activerecord-cassandra-adapter 0.1.0
Sign up to get free protection for your applications and to get access to all the features.
- data/README.rdoc +69 -0
- data/lib/active_cassandra/cf.rb +88 -0
- data/lib/active_cassandra/sqlparser.tab.rb +790 -0
- data/lib/active_cassandra/sqlparser.y +298 -0
- data/lib/active_record/connection_adapters/cassandra_adapter.rb +269 -0
- metadata +69 -0
@@ -0,0 +1,298 @@
|
|
1
|
+
class SQLParser
|
2
|
+
options no_result_var
|
3
|
+
rule
|
4
|
+
sql : create_statement
|
5
|
+
| read_statemant
|
6
|
+
| update_statemant
|
7
|
+
| delete_statemant
|
8
|
+
|
9
|
+
create_statement : INSERT INTO id '(' id_list ')' VALUES '(' value_list ')'
|
10
|
+
{
|
11
|
+
{:command => :insert, :table => val[2], :column_list => val[4], :value_list => val[8]}
|
12
|
+
}
|
13
|
+
|
14
|
+
read_statemant : SELECT select_list FROM id where_clause order_by_clause limit_clause offset_clause
|
15
|
+
{
|
16
|
+
{:command => :select, :table => val[3], :select_list => val[1], :condition => val[4], :order => val[5], :limit => val[6], :offset => val[7]}
|
17
|
+
}
|
18
|
+
| SELECT DISTINCT id FROM id where_clause
|
19
|
+
{
|
20
|
+
{:command => :select, :table => val[4], :select_list => val[2], :distinct => val[2], :condition => val[5]}
|
21
|
+
}
|
22
|
+
| SELECT count_clause FROM id where_clause order_by_clause limit_clause offset_clause
|
23
|
+
{
|
24
|
+
{:command => :select, :table => val[3], :count => val[1], :condition => val[4], :order => val[5], :limit => val[6], :offset => val[7]}
|
25
|
+
}
|
26
|
+
|
27
|
+
count_clause : COUNT '(' count_arg ')'
|
28
|
+
{
|
29
|
+
"count_all"
|
30
|
+
}
|
31
|
+
| COUNT '(' count_arg ')' AS id
|
32
|
+
{
|
33
|
+
val[5]
|
34
|
+
}
|
35
|
+
|
36
|
+
count_arg : '*'
|
37
|
+
| id
|
38
|
+
|
39
|
+
select_list : '*'
|
40
|
+
{
|
41
|
+
[]
|
42
|
+
}
|
43
|
+
| id_list
|
44
|
+
|
45
|
+
where_clause :
|
46
|
+
{
|
47
|
+
[]
|
48
|
+
}
|
49
|
+
| WHERE id_search_condition
|
50
|
+
{
|
51
|
+
val[1]
|
52
|
+
}
|
53
|
+
| WHERE search_condition
|
54
|
+
{
|
55
|
+
val[1]
|
56
|
+
}
|
57
|
+
|
58
|
+
id_search_condition : id_predicate
|
59
|
+
| '(' id_predicate ')'
|
60
|
+
{
|
61
|
+
val[1]
|
62
|
+
}
|
63
|
+
|
64
|
+
id_predicate : ID '=' value
|
65
|
+
{
|
66
|
+
val[2]
|
67
|
+
}
|
68
|
+
| ID IN '(' value_list ')'
|
69
|
+
{
|
70
|
+
val[3]
|
71
|
+
}
|
72
|
+
|
73
|
+
search_condition : boolean_primary
|
74
|
+
{
|
75
|
+
[val[0]].flatten
|
76
|
+
}
|
77
|
+
| search_condition AND boolean_primary
|
78
|
+
{
|
79
|
+
(val[0] << val[2]).flatten
|
80
|
+
}
|
81
|
+
|
82
|
+
boolean_primary : predicate
|
83
|
+
| '(' search_condition ')'
|
84
|
+
{
|
85
|
+
val[1]
|
86
|
+
}
|
87
|
+
|
88
|
+
predicate : id op value
