1 | package org.hathitrust.extractedfeatures;
|
---|
2 |
|
---|
3 | import java.io.BufferedInputStream;
|
---|
4 | import java.io.FileInputStream;
|
---|
5 | import java.io.FileNotFoundException;
|
---|
6 | import java.io.IOException;
|
---|
7 | import java.io.Serializable;
|
---|
8 | import java.util.ArrayList;
|
---|
9 |
|
---|
10 | import org.apache.commons.cli.*;
|
---|
11 | import org.apache.hadoop.io.Text;
|
---|
12 | import org.apache.spark.api.java.*;
|
---|
13 | import org.apache.spark.util.DoubleAccumulator;
|
---|
14 | import org.apache.spark.util.LongAccumulator;
|
---|
15 | import org.hathitrust.extractedfeatures.PerPageJSONFlatmap;
|
---|
16 | import org.json.JSONObject;
|
---|
17 | import org.apache.spark.SparkConf;
|
---|
18 |
|
---|
19 | public class ProcessForSolrIngestJSONFilelist implements Serializable
|
---|
20 | {
|
---|
21 | private static final long serialVersionUID = 1L;
|
---|
22 |
|
---|
23 | protected static final int DEFAULT_NUM_CORES = 10;
|
---|
24 | protected static final int MINIMUM_NUM_PARTITIONS = 10*DEFAULT_NUM_CORES;
|
---|
25 |
|
---|
26 | protected static final int DEFAULT_FILES_PER_PARTITION = 3000;
|
---|
27 |
|
---|
28 | protected String _input_dir;
|
---|
29 | protected String _json_list_filename;
|
---|
30 |
|
---|
31 | protected String _solr_base_url;
|
---|
32 | protected String _solr_collection;
|
---|
33 |
|
---|
34 | protected String _whitelist_filename;
|
---|
35 | protected String _langmap_directory;
|
---|
36 |
|
---|
37 | //protected String _solr_url;
|
---|
38 | protected String _output_dir;
|
---|
39 |
|
---|
40 | protected int _verbosity;
|
---|
41 |
|
---|
42 | public ProcessForSolrIngestJSONFilelist(String input_dir, String json_list_filename,
|
---|
43 | String solr_collection, String solr_base_url,
|
---|
44 | String output_dir, int verbosity)
|
---|
45 | {
|
---|
46 | _input_dir = input_dir;
|
---|
47 | _json_list_filename = json_list_filename;
|
---|
48 |
|
---|
49 | _solr_collection = solr_collection;
|
---|
50 | _solr_base_url = solr_base_url;
|
---|
51 |
|
---|
52 | boolean use_whitelist = Boolean.getBoolean("wcsa-ef-ingest.use-whitelist");
|
---|
53 | _whitelist_filename = (use_whitelist) ? System.getProperty("wcsa-ef-ingest.whitelist-filename") : null;
|
---|
54 |
|
---|
55 | boolean use_langmap = Boolean.getBoolean("wcsa-ef-ingest.use-langmap");
|
---|
56 | _langmap_directory = (use_langmap) ? System.getProperty("wcsa-ef-ingest.langmap-directory") : null;
|
---|
57 |
|
---|
58 | _output_dir = output_dir;
|
---|
59 | _verbosity = verbosity;
|
---|
60 | }
|
---|
61 |
|
---|
62 | protected String generateSparkAppName(String exec_mode)
|
---|
63 | {
|
---|
64 | String spark_app_name = "[" + exec_mode + "] Extract Features: Process for Solr Ingest from JSON-filelist";
|
---|
65 | spark_app_name += " [" + _solr_collection + "]";
|
---|
66 |
|
---|
67 | if (_solr_base_url != null) {
|
---|
68 | spark_app_name += " solr_base_url=" + _solr_base_url;
|
---|
69 | }
|
---|
70 |
|
---|
71 | if (_output_dir != null) {
|
---|
72 | spark_app_name += " output_dir=" + _output_dir;
|
---|
73 | }
|
---|
74 |
|
---|
75 | return spark_app_name;
|
---|
76 | }
|
---|
77 |
|
---|
78 | public ArrayList<String> extrapolateSolrEndpoints(String solr_collection)
|
---|
79 | {
|
---|
80 | ArrayList<String> solr_endpoints = new ArrayList<String>();
|
---|
81 |
|
---|
82 | if (_solr_base_url != null) {
|
---|
83 | String solr_url = _solr_base_url + "/" + solr_collection + "/update";
|
---|
84 |
|
---|
85 | String solr_cloud_nodes = System.getProperty("wcsa-ef-ingest.solr-cloud-nodes",null);
|
---|
86 | if (solr_cloud_nodes != null) {
|
---|
87 | String [] cloud_nodes = solr_cloud_nodes.split(",");
|
---|
88 | for (String cn : cloud_nodes) {
|
---|
89 | String solr_endpoint = solr_url.replaceFirst("//.*?:\\d+/", "//"+cn+"/");
