Whitespace

The whitespace tokenizer splits text at five ASCII whitespace characters: tab, line feed, form feed, carriage return, and space.

Tokenization rules

The whitespace tokenizer splits text only at the following five ASCII whitespace characters:

Character	Name	Unicode code point
\t	Horizontal tab	U+0009
\n	Line feed	U+000A
\x0C or \f	Form feed	U+000C
\r	Carriage return	U+000D
' '	Space	U+0020

These separators are discarded, and consecutive separators do not produce empty tokens. Punctuation and other characters remain in the tokens. In particular, vertical tab (\x0B, U+000B), no-break space (\u00A0), and ideographic space (\u3000) do not trigger splitting.

This set follows Rust's char::is_ascii_whitespace(), which excludes other Unicode whitespace characters.

The following examples use {"tokenizer": "whitespace"} with no filters. Inputs and outputs use Python string notation: escape sequences such as \t and \u00A0 represent the actual characters.

Input	Output tokens
"a\tb\nc\x0Cd\re f"	["a", "b", "c", "d", "e", "f"]
"Hello,World! foo_bar"	["Hello,World!", "foo_bar"]
"a\x0Bb"	["a\x0Bb"]
"a\u00A0b"	["a\u00A0b"]
"a\u3000b"	["a\u3000b"]
"\x20a\x20\x20b\x20"	["a", "b"]

Configuration

To configure an analyzer using the whitespace tokenizer, set tokenizer to whitespace in analyzer_params.

Python

python

analyzer_params = {
    "tokenizer": "whitespace",
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "whitespace");

Go

go

analyzerParams := map[string]any{"tokenizer": "whitespace"}

Rust

rust

let analyzer_params = serde_json::json!({"tokenizer": "whitespace"});

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "whitespace"}
};

NodeJS

javascript

const analyzer_params = {
    "tokenizer": "whitespace"
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "whitespace"
}'

The whitespace tokenizer can work in conjunction with one or more filters. For example, the following code defines an analyzer that uses the whitespace tokenizer and lowercase filter:

Python

python

analyzer_params = {
    "tokenizer": "whitespace",
    "filter": ["lowercase"]
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "whitespace");
analyzerParams.put("filter", Collections.singletonList("lowercase"));

Go

go

analyzerParams = map[string]any{"tokenizer": "whitespace", "filter": []any{"lowercase"}}

Rust

rust

let analyzer_params = serde_json::json!({"tokenizer": "whitespace", "filter": ["lowercase"]});

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "whitespace"},
    {"filter", {"lowercase"}}
};

NodeJS

javascript

const analyzer_params = {
    "tokenizer": "whitespace",
    "filter": ["lowercase"]
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "whitespace",
  "filter": [
    "lowercase"
  ]
}'

After defining analyzer_params, you can apply them to a VARCHAR field when defining a collection schema. This allows Zilliz Cloud to process the text in that field using the specified analyzer for efficient tokenization and filtering. For details, refer to Example use.

Examples

Before applying the analyzer configuration to your collection schema, verify its behavior using the run_analyzer method.

Analyzer configuration

Python

python

analyzer_params = {
    "tokenizer": "whitespace",
    "filter": ["lowercase"]
}

Java

java

Map<String, Object> analyzerParams = new HashMap<>();
analyzerParams.put("tokenizer", "whitespace");
analyzerParams.put("filter", Collections.singletonList("lowercase"));

Go

go

analyzerParams = map[string]any{"tokenizer": "whitespace", "filter": []any{"lowercase"}}

Rust

rust

let analyzer_params = serde_json::json!({"tokenizer": "whitespace", "filter": ["lowercase"]});

C++

c++

nlohmann::json analyzer_params = {
    {"tokenizer", "whitespace"},
    {"filter", {"lowercase"}}
};

NodeJS

javascript

const analyzer_params = {
    "tokenizer": "whitespace",
    "filter": ["lowercase"]
};

cURL

bash

# restful
analyzerParams='{
  "tokenizer": "whitespace",
  "filter": [
    "lowercase"
  ]
}' 

Verification using run_analyzer

Python

python

from pymilvus import (
    MilvusClient,
)

client = MilvusClient(
    uri="YOUR_CLUSTER_ENDPOINT",
    token="YOUR_CLUSTER_TOKEN"
)

