1 | package org.hathitrust.extractedfeatures;
|
---|
2 |
|
---|
3 | import java.io.Serializable;
|
---|
4 | import org.apache.commons.cli.*;
|
---|
5 |
|
---|
6 | import org.apache.spark.api.java.*;
|
---|
7 | import org.apache.spark.util.DoubleAccumulator;
|
---|
8 | import org.hathitrust.extractedfeatures.PagedJSON;
|
---|
9 | import org.json.JSONObject;
|
---|
10 | import org.apache.spark.SparkConf;
|
---|
11 |
|
---|
12 | public class ProcessForSolrIngest implements Serializable
|
---|
13 | {
|
---|
14 | private static final long serialVersionUID = 1L;
|
---|
15 |
|
---|
16 | // Following details on number of partitions to use given in
|
---|
17 | // "Parallelized collections" section of:
|
---|
18 | // https://spark.apache.org/docs/2.0.1/programming-guide.html
|
---|
19 | //
|
---|
20 | // For a more detailed discussion see:
|
---|
21 | // http://blog.cloudera.com/blog/2015/03/how-to-tune-your-apache-spark-jobs-part-2/
|
---|
22 |
|
---|
23 | public static final int NUM_CORES = 6;
|
---|
24 | public static final int NUM_PARTITIONS = 2*NUM_CORES; // default would appear to be 2
|
---|
25 |
|
---|
26 | protected String _input_dir;
|
---|
27 | protected String _json_list_filename;
|
---|
28 | protected String _solr_url;
|
---|
29 | protected String _output_dir;
|
---|
30 |
|
---|
31 | protected int _verbosity;
|
---|
32 |
|
---|
33 | public ProcessForSolrIngest(String input_dir, String json_list_filename,
|
---|
34 | String solr_url, String output_dir, int verbosity)
|
---|
35 | {
|
---|
36 | _input_dir = input_dir;
|
---|
37 | _json_list_filename = (json_list_filename != null) ? json_list_filename : input_dir;
|
---|
38 |
|
---|
39 | _solr_url = solr_url;
|
---|
40 | _output_dir = output_dir;
|
---|
41 | _verbosity = verbosity;
|
---|
42 | }
|
---|
43 |
|
---|
44 | public void execPerVolume()
|
---|
45 | {
|
---|
46 | String spark_app_name = "[Per Volume] Extract Features: Process for Solr Ingest";
|
---|
47 | spark_app_name += " [" + _json_list_filename + "]";
|
---|
48 |
|
---|
49 | SparkConf conf = new SparkConf().setAppName(spark_app_name);
|
---|
50 | JavaSparkContext jsc = new JavaSparkContext(conf);
|
---|
51 |
|
---|
52 | if (_verbosity >= 2) {
|
---|
53 | System.out.println("Default Minimum Partions: " + jsc.defaultMinPartitions());
|
---|
54 | System.out.println("Default Parallelism: " + jsc.defaultParallelism());
|
---|
55 | }
|
---|
56 |
|
---|
57 | JavaRDD<String> json_list_data = jsc.textFile(_json_list_filename,NUM_PARTITIONS).cache();
|
---|
58 |
|
---|
59 | long num_volumes = json_list_data.count();
|
---|
60 | double per_vol = 100.0/(double)num_volumes;
|
---|
61 |
|
---|
62 | DoubleAccumulator progress_accum = jsc.sc().doubleAccumulator("Progress Percent");
|
---|
63 |
|
---|
64 | System.err.println();
|
---|
65 | System.err.println();
|
---|
66 | System.err.println();
|
---|
67 | System.err.println("****##### _input_dir = " + _input_dir);
|
---|
68 | System.err.println();
|
---|
69 | System.err.println();
|
---|
70 | System.err.println();
|
---|
71 |
|
---|
72 | PerVolumeJSON per_vol_json = new PerVolumeJSON(_input_dir,_solr_url,_output_dir,_verbosity, progress_accum,per_vol);
|
---|
73 |
|
---|
74 | json_list_data.foreach(per_vol_json);
|
---|
75 |
|
---|
