Перейти к содержанию

ConfDB Query Language

ConfDB offers flexible query language based on predicate logic. ConfDB queries may be used for config processing, including fetching, classification and validation.

Common Concepts


Query context is the couple of variables and their values. It is a simple key-value structure implemented over python's dict. Context represents a possible state of pipeline. Contexts are grouped together representing possibilities on each stage of pipeline. Within the possibilities, each context is independent on each other.

Variables can be either bound (known values) or unbound (superposition of possible values).


Predicate is the logic function, defined by its arguments and accepting possibilities on input, evaluating them and passing outcomes as output. Predicates evaluate input context independently. Each input context may be evaluated to one or more output context.

graph LR
    Possibilities --> Predicate
    Predicate --> Outcomes


Predicates may be combined together into sequential (and) and parallel (or) chains. Chains can be grouped together by bracket operator. Resulting chains build pipeline which is valid predicate too.


Query is the pipeline, which can be applied to exact ConfDB state and input context to produce output contexts with possible outcomes


Query is the Python expression


Variable is coded as plain python variables


Predicate is coded as python function call

Match('interface', X, 'description', Y)

See Query Builtin-predicates for possible predicates and examples

Sequential chain

Sequential chain is the combination of two predicates when output of first predicate serves as input to the second one

graph LR
    Input --> P1
    P1 --> P2
    P2 --> Output

Sequential chain is coded by boolean and operator

P1() and P2()

Sequential chain can contain more than two operators like

P1() and P2() and P3() and P4()

Parallel chain

Parallel chain consists of two or more predicates independentently accepting same input and combining and deduplicating resulting outputs

graph LR
    Input --> P1
    Input --> P2
    P1 --> Output
    P2 --> Output

Parallel chains are coded by or operator

P1() or P2()

Chain Grouping

Chains can be grouped using brackets

(P1() and P2()) or P3() or (P4() and P5())

Built-in predicates

Simple Logic


Always True, pass context unmodified


Always False, breaks predicate chain

Context Manipulation


Add or modify variables of context. If variable value is a list, expand the list and apply production

Set(X=[1, 2, 3])


Delete variables from context. Deduplicate contexts when necessary


ConfDB Matching


Match *args against ConfDB. Bind unbound variables on match

ConfDB path
Match('interface', X, 'description', Y)

NotMatch(self, _input, *args)

Pass only if *args is not matched against ConfDB. Bind unbound variables when possible

ConfDB path
NotMatch('interface', X, 'description')

ConfDB Manipulation


Set Fact to database

:param *args: ConfDB path of fact, eigther constants or bound variables
Fact('interface', X, 'hints', 'test')

Filtering and Checking


Pass context only if `expr` is evaluated as True

:param expr: Python expression

Filter(X % 2 == 0)

Re(pattern, name, ignore_case=None):

Match variable name against regular expression pattern. Pass context further if matched. If regular expression contains named groups, i.e. (?P....), apply them as context variables

:param pattern: Regular expression pattern
:param name: Variable name
:param ignore_case: Ignore case during match
Re("a+", X)
Re("a+", X, ignore_case=True)
Re("-(?P<abs>\d+)", X)

HasVLAN(vlan_filter, vlan_id):

Check `vlan_id` is within `vlan_filter` expression

:param vlan_filter: VC Filter expression
:param vlan_id: Vlan Id or bound variable
HasVlan("1-99,200-299", X)


Group(stack=None, args, kwargs):

Group input context on given variables
    Match("interfaces", name)
    or Match("interfaces", name, "type", type)
    or Match("interfaces", name, "description", description)
    or Match("interfaces", name, "admin-status", admin_status)
) and Group("name")

Collapse(args, *kwargs):

Collapse multiple keys to a single one following rules

:param kwargs: One of collapse operation should be specified
    * join=<sep> -- join lines with separator sep
    * joinrange=<sep> -- join lines with separator sep and apply range optimization
Collapse("interfaces", X, "tagged-vlans", join=",")
Collapse("interfaces", X, "tagged-vlans", joinrange=",")



Dump current context to stdout and pass unmodified

:param message: Optional message

Sprintf(name, fmt, *args):

Perform string formatting and apply result to context variable

:param name: Target variable name
:param fmt: String format
:param args: Values or bound variables
Sprintf(y, 'x = %s, y = %s', x, '2')


Fetch all interfaces descriptions

Match("interfaces", name, "description", descr)