|
89
|
+
{
|
90
|
+
{:name => val[0], :op => val[1], :expr => val[2]}
|
91
|
+
}
|
92
|
+
| NOT id op value
|
93
|
+
{
|
94
|
+
{:name => val[1], :op => val[2], :expr => val[3], :not => true}
|
95
|
+
}
|
96
|
+
| id op '(' value_list ')'
|
97
|
+
{
|
98
|
+
{:name => val[0], :op => val[1], :expr => val[3]}
|
99
|
+
}
|
100
|
+
| NOT id op '(' value_list ')'
|
101
|
+
{
|
102
|
+
{:name => val[1], :op => val[2], :expr => val[4], :not => true}
|
103
|
+
}
|
104
|
+
| between_predicate
|
105
|
+
| not_in_predicate
|
106
|
+
|
107
|
+
between_predicate : id BETWEEN value AND value
|
108
|
+
{
|
109
|
+
{:name => val[0], :op => '$bt', :expr => [val[2], val[4]]}
|
110
|
+
}
|
111
|
+
|
112
|
+
not_in_predicate : id NOT IN '(' value_list ')'
|
113
|
+
{
|
114
|
+
{:name => val[0], :op => '$in', :expr => val[4], :not => true}
|
115
|
+
}
|
116
|
+
|
117
|
+
order_by_clause :
|
118
|
+
{
|
119
|
+
nil
|
120
|
+
}
|
121
|
+
| ORDER BY id ordering_spec
|
122
|
+
{
|
123
|
+
{:name => val[2], :type => val[3]}
|
124
|
+
}
|
125
|
+
|
126
|
+
ordering_spec :
|
127
|
+
{
|
128
|
+
:asc
|
129
|
+
}
|
130
|
+
| order_spec
|
131
|
+
|
132
|
+
limit_clause :
|
133
|
+
{
|
134
|
+
nil
|
135
|
+
}
|
136
|
+
| LIMIT NUMBER
|
137
|
+
{
|
138
|
+
val[1].to_i
|
139
|
+
}
|
140
|
+
| LIMIT STRING
|
141
|
+
{
|
142
|
+
val[1]
|
143
|
+
}
|
144
|
+
|
145
|
+
offset_clause :
|
146
|
+
{
|
147
|
+
nil
|
148
|
+
}
|
149
|
+
| OFFSET NUMBER
|
150
|
+
{
|
151
|
+
val[1].to_i
|
152
|
+
}
|
153
|
+
| OFFSET STRING
|
154
|
+
{
|
155
|
+
val[1]
|
156
|
+
}
|
157
|
+
|
158
|
+
update_statemant : UPDATE id SET set_clause_list where_clause
|
159
|
+
{
|
160
|
+
{:command => :update, :table => val[1], :set_clause_list => val[3], :condition => val[4]}
|
161
|
+
}
|
162
|
+
|
163
|
+
set_clause_list : set_clause
|
164
|
+
| set_clause_list ',' set_clause
|
165
|
+
{
|
166
|
+
val[0].merge val[2]
|
167
|
+
}
|
168
|
+
|
169
|
+
set_clause : id '=' value
|
170
|
+
{
|
171
|
+
{val[0] => val[2]}
|
172
|
+
}
|
173
|
+
|
174
|
+
delete_statemant : DELETE FROM id where_clause
|
175
|
+
{
|
176
|
+
{:command => :delete, :table => val[2], :condition => val[3]}
|
177
|
+
}
|
178
|
+
|
179
|
+
id : IDENTIFIER
|
180
|
+
|
181
|
+
id_list : id
|
182
|
+
{
|
183
|
+
[val[0]]
|
184
|
+
}
|
185
|
+
| id_list ',' id
|
186
|
+
{
|
187
|
+
val[0] << val[2]
|
188
|
+
}
|
189
|
+
|
190
|
+
value : STRING
|
191
|
+
| NUMBER
|
192
|
+
| NULL
|
193
|
+
|
194
|
+
value_list : value
|
195
|
+
{
|
196
|
+
[val[0]]
|
197
|
+
}
|
198
|
+
| value_list ',' value
|
199
|
+
{
|
200
|
+
val[0] << val[2]
|
201
|
+
}
|
202
|
+
|
203
|
+
op : IN { '$in' }
|
204
|
+
| REGEXP { '$regexp' }
|
205
|
+
| '<>' { :'!=' }
|
206
|
+
| '!=' { :'!=' }
