JEP-16 String manipulation (GorillaStack)

[springcomp]

String Manipulation (GorillaStack)

JEP	16
Author	Chris Armstrong (Gorilla Stack)
Status	draft
Created	31-10-2019

Abstract

This document proposes expanding the list of available string manipulation functions in the spec.

Motivation

The specification only contains basic string search and manipulation functions, such as starts_with() and contains(), which are insufficient for other use cases, such as replacing parts of strings, splitting strings, using regular expressions for searching or manipulation of strings, or other common string manipulations found in the common libraries of mainstream languages such as Python or JavaScript.

Specification

The following functions will be added to the specification:

split

array[string] split(string $subject, string $split_str, number $max = -1)

Given string $subject, split will split on occurrences string $split_str, up to and including $max times (if specified).

The result is an array containing each partial string between occurrences of $split_str. If the string contains no occurrences of $split_str, an array containing just the original $subject string will be returned.

$max is a optional parameter. When specified as an integer greater than zero, it will split the string up to ($max - 1) times, with the last string in the array containing the remaining contents of $subject unmodified.

If not specified or it is less than zero, the string will be split on every occurrence of $split_str.

When $max is equal to zero, it will behave as if the string is not split at all, and return an array containing one element, the original string.

Examples

Expression	Result
split('average\|-\|min\|-\|max\|-\|mean\|-\|mode\|-\|median', '\|-\|')	["average", "min", "max", "mean", "mode", "median"]
split('average\|-\|min\|-\|max\|-\|mean\|-\|mode\|-\|median', '\|-\|', 3)	["average", "min", "max\|-\|mean\|-\|mode\|-\|median"]
split('average\|-\|min\|-\|max\|-\|mean\|-\|mode\|-\|median', '-')	["average\|", "\|min\|", "\|max\|", "\|mean\|", "\|mode\|", "\|median"]
split('average\|-\|min\|-\|max\|-\|mean\|-\|mode\|-\|median', '\|-\|', 0)	["average\|-\|min\|-\|max\|-\|mean\|-\|mode\|-\|median"]
split('every character', '')	["e", "v", "e", "r", "y", " ", "c", "h", "a", "r", "a", "c", "t", "e", "r"]

Impact

There should be no impact to existing users, as these are new functions.

[springcomp]

Consider merging similar proposal from the Python repository.

[springcomp]

Consider studying this discussion which might be relevant.

[springcomp]

If we were to add string manipulation functions, would that be good to at least specify a first reasonable batch, instead of one function at a time?

I’m think about:

• split obviously
• substring which might be better served by overloading the slice function.
• starts_with and ends_with functions, that could useful in filter-expressions.
• to_lower and to_upper functions although there seems to be some conventions over using the to_ prefix.
• replace function.
• C#-style IndexOf or python-style find() functions to search for a substring in a larger string.

I specifically do not think we would need a trim function as I expect input JSON documents to be properly structured.

However, this contradicts one of the tenets of JMESPath which is to drive changes from real-world scenarios.

[springcomp]

Closed in favor of corresponding discussion