Skip to content

Latest commit

 

History

History
314 lines (233 loc) · 7.87 KB

README.md

File metadata and controls

314 lines (233 loc) · 7.87 KB
Raw

Elastic Lens

Crates.io Rust

An opinionated framework to work with Elasticsearch.

About

Simple DSL framework to build Elasticsearch requests as well as parse the responses back. There is a STRONG chance not every piece of functionality is available so read the docs and make sure it's the right fit for you.

This project is in it's infancy and is currently supporting a real work project. This is what is driving it's development for now; however, if you have suggestions or edits please feel free to open an issue 👍.

Getting Started

In your Cargo.toml file:

# You must pick one of the currently two supported adapters
# - "official_es7"
# - "official_es8"
elastic_lens = { version = "0.1.8", features = ["official_es7"] }
tokio = { version = "1", features = ["full"] }
serde = { version = "1.0", features = ["derive"] }

Functionality Tour

All of these samples can also be found in examples/sample_code.rs.

Creating a Client

use elastic_lens::{
    client::{Client, DefaultAdapter},
    Error,
};

pub fn create_client() -> Result<Client<DefaultAdapter>, Error> {
    Ok(Client::default_builder()
        .host("http://localhost:9200")
        .index("sample-index")
        .credentials("username", "secret")
        .default_limit(20)
        .build()?)
}

Simple Search

// See `examples/inventory_item.rs` for definition
use super::inventory_item::*;

use elastic_lens::{prelude::*, response::SearchResults, Error};

pub async fn clothing_inventory() -> Result<SearchResults<InventoryItem>, Error> {
    let client = create_client()?;

    let mut search = Search::default();
    search.with(CATEGORY.contains("clothing"));
    search.with(COST.greater_than(500));

    Ok(client.search(&search).await?)
}

Complex Search

use elastic_lens::{prelude::*, request::search::GeoPoint, response::SearchResults, Error};
use serde_json::Value;

// Getting the documents back as serde JSON
// The client will deserialize the `_source` to any type
// that implements for it.  See `examples/inventory_item.rs`
// for an example.
pub async fn complex_search() -> Result<SearchResults<Value>, Error> {
    let client = create_client()?;

    let mut search = Search::default();

    search.with(
        field("server.location")
            .within(500)
            .miles()
            .of(GeoPoint::new(12.2, 18.9)),
    );

    search.with(field("log.level").has_any_of(["error", "warning"]));
    search.with(field("log.trace").exists());

    search.with(if_any_match(|any| {
        any.with(field("service").contains("backend-core"));

        any.with(if_all_match(|all| {
            all.with(field("service").contains("frontend-core"));
            all.with(field("tags").has_any_of(["market-place", "registration"]));
        }));
    }));

    Ok(client.search(&search).await?)
}

MultiSearch

use super::inventory_item::*;
use elastic_lens::{prelude::*, Error};

pub async fn report_clothing_and_office() -> Result<(), Error> {
    let client = create_client()?;

    let mut clothing = Search::default();
    clothing.with(CATEGORY.contains("clothing"));

    let mut office = Search::default();
    office.with(CATEGORY.contains("office"));

    let results = client
        .multi_search::<InventoryItem>(&[clothing, office])
        .await?;

    println!("Clothing:");

    for doc in results[0].docs() {
        println!("{doc:?}");
    }

    println!("\nOffice:");

    for doc in results[1].docs() {
        println!("{doc:?}");
    }

    Ok(())
}

Simple Field Sort

use elastic_lens::{prelude::*, response::SearchResults, Error};
use serde_json::Value;

pub async fn five_cheapest_items() -> Result<SearchResults<Value>, Error> {
    let client = create_client()?;
    let mut search = Search::default();

    search.sort(by_field("cost").ascending().with_missing_values_last());

    search.set_limit(5);

    Ok(client.search(&search).await?)
}

Sorting by GeoDistance

use elastic_lens::{prelude::*, request::search::GeoPoint, response::SearchResults, Error};
use serde_json::Value;

pub async fn nearest_allies() -> Result<SearchResults<Value>, Error> {
    let client = create_client()?;

    let mut search = Search::default();

    search.with(field("user.is_ally").contains(true));

    search.sort(
        by_field("user.location")
            .by_distance_from(GeoPoint::new(1.1, 2.2))
            .in_ascending_order()
            .ignore_unmapped_documents(),
    );

    Ok(client.search(&search).await?)
}

Script Score Sorting

use super::inventory_item::*;
use elastic_lens::{prelude::*, response::SearchResults, Error};

pub async fn some_cheaper_first() -> Result<SearchResults<InventoryItem>, Error> {
    let client = create_client()?;
    let mut search = Search::default();

    search.with(!SUB_CATEGORY.contains("beanie"));

    search.sort(
        by_script(
            r#"
              if ( doc['cost'].value > params.breakpoint ) {
                  doc['cost'].value / 100
              } else {
                  doc['cost'].value * 100
              }
            "#,
        )
        .with_params([("breakpoint", 1300)]),
    );

    Ok(client.search(&search).await?)
}

Term Aggregations

use elastic_lens::{prelude::*, Error};
use serde_json::Value;

pub async fn category_aggs_with_sub_categories() -> Result<StringTerms, Error> {
    let client = create_client()?;
    let mut search = Search::default();

    search
        .create_aggregation("categories")
        .for_field("category")
        .count_terms()
        .with_sub_aggregations(|sub| {
            sub.create_aggregation("sub-categories")
                .for_field("sub_category")
                .count_terms()
                .for_top(20);
        });

    let mut results = client.search::<Value>(&search).await?;
    Ok(results.aggs_mut().take("categories")?)
}

Stats Aggregations

use elastic_lens::{prelude::*, Error};
use serde_json::Value;

pub async fn collect_price_stats() -> Result<Stats, Error> {
    let client = create_client()?;
    let mut search = Search::default();

    search
        .create_aggregation("price-stats")
        .for_field("item.price")
        .collect_stats();

    let mut results = client.search::<Value>(&search).await?;
    Ok(results.aggs_mut().take("price-stats")?)
}

Filter Aggregations

use super::inventory_item::*;
use elastic_lens::{prelude::*, Error};

pub async fn less_than_20_report() -> Result<Filtered, Error> {
    let client = create_client()?;

    let mut search = Search::default();

    search
        .create_aggregation("under-twenty")
        .filtered_by(|search| search.with(COST.less_than(20_00)))
        .with_sub_aggregations(|aggs| {
            aggs.create_aggregation("categories")
                .for_field("category")
                .count_terms()
                .for_top(20);
        });

    // we don't care about documents
    search.set_limit(0);

    // since we know we won't have any results we can use `()`
    // as the type since it doesn't matter
    let mut results = client.search::<()>(&search).await?;
    Ok(results.aggs_mut().take::<Filtered>("under-twenty")?)
}

Playing with the Examples

You'll need Elasticsearch running on port 9200. You can run it dockerized with:

docker run -it -p 9200:9200 -e "discovery.type=single-node" elasticsearch:7.11.2

Then, run bin/setup_index.sh to bootstrap some data that the examples will query against.

The examples are found in the examples/ directory and can be run with cargo: cargo run --example <name> (..args..) Here is current list of examples:

  • fetch_a_document
  • simple_search
  • simple_aggs
  • filter_aggs
  • multi_search
  • simple_sort