...
|
96
|
+
end
|
97
|
+
end
|
98
|
+
|
99
|
+
stmt.close
|
100
|
+
|
101
|
+
# or, use a block
|
102
|
+
|
103
|
+
db.prepare( "select * from person" ) do |stmt|
|
104
|
+
1000.times do
|
105
|
+
stmt.execute do |result|
|
106
|
+
...
|
107
|
+
end
|
108
|
+
end
|
109
|
+
end
|
110
|
+
```
|
111
|
+
|
112
|
+
|
113
|
+
This is made more useful by the ability to bind variables to placeholders
|
114
|
+
via the `Statement#bind_param` and `Statement#bind_params` methods. (See the
|
115
|
+
next FAQ for details.)
|
116
|
+
|
117
|
+
## How do I use placeholders in an SQL statement?
|
118
|
+
|
119
|
+
Placeholders in an SQL statement take any of the following formats:
|
120
|
+
|
121
|
+
|
122
|
+
* `?`
|
123
|
+
* `?_nnn_`
|
124
|
+
* `:_word_`
|
125
|
+
|
126
|
+
|
127
|
+
Where _n_ is an integer, and _word_ is an alpha-numeric identifier (or
|
128
|
+
number). When the placeholder is associated with a number, that number
|
129
|
+
identifies the index of the bind variable to replace it with. When it
|
130
|
+
is an identifier, it identifies the name of the corresponding bind
|
131
|
+
variable. (In the instance of the first format--a single question
|
132
|
+
mark--the placeholder is assigned a number one greater than the last
|
133
|
+
index used, or 1 if it is the first.)
|
134
|
+
|
135
|
+
|
136
|
+
For example, here is a query using these placeholder formats:
|
137
|
+
|
138
|
+
|
139
|
+
```sql
|
140
|
+
select *
|
141
|
+
from table
|
142
|
+
where ( c = ?2 or c = ? )
|
143
|
+
and d = :name
|
144
|
+
and e = :1
|
145
|
+
```
|
146
|
+
|
147
|
+
|
148
|
+
This defines 5 different placeholders: 1, 2, 3, and "name".
|
149
|
+
|
150
|
+
|
151
|
+
You replace these placeholders by _binding_ them to values. This can be
|
152
|
+
accomplished in a variety of ways.
|
153
|
+
|
154
|
+
|
155
|
+
The `Database#execute`, and `Database#execute2` methods all accept additional
|
156
|
+
arguments following the SQL statement. These arguments are assumed to be
|
157
|
+
bind parameters, and they are bound (positionally) to their corresponding
|
158
|
+
placeholders:
|
159
|
+
|
160
|
+
|
161
|
+
```ruby
|
162
|
+
db.execute( "select * from table where a = ? and b = ?",
|
163
|
+
"hello",
|
164
|
+
"world" )
|
165
|
+
```
|
166
|
+
|
167
|
+
|
168
|
+
The above would replace the first question mark with 'hello' and the
|
169
|
+
second with 'world'. If the placeholders have an explicit index given, they
|
170
|
+
will be replaced with the bind parameter at that index (1-based).
|
171
|
+
|
172
|
+
|
173
|
+
If a Hash is given as a bind parameter, then its key/value pairs are bound
|
174
|
+
to the placeholders. This is how you bind by name:
|
175
|
+
|
176
|
+
|
177
|
+
```ruby
|
178
|
+
db.execute( "select * from table where a = :name and b = :value",
|
179
|
+
"name" => "bob",
|
180
|
+
"value" => "priceless" )
|
181
|
+
```
|
182
|
+
|
183
|
+
|
184
|
+
You can also bind explicitly using the `Statement` object itself. Just pass
|
185
|
+
additional parameters to the `Statement#execute` statement:
|
186
|
+
|
187
|
+
|
188
|
+
```ruby
|
189
|
+
db.prepare( "select * from table where a = :name and b = ?" ) do |stmt|
|
190
|
+
stmt.execute "value", "name" => "bob"
|
191
|
+
end
|
192
|
+
```
|
193
|
+
|
194
|
+
|
195
|
+
Or do a `Database#prepare` to get the `Statement`, and then use either
|
196
|
+
`Statement#bind_param` or `Statement#bind_params`:
|
197
|
+
|
198
|
+
|
199
|
+
```ruby
|
200
|
+
stmt = db.prepare( "select * from table where a = :name and b = ?" )
|
201
|
+
|
202
|
+
stmt.bind_param( "name", "bob" )
|
203
|
+
stmt.bind_param( 1, "value" )
|
204
|
+
|
205
|
+
# or
|
206
|
+
|
207
|
+
stmt.bind_params( "value", "name" => "bob" )