|
207
|
+
| '>=' { :'>=' }
|
208
|
+
| '<=' { :'<=' }
|
209
|
+
| '>' { :'>' }
|
210
|
+
| '<' { :'<' }
|
211
|
+
| '=' { :'==' }
|
212
|
+
|
213
|
+
order_spec : ASC { :asc }
|
214
|
+
| DESC { :desc }
|
215
|
+
|
216
|
+
end
|
217
|
+
|
218
|
+
---- header
|
219
|
+
|
220
|
+
require 'strscan'
|
221
|
+
|
222
|
+
module ActiveCassandra
|
223
|
+
|
224
|
+
---- inner
|
225
|
+
|
226
|
+
KEYWORDS = %w(
|
227
|
+
AND
|
228
|
+
AS
|
229
|
+
ASC
|
230
|
+
BETWEEN
|
231
|
+
BY
|
232
|
+
COUNT
|
233
|
+
DELETE
|
234
|
+
DESC
|
235
|
+
DISTINCT
|
236
|
+
FROM
|
237
|
+
IN
|
238
|
+
INSERT
|
239
|
+
INTO
|
240
|
+
LIMIT
|
241
|
+
NOT
|
242
|
+
OFFSET
|
243
|
+
ORDER
|
244
|
+
REGEXP
|
245
|
+
SELECT
|
246
|
+
SET
|
247
|
+
UPDATE
|
248
|
+
VALUES
|
249
|
+
WHERE
|
250
|
+
)
|
251
|
+
|
252
|
+
KEYWORD_REGEXP = Regexp.compile("(?:#{KEYWORDS.join '|'})\\b", Regexp::IGNORECASE)
|
253
|
+
|
254
|
+
def initialize(obj)
|
255
|
+
src = obj.is_a?(IO) ? obj.read : obj.to_s
|
256
|
+
@ss = StringScanner.new(src)
|
257
|
+
end
|
258
|
+
|
259
|
+
def scan
|
260
|
+
piece = nil
|
261
|
+
|
262
|
+
until @ss.eos?
|
263
|
+
if (tok = @ss.scan /\s+/)
|
264
|
+
# nothing to do
|
265
|
+
elsif (tok = @ss.scan /(?:<>|!=|>=|<=|>|<|=)/)
|
266
|
+
yield tok, tok
|
267
|
+
elsif (tok = @ss.scan KEYWORD_REGEXP)
|
268
|
+
yield tok.upcase.to_sym, tok
|
269
|
+
elsif (tok = @ss.scan /NULL\b/i)
|
270
|
+
yield :NULL, nil
|
271
|
+
elsif (tok = @ss.scan /'(?:[^']|'')*'/) #'
|
272
|
+
yield :STRING, tok.slice(1...-1).gsub(/''/, "'")
|
273
|
+
elsif (tok = @ss.scan /-?(?:0|[1-9]\d*)(?:\.\d+)/)
|
274
|
+
yield :NUMBER, tok.to_f
|
275
|
+
elsif (tok = @ss.scan /-?(?:0|[1-9]\d*)/)
|
276
|
+
yield :NUMBER, tok.to_i
|
277
|
+
elsif (tok = @ss.scan /[,\(\)\*]/)
|
278
|
+
yield tok, tok
|
279
|
+
elsif (tok = @ss.scan /(?:[a-z_]\w+\.|[a-z]\.)*ID\b/i)
|
280
|
+
yield :ID, tok
|
281
|
+
elsif (tok = @ss.scan /(?:[a-z_]\w+\.|[a-z]\.)*(?:[a-z_]\w+|[a-z])/i)
|
282
|
+
yield :IDENTIFIER, tok
|
283
|
+
else
|
284
|
+
raise Racc::ParseError, ('parse error on value "%s"' % @ss.rest.inspect)
|
285
|
+
end
|
286
|
+
end
|
287
|
+
|
288
|
+
yield false, '$'
|
289
|
+
end
|
290
|
+
private :scan
|
291
|
+
|
292
|
+
def parse
|
293
|
+
yyparse self, :scan
|
294
|
+
end
|
295
|
+
|
296
|
+
---- footer
|
297
|
+
|
298
|
+
end # module ActiveCassandra
|
@@ -0,0 +1,269 @@
|
|
1
|
+
require 'active_record/base'
|
2
|
+
require 'active_record/connection_adapters/abstract_adapter'
|
3
|
+
require 'cassandra'
|
4
|
+
require 'active_cassandra/cf'
|
5
|
+
require 'active_cassandra/sqlparser.tab'
|
6
|
+
|
7
|
+
module ActiveRecord
|
8
|
+
class Base
|
9
|
+
def self.cassandra_connection(config)
|
10
|
+
config.symbolize_keys!
