JSON configuration files need to be parsed to simplify Java programs and processing logic. But Hadoop itself doesn't seem to have built-in JSON file parsing function, so it has to turn to the third-party JSON toolkit. Json-simple is selected here to realize this function. Brief introduction:

(1): The command to execute Java programs on Hadoop is as follows:

(2): My-MapReduce。 Jar is a MapReduce program for log processing. Now suppose you need to deal with a configuration file in JSON format. Here, we ignore the details of how to read files in Hadoop cluster and only focus on how to use JSON toolkit. The following is a simple HelloWorld program:

In the HelloWorld program, you only need to modify the JSON object and print out its contents, thus verifying the process of parsing and modifying the JSON content.

Second compilation:

Because MapReduce programs need to be submitted to Hadoop cluster for execution, if there is no corresponding jar package on the cluster, the json-simple package that HelloWorld depends on must exist in the classpath of the cluster. The following exception occurs when executing HelloWorld:

Exception in thread "main" java.lang.noclassDefoundError: org/json/simple/JSONObject.

The simple solution is to package the json-simple package directly with the compilation result of HelloWorld, and then execute the jar package that you need to decompress json-simple with the command hadoop jar HelloWorld.jar, and package it with HelloWorld.

The compilation command is as follows:

Third, realize HelloWorld: