Changeset 31260 for other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/ProcessForPOSCount.java

Timestamp:

2016-12-21T00:12:10+13:00 (7 years ago)

Author:

davidb

Message:

Language counting

File:

• other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/ProcessForPOSCount.java (modified) (2 diffs)

other-projects/hathitrust/wcsa/extracted-features-solr/trunk/solr-ingest/src/main/java/org/hathitrust/extractedfeatures/ProcessForPOSCount.java

@@ -45 +45 @@
     protected String generateSparkAppName(String exec_mode)
     {
-        String spark_app_name = "[" + exec_mode + "] Extracted Features: Process for POS";
+        String spark_app_name = "[" + exec_mode + "] Extracted Features: Process for POS Count";
         spark_app_name += " [" + _json_list_filename + "]";
     
@@ -79 +79 @@
         //boolean icu_tokenize = Boolean.getBoolean("wcsa-ef-ingest.icu-tokenize");
         
-        PerVolumePOSStreamFlatmap paged_solr_wordfreq_flatmap 
+        PerVolumePOSStreamFlatmap paged_solr_posfreq_flatmap 
             = new PerVolumePOSStreamFlatmap(_input_dir,_verbosity, 
                                      per_vol_progress_accum,per_vol,
                                      strict_file_io);
-        JavaRDD<String> pos_list = json_list_data.flatMap(paged_solr_wordfreq_flatmap); 
+        JavaRDD<String> pos_list = json_list_data.flatMap(paged_solr_posfreq_flatmap); 
         pos_list.setName("pos-stream");