Changeset 30942
- Timestamp:
- 2016-10-26T14:27:44+13:00 (7 years ago)
- Location:
- other-projects/hathitrust/solr-extracted-features/trunk/src/main/java/org/hathitrust
- Files:
-
- 2 edited
Legend:
- Unmodified
- Added
- Removed
-
other-projects/hathitrust/solr-extracted-features/trunk/src/main/java/org/hathitrust/PagedJSON.java
r30937 r30942 24 24 25 25 protected String _input_dir; 26 protected int _verbosity; 26 27 27 public PