6.3.6. Ellipse Path

Inspired by jq simple selectors, but over elps objects! This library allows concise and easy manipulation of elps data. parse_tests.go has a comprehensive set of tests to demonstrate typical queries. You can readily compare these results against jq using the jq playground: https://jqplay.org.

We only support simple selectors that output a single value. All paths are strings that must begin with . A path consists of a series of selectors, and performs operations (both mutating and non-mutating versions): 1) Get, 2) Set, 3) Delete, 4) Nil. These operations take ELPS LVals as input, and return an elps LVal.

In the below examples we use JSON to represent elps data structures.

Index & Dot Selectors

Index and dot examples: .regularKey, ["$special key"], [0]

Object: {"hello":"world"}

Path: .hello

Output: "world"

Range Selectors

Range examples: [1:2], [:2], [1:], [:]

Object: ["a","b","c"]

Get Path: .[1:]

Output: ["b","c"]

Iterator Selectors

Iterator selectors execute a path selector on each element of an array.

Object: {"a":[ { "b": 2 }, 0, {"b": 3 } ] }

Get Path: .a[].b

Output: [2,null,3]

Note: jq would throw an error since .b on 0 is an error, however elpspath iterators returns null when the selector throws an error.

Like in jq, nested iterators are collapsed in the return results:

Object: [ {"a":[ { "b": 2 }, {"b": 3 } ] }, {"a":[ { "b": 4 } ]} ]

Get Path: .[].a[].b Output: [2,3,4]

Differences from jq

• Range selectors such as [:2] is treated as a query of two paths: [0], [1], where the query results are wrapped in array [...]
• elpspath supports [:] notation which is equivalent to [0:].
• Iterator notation [] will skip over some errors that jq will throw an error.
• jq iterators have a notation of "multiple outputs" where each output of an iterator is printed on a new line. elpspath iterators return multiple outputs wrapped in an array, as jq would if there was an outer braces [a[].b] array constructor surrounding the iterator query.