Changes regarding Meilisearch v1.4.0

[brunoocasali]

This issue gathers the changes related to the v1.4.0 of Meilisearch that will impact the integrations team.

📅 Release date: Sep 25th, 2023

The whole milestone of v1.4.0 is here!

⚠️ The SDKs, which should be done by the release date, are only tier #1. Some of the features described may be done for tier #2 but will not be ready for the release day. Check the features below to understand which tier will receive which feature.

🖖 Click here to check the current tiers state of the integrations:

Understand everything about tiers here.

Integration	Tier
Javascript	#1
PHP	#1
meilisearch-js-plugins	#1
Cloud Providers	#1
Kubernetes Chart	#1
Python	#2
Ruby	#2
Go	#2
Strapi	#2
.NET	#2
Rails	#2
Rust	#2
Symfony	#3
Java	#3
Firebase	#3
docs-searchbar.js	#3
docs-scraper	#3
Dart	#3
Swift	#3
Vuepress	#3
Gatsby	#3
Flutter	#3

⚠️ If no change in the public API is added during the bump pull requests, there is no need for a new release, according to the Integrations team's versioning policy guideline.

---

Customize text separator

Related to:

• issue: Customize text separator meilisearch#3610
• spec: Separators settings api specifications#254

Adapts Meilisearch to specific term nomenclature and thus covers more use cases by adding customization to the engine.

Make Meilisearch more customizable and adaptable to different types of data set content and, therefore, different use cases by letting them configure the separators to be indexed.

What needs to be changed:

• Add two new settings: separatorTokens and nonSeparatorTokens with get, update, and reset methods
• Update the /settings API to enable the possibility to receive in the payload the two new configurations separatorTokens and nonSeparatorTokens

blueprint of the methods:

client.index('indexName').getSeparatorTokens();
client.index('indexName').updateSeparatorTokens(['|', '/', '&sep']);
client.index('indexName').resetSeparatorTokens();

client.index('indexName').getNonSeparatorTokens();
client.index('indexName').updateNonSeparatorTokens(['@', '#']);
client.index('indexName').resetNonSeparatorTokens();

---

Load a user-defined dictionary

Related to:

• issue: Load a user-defined dictionary  meilisearch#3917
• spec: User dictionary settings API specifications#255

Some Languages are not always space-separated (Chinese, Japanese, Korean...), forcing Meilisearch to rely on common word dictionaries to segment a text into words. However, this strategy only segments common words and struggles when a text contains domain-specific, technical, or People Nouns.

Because it is impossible to have an exhaustive list of words of all the possible domains or a complete list of Nouns, Meilisearch should allow users to add their list of words specialized for their use case.

What needs to be changed:

• Add the new setting: dictionary with get, update, and reset methods
• Update the /settings API to enable the possibility of receiving in the payload for dictionary

client.index('indexName').getDictionary();
client.index('indexName').updateDictionary(['|', '/', '&sep']);
client.index('indexName').resetDictionary();

[curquiza]

I close this issue since everything has been done for v1.4.0 on integration side
Issues are opened on the repo that are missing the feature.