|
208
|
+
```
|
209
|
+
|
210
|
+
## How do I discover metadata about a query?
|
211
|
+
|
212
|
+
If you ever want to know the names or types of the columns in a result
|
213
|
+
set, you can do it in several ways.
|
214
|
+
|
215
|
+
|
216
|
+
The first way is to ask the row object itself. Each row will have a
|
217
|
+
property "fields" that returns an array of the column names. The row
|
218
|
+
will also have a property "types" that returns an array of the column
|
219
|
+
types:
|
220
|
+
|
221
|
+
|
222
|
+
```ruby
|
223
|
+
rows = db.execute( "select * from table" )
|
224
|
+
p rows[0].fields
|
225
|
+
p rows[0].types
|
226
|
+
```
|
227
|
+
|
228
|
+
|
229
|
+
Obviously, this approach requires you to execute a statement that actually
|
230
|
+
returns data. If you don't know if the statement will return any rows, but
|
231
|
+
you still need the metadata, you can use `Database#query` and ask the
|
232
|
+
`ResultSet` object itself:
|
233
|
+
|
234
|
+
|
235
|
+
```ruby
|
236
|
+
db.query( "select * from table" ) do |result|
|
237
|
+
p result.columns
|
238
|
+
p result.types
|
239
|
+
...
|
240
|
+
end
|
241
|
+
```
|
242
|
+
|
243
|
+
|
244
|
+
Lastly, you can use `Database#prepare` and ask the `Statement` object what
|
245
|
+
the metadata are:
|
246
|
+
|
247
|
+
|
248
|
+
```ruby
|
249
|
+
stmt = db.prepare( "select * from table" )
|
250
|
+
p stmt.columns
|
251
|
+
p stmt.types
|
252
|
+
```
|
253
|
+
|
254
|
+
## I'd like the rows to be indexible by column name.
|
255
|
+
|
256
|
+
By default, each row from a query is returned as an `Array` of values. This
|
257
|
+
means that you can only obtain values by their index. Sometimes, however,
|
258
|
+
you would like to obtain values by their column name.
|
259
|
+
|
260
|
+
|
261
|
+
The first way to do this is to set the Database property `results_as_hash`
|
262
|
+
to true. If you do this, then all rows will be returned as Hash objects,
|
263
|
+
with the column names as the keys. (In this case, the `fields` property
|
264
|
+
is unavailable on the row, although the "types" property remains.)
|
265
|
+
|
266
|
+
|
267
|
+
```ruby
|
268
|
+
db.results_as_hash = true
|
269
|
+
db.execute( "select * from table" ) do |row|
|
270
|
+
p row['column1']
|
271
|
+
p row['column2']
|
272
|
+
end
|
273
|
+
```
|
274
|
+
|
275
|
+
|
276
|
+
The other way is to use Ara Howard's
|
277
|
+
[`ArrayFields`](http://rubyforge.org/projects/arrayfields)
|
278
|
+
module. Just `require "arrayfields"`, and all of your rows will be indexable
|
279
|
+
by column name, even though they are still arrays!
|
280
|
+
|
281
|
+
|
282
|
+
```ruby
|
283
|
+
require 'arrayfields'
|
284
|
+
|
285
|
+
...
