I am trying to get documents with exact matches. Here is my code:

self.typesense_client.collections[self.default_collection_name].documents.search({
    'q': '*',
    'filter_by': 'url_without_anchor:=`' + url +'`'
})

There is the schema:

{
  "name": "url_without_anchor",
  "type": "string",
  "facet": true,
  "index": true,
  "store": true
}

Problem is when I am searching for url: /en/products/transducers/inertial-sensors/inertial-measurement-units--imu-/3dm-cv5-imu It returns /en/products/transducers/inertial-sensors/inertial-measurement-units--imu-/3dm-cv5-imu/p-xxx /en/products/transducers/inertial-sensors/inertial-measurement-units--imu-/3dm-cv5-imu/p-yyy etc. But search is working fine for following urls • /en/products/transducers/force/c10 • /en/products/instruments/sound-vibration-daq/microphone--calibration/9721-b For those it returns exact matching document, not /en/products/transducers/force/c10/p-.... Please help to to identity the problem, or better way to query. If you required more information about my setup let me know. Thanks in Advance

Could you try setting

"token_separators": ["/", ".", "-"]

in the collection schema?

@Jason Bosco token separators are already added.

Could you share a set of curl commands like this that replicates the issue with a minimal collection and a few sample documents?

Please find the schema, documents, curl, response in this zip

Hey Rohan, The documents file you shared seems to be from a search response. Could you use the export API to export the documents as they are indexed in the collection?

hbkworld_documents.jsonl

Could you share how many documents it should be returning?

Okay, Let me try this. and I'll let you know.

Could you provide a script that indexes the data, creates the collection and searches? It worked on my testing.

call we have a short call?

Due to our bandwidth, we can't provide calls to users in the public Slack community. If you need prioritized support, you can sign up for a support plan on Typesense Cloud: https://cloud.typesense.org/support-plans

Tried all the solution you have mentioned earlier. Still not working. Can you suggest me an workaround for this ?

Like Jason mentioned, we'd need a reproducible example like this. On my testing, adding

:

to your collection's

symbols_to_index

parameter fixed the issue. There may be other factors to your current setup that affect this

Reproduced the issue using the following file. Note: Please check the last two curls

Hey Rohan, thank you for this. I've identified the issue being related to the truncation logic that occurs if a word token is larger than 100 characters when indexing and filtering. We'll mention you when we have a fix

@Fanis Tharropoulos Thank you so much, this really comes as a relief.

It's this PR: https://github.com/typesense/typesense/pull/2549. You can set notifications by subscribing to the updates of the PR.