# Sample text to analyze
sample_text = "The Milvus vector database is built for scale!"

# Run the whitespace analyzer with the defined configuration
result = client.run_analyzer(sample_text, analyzer_params)
print("Whitespace analyzer output:", result)

Java

java

import io.milvus.v2.client.ConnectConfig;
import io.milvus.v2.client.MilvusClientV2;
import io.milvus.v2.service.vector.request.RunAnalyzerReq;
import io.milvus.v2.service.vector.response.RunAnalyzerResp;
import java.util.*;

ConnectConfig config = ConnectConfig.builder()
        .uri("YOUR_CLUSTER_ENDPOINT")
        .token("YOUR_CLUSTER_TOKEN")
        .build();
MilvusClientV2 client = new MilvusClientV2(config);

List<String> texts = new ArrayList<>();
texts.add("The Milvus vector database is built for scale!");

RunAnalyzerResp resp = client.runAnalyzer(RunAnalyzerReq.builder()
        .texts(texts)
        .analyzerParams(analyzerParams)
        .build());
List<RunAnalyzerResp.AnalyzerResult> results = resp.getResults();

Go

go

import (
    "context"
    "fmt"

    "github.com/milvus-io/milvus/client/v3/milvusclient"
)

ctx := context.Background()
client, err := milvusclient.New(ctx, &milvusclient.ClientConfig{
    Address: "YOUR_CLUSTER_ENDPOINT",
    APIKey:  "YOUR_CLUSTER_TOKEN",
})
if err != nil {
    fmt.Println(err.Error())
    // handle error
}

texts := []string{"The Milvus vector database is built for scale!"}
option := milvusclient.NewRunAnalyzerOption(texts...).
    WithAnalyzerParams(analyzerParams)

result, err := client.RunAnalyzer(ctx, option)
if err != nil {
    fmt.Println(err.Error())
    // handle error
}

Rust

rust

use milvus::v2::prelude::*;

let client = ClientV2::new(&ConnectConfig::new().uri("YOUR_CLUSTER_ENDPOINT").token("YOUR_CLUSTER_TOKEN")).await?;

let sample_text = "The Milvus vector database is built for scale!";
client
    .run_analyzer(
        RunAnalyzerRequest::builder()
            .texts([sample_text])
            .analyzer_params(analyzer_params.clone())
            .build()?,
    )
    .await?;

C++

c++

#include <iostream>
#include "milvus/MilvusClientV2.h"

auto client = milvus::MilvusClientV2::Create();

milvus::ConnectParam connect_param{"YOUR_CLUSTER_ENDPOINT", "YOUR_CLUSTER_TOKEN"};
auto status = client->Connect(connect_param);
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}

std::string text = "The Milvus vector database is built for scale!";
auto request = milvus::RunAnalyzerRequest()
                       .AddText(text)
                       .WithAnalyzerParams(analyzer_params);

milvus::RunAnalyzerResponse response;
status = client->RunAnalyzer(request, response);
if (!status.IsOk()) {
    std::cout << status.Message() << std::endl;
}

NodeJS

javascript

import { MilvusClient } from "@zilliz/milvus2-sdk-node";

const client = new MilvusClient({
    address: "YOUR_CLUSTER_ENDPOINT",
    token: "YOUR_CLUSTER_TOKEN"
});

// Sample text to analyze
const sample_text = "The Milvus vector database is built for scale!";

// Run the whitespace analyzer with the defined configuration
const result = await client.runAnalyzer({
    text: sample_text,
    analyzer_params: analyzer_params
});

cURL

bash

# restful
export MILVUS_HOST="YOUR_CLUSTER_ENDPOINT"
export SAMPLE_TEXT="The Milvus vector database is built for scale!"
curl -X POST "http://${MILVUS_HOST}/v2/vectordb/common/run_analyzer" \
  -H "Content-Type: application/json" \
  -H "Request-Timeout: 10" \
  -d '{
    "text": ["'"${SAMPLE_TEXT}"'"],
    "analyzerParams": "{\"tokenizer\":\"whitespace\",\"filter\":[\"lowercase\"]}"
  }' 

Expected output

sql

['the', 'milvus', 'vector', 'database', 'is', 'built', 'for', 'scale!']