[davsclaus]

Stacked on #28250. This is the object storage group of the catalog body types (CAMEL-24961): aws2-s3, azure-storage-blob, dropbox, google-storage, ibm-cos, minio and azure-files. Of the components the human-written routes use, these are the storage ones used most: 17, 12, 8, 6, 6, 4 and 4 times.

What the catalog says now

Component	Consumer	Producer input	Producer output
aws2-s3	bytes; a stream with includeBody=false, no body with ignoreBody=true	without an operation, the content to upload: text, bytes, stream or file	unchanged after an upload. By operation: getObject a stream, the lists a list, the links text, deleteObject true, the SDK response for the others
azure-storage-blob	a stream, or with fileDir the file	any. The uploads take text, bytes, stream or file, the block lists a list	a list for listBlobContainers (the default) and the other lists. getBlob a stream, downloadBlobToFile a file, true for the writes
dropbox	get: the content, or a map for a folder. search: a list	any: localPath or a header can name the file to upload	put: OK or KO. get: as the consumer. del and move: the path. search: a list
google-storage	the content, a file with downloadFileName, the Blob with includeBody=false	the content to upload	the Blob of the upload. getObject bytes, the lists a list, the deletes true or false
ibm-cos	a stream; no body with includeBody=false	the content to upload with putObject	unchanged after putObject. getObject a stream, the lists a list, headBucket a Boolean
minio	bytes; no body with includeBody=false	the content to upload	unchanged after an upload. getObject a stream, listObjects an Iterable and listBuckets an Iterator: iterator
azure-files	the file as read, as for ftp	text, bytes, stream or file	unchanged

In every one of them, an upload of a Map or a List fails rather than sending its toString(). It is an InvalidPayloadException, or an IllegalArgumentException in azure-storage-blob. The descriptions say to marshal it first.

Checked at runtime. Routes logged the class of the body after each step, against camel infra run aws (S3 on LocalStack) and camel infra run azure (Azurite):

• aws2-s3:
  • every operation;
  • the upload of text, bytes, a stream and a file, and the failing Map, List and null;
  • pojoRequest, multiPartUpload and streamingUploadMode;
  • the consumer with includeBody and ignoreBody.
    With stream caching on, the next step after getObject sees the stream cache, not the SDK's ResponseInputStream.
• ibm-cos and minio: against the same S3, ibm-cos through an AmazonS3 client bean with HMAC keys, and minio through its endpoint option. The image of camel infra run minio needs a login on quay.io now.
• azure-storage-blob: 16 operations, and the consumer with and without fileDir.
• dropbox, google-storage and azure-files: these have no local server, so they were read from the code.

Options the component can set. In aws2-s3, azure-storage-blob, google-storage, ibm-cos and minio, every option is a component option too, as the component shares its configuration. The validator counts a component option only when the endpoint sets it. So it checks a step only when the endpoint names the operation: after aws2-s3:reports?operation=getObject the body is a stream, and aws2-s3:reports says nothing. At runtime the operation header (such as CamelAwsS3Operation) wins over the option, as it does for aws2-sqs.

One validator change. The validator called a body "the file as it was read (a GenericFile)" whenever file was one of its carriers. Now that wording, and the file properties such as ${body.fileName}, apply only when the file is the only carrier. A blob that is a stream or a downloaded file is a payload like any other. Before this group, no side in the catalog mixed file with another payload carrier.

Docs. The pages of aws2-s3, google-storage and minio said getObject returns an S3Object, a Blob or a MinioObject instance. It returns the content of the object. The aws2-s3 consumer example said includeBody=false gives only the metadata. It gives the stream, and ignoreBody=true gives the metadata only.

Measured. I ran the validator over the 1204 human-written and 1209 model-written route files, and the reports are the same as with #28250. The human files use these components 57 times. Most are Kamelets that pass the body to kamelet:sink, or uploads, or getObject through toD.

Tests

• TextBodyFlowTest.testTheBodyOfObjectStorage covers four routes:
  • a field read after getObject is reported;
  • a field read after the blob consumer is reported;
  • after an upload, which names no operation, nothing is reported;
  • nothing is reported on the List of dropbox search.
• These pass:
  • the 281 tests of camel-yaml-dsl-validator;
  • the 265 validator and catalog doc tests of camel-jbang-core.
• The answer of camel_catalog_doc stays under its 18 000 characters: aws2-s3, the largest, is 17 704.

Claude Code on behalf of davsclaus

🤖 Generated with Claude Code

[github-actions]

🌟 Thank you for your contribution to the Apache Camel project! 🌟
🤖 CI automation will test this PR automatically.

🐫 Apache Camel Committers, please review the following items:

• First-time contributors require MANUAL approval for the GitHub Actions to run
• You can use the command /component-test (camel-)component-name1 (camel-)component-name2.. to request a test from the test bot although they are normally detected and executed by CI.
• You can label PRs using skip-tests and test-dependents to fine-tune the checks executed by this PR.
• Build and test logs are available in the summary page. Only Apache Camel committers have access to the summary.

⚠️ Be careful when sharing logs. Review their contents before sharing them publicly.

[gnodet-bot]

LGTM. The @BodyTypes annotations are consistent across all 7 components, the runtime-verified descriptions match the catalog JSON, and the file() method fix (exact-list equality vs. contains) is correct and intentional. Docs fixes for getObject return types are accurate.

One behavior-doc note for awareness: the aws2-s3-consumer-examples.adoc previously stated includeBody=false gives "only metadata" — the new wording correctly says it gives the stream of the object (and points to ignoreBody=true for metadata-only). No action needed.

This review was generated by an AI agent, Hermès on behalf of @gnodet.