Skip to content

GH-4331 improve token estimation for binary data #4332

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 2 commits into from
Closed

GH-4331 improve token estimation for binary data #4332

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3d14a0d
fix(tokenizer): use Base64 encoding for binary data token estimation
Hyeri1ee Sep 7, 2025
89c3e60
chore(tokenizer): apply checkStyleon JTokkitTokenCountEstimator.java
Hyeri1ee Sep 7, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
25 changes: 18 additions & 7 deletions ...ai-commons/src/main/java/org/springframework/ai/tokenizer/JTokkitTokenCountEstimator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,6 +16,8 @@

package org.springframework.ai.tokenizer;

import java.util.Base64;

import com.knuddels.jtokkit.Encodings;
import com.knuddels.jtokkit.api.Encoding;
import com.knuddels.jtokkit.api.EncodingType;
Expand All @@ -34,43 +36,52 @@
*/
public class JTokkitTokenCountEstimator implements TokenCountEstimator {

/**
* The JTokkit encoding instance used for token counting.
*/
private final Encoding estimator;

/**
* Creates a new JTokkitTokenCountEstimator with default CL100K_BASE encoding.
*/
public JTokkitTokenCountEstimator() {
this(EncodingType.CL100K_BASE);
}

public JTokkitTokenCountEstimator(EncodingType tokenEncodingType) {
/**
* Creates a new JTokkitTokenCountEstimator with the specified encoding type.
* @param tokenEncodingType the encoding type to use for token counting
*/
public JTokkitTokenCountEstimator(final EncodingType tokenEncodingType) {
this.estimator = Encodings.newLazyEncodingRegistry().getEncoding(tokenEncodingType);
}

@Override
public int estimate(String text) {
public int estimate(final String text) {
if (text == null) {
return 0;
}
return this.estimator.countTokens(text);
}

@Override
public int estimate(MediaContent content) {
public int estimate(final MediaContent content) {
int tokenCount = 0;

if (content.getText() != null) {
tokenCount += this.estimate(content.getText());
}

if (!CollectionUtils.isEmpty(content.getMedia())) {

for (Media media : content.getMedia()) {

tokenCount += this.estimate(media.getMimeType().toString());

if (media.getData() instanceof String textData) {
tokenCount += this.estimate(textData);
}
else if (media.getData() instanceof byte[] binaryData) {
tokenCount += binaryData.length; // This is likely incorrect.
String base64 = Base64.getEncoder().encodeToString(binaryData);
tokenCount += this.estimate(base64);
}
}
}
Expand All @@ -79,7 +90,7 @@ else if (media.getData() instanceof byte[] binaryData) {
}

@Override
public int estimate(Iterable<MediaContent> contents) {
public int estimate(final Iterable<MediaContent> contents) {
int totalSize = 0;
for (MediaContent mediaContent : contents) {
totalSize += this.estimate(mediaContent);
Expand Down