|
286
|
+
db.execute( "select * from table" ) do |row|
|
287
|
+
p row[0] == row['column1']
|
288
|
+
p row[1] == row['column2']
|
289
|
+
end
|
290
|
+
```
|
291
|
+
|
292
|
+
## How do I insert binary data into the database?
|
293
|
+
|
294
|
+
Use blobs. Blobs are new features of SQLite3. You have to use bind
|
295
|
+
variables to make it work:
|
296
|
+
|
297
|
+
|
298
|
+
```ruby
|
299
|
+
db.execute( "insert into foo ( ?, ? )",
|
300
|
+
SQLite3::Blob.new( "\0\1\2\3\4\5" ),
|
301
|
+
SQLite3::Blob.new( "a\0b\0c\0d ) )
|
302
|
+
```
|
303
|
+
|
304
|
+
|
305
|
+
The blob values must be indicated explicitly by binding each parameter to
|
306
|
+
a value of type `SQLite3::Blob`.
|
307
|
+
|
308
|
+
## How do I do a DDL (insert, update, delete) statement?
|
309
|
+
|
310
|
+
You can actually do inserts, updates, and deletes in exactly the same way
|
311
|
+
as selects, but in general the `Database#execute` method will be most
|
312
|
+
convenient:
|
313
|
+
|
314
|
+
|
315
|
+
```ruby
|
316
|
+
db.execute( "insert into table values ( ?, ? )", *bind_vars )
|
317
|
+
```
|
318
|
+
|
319
|
+
## How do I execute multiple statements in a single string?
|
320
|
+
|
321
|
+
The standard query methods (`Database#execute`, `Database#execute2`,
|
322
|
+
`Database#query`, and `Statement#execute`) will only execute the first
|
323
|
+
statement in the string that is given to them. Thus, if you have a
|
324
|
+
string with multiple SQL statements, each separated by a string,
|
325
|
+
you can't use those methods to execute them all at once.
|
326
|
+
|
327
|
+
|
328
|
+
Instead, use `Database#execute_batch`:
|
329
|
+
|
330
|
+
|
331
|
+
```ruby
|
332
|
+
sql = <<SQL
|
333
|
+
create table the_table (
|
334
|
+
a varchar2(30),
|
335
|
+
b varchar2(30)
|
336
|
+
);
|
337
|
+
|
338
|
+
insert into the_table values ( 'one', 'two' );
|
339
|
+
insert into the_table values ( 'three', 'four' );
|
340
|
+
insert into the_table values ( 'five', 'six' );
|
341
|
+
SQL
|
342
|
+
|
343
|
+
db.execute_batch( sql )
|
344
|
+
```
|
345
|
+
|
346
|
+
|
347
|
+
Unlike the other query methods, `Database#execute_batch` accepts no
|
348
|
+
block. It will also only ever return `nil`. Thus, it is really only
|
349
|
+
suitable for batch processing of DDL statements.
|
350
|
+
|
351
|
+
## How do I begin/end a transaction
|
352
|
+
|
353
|
+
Use `Database#transaction` to start a transaction. If you give it a block,
|
354
|
+
the block will be automatically committed at the end of the block,
|
355
|
+
unless an exception was raised, in which case the transaction will be
|
356
|
+
rolled back. (Never explicitly call `Database#commit` or `Database#rollback`
|
357
|
+
inside of a transaction block--you'll get errors when the block
|
358
|
+
terminates!)
|
359
|
+
|
360
|
+
|
361
|
+
```ruby
|
362
|
+
database.transaction do |db|
|
363
|
+
db.execute( "insert into table values ( 'a', 'b', 'c' )" )
|
364
|
+
...
|
365
|
+
end
|
366
|
+
```
|
367
|
+
|
368
|
+
|
369
|
+
Alternatively, if you don't give a block to `Database#transaction`, the
|
370
|
+
transaction remains open until you explicitly call `Database#commit` or
|
371
|
+
`Database#rollback`.
|
372
|
+
|
373
|
+
|
374
|
+
```ruby
|
375
|
+
db.transaction
|
376
|
+
db.execute( "insert into table values ( 'a', 'b', 'c' )" )
|
377
|
+
db.commit
|
378
|
+
```
|
379
|
+
|
380
|
+
|
381
|
+
Note that SQLite does not allow nested transactions, so you'll get errors
|
382
|
+
if you try to open a new transaction while one is already active. Use
|
383
|
+
`Database#transaction_active?` to determine whether a transaction is
|
384
|
+
active or not.
