空解析器和 Tika 服务器模式

Question

I am having trouble understanding how parsers are loaded into Tika. From their documentation it appears that Tika-app comes prepackaged with the parsers ( https://tika.apache.org/1.17/gettingstarted.html ). When I run this command to start the server though

    ./.java-buildpack/open_jdk_jre/bin/java -jar ./lib/tika-app-1.24.1.jar -s --port ${PORT}

    2020-11-02T13:30:26.04-0600 [APP/PROC/WEB/0] ERR Nov 02, 2020 7:30:26 PM org.apache.tika.config.InitializableProblemHandler$3 handleInitializableProblem
   2020-11-02T13:30:26.04-0600 [APP/PROC/WEB/0] ERR WARNING: J2KImageReader not loaded. JPEG2000 files will not be processed.
   2020-11-02T13:30:26.04-0600 [APP/PROC/WEB/0] ERR See https://pdfbox.apache.org/2.0/dependencies.html#jai-image-io
   2020-11-02T13:30:26.04-0600 [APP/PROC/WEB/0] ERR for optional dependencies.
   2020-11-02T13:30:26.53-0600 [APP/PROC/WEB/0] ERR Nov 02, 2020 7:30:26 PM org.apache.tika.config.InitializableProblemHandler$3 handleInitializableProblem
   2020-11-02T13:30:26.53-0600 [APP/PROC/WEB/0] ERR WARNING: org.xerial's sqlite-jdbc is not loaded.
   2020-11-02T13:30:26.53-0600 [APP/PROC/WEB/0] ERR Please provide the jar on your classpath to parse sqlite files.
   2020-11-02T13:30:26.53-0600 [APP/PROC/WEB/0] ERR See tika-parsers/pom.xml for the correct version.
   2020-11-02T13:30:26.80-0600 [APP/PROC/WEB/0] OUT Successfully started tika-app's server on port: 8080
   2020-11-02T13:30:26.80-0600 [APP/PROC/WEB/0] ERR WARNING: The server option in tika-app is deprecated and will be removed
   2020-11-02T13:30:26.80-0600 [APP/PROC/WEB/0] ERR by Tika 2.0 if not shortly after Tika 1.14.
   2020-11-02T13:30:26.80-0600 [APP/PROC/WEB/0] ERR Please migrate to the JAX-RS tika-server package.
   2020-11-02T13:30:26.80-0600 [APP/PROC/WEB/0] ERR See https://wiki.apache.org/tika/TikaJAXRS for usage.
   2020-11-02T13:31:25.66-0600 [HEALTH/0] ERR Failed to make HTTP request to '/version' on port 8080: timed out after 1.00 seconds
   2020-11-02T13:31:25.66-0600 [CELL/0] ERR Timed out after 1m0s: health check never passed.

I have the most recent tika version 1.24.1. Their documentation mentions downloading tika-server and passing classpath at runtime to point to a tika-parsers.jar ( https://cwiki.apache.org/confluence/display/TIKA/Troubleshooting+Tika#TroubleshootingTika-ParsersMissing ) but I can't find the parsers.jar file anywhere. I am using openjdk-jre-1.8.0 to run this.

Answer 1

The parsers should be bundled by default. Tika App in server mode (-s) is socket based server. You can confirm it is working by using netcat and seeing if you get a response:

nc localhost 8080 -q2 < test.pdf

To use this in Python you would need to write custom code open a socket and send the input in, send a SHUT_WR, and read the output back.

If you are using tika-python library, it is expecting to use a Tika Server which is in the tika-server JAR not the tika-app JAR. It has some helper settings so you can point to the JAR, or you can host your own instance (self run or docker) and give it the URL.