feat(idp extraction connector): implement new idp extraction outbound connector #3482
Conversation
|
|
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
...s/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/caller/BedrockCaller.java
Fixed
Show fixed
Hide fixed
...s/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/caller/BedrockCaller.java
Fixed
Show fixed
Hide fixed
...s/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/caller/BedrockCaller.java
Fixed
Show fixed
Hide fixed
...ectors/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/utils/AwsS3Util.java
Fixed
Show fixed
Hide fixed
...ectors/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/utils/AwsS3Util.java
Fixed
Show fixed
Hide fixed
...ectors/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/utils/AwsS3Util.java
Fixed
Show fixed
Hide fixed
| Message message = | ||
| Message.builder() | ||
| .content(ContentBlock.fromText(prompt)) | ||
| .role(ConversationRole.USER) | ||
| .build(); |
Check notice
Code scanning / CodeQL
Unread local variable Note
...s/idp-extraction/src/main/java/io/camunda/connector/idp/extraction/caller/BedrockCaller.java
Fixed
Show fixed
Hide fixed
| ConverseResponse response = | ||
| bedrockRuntimeClient.converse( | ||
| request -> | ||
| request | ||
| .modelId(converseData.modelId()) | ||
| .messages(message) | ||
| .inferenceConfig( | ||
| config -> | ||
| config | ||
| .maxTokens(converseData.maxTokens()) | ||
| .temperature(converseData.temperature()) | ||
| .topP(converseData.topP()))); |
Check notice
Code scanning / CodeQL
Unread local variable Note
| private static final Logger LOGGER = LoggerFactory.getLogger(AwsS3Util.class); | ||
|
|
||
| private static String uploadNewFileFromUrl( | ||
| final String documentUrl, final String bucketName, final S3AsyncClient s3AsyncClient) |
Check notice
Code scanning / CodeQL
Useless parameter Note
| PutObjectRequest putObjectRequest = | ||
| PutObjectRequest.builder().bucket(bucketName).key(documentKey).build(); |
Check notice
Code scanning / CodeQL
Unread local variable Note
| AsyncRequestBody asyncRequestBody = | ||
| AsyncRequestBody.fromInputStream( | ||
| body -> | ||
| body.executor(executorService) | ||
| .contentLength(contentLength) | ||
| .inputStream(inputStream) | ||
| .build()); |
Check notice
Code scanning / CodeQL
Unread local variable Note
| import org.apache.commons.lang3.builder.EqualsBuilder; | ||
| import org.apache.commons.lang3.builder.HashCodeBuilder; | ||
|
|
||
| public class ExtractionRequest extends AwsBaseRequest { |
There was a problem hiding this comment.
To be more flexible in terms of cloud providers, I would recommend to create another class in your model like AwsRequest and just reference it ExtractionRequest so you dont have to inherit from the AwsBaseRequest here.
There was a problem hiding this comment.
It would also allow you to change this one to a record.
There was a problem hiding this comment.
Created a new class BaseRequest to extend from AwsBaseRequest .
Converted ExtractionRequest to a record.
|
@sahilbhatoacamunda Congrats to your first Connector 🥳 and great work. Added just a smaller comments regarding the model. From my perspective its fine for now that you implemented your own polling although its somewhat duplicated in the textract connector right now but we can generalize later rather than now. |
| .role(ConversationRole.USER) | ||
| .build(); | ||
|
|
||
| ConverseData converseData = extractionRequest.input().converseData(); |
Check notice
Code scanning / CodeQL
Unread local variable Note
| */ | ||
| package io.camunda.connector.idp.extraction.model; | ||
|
|
||
| public record ExtractionResult(String response) {} |
There was a problem hiding this comment.
Using an Object would allow to return a json object.
…g to allow json objects
Description
This PR introduces the integration of an outbound IDP extraction connector. This connector uses AWS Bedrock and Textract/Apache PdfBox to extract the text and then analyze it based on the input taxonomy.
Related issues
closes https:/camunda/product-hub/issues/2527
Checklist
no milestonelabel.