|
385
|
+
|
386
|
+
## How do I discover metadata about a table/index?
|
387
|
+
|
388
|
+
## How do I do tweak database settings?
|
data/INSTALLATION.md
ADDED
@@ -0,0 +1,267 @@
|
|
1
|
+
|
2
|
+
# Installation and Using SQLite3 extensions
|
3
|
+
|
4
|
+
This document will help you install the `sqlite3` ruby gem. It also contains instructions on loading database extensions and building against drop-in replacements for sqlite3.
|
5
|
+
|
6
|
+
## Installation
|
7
|
+
|
8
|
+
### Native Gems (recommended)
|
9
|
+
|
10
|
+
In v2.0.0 and later, native (precompiled) gems are available for recent Ruby versions on these platforms:
|
11
|
+
|
12
|
+
- `aarch64-linux-gnu` (requires: glibc >= 2.29)
|
13
|
+
- `aarch64-linux-musl`
|
14
|
+
- `arm-linux-gnu` (requires: glibc >= 2.29)
|
15
|
+
- `arm-linux-musl`
|
16
|
+
- `arm64-darwin`
|
17
|
+
- `x64-mingw32` / `x64-mingw-ucrt`
|
18
|
+
- `x86-linux-gnu` (requires: glibc >= 2.17)
|
19
|
+
- `x86-linux-musl`
|
20
|
+
- `x86_64-darwin`
|
21
|
+
- `x86_64-linux-gnu` (requires: glibc >= 2.17)
|
22
|
+
- `x86_64-linux-musl`
|
23
|
+
|
24
|
+
⚠ Ruby 3.0 linux users must use Rubygems >= 3.3.22 in order to use these gems.
|
25
|
+
|
26
|
+
⚠ Musl linux users should update to Bundler >= 2.5.6 to avoid https://github.com/rubygems/rubygems/issues/7432
|
27
|
+
|
28
|
+
If you are using one of these Ruby versions on one of these platforms, the native gem is the recommended way to install sqlite3-ruby.
|
29
|
+
|
30
|
+
For example, on a linux system running Ruby 3.1:
|
31
|
+
|
32
|
+
``` text
|
33
|
+
$ ruby -v
|
34
|
+
ruby 3.1.2p20 (2022-04-12 revision 4491bb740a) [x86_64-linux]
|
35
|
+
|
36
|
+
$ time gem install sqlite3
|
37
|
+
Fetching sqlite3-1.5.0-x86_64-linux.gem
|
38
|
+
Successfully installed sqlite3-1.5.0-x86_64-linux
|
39
|
+
1 gem installed
|
40
|
+
|
41
|
+
real 0m4.274s
|
42
|
+
user 0m0.734s
|
43
|
+
sys 0m0.165s
|
44
|
+
```
|
45
|
+
|
46
|
+
#### Avoiding the precompiled native gem
|
47
|
+
|
48
|
+
The maintainers strongly urge you to use a native gem if at all possible. It will be a better experience for you and allow us to focus our efforts on improving functionality rather than diagnosing installation issues.
|
49
|
+
|
50
|
+
If you're on a platform that supports a native gem but you want to avoid using it in your project, do one of the following:
|
51
|
+
|
52
|
+
- If you're not using Bundler, then run `gem install sqlite3 --platform=ruby`
|
53
|
+
- If you are using Bundler
|
54
|
+
- version 2.3.18 or later, you can specify [`gem "sqlite3", force_ruby_platform: true`](https://bundler.io/v2.3/man/gemfile.5.html#FORCE_RUBY_PLATFORM)
|
55
|
+
- version 2.1 or later, then you'll need to run `bundle config set force_ruby_platform true`
|
56
|
+
- version 2.0 or earlier, then you'll need to run `bundle config force_ruby_platform true`
|
57
|
+
|
58
|
+
|
59
|
+
### Compiling the source gem
|
60
|
+
|
61
|
+
If you are on a platform or version of Ruby that is not covered by the Native Gems, then the vanilla "ruby platform" (non-native) gem will be installed by the `gem install` or `bundle` commands.