76 | long num_ids = num_volumes;
|
---|
77 |
|
---|
78 | System.out.println("");
|
---|
79 | System.out.println("############");
|
---|
80 | System.out.println("# Number of volume ids: " + num_ids);
|
---|
81 | System.out.println("############");
|
---|
82 | System.out.println("");
|
---|
83 |
|
---|
84 | jsc.close();
|
---|
85 | }
|
---|
86 |
|
---|
87 | public void execPerPage()
|
---|
88 | {
|
---|
89 | String spark_app_name = "[Per Page] Extract Features: Process for Solr Ingest";
|
---|
90 | spark_app_name += " [" + _json_list_filename + "]";
|
---|
91 |
|
---|
92 | SparkConf conf = new SparkConf().setAppName(spark_app_name);
|
---|
93 | JavaSparkContext jsc = new JavaSparkContext(conf);
|
---|
94 |
|
---|
95 | if (_verbosity >= 2) {
|
---|
96 | System.out.println("Default Minimum Partions: " + jsc.defaultMinPartitions());
|
---|
97 | System.out.println("Default Parallelism: " + jsc.defaultParallelism());
|
---|
98 | }
|
---|
99 |
|
---|
100 | JavaRDD<String> json_list_data = jsc.textFile(_json_list_filename,NUM_PARTITIONS).cache();
|
---|
101 |
|
---|
102 | long num_volumes = json_list_data.count();
|
---|
103 | double per_vol = 100.0/(double)num_volumes;
|
---|
104 |
|
---|
105 | DoubleAccumulator progress_accum = jsc.sc().doubleAccumulator("Progress Percent");
|
---|
106 |
|
---|
107 | PagedJSON paged_json = new PagedJSON(_input_dir,_solr_url,_output_dir,_verbosity, progress_accum,per_vol);
|
---|
108 | JavaRDD<JSONObject> json_per_page_ids = json_list_data.flatMap(paged_json).cache();
|
---|
109 |
|
---|
110 | PagedJSONForeach paged_json_foreach = new PagedJSONForeach(_input_dir,_solr_url,_output_dir,_verbosity, progress_accum,per_vol);
|
---|
111 | json_per_page_ids.foreach(paged_json_foreach);
|
---|
112 |
|
---|
113 | /*
|
---|
114 | System.out.println("");
|
---|
115 | System.out.println("############");
|
---|
116 | System.out.println("# Progress Accumulator: " + progress_accum.value());
|
---|
117 | System.out.println("############");
|
---|
118 | System.out.println("");
|
---|
119 | */
|
---|
120 |
|
---|
121 | long num_page_ids = json_per_page_ids.count();
|
---|
122 |
|
---|
123 | System.out.println("");
|
---|
124 | System.out.println("############");
|
---|
125 | System.out.println("# Number of page ids: " + num_page_ids);
|
---|
126 | System.out.println("############");
|
---|
127 | System.out.println("");
|
---|
128 |
|
---|
129 | /*
|
---|
130 | if (_output_dir != null) {
|
---|
131 | String rdd_save_file = "rdd-solr-json-page-files";
|
---|
132 | json_ids.saveAsTextFile(rdd_save_file);
|
---|
133 | System.out.println("############");
|
---|
134 | System.out.println("# Saved RDD of Solr JSON page files, top-level, as:");
|
---|
135 | System.out.println("# " + rdd_save_file);
|
---|
136 | System.out.println("############");
|
---|
137 | System.out.println("");
|
---|
138 | }
|
---|
139 | */
|
---|
140 |
|
---|
141 | jsc.close();
|
---|
142 | }
|
---|
143 |
|
---|
144 |
|
---|
145 |
|
---|
146 |
|
---|
147 | public static void print_usage(HelpFormatter formatter, Options options)
|
---|
148 | {
|
---|
149 | formatter.printHelp("RUN.bash [options] input-dir json-filelist.txt", options);
|
---|
150 | }
|
---|
151 |
|
---|
152 | public static void main(String[] args) {
|
---|
