source: other-projects/hathitrust/solr-extracted-features/trunk/src/main/java/org/hathitrust/extractedfeatures/PerPageJSONMap.java@ 31013

Last change on this file since 31013 was 31013, checked in by davidb, 7 years ago

Accumulator for PerPageMap

  • Property svn:executable set to *
File size: 2.1 KB
Line 
1package org.hathitrust.extractedfeatures;
2
3import org.apache.spark.api.java.function.Function;
4import org.apache.spark.api.java.function.MapFunction;
5import org.apache.spark.util.DoubleAccumulator;
6import org.json.JSONObject;
7
8
9class PerPageJSONMap implements Function<JSONObject, String>
10{
11 private static final long serialVersionUID = 1L;
12
13 protected String _input_dir;
14 protected String _solr_url;
15 protected String _output_dir;
16 protected int _verbosity;
17
18 protected DoubleAccumulator _progress_accum;
19 protected double _progress_step;
20
21
22 public PerPageJSONMap(String input_dir, String solr_url, String output_dir, int verbosity,
23 DoubleAccumulator progress_accum, double progress_step)
24 {
25 _input_dir = input_dir;
26 _solr_url = solr_url;
27 _output_dir = output_dir;
28 _verbosity = verbosity;
29
30 _progress_accum = progress_accum;
31 _progress_step = progress_step;
32 }
33
34 public String call(JSONObject solr_add_doc_json)
35 {
36 String output_json_bz2 = solr_add_doc_json.getString("filename_json_bz2");
37 solr_add_doc_json.remove("filename_json_bz2");
38
39 boolean random_test = (Math.random()>0.999); // every 1000
40
41 if ((_verbosity >=2) && (random_test)) {
42 System.out.println("==================");
43 System.out.println("Sample output Solr add JSON [random test 1/1000]: " + solr_add_doc_json.toString());
44 System.out.println("==================");
45 }
46
47
48 if (_solr_url != null) {
49 if ((_verbosity >=2) && (random_test)) {
50 System.out.println("==================");
51 System.out.println("Posting to: " + _solr_url);
52 System.out.println("==================");
53 }
54 SolrDocJSON.postSolrDoc(_solr_url, solr_add_doc_json);
55 }
56
57 if (_output_dir != null) {
58 if ((_verbosity >=2) && (random_test)) {
59 System.out.println("==================");
60 System.out.println("Saving to: " + _output_dir);
61 System.out.println("==================");
62 }
63 SolrDocJSON.saveSolrDoc(solr_add_doc_json, _output_dir + "/" + output_json_bz2);
64 }
65
66 _progress_accum.add(_progress_step);
67
68 return output_json_bz2;
69 }
70
71}
72
Note: See TracBrowser for help on using the repository browser.