Croway opened a new pull request, #26102:
URL: https://github.com/apache/camel/pull/26102

   ## Summary
   
   `DoclingEndpoint` stores the `docling:<operationId>` URI path segment (e.g. 
`docling:EXTRACT_STRUCTURED_DATA`, matching the component's own documented 
syntax `docling:operationId`), but `DoclingComponent` never applied it to 
`DoclingConfiguration.operation`, and `DoclingProducer.getOperation()` only 
consults the `CamelDoclingOperation` header or the configured/default operation 
(`CONVERT_TO_MARKDOWN`).
   
   As a result, an endpoint such as:
   ```
   
docling:EXTRACT_STRUCTURED_DATA?useDoclingServe=true&doclingServeUrl=...&outputFormat=json
   ```
   silently performs `CONVERT_TO_MARKDOWN` instead - the URI path segment is 
effectively ignored unless the operation is *also* passed via the `operation` 
query parameter or the `CamelDoclingOperation` header (the pattern already used 
by this component's own docs/tests, e.g. 
`docling:convert?operation=CONVERT_TO_MARKDOWN`).
   
   I hit this concretely while testing the `docling` example in 
[camel-spring-boot-examples](https://github.com/apache/camel-spring-boot-examples):
 a `document-metadata-extractor` route using 
`docling:EXTRACT_STRUCTURED_DATA?...&outputFormat=json` was writing markdown 
text into `.json` files (invalid JSON), because it was actually always running 
`CONVERT_TO_MARKDOWN`.
   
   - `DoclingComponent.createEndpoint()` now parses `operationId` into a 
`DoclingOperations` value and applies it to the endpoint's configuration 
*before* `setProperties()` runs, so an explicit `?operation=...` query 
parameter still takes precedence, and an `operationId` that isn't a recognized 
operation name (used as a purely descriptive endpoint id) is silently ignored 
rather than failing endpoint creation.
   - Added regression tests in `DoclingComponentTest` covering: `operationId` 
selecting the operation, an explicit `operation` parameter overriding a 
recognized `operationId`, and an unrecognized `operationId` being ignored in 
favor of the `operation` parameter.
   
   Related, not fixed here: once the operation is dispatched correctly, 
`EXTRACT_STRUCTURED_DATA` (and `CONVERT_TO_JSON` in docling-serve mode) sets 
the exchange body to a `DoclingDocument` POJO regardless of `outputFormat`, and 
there's no registered type converter to `String`/bytes for it - callers need to 
marshal it explicitly (e.g. via `camel-jackson`) before handing it to a 
byte-oriented endpoint like `file:`. Happy to file that as a separate follow-up 
if useful.
   
   ## Test plan
   
   - [x] `mvn -o test -Dtest='!*IT'` in `components/camel-ai/camel-docling` - 
all unit tests pass, including the new regression tests
   - [x] Reproduced against a live `docling-serve` container via the 
`camel-spring-boot-examples` docling example: before the fix, 
`EXTRACT_STRUCTURED_DATA` silently ran `CONVERT_TO_MARKDOWN`; after the fix, it 
correctly dispatches to structured-data extraction
   - [x] Verified the pre-existing `docling:convert?operation=...` style (used 
elsewhere in this component's tests/docs) is unaffected
   
   🤖 Generated with [Claude Code](https://claude.com/claude-code)
   
   https://claude.ai/code/session_01Hs6c6LpuL8ZZ3gaSXwt7CS


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to