|
62
|
+
|
63
|
+
|
64
|
+
#### Packaged libsqlite3
|
65
|
+
|
66
|
+
By default, as of v1.5.0 of this library, the latest available version of libsqlite3 is packaged with the gem and will be compiled and used automatically. This takes a bit longer than the native gem, but will provide a modern, well-supported version of libsqlite3.
|
67
|
+
|
68
|
+
For example, on a linux system running Ruby 2.5:
|
69
|
+
|
70
|
+
``` text
|
71
|
+
$ ruby -v
|
72
|
+
ruby 2.5.9p229 (2021-04-05 revision 67939) [x86_64-linux]
|
73
|
+
|
74
|
+
$ time gem install sqlite3
|
75
|
+
Building native extensions. This could take a while...
|
76
|
+
Successfully installed sqlite3-1.5.0
|
77
|
+
1 gem installed
|
78
|
+
|
79
|
+
real 0m20.620s
|
80
|
+
user 0m23.361s
|
81
|
+
sys 0m5.839s
|
82
|
+
```
|
83
|
+
|
84
|
+
##### Controlling compilation flags for sqlite
|
85
|
+
|
86
|
+
Upstream sqlite allows for the setting of some parameters at compile time. If you're an expert and would like to set these, you may do so at gem install time in two different ways ...
|
87
|
+
|
88
|
+
**If you're installing the gem using `gem install`** then you can pass in these compile-time flags like this:
|
89
|
+
|
90
|
+
``` sh
|
91
|
+
gem install sqlite3 --platform=ruby -- \
|
92
|
+
--with-sqlite-cflags="-DSQLITE_DEFAULT_CACHE_SIZE=9999 -DSQLITE_DEFAULT_PAGE_SIZE=4444"
|
93
|
+
```
|
94
|
+
|
95
|
+
or the equivalent:
|
96
|
+
|
97
|
+
``` sh
|
98
|
+
CFLAGS="-DSQLITE_DEFAULT_CACHE_SIZE=9999 -DSQLITE_DEFAULT_PAGE_SIZE=4444" \
|
99
|
+
gem install sqlite3 --platform=ruby
|
100
|
+
```
|
101
|
+
|
102
|
+
**If you're installing the gem using `bundler`** then you should first pin the gem to the "ruby" platform gem, so that you are compiling from source:
|
103
|
+
|
104
|
+
``` ruby
|
105
|
+
# Gemfile
|
106
|
+
gem "sqlite3", force_ruby_platform: true # requires bundler >= 2.3.18
|
107
|
+
```
|
108
|
+
|
109
|
+
and then set up a bundler config parameter for `build.sqlite3`:
|
110
|
+
|
111
|
+
``` sh
|
112
|
+
bundle config set build.sqlite3 \
|
113
|
+
"--with-sqlite-cflags='-DSQLITE_DEFAULT_CACHE_SIZE=9999 -DSQLITE_DEFAULT_PAGE_SIZE=4444'"
|
114
|
+
```
|
115
|
+
|
116
|
+
NOTE the use of single quotes within the double-quoted string to ensure the space between compiler flags is interpreted correctly. The contents of your `.bundle/config` file should look like:
|
117
|
+
|
118
|
+
``` yaml
|
119
|
+
---
|
120
|
+
BUNDLE_BUILD__SQLITE3: "--with-sqlite-cflags='-DSQLITE_DEFAULT_CACHE_SIZE=9999 -DSQLITE_DEFAULT_PAGE_SIZE=4444'"
|
121
|
+
```
|
122
|
+
|
123
|
+
|
124
|
+
#### System libsqlite3
|
125
|
+
|
126
|
+
If you would prefer to build the sqlite3-ruby gem against your system libsqlite3, which requires that you install libsqlite3 and its development files yourself, you may do so by using the `--enable-system-libraries` flag at gem install time.
