jbonofre opened a new issue, #727: URL: https://github.com/apache/camel-karaf/issues/727
## Problem In Karaf, the `tika:parse` operation returns **empty content** regardless of input, with no exception and no warning. The `camel-tika` feature installs and the route runs end-to-end, but nothing is ever extracted from the document. ## Root cause Camel's `TikaProducer` builds a `new AutoDetectParser(TikaConfig)`. `AutoDetectParser` discovers `org.apache.tika.parser.Parser` implementations through the JDK `ServiceLoader` (`META-INF/services/org.apache.tika.parser.Parser`), which does not cross OSGi bundle boundaries. The parser implementations live in separate bundles (`tika-parser-html-module`, `tika-parser-text-module`, …) that the `tika-core` bundle's classloader cannot see, so `AutoDetectParser` finds no parser and falls back to `EmptyParser` → empty output. Verified in a Pax Exam container: the route logs ``` Will parse: <html>...The quick brown fox...</html> Parsed: ``` Adding the Karaf `spifly` feature to `camel-tika` alone does **not** fix it — the Tika parser-module bundles also need SPI-Provider wiring for SPIFly to register them. ## Scope This is a Karaf-specific OSGi integration gap in the `camel-tika` feature, separate from #713 (which only fixed the missing `juniversalchardet` transitive bundle so the parser modules resolve). The itest added in #716 deliberately does not assert on extracted content for this reason. ## Possible directions - Wrap the `tika-parser-*` module bundles with the appropriate `SPI-Provider` / `Require-Capability osgi.serviceloader` clauses and add `spifly` to the feature so `AutoDetectParser`'s `ServiceLoader.load` is mediated by SPIFly. - Or provide an OSGi-aware `TikaConfig`/parser wiring in `camel-core-osgi` / the `camel-tika` packaging. Once resolved, the `camel-tika` itest can be tightened to assert on the extracted text. -- 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]