|
11
|
+
host = config[:host] || '127.0.0.1'
|
12
|
+
port = config[:port] || 9160
|
13
|
+
|
14
|
+
unless (keyspace = config[:keyspace] || config[:database])
|
15
|
+
raise ArgumentError, "No database file specified. Missing argument: keyspace"
|
16
|
+
end
|
17
|
+
|
18
|
+
thrift_client_options = config.dup
|
19
|
+
[:adapter, :host, :port, :keyspace, :database].each {|i| thrift_client_options.delete(i) }
|
20
|
+
|
21
|
+
client = Cassandra.new(keyspace, "#{host}:#{port}", thrift_client_options)
|
22
|
+
ConnectionAdapters::CassandraAdapter.new(client, logger, config)
|
23
|
+
end
|
24
|
+
end # class Base
|
25
|
+
|
26
|
+
module ConnectionAdapters
|
27
|
+
class CassandraAdapter < AbstractAdapter
|
28
|
+
def initialize(client, logger, config)
|
29
|
+
super(client, logger)
|
30
|
+
@config = config
|
31
|
+
end
|
32
|
+
|
33
|
+
def supports_count_distinct?
|
34
|
+
false
|
35
|
+
end
|
36
|
+
|
37
|
+
def select(sql, name = nil)
|
38
|
+
log(sql, name)
|
39
|
+
|
40
|
+
parsed_sql = ActiveCassandra::SQLParser.new(sql).parse
|
41
|
+
|
42
|
+
cf = parsed_sql[:table].to_sym
|
43
|
+
cond = parsed_sql[:condition]
|
44
|
+
count = parsed_sql[:count]
|
45
|
+
# not implemented:
|
46
|
+
# distinct = parsed_sql[:distinct]
|
47
|
+
sqlopts, casopts = rowopts(parsed_sql)
|
48
|
+
|
49
|
+
if count and cond.empty? and sqlopts.empty?
|
50
|
+
[{count => @connection.count_range(cf, casopts)}]
|
51
|
+
elsif is_id?(cond)
|
52
|
+
ks = [cond].flatten
|
53
|
+
@connection.multi_get(cf, ks, casopts).values
|
54
|
+
else
|
55
|
+
rows = @connection.get_range(cf, casopts).select {|i| i.columns.length > 0 }.map do |key_slice|
|
56
|
+
key_slice_to_hash(key_slice)
|
57
|
+
end
|
58
|
+
|
59
|
+
unless cond.empty?