|
127
|
+
|
128
|
+
PLEASE NOTE:
|
129
|
+
|
130
|
+
- you must avoid installing a precompiled native gem (see [previous section](#avoiding-the-precompiled-native-gem))
|
131
|
+
- only versions of libsqlite3 `>= 3.5.0` are supported,
|
132
|
+
- and some library features may depend on how your libsqlite3 was compiled.
|
133
|
+
|
134
|
+
For example, on a linux system running Ruby 2.5:
|
135
|
+
|
136
|
+
``` text
|
137
|
+
$ time gem install sqlite3 -- --enable-system-libraries
|
138
|
+
Building native extensions with: '--enable-system-libraries'
|
139
|
+
This could take a while...
|
140
|
+
Successfully installed sqlite3-1.5.0
|
141
|
+
1 gem installed
|
142
|
+
|
143
|
+
real 0m4.234s
|
144
|
+
user 0m3.809s
|
145
|
+
sys 0m0.912s
|
146
|
+
```
|
147
|
+
|
148
|
+
If you're using bundler, you can opt into system libraries like this:
|
149
|
+
|
150
|
+
``` sh
|
151
|
+
bundle config build.sqlite3 --enable-system-libraries
|
152
|
+
```
|
153
|
+
|
154
|
+
If you have sqlite3 installed in a non-standard location, you may need to specify the location of the include and lib files by using `--with-sqlite-include` and `--with-sqlite-lib` options (or a `--with-sqlite-dir` option, see [MakeMakefile#dir_config](https://ruby-doc.org/stdlib-3.1.1/libdoc/mkmf/rdoc/MakeMakefile.html#method-i-dir_config)). If you have pkg-config installed and configured properly, this may not be necessary.
|
155
|
+
|
156
|
+
``` sh
|
157
|
+
gem install sqlite3 -- \
|
158
|
+
--enable-system-libraries \
|
159
|
+
--with-sqlite3-include=/opt/local/include \
|
160
|
+
--with-sqlite3-lib=/opt/local/lib
|
161
|
+
```
|
162
|
+
|
163
|
+
|
164
|
+
#### System libsqlcipher
|
165
|
+
|
166
|
+
If you'd like to link against a system-installed libsqlcipher, you may do so by using the `--with-sqlcipher` flag:
|
167
|
+
|
168
|
+
``` text
|
169
|
+
$ time gem install sqlite3 -- --with-sqlcipher
|
170
|
+
Building native extensions with: '--with-sqlcipher'
|
171
|
+
This could take a while...
|
172
|
+
Successfully installed sqlite3-1.5.0
|
173
|
+
1 gem installed
|
174
|
+
|
175
|
+
real 0m4.772s
|
176
|
+
user 0m3.906s
|
177
|
+
sys 0m0.896s
|
178
|
+
```
|
179
|
+
|
180
|
+
If you have sqlcipher installed in a non-standard location, you may need to specify the location of the include and lib files by using `--with-sqlite-include` and `--with-sqlite-lib` options (or a `--with-sqlite-dir` option, see [MakeMakefile#dir_config](https://ruby-doc.org/stdlib-3.1.1/libdoc/mkmf/rdoc/MakeMakefile.html#method-i-dir_config)). If you have pkg-config installed and configured properly, this may not be necessary.
|
181
|
+
|
182
|
+
|
183
|
+
## Using SQLite3 extensions
|
184
|
+
|
185
|
+
### How do I load a sqlite extension?
|
186
|
+
|
187
|
+
Some add-ons are available to sqlite as "extensions". The instructions that upstream sqlite provides at https://www.sqlite.org/loadext.html are the canonical source of advice, but here's a brief example showing how you can do this with the `sqlite3` ruby gem.
