You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Mirza Aliev (Jira)" <ji...@apache.org> on 2023/04/05 08:18:00 UTC

[jira] [Assigned] (IGNITE-18954) Design a language for parsing of a filter expression

     [ https://issues.apache.org/jira/browse/IGNITE-18954?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mirza Aliev reassigned IGNITE-18954:
------------------------------------

    Assignee: Mirza Aliev

> Design a language for parsing of a filter expression
> ----------------------------------------------------
>
>                 Key: IGNITE-18954
>                 URL: https://issues.apache.org/jira/browse/IGNITE-18954
>             Project: Ignite
>          Issue Type: Improvement
>            Reporter: Mirza Aliev
>            Assignee: Mirza Aliev
>            Priority: Major
>              Labels: ignite-3
>
> {*}Motivation{*}:
> We need to parse and be able to use filters, that user set for the distribution zone
> *Definition of done:*
>  
> * Filters are parsed and it is possible to filter arbitrary set of attributes
> * Validator is implemented, so we can validate a string representation of a filter expression
> *Implementation details:*
>  Language must allow to perform simple operations, like 
> * “ “ – default empty filter, means that all nodes match
> * () expressions with parentheses 
> * key1 = “A” && key2 = “B”, () && ()
> * key1 = “A” || key2 = “B”, () || ()
> * key1 != "A"
> Attributes representation must look like key -> value pairs, like 
> {noformat}
> (region = EU || region = US) && (storage != HDD || storage = SSD)
> {noformat}
> After some investigation, I propose to not implement our own language for the filters and attributes, but reuse JSON functionality.
> For nodes, user just pass JSON representation of attributes through the {{ignite-config.json}}:
> {code:java}
> {
>     "network": {
>         "port": 3344
>     },
>     "rest": {
>         "port": 10300
>     },
>     "nodeAttributes": {
>         "nodeName": "node1"
>         "region": "EU",
>         "storage": "SSD",
>         "dataRegion" : 10
>     }
> }
> {code}
> For parsing, we can use https://github.com/json-path/JsonPath, which is already used in ignite {{sql-engine}} module, so we don't need to bring a new dependancy.
> {{JsonPath}} has a wide functionality to work with JSON, we are interested in filtering queries. See all list of available functionality here https://github.com/json-path/JsonPath#functions. Example of usages you can find here https://rows.com/docs/filtering-with-jsonpath
> In general, it will allow us to make all operations, that we have listed before.
> Lets provide some examples with our node attributes:
> {{['nodeName'][?(@.['region'] == 'EU')]}} -- will give all nodeName, that have region equals to 'EU'
> {{['nodeName'].[?(@.['dataRegion'] > 10 && @.[storage] != 'HDD')]}} -- will give all nodeNames, that has dataRegion > 10 and storage not equal to HDD
>  



--
This message was sent by Atlassian Jira
(v8.20.10#820010)