private static DataSet<Tuple3<Integer, String, Integer>> getRanksDataSet(
ExecutionEnvironment env, ParameterTool params) {
// Create DataSet for ranks relation (Rank, URL, Avg-Visit-Duration)
if (params.has("ranks")) {
return env.readCsvFile(params.get("ranks"))
.fieldDelimiter("|")
.types(Integer.class, String.class, Integer.class);
} else {
System.out.println("Executing WebLogAnalysis example with default ranks data set.");
System.out.println("Use --ranks to specify file input.");
return WebLogData.getRankDataSet(env);
}
}
private static DataSet<Tuple2<String, String>> getDocumentsDataSet(
ExecutionEnvironment env, ParameterTool params) {
// Create DataSet for documents relation (URL, Doc-Text)
if (params.has("documents")) {
return env.readCsvFile(params.get("documents"))
.fieldDelimiter("|")
.types(String.class, String.class);
} else {
System.out.println("Executing WebLogAnalysis example with default documents data set.");
System.out.println("Use --documents to specify file input.");
return WebLogData.getDocumentDataSet(env);
}
}