|
188
|
+
|
189
|
+
In this example, I'll be loading the ["spellfix" extension](https://www.sqlite.org/spellfix1.html):
|
190
|
+
|
191
|
+
``` text
|
192
|
+
# download spellfix.c from somewherehttp://www.sqlite.org/src/finfo?name=ext/misc/spellfix.c
|
193
|
+
$ wget https://raw.githubusercontent.com/sqlite/sqlite/master/ext/misc/spellfix.c
|
194
|
+
spellfix.c 100%[=================================================>] 100.89K --.-KB/s in 0.09s
|
195
|
+
|
196
|
+
# follow instructions at https://www.sqlite.org/loadext.html
|
197
|
+
# (you will need sqlite3 development packages for this)
|
198
|
+
$ gcc -g -fPIC -shared spellfix.c -o spellfix.o
|
199
|
+
|
200
|
+
$ ls -lt
|
201
|
+
total 192
|
202
|
+
-rwxrwxr-x 1 flavorjones flavorjones 87984 2023-05-24 10:44 spellfix.o
|
203
|
+
-rw-rw-r-- 1 flavorjones flavorjones 103310 2023-05-24 10:43 spellfix.c
|
204
|
+
```
|
205
|
+
|
206
|
+
Then, in your application, use that `spellfix.o` file like this:
|
207
|
+
|
208
|
+
``` ruby
|
209
|
+
require "sqlite3"
|
210
|
+
|
211
|
+
db = SQLite3::Database.new(':memory:')
|
212
|
+
db.enable_load_extension(true)
|
213
|
+
db.load_extension("/path/to/sqlite/spellfix.o")
|
214
|
+
db.execute("CREATE VIRTUAL TABLE demo USING spellfix1;")
|
215
|
+
```
|
216
|
+
|
217
|
+
### How do I use my own sqlite3 shared library?
|
218
|
+
|
219
|
+
Some folks have strong opinions about what features they want compiled into sqlite3; or may be using a package like SQLite Encryption Extension ("SEE"). This section will explain how to get your Ruby application to load that specific shared library.
|
220
|
+
|
221
|
+
If you've installed your alternative as an autotools-style installation, the directory structure will look like this:
|
222
|
+
|
223
|
+
```
|
224
|
+
/opt/sqlite3
|
225
|
+
├── bin
|
226
|
+
│ └── sqlite3
|
227
|
+
├── include
|
228
|
+
│ ├── sqlite3.h
|
229
|
+
│ └── sqlite3ext.h
|
230
|
+
├── lib
|
231
|
+
│ ├── libsqlite3.a
|
232
|
+
│ ├── libsqlite3.la
|
233
|
+
│ ├── libsqlite3.so -> libsqlite3.so.0.8.6
|
234
|
+
│ ├── libsqlite3.so.0 -> libsqlite3.so.0.8.6
|
235
|
+
│ ├── libsqlite3.so.0.8.6
|
236
|
+
│ └── pkgconfig
|
237
|
+
│ └── sqlite3.pc
|
238
|
+
└── share
|
239
|
+
└── man
|
240
|
+
└── man1
|
241
|
+
└── sqlite3.1
|
242
|
+
```
|
243
|
+
|
244
|
+
You can build this gem against that library like this:
|
245
|
+
|
246
|
+
```
|
247
|
+
gem install sqlite3 --platform=ruby -- \
|
248
|
+
--enable-system-libraries \
|
249
|
+
--with-opt-dir=/opt/sqlite
|
250
|
+
```
|
251
|
+
|
252
|
+
Explanation:
|
253
|
+
|
254
|
+
- use `--platform=ruby` to avoid the precompiled native gems (see the README)
|
255
|
+
- the `--` separates arguments passed to "gem install" from arguments passed to the C extension builder
|
256
|
+
- use `--enable-system-libraries` to avoid the vendored sqlite3 source
|
257
|
+
- use `--with-opt-dir=/path/to/installation` to point the build process at the desired header files and shared object files
|
258
|
+
|
259
|
+
Alternatively, if you've simply downloaded an "amalgamation" and so your compiled library and header files are in arbitrary locations, try this more detailed command:
|
260
|
+
|
261
|
+
```
|
262
|
+
gem install sqlite3 --platform=ruby -- \
|
263
|
+
--enable-system-libraries \
|
264
|
+
--with-opt-include=/path/to/include \
|
265
|
+
--with-opt-lib=/path/to/lib
|
266
|
+
```
|
267
|
+
|