153 | Options options = new Options();
|
---|
154 |
|
---|
155 | Option verbosity_opt = new Option("v", "verbosity", true,
|
---|
156 | "Set to control the level of debugging output [0=none, 1=some, 2=lots]");
|
---|
157 | verbosity_opt.setRequired(false);
|
---|
158 | options.addOption(verbosity_opt);
|
---|
159 |
|
---|
160 | Option output_dir_opt = new Option("o", "output-dir", true,
|
---|
161 | "If specified, save BZipped Solr JSON files to this directory");
|
---|
162 | output_dir_opt.setRequired(false);
|
---|
163 | options.addOption(output_dir_opt);
|
---|
164 |
|
---|
165 | Option solr_url_opt = new Option("u", "solr-url", true,
|
---|
166 | "If specified, the URL to post the Solr JSON data to");
|
---|
167 | solr_url_opt.setRequired(false);
|
---|
168 | options.addOption(solr_url_opt);
|
---|
169 |
|
---|
170 | Option read_only_opt = new Option("r", "read-only", false,
|
---|
171 | "Used to initiate a run where the files are all read in, but nothing is ingested/saved");
|
---|
172 | read_only_opt.setRequired(false);
|
---|
173 | options.addOption(read_only_opt);
|
---|
174 |
|
---|
175 | // Need to work with CLI v1.2 as this is the JAR that is bundled with Hadoop/Spark
|
---|
176 | CommandLineParser parser = new GnuParser();
|
---|
177 | //CommandLineParser parser = new DefaultParser(); // if working with CLI v1.3 and above
|
---|
178 |
|
---|
179 | HelpFormatter formatter = new HelpFormatter();
|
---|
180 | CommandLine cmd = null;
|
---|
181 |
|
---|
182 | try {
|
---|
183 | cmd = parser.parse(options, args);
|
---|
184 | }
|
---|
185 | catch (ParseException e) {
|
---|
186 | System.err.println(e.getMessage());
|
---|
187 | print_usage(formatter,options);
|
---|
188 | System.exit(1);
|
---|
189 | }
|
---|
190 |
|
---|
191 |
|
---|
192 | String verbosity_str = cmd.getOptionValue("verbosity","0");
|
---|
193 | int verbosity = Integer.parseInt(verbosity_str);
|
---|
194 |
|
---|
195 | String output_dir = cmd.getOptionValue("output-dir",null);
|
---|
196 | String solr_url = cmd.getOptionValue("solr-url",null);
|
---|
197 | boolean read_only = cmd.hasOption("read-only");
|
---|
198 |
|
---|
199 | String[] filtered_args = cmd.getArgs();
|
---|
200 |
|
---|
201 | if (filtered_args.length != 2) {
|
---|
202 | print_usage(formatter,options);
|
---|
203 | System.exit(1);
|
---|
204 | }
|
---|
205 |
|
---|
206 | if (!read_only && ((output_dir == null) && (solr_url==null))) {
|
---|
207 | System.err.println("Need to specify either --solr-url or --output-dir otherwise generated files are not ingested/saved");
|
---|
208 | print_usage(formatter,options);
|
---|
209 | System.exit(1);
|
---|
210 | }
|
---|
211 | if (read_only) {
|
---|
212 | // For this case, need to ensure solr-url and output-dir are null
|
---|
213 | output_dir = null;
|
---|
214 | solr_url = null;
|
---|
215 | }
|
---|
216 |
|
---|
217 | String input_dir = filtered_args[0];
|
---|
218 | String json_list_filename = filtered_args[1];
|
---|
219 |
|
---|
220 | ProcessForSolrIngest prep_for_ingest
|
---|
221 | = new ProcessForSolrIngest(input_dir,json_list_filename,solr_url,output_dir,verbosity);
|
---|
222 |
|
---|
223 | //prep_for_ingest.execPerVolume();
|
---|
224 | prep_for_ingest.execPerPage();
|
---|
225 | }
|
---|
226 | }
|
---|