|
---|
90 | solr_endpoints.add(solr_endpoint);
|
---|
91 | }
|
---|
92 | }
|
---|
93 | else {
|
---|
94 | solr_endpoints.add(solr_url);
|
---|
95 | }
|
---|
96 | }
|
---|
97 |
|
---|
98 | return solr_endpoints;
|
---|
99 | }
|
---|
100 |
|
---|
101 | public void execPerVolumeJSONFilelist()
|
---|
102 | {
|
---|
103 | String spark_app_name = generateSparkAppName("Per Volume");
|
---|
104 |
|
---|
105 | SparkConf spark_conf = new SparkConf().setAppName(spark_app_name);
|
---|
106 | JavaSparkContext jsc = new JavaSparkContext(spark_conf);
|
---|
107 |
|
---|
108 | //String filename_root = _json_list_filename.replaceAll(".*/","").replaceAll("\\..*$","");
|
---|
109 |
|
---|
110 | //int num_partitions = Integer.getInteger("wcsa-ef-ingest.num-partitions", DEFAULT_NUM_PARTITIONS);
|
---|
111 | int files_per_partition = Integer.getInteger("wcsa-ef-ingest.files-per-partition", DEFAULT_FILES_PER_PARTITION);
|
---|
112 |
|
---|
113 | JavaRDD<String> json_list_data = jsc.textFile(_json_list_filename).cache();
|
---|
114 | json_list_data.setName("JSON-file-list");
|
---|
115 |
|
---|
116 | long num_volumes = json_list_data.count();
|
---|
117 | double per_vol = 100.0/(double)num_volumes;
|
---|
118 |
|
---|
119 | int num_partitions = (int)(num_volumes/files_per_partition)+1;
|
---|
120 | if (num_partitions < MINIMUM_NUM_PARTITIONS) {
|
---|
121 | num_partitions = MINIMUM_NUM_PARTITIONS;
|
---|
122 | }
|
---|
123 | JavaRDD<String> json_list_data_rp = json_list_data.repartition(num_partitions);
|
---|
124 | json_list_data_rp.setName("JSON-file-list--repartitioned");
|
---|
125 |
|
---|
126 | DoubleAccumulator progress_accum = jsc.sc().doubleAccumulator("Per Volume Progress Percent");
|
---|
127 |
|
---|
128 | boolean icu_tokenize = Boolean.getBoolean("wcsa-ef-ingest.icu-tokenize");
|
---|
129 | boolean strict_file_io = Boolean.getBoolean("wcsa-ef-ingest.strict-file-io");
|
---|
130 |
|
---|
131 | ArrayList<String> solr_endpoints = extrapolateSolrEndpoints(_solr_collection);
|
---|
132 |
|
---|
133 | PerVolumeJSONList per_vol_json = new PerVolumeJSONList(_input_dir,_whitelist_filename, _langmap_directory,
|
---|
134 | solr_endpoints,_output_dir,_verbosity,
|
---|
135 | icu_tokenize,strict_file_io);
|
---|
136 |
|
---|
137 | JavaRDD<Integer> per_volume_page_count = json_list_data_rp.map(per_vol_json);
|
---|
138 | per_volume_page_count.setName("volume-page-counts");
|
---|
139 |
|
---|
140 | //Integer total_page_count = volume_page_counts.reduce((a, b) -> a + b);
|
---|
141 | long num_vol_ids = per_volume_page_count.count();
|
---|
142 |
|
---|
143 | System.out.println("");
|
---|
144 | System.out.println("############");
|
---|
145 | System.out.println("# Number of volume ids: " + num_vol_ids);
|
---|
146 | System.out.println("############");
|
---|
147 | System.out.println("");
|
---|
148 |
|
---|
149 | jsc.close();
|
---|
150 | }
|
---|
151 |
|
---|
152 | public static void print_usage(HelpFormatter formatter, Options options)
|
---|
153 | {
|
---|
154 | formatter.printHelp("RUN.bash [options] input-dir json-filelist.txt solr-collection", options);
|
---|
155 | }
|
---|
156 |
|
---|
157 | public static void main(String[] args) {
|
---|
158 | Options options = new Options();
|
---|
159 |
|
---|
160 | Option verbosity_opt = new Option("v", "verbosity", true,
|
---|
161 | "Set to control the level of debugging output [0=none, 1=some, 2=lots]");
|
---|
162 | verbosity_opt.setRequired(false);
|
---|
163 | options.addOption(verbosity_opt);
|
---|
164 |
|
---|
165 | Option properties_opt = new Option("p", "properties", true,
|
---|
166 | "Read in the specified Java properties file");
|
---|
167 | properties_opt.setRequired(false);
|
---|
168 | options.addOption(properties_opt);
|
---|
169 |
|
---|
170 | Option output_dir_opt = new Option("o", "output-dir", true,
|
---|
