allocationfilterutil

Documentation

Overview

allocationfilterutil provides functionality for parsing V2 of the Kubecost filter language for Allocation types.

e.g. "filter=namespace:kubecost+controllerkind:deployment"

Index

• func ParseAllocationFilter(filter string) (kubecost.AllocationFilter, error)

Functions

func ParseAllocationFilter

func ParseAllocationFilter(filter string) (kubecost.AllocationFilter, error)

ParseAllocationFilter converts a string of the V2 Allocation Filter language into a kubecost.AllocationFilter.

Example queries:

namespace:"kubecost"
label[app]:"cost-analyzer"
node!:"node1","node2"
cluster:"cluster-one"+namespace!:"kube-system"

The grammar is approximately as follows:

Original design doc [1] contains first grammar. This is a slight modification of that grammar to help guide the implementation of the parser.

[1] https://docs.google.com/document/d/1HKkp2bv3mnvfQoBZlpHjfZwQ0FzDLOHKpnwV9gQ_KgU/edit?pli=1

<filter> ::= <comparison> ('+' <comparison>)*

NOTE: Language can be extended to support ORs between
comparisons by adding a '|' operator in between comparisons,
though precedence will have to be carefully defined and it may
require adding support for ()-enclosed statements to deal with
precedence.
This would allow for queries like:
  namespace:"x"|label[app]="foo"

<comparison> ::= <filter-key> <filter-op> <filter-value>

<filter-key> ::= <filter-field-2> <keyed-access>

| <filter-field-1>

<filter-op> ::= ':' | '!:'

<filter-value> ::= '"' <identifier> '"' (',' <filter-value>)*

<filter-field-2> ::= 'label' | 'annotation'

<filter-field-1> ::= 'cluster' | 'node' | 'namespace'

| 'controllerName' | 'controllerKind'
| 'container' | 'pod' | 'services'

<keyed-access> ::= '[' <identifier> ']'

<identifier> ::= --- valid K8s name or Prom-sanitized K8s name