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¶

Contexts¶

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).

Predicates¶

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

Pipelining¶

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¶

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

Syntax¶

Query is the Python expression

Variable¶

Variable is coded as plain python variables

Predicate¶

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¶

True()¶

Always True, pass context unmodified

False()¶

Always False, breaks predicate chain

Context Manipulation¶

Set(**kwargs)¶

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

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

Del(*args)¶

Delete variables from context. Deduplicate contexts when necessary

Del(X)

ConfDB Matching¶

Match(*args)¶

Match *args against ConfDB. Bind unbound variables on match

Params:

Name	Description
`*args`	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

Params:

Name	Description
`*args`	ConfDB path

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

ConfDB Manipulation¶

Fact(*args)¶

Set Fact to database

Params:

Name	Description
`*args`	ConfDB path of fact, eigther constants or bound variables

Fact('interface', X, 'hints', 'test')

Filtering and Checking¶

Filter(expr):¶

Pass context only if expr is evaluated as True

Name	Description
`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

Name	Description
`pattern`	Regular expression pattern
`name`	Variable name
`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

Params:

Name	Description
`vlan_filter`	VC Filter expression
`vlan_id`	Vlan Id or bound variable

HasVlan("1-99,200-299", X)

Aggregation¶

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

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=",")

Debugging¶

Dump(message=None):¶

Dump current context to stdout and pass unmodified

Name	Description
`message`	Optional message

Dump()
Dump("Point1")

Sprintf(name, fmt, *args):¶

Perform string formatting and apply result to context variable.

Params:

Name	Description
`name`	Target variable name
`fmt`	String format
`args`	Values or bound variables

Sprintf(y, 'x = %s, y = %s', x, '2')

Examples¶

Fetch all interfaces descriptions

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