Ruby Proc filter plugin for Embulk
This plugin is inspired by mgi166/embulk-filter-eval: Eval ruby code on filtering
This plugin apply ruby proc to each record.
Overview
- Plugin type: filter
Configuration
- columns: filter definition (hash, required)
-
requires: pre required libraries (array, default:
[])
Example
input
id,account,time,purchase,comment,data
1,32864,2015-01-27 19:23:49,20150127,embulk,"{\"foo\": \"bar\", \"events\": [{\"id\": 1, \"name\": \"Name1\"}, {\"id\": 2, \"name\": \"Name2\"}]}"
2,14824,2015-01-27 19:01:23,20150127,embulk jruby,NULL
3,27559,2015-01-28 02:20:02,20150128,"Embulk ""csv"" parser plugin",NULL
4,11270,2015-01-29 11:54:36,20150129,NULL,NULL
config
# ...
filters:
- type: ruby_proc
requires:
- cgi
variables:
multiply: 3
before:
- proc: |
-> do
puts "before proc"
@started_at = Time.now
end
after:
- proc: |
-> do
puts "after proc"
p Time.now - @started_at
end
rows:
- proc: |
->(record) do
[record.dup, record.dup.tap { |r| r["id"] += 10 }]
end
skip_rows:
- proc: |
->(record) do
record["id"].odd?
end
columns:
- name: data
proc: |
->(data) do
data["events"] = data["events"].map.with_index do |e, idx|
e.tap { |e_| e_["idx"] = idx }
end
data
end
- name: id
proc: |
->(id) do
id * variables["multiply"]
end
type: string
- name: comment
proc_file: comment_upcase.rb
skip_nil: false
type: json
pages:
- proc: |
->(records) do
records.map do |record|
record.tap { |r| r["id"] += 1 }
end
end
# ...
If you want to skip record in "rows proc" or "columns proc", use throw :skip_record.
# comment_upcase.rb
->(comment, record) do
return [record["account"].to_s].to_json unless comment
comment.upcase.split(" ").map { |s| CGI.escape(s) }
end-
beforeandafteris executed at once - procs is evaluated on same binding (instance of Evaluator class)
- instance variable is shared
- rows proc must return record hash or array of record hash.
- user must take care of object identity. Otherwise, error may be occurred when plugin applys column procs.
- pages proc must return array of record hash.
- use
page_sizeoption to increase size of processing record (ex.-X page_size=64KB)
- use
proc execution order
- before procs
- per record
- all row procs
- per record applied row procs
- all skip_row procs
- column procs
- per page procs
- after procs
preview
+-----------+--------------+-------------------------+-------------------------+------------------------------------------+------------------------------------------------------------------------------------------+
| id:string | account:long | time:timestamp | purchase:timestamp | comment:json | data:json |
+-----------+--------------+-------------------------+-------------------------+------------------------------------------+------------------------------------------------------------------------------------------+
| 3 | 32,864 | 2015-01-27 19:23:49 UTC | 2015-01-27 00:00:00 UTC | ["EMBULK"] | {"events":[{"id":1,"name":"Name1","idx":0},{"id":2,"name":"Name2","idx":1}],"foo":"bar"} |
| 34 | 32,864 | 2015-01-27 19:23:49 UTC | 2015-01-27 00:00:00 UTC | ["EMBULK"] | {"events":[{"id":1,"name":"Name1","idx":0},{"id":2,"name":"Name2","idx":1}],"foo":"bar"} |
| 6 | 14,824 | 2015-01-27 19:01:23 UTC | 2015-01-27 00:00:00 UTC | ["EMBULK","JRUBY"] | |
| 37 | 14,824 | 2015-01-27 19:01:23 UTC | 2015-01-27 00:00:00 UTC | ["EMBULK","JRUBY"] | |
| 9 | 27,559 | 2015-01-28 02:20:02 UTC | 2015-01-28 00:00:00 UTC | ["EMBULK","%22CSV%22","PARSER","PLUGIN"] | |
| 40 | 27,559 | 2015-01-28 02:20:02 UTC | 2015-01-28 00:00:00 UTC | ["EMBULK","%22CSV%22","PARSER","PLUGIN"] | |
| 12 | 11,270 | 2015-01-29 11:54:36 UTC | 2015-01-29 00:00:00 UTC | ["11270"] | |
| 43 | 11,270 | 2015-01-29 11:54:36 UTC | 2015-01-29 00:00:00 UTC | ["11270"] | |
+-----------+--------------+-------------------------+-------------------------+------------------------------------------+------------------------------------------------------------------------------------------+
Build
$ rake