|
60
|
+
rows = filter(cond).call(rows)
|
61
|
+
end
|
62
|
+
|
63
|
+
if (offset = sqlopts[:offset])
|
64
|
+
rows = rows.slice(offset..-1)
|
65
|
+
end
|
66
|
+
|
67
|
+
if (limit = sqlopts[:limit])
|
68
|
+
rows = rows.slice(0, limit)
|
69
|
+
end
|
70
|
+
|
71
|
+
count ? [{count => rows.length}] : rows
|
72
|
+
end
|
73
|
+
end
|
74
|
+
|
75
|
+
def insert_sql(sql, name = nil, pk = nil, id_value = nil, sequence_name = nil)
|
76
|
+
log(sql, name)
|
77
|
+
|
78
|
+
parsed_sql = ActiveCassandra::SQLParser.new(sql).parse
|
79
|
+
table = parsed_sql[:table]
|
80
|
+
cf = table.to_sym
|
81
|
+
column_list = parsed_sql[:column_list]
|
82
|
+
value_list = parsed_sql[:value_list]
|
83
|
+
|
84
|
+
class_name = ActiveRecord::Base.class_name(table)
|
85
|
+
rowid = Module.const_get(class_name).__identify.to_s
|
86
|
+
|
87
|
+
nvs = {}
|
88
|
+
column_list.zip(value_list).each {|n, v| nvs[n] = v.to_s }
|
89
|
+
|
90
|
+
@connection.insert(cf, rowid, nvs)
|
91
|
+
|
92
|
+
return rowid
|
93
|
+
end
|
94
|
+
|
95
|
+
def update_sql(sql, name = nil)
|
96
|
+
log(sql, name)
|
97
|
+
parsed_sql = ActiveCassandra::SQLParser.new(sql).parse
|
98
|
+
cf = parsed_sql[:table].to_sym
|
99
|
+
cond = parsed_sql[:condition]
|
100
|
+
|
101
|
+
nvs = {}
|
102
|
+
parsed_sql[:set_clause_list].each do |n, v|
|
103
|
+
n = n.split('.').last
|
104
|
+
nvs[n] = v.to_s
|
105
|
+
end
|
106
|
+
|
107
|
+
n = 0
|
108
|
+
|
109
|
+
if is_id?(cond)
|
110
|
+
ks = [cond].flatten
|
111
|
+
rs = @connection.multi_get(cf, ks)
|
112
|
+
|
113
|
+
ks.each do |key|
|
114
|
+
row = rs[key]
|
115
|
+
@connection.insert(cf, key, row.merge(nvs))
|
116
|
+
n += 1
|
117
|
+
end
|
118
|
+
else
|
119
|
+
rows = @connection.get_range(cf).select {|i| i.columns.length > 0 }.map do |key_slice|
|
120
|
+
key_slice_to_hash(key_slice)
|
121
|
+
end
|
122
|
+
|
123
|
+
unless cond.empty?
|
124
|
+
rows = filter(cond).call(rows)
|
125
|
+
end
|
126
|
+
|
127
|
+
rows.each do |row|
|
128
|
+
@connection.insert(cf, row['id'], row.merge(nvs))
|
129
|
+
n += 1
|
130
|
+
end
|
131
|
+
end
|
132
|
+
|
133
|
+
return n
|
134
|
+
end
|
135
|
+
|
136
|
+
def delete_sql(sql, name = nil)
|
137
|
+
log(sql, name)
|
138
|
+
|
139
|
+
parsed_sql = ActiveCassandra::SQLParser.new(sql).parse
|
140
|
+
cf = parsed_sql[:table].to_sym
|
141
|
+
cond = parsed_sql[:condition]
|
142
|
+
|
143
|
+
n = 0
|
144
|
+
|
145
|
+
if is_id?(cond)
|
146
|
+
[cond].flatten.each do |key|
|
147
|
+
@connection.remove(cf, key)
|
148
|
+
n += 1
|
149
|
+
end
|
150
|
+
else
|
151
|
+
rows = @connection.get_range(cf).select {|i| i.columns.length > 0 }
|
152
|
+
|
153
|
+
unless cond.empty?
