# Filter plugin for modifying each event record for [Fluentd](http://fluentd.org)
Adding arbitary field to event record without customizing existence plugin.
For example, generated event from *in_tail* doesn't contain "hostname" of running machine.
In this case, you can use *record_modifier* to add "hostname" field to event record.
## Requirements
| fluent-plugin-record-modifier | fluentd | ruby |
|--------------------------------|---------|------|
| >= 1.0.0 | >= v0.14.0 | >= 2.1 |
| < 1.0.0 | >= v0.12.0 | >= 1.9 |
NOTE: fluent-plugin-record-modifier v1.0.0 is now RC. We will release stable v1.0.0 soon.
## Installation
Use RubyGems:
gem install fluent-plugin-record-modifier
## Configuration
Use `record_modifier` filter.
@type record_modifier
gen_host "#{Socket.gethostname}"
foo bar
If following record is passed:
```js
{"message":"hello world!"}
```
then you got new record like below:
```js
{"message":"hello world!", "gen_host":"oreore-mac.local", "foo":"bar"}
```
You can also use `record_transformer` like `${xxx}` placeholders and access `tag`, `time`, `record` and `tag_parts` values by Ruby code.
@type record_modifier
tag ${tag}
tag_extract ${tag_parts[0]}-${tag_parts[1]}-foo
formatted_time ${Time.at(time).to_s}
new_field foo:${record['key1'] + record['dict']['key']}
`record_modifier` is faster than `record_transformer`. See [this comment](https://github.com/repeatedly/fluent-plugin-record-modifier/pull/7#issuecomment-169843012).
But unlike `record_transformer`, `record_modifier` doesn't support following features for now.
- tag_suffix and tag_prefix
- dynamic key placeholder
### char_encoding
Fluentd including some plugins treats logs as a BINARY by default to forward.
But a user sometimes wants to process logs depends on their requirements, e.g. handling char encoding correctly.
`char_encoding` parameter is useful for this case.
```conf
@type record_modifier
# set UTF-8 encoding information to string.
char_encoding utf-8
# change char encoding from 'UTF-8' to 'EUC-JP'
char_encoding utf-8:euc-jp
```
In `char_encoding from:to` case, it replaces invalid character with safe character.
### remove_keys
The logs include needless record keys in some cases.
You can remove it by using `remove_keys` parameter.
```conf
@type record_modifier
# remove key1 and key2 keys from record
remove_keys key1,key2
```
If following record is passed:
```js
{"key1":"hoge", "key2":"foo", "key3":"bar"}
```
then you got new record like below:
```js
{"key3":"bar"}
```
### whitelist_keys
If you want to handle the set of explicitly specified keys, you can use `whitelist_keys` of this plugin. It's exclusive with `remove_keys`.
```conf
@type record_modifier
# remove all keys except for key1 and key2
whitelist_keys key1,key2
```
If following record is passed:
```js
{"key1":"hoge", "key2":"foo", "key3":"bar"}
```
then you got new record like below:
```js
{"key1":"hoge", "key2":"foo"}
```
### Ruby code trick for complex logic
If you need own complex logic in filter, writing filter plugin is better. But if you don't want to write new plugin, you can use temporal key trick like below:
```
@type record_modifier
remove_keys _dummy_
_dummy_ ${if record.has_key?('foo'); record['bar'] = 'Hi!'; end; nil}
}
```
### record_modifier output
In v0.10, you can use `record_modifier` output to emulate filter. `record_modifier` output doesn't support `` way.
type record_modifier
tag foo.filtered
gen_host "#{Socket.gethostname}"
foo bar
## TODO
* Adding following features if needed
* Replace record value
## Copyright