Trying to match array index pattern in string using Regex

Question

I'm using regex to get the indices of variables accessing pattern represented as string, for example:

string = "test[2][56].var1[4]"

The regex match result in the groups 2, 56, 4 captured.

matchGroups = ["2", "56", "4"]

The regex below works.

\[([^\.\]]+)\]

But I can't allow cases like:

"test.[3].var1" or "test[3]/"

I tried to limit the characters allowed before and after each group using the regex below:

[\]a-zA-Z0-9]\[([^\.\]]+)\]([\[a-zA-Z0-9])?

But some cases stopped working like the case "test[0].var7[3][4]"(4 is not captured).

I need help to make this work with all cases again(cases are in the link below).

https://regex101.com/r/mOgOHn/2

• trincot: "But I can't allow cases like ... "test[3]/"": that puzzles me. So the / should make it not match? Which other characters should make it not match? What if "test[3][4]/"? Does then the [3] still match, but not the [4]? What about a space, a line break, a comma instead of the slash....etc? What should match in "test[1][hello][2][#][3], next[4]"?
• Briuor: Sorry @trincot I didn't specified it, but the text input should be a valid javascript array access pattern, test[3][4]/ is not valid in javascript for example.
• trincot: So would you match test[foo[3]] and test["foo"] and test[4/2]?
• Briuor: Thank you for asking these questions I didn't clarify well, I should allow only numeric indexes like test[4], not expressions or object property, I also can't allow test[foo[3]], I didn't think about this case, I should avoid it too.
• trincot: So if there is a mix of those, do you still want to match the numeric indexes, or should nothing match? Like for test[1].view[2]["foo"][3][4/2] should it match 1, 2 and 3, or should it match nothing? What if there is a semicolon after this expression, like test[1].view[2]["foo"][3][4/2]; ... should it then match nothing?

Answer 1

I would replace the string, then do matching

Replace Regex:

/.*?\[(\d+)\].*?/

Replace the string where any number between brackets is replaced by a number followed by ,

Match Regex:

/\d+(?=,)/

Look ahead positive; which Find any number followed by ,

const captureNumber = str => str.replace(/.*?\[(\d+)\].*?/g, "1,ドル").match(/\d+(?=,)/g)
console.log(captureNumber("test[0].var7[3][4]")); //[ "0", "3", "4" ]
console.log(captureNumber("test[3]/")); //[ "3" ] 
console.log(captureNumber("test.[3].var1")); //[ "3" ] 
console.log(captureNumber("test[2][56].var1[4]")); //[ "2", "56", "4" ]

Answer 2

the simplest would be to use a regexp using a positive lookbehind and a positive lookahead. To this, you can add safety on spaces in case you have a case like "test[ 2 ][ 56 ].var1[ 4 ]", to which a .trim() method string must be applied. Here is a snippet illustrating this:

const getIndex = str => str.match(/(?<=\[) *\d+ *(?=\])/g).map(val => val.trim());
console.log(getIndex("test[0].var7[3][4]")); //[ "0", "3", "4" ]
console.log(getIndex("test[3]/")); //[ "3" ]
console.log(getIndex("test.[3].var1")); //[ "3" ]
console.log(getIndex("test[2 ][ 56 ].var1[4 ]")); //[ "2", "56", "4" ]

Good luck !