171 | "If specified, save BZipped Solr JSON files to this directory");
|
---|
172 | output_dir_opt.setRequired(false);
|
---|
173 | options.addOption(output_dir_opt);
|
---|
174 |
|
---|
175 | Option solr_base_url_opt = new Option("u", "solr-base-url", true,
|
---|
176 | "If specified, the base URL to post the Solr JSON data to");
|
---|
177 | solr_base_url_opt.setRequired(false);
|
---|
178 | options.addOption(solr_base_url_opt);
|
---|
179 |
|
---|
180 | Option read_only_opt = new Option("r", "read-only", false,
|
---|
181 | "Used to initiate a run where the files are all read in, but nothing is ingested/saved");
|
---|
182 | read_only_opt.setRequired(false);
|
---|
183 | options.addOption(read_only_opt);
|
---|
184 |
|
---|
185 | // Need to work with CLI v1.2 as this is the JAR that is bundled with Hadoop/Spark
|
---|
186 | CommandLineParser parser = new GnuParser();
|
---|
187 | //CommandLineParser parser = new DefaultParser(); // if working with CLI v1.3 and above
|
---|
188 |
|
---|
189 | HelpFormatter formatter = new HelpFormatter();
|
---|
190 | CommandLine cmd = null;
|
---|
191 |
|
---|
192 | try {
|
---|
193 | cmd = parser.parse(options, args);
|
---|
194 | }
|
---|
195 | catch (ParseException e) {
|
---|
196 | System.err.println(e.getMessage());
|
---|
197 | print_usage(formatter,options);
|
---|
198 | System.exit(1);
|
---|
199 | }
|
---|
200 |
|
---|
201 |
|
---|
202 | String verbosity_str = cmd.getOptionValue("verbosity","1");
|
---|
203 | int verbosity = Integer.parseInt(verbosity_str);
|
---|
204 |
|
---|
205 | String property_filename = cmd.getOptionValue("properties",null);
|
---|
206 |
|
---|
207 | String output_dir = cmd.getOptionValue("output-dir",null);
|
---|
208 | String solr_base_url = cmd.getOptionValue("solr-base-url",null);
|
---|
209 | boolean read_only = cmd.hasOption("read-only");
|
---|
210 |
|
---|
211 | String[] filtered_args = cmd.getArgs();
|
---|
212 |
|
---|
213 | if (filtered_args.length != 3) {
|
---|
214 | print_usage(formatter,options);
|
---|
215 | System.exit(1);
|
---|
216 | }
|
---|
217 |
|
---|
218 | if (property_filename != null) {
|
---|
219 | try {
|
---|
220 | FileInputStream fis = new FileInputStream(property_filename);
|
---|
221 | BufferedInputStream bis = new BufferedInputStream(fis);
|
---|
222 |
|
---|
223 | System.getProperties().load(bis);
|
---|
224 | }
|
---|
225 | catch (FileNotFoundException e) {
|
---|
226 | // TODO Auto-generated catch block
|
---|
227 | e.printStackTrace();
|
---|
228 | System.err.println("File not found: '" + property_filename + "'. Skipping property file read");
|
---|
229 | }
|
---|
230 | catch (IOException e) {
|
---|
231 | System.err.println("IO Exception for: '" + property_filename + "'. Malformed syntax? Skipping property file read");
|
---|
232 | }
|
---|
233 | }
|
---|
234 |
|
---|
235 | if (!read_only && ((output_dir == null) && (solr_base_url==null))) {
|
---|
236 | System.err.println("Need to specify either --solr-base-url or --output-dir otherwise generated files are not ingested/saved");
|
---|
237 | print_usage(formatter,options);
|
---|
238 | System.exit(1);
|
---|
239 | }
|
---|
240 | if (read_only) {
|
---|
241 | // For this case, need to ensure solr-url and output-dir are null
|
---|
242 | output_dir = null;
|
---|
243 | solr_base_url = null;
|
---|
244 | }
|
---|
245 |
|
---|
246 | String input_dir = filtered_args[0];
|
---|
247 | String json_file_list = filtered_args[1];
|
---|
248 | String solr_collection = filtered_args[2];
|
---|
249 |
|
---|
250 | ProcessForSolrIngestJSONFilelist prep_for_ingest
|
---|
251 | = new ProcessForSolrIngestJSONFilelist(input_dir,json_file_list,
|
---|
252 | solr_collection,solr_base_url,
|
---|
253 | output_dir,verbosity);
|
---|
254 |
|
---|
255 | prep_for_ingest.execPerVolumeJSONFilelist();
|
---|
256 |
|
---|
257 | }
|
---|
258 | }
|
---|