|
154
|
+
rows = rows.map {|i| key_slice_to_hash(i) }
|
155
|
+
rows = filter(cond).call(rows)
|
156
|
+
|
157
|
+
rows.each do |row|
|
158
|
+
@connection.remove(cf, row['id'])
|
159
|
+
n += 1
|
160
|
+
end
|
161
|
+
else
|
162
|
+
rows.each do |key_slice|
|
163
|
+
@connection.remove(cf, key_slice.key)
|
164
|
+
n += 1
|
165
|
+
end
|
166
|
+
end
|
167
|
+
end
|
168
|
+
|
169
|
+
return n
|
170
|
+
end
|
171
|
+
|
172
|
+
def add_limit_offset!(sql, options)
|
173
|
+
if (limit = options[:limit])
|
174
|
+
if limit.kind_of?(Numeric)
|
175
|
+
sql << " LIMIT #{limit.to_i}"
|
176
|
+
else
|
177
|
+
sql << " LIMIT #{quote(limit)}"
|
178
|
+
end
|
179
|
+
end
|
180
|
+
|
181
|
+
if (offset = options[:offset])
|
182
|
+
if offset.kind_of?(Numeric)
|
183
|
+
sql << " OFFSET #{offset.to_i}"
|
184
|
+
else
|
185
|
+
sql << " OFFSET #{quote(offset)}"
|
186
|
+
end
|
187
|
+
end
|
188
|
+
end
|
189
|
+
|
190
|
+
private
|
191
|
+
def key_slice_to_hash(key_slice)
|
192
|
+
hash = {'id' => key_slice.key}
|
193
|
+
|
194
|
+
key_slice.columns.each do |i|
|
195
|
+
column = i.column
|
196
|
+
hash[column.name] = column.value
|
197
|
+
end
|
198
|
+
|
199
|
+
return hash
|
200
|
+
end
|
201
|
+
|
202
|
+
def is_id?(cond)
|
203
|
+
not cond.kind_of?(Array) or not cond.all? {|i| i.kind_of?(Hash) }
|
204
|
+
end
|
205
|
+
|
206
|
+
def filter(cond)
|
207
|
+
fs = []
|
208
|
+
|
209
|
+
cond.each do |c|
|
210
|
+
name, op, expr, has_not = c.values_at(:name, :op, :expr, :not)
|
211
|
+
name = name.split('.').last
|
212
|
+
expr = Regexp.compile(expr) if op == '$regexp'
|
213
|
+
|
214
|
+
func = case op
|
215
|
+
when '$in'
|
216
|
+
lambda {|i| expr.include?(i) }
|
217
|
+
when '$bt'
|
218
|
+
lambda {|i| expr[0] <= i and i <= expr[1] }
|
219
|
+
when '$regexp'
|
220
|
+
lambda {|i| i =~ Regexp.compile(expr) }
|
221
|
+
when :'>=', :'<=', :'>', :'<'
|
222
|
+
lambda {|i| i.to_i.send(op, expr.to_i) }
|
223
|
+
else
|
224
|
+
lambda {|i| i.send(op, expr) }
|
225
|
+
end
|
226
|
+
|
227
|
+
fs << (has_not ? lambda {|row| not func.call(row[name]) } : lambda {|row| func.call(row[name])})
|
228
|
+
end
|
229
|
+
|
230
|
+
lambda do |rows|
|
231
|
+
fs.inject(rows) {|r, f| r.select {|i| f.call(i) } }
|
232
|
+
end
|
233
|
+
end
|
234
|
+
|
235
|
+
def rowopts(parsed_sql)
|
236
|
+
order, limit, offset = parsed_sql.values_at(:order, :limit, :offset)
|
237
|
+
sqlopts = {}
|
238
|
+
casopts = {}
|
239
|
+
|
240
|
+
# not implemented:
|
241
|
+
# if order
|
242
|
+
# name, type = order.values_at(:name, :type)
|
243
|
+
# ...
|
244
|
+
# end
|
245
|
+
|
246
|
+
if offset
|
247
|
+
if offset.kind_of?(Numeric)
|
248
|
+
sqlopts[:offset] = offset
|
249
|
+
else
|
250
|
+
# XXX: offset is not equals to SQL OFFSET
|
251
|
+
casopts[:start] = offset
|
252
|
+
end
|
253
|
+
end
|
254
|
+
|
255
|
+
if limit
|
256
|
+
if limit.kind_of?(Numeric)
|
257
|
+
sqlopts[:limit] = limit
|
258
|
+
else
|
259
|
+
# XXX: limit is not equals to SQL LIMIT
|
260
|
+
casopts[:finish] = limit
|
261
|
+
end
|
262
|
+
end
|
263
|
+
|
264
|
+
return [sqlopts, casopts]
|
265
|
+
end
|
266
|
+
|
267
|
+
end # class CassandraAdapter
|
268
|
+
end # module ConnectionAdapters
|
269
|
+
end # module ActiveRecord
|