jmespath/chat

glenveegee

@glenveegee

Can't be done with the standard jmespath spec. Some "augmented" jmespath libraries add this functionality but not available in all languages

PeiSong

@naivefun

oh I need use custom func to impl myself

glenveegee

@glenveegee

There are libraries out there that provide the ability to add your own custom functions to the existing jmespath spec. I'm afraid this is the only way I have found to make it possible to support the kind of operations I've required in the past.

I've also created proposals as suggested by the author @jamesls here (https://github.com/jmespath/jmespath.jep) in the hope that they get added to the spec

Unfortunately @jamesls seems quite busy at the moment as there hasn't been any feedback from him in the last 6 months

glenveegee

@glenveegee

If you could... would you want to be able to do something like this:

expect(
    search([6.1, 4.2, 6.3], "_groupBy(@, as_lambda('Math.floor'))")
).toStrictEqual({
      '4': [4.2],
      '6': [6.1, 6.3],
    });

or perhaps

expect(
    search([1, 2], "_flatMap(@, as_lambda('x => [x, x]'))")
).toStrictEqual([1, 1, 2, 2]);

Thoughts?

glenveegee

@glenveegee

Meh... did it anyway. _flatMap, _groupBy and 184 other (fully typed) functions have been built into JMESPath expressions in our https://www.npmjs.com/package/@metrichor/jmespath-plus library. Level-up your JMESPath expressions with the power of lodash functions. Check out https://github.com/nanoporetech/jmespath-plus/releases/tag/v0.3.0 for a complete list of supported lodash functions

innocuoussoul

@innocuoussoul

is it possible to filter expected values so the resulting response is only unique?

{
  "subject": [
    "Fwd: Please complete your assigned training"
  ],
  "labelIds": [
    "UNREAD",
    "CATEGORY_PERSONAL",
    "Label_6568222846829069483",
    "INBOX"
  ]
}

{subject:payload.headers[?name=='Subject'].value,labelIds:labelIds}

i want to filter UNREAD, CATEGORY_PERSONAL, AND INBOX... so only that unique label remains.

glenveegee

@glenveegee

{subject: payload.headers[?name=='Subject'].value, labelIds:map(&(!contains(['UNREAD', 'CATEGORY_PERSONAL', 'INBOX'], @) && [@]) || [] ,labelIds)[]}

{subject: payload.headers[?name=='Subject'].value, labelIds:labelIds[*][(!contains(['UNREAD', 'CATEGORY_PERSONAL', 'INBOX'], @) && @) ||[]][]}

Moisés Jiménez

@Mo_Jim_twitter

Hi everyone, I have a question. Is it possible to use the result of a query to search again? For example, say we have the slightly modified example from jmespath.org

{
"locations": [
{"name": "Seattle", "state": "WA"},
{"name": "New York", "state": "NY"},
{"name": "Bellevue", "state": "WA"},
{"Seattle": "Olympia", "state": "WA"}
]
}

could this query
{value: @, key: keys(@)[?ends_with(@, 'Seattle')]} | @.value.[@.key]

return "Olympia"??

glenveegee

@glenveegee

You should be able to pipe your projection but the query is probably not working in the way you want

What are you actually trying to do

Moisés Jiménez

@Mo_Jim_twitter

so what I want to do is find all the keys in a dictionary that end with a certain string and then use them to extract the values

glenveegee

@glenveegee

something like this?

locations[?ends_with(state, 'A')].name

Moisés Jiménez

@Mo_Jim_twitter

yeah this gives me just the key names, same as keys(@)[?ends_with(@, 'Seattle')], now I need to use that to extract the values

glenveegee

@glenveegee

Could you write the result of the query you'd like (what you hope to get out). I'm afraid I still don't follow

Moisés Jiménez

@Mo_Jim_twitter

lets say we have this:
{"key_a_0": "a", "key_b_0": "b", "key_c_1": "c", "key_d_2": d, "key_e_0": "e"}
and we do
{value: @, key: keys(@)[?ends_with(@, '_0')]} | @.value.[@.key]
I would like the result to be
"a", "b", "e"

stiller_leser

@stiller_leser_twitter

Is it possible to filter by timestamp? Say I have a timestamp in my json and I would only like to retrieve the entries from the last 5 minutes?

glenveegee

@glenveegee

@Mo_Jim_twitter I don't think that's possible. What language are you coding in?

@stiller_leser_twitter If it's a formatted date then no. If it's in milliseconds then yes but only if you hard code now time

What language are you working with?

glenveegee

@glenveegee

@jamesls please, please, please can we look at a proposal for an entries function in the spec so we can get object keys and values in the same execution context.

This isn't the first time this sort of query has arisen and I'm certain it won't be the last

stiller_leser

@stiller_leser_twitter

Hm ok @glenveegee is there a way to only get the last x entries then? Currently my query looks like this: resources[?involvedObject.apiVersion == v1 && involvedObject.kind == Service && involvedObject.name==my-name].{ firstTimestamp: firstTimestamp, lastTimestamp: lastTimestamp, count: count, reason: reason, message: message }

glenveegee

@glenveegee

Yes you can use slices

https://jmespath.org/specification.html#slices

I your case search([-x:], [0, 1, 2, 3]) -> [2, 3] where x is a number - the last x entries

stiller_leser

@stiller_leser_twitter

How would I wrap this around my existing search?

glenveegee

@glenveegee

for the last 5 you'd do something like

resources[?involvedObject.apiVersion == v1 && involvedObject.kind == Service && involvedObject.name==my-name].{ firstTimestamp: firstTimestamp, lastTimestamp: lastTimestamp, count: count, reason: reason, message: message }[-5:]

...at least that's what I think you'd have to do

stiller_leser

@stiller_leser_twitter

That's correct. I kept trying to get it too work with search :D Thank you for the help here!

meschiany

@meschiany

Hello, I have a structure query i could not figure out, is this the place?

{
"tags": {
"private_password_reset": {
"tag_name": "PasswordResetSettings"
},
"users": {
"tag_name": "users"
}
}
}

result should be: PasswordResetSettings, users

glenveegee

@glenveegee

tags.*.tag_name

join(', ', tags.*.tag_name)

meschiany

@meschiany

wow, you made it look very easy, thank you

varajgandhi

@varajgandhi

^[8-9]\d{7}$ this regx not working for json params..mobile number only starting with 8 or 9 and totally 8 digits

kindly advise anyone

also single \ throws error in JSON

glenveegee

@glenveegee

Regex search and filter is not part of the jmespath spec as far as I'm aware. Where's this coming from?

varajgandhi

@varajgandhi

am passing value for angular form

glenveegee

@glenveegee

'angular' so I'm assuming JS/TS then what jmespath library are you using?

What is your input to the jmespath search function and what does your expression look like

varajgandhi

@varajgandhi

currently, params take 0-9 as the first digit for a phone number but I need to change the pattern for the starting digit , it must accept either 8 or 9

glenveegee

@glenveegee

Ah. Don't think this has anything to do with JMESPath but fwiw your regex could be written as ^[89][0-9\-]*$. The regex in the image isn't particularly good because it permits a telephone number like 8------------------------5

You might want to rewrite that regex anyway to make it stricter

Where communities thrive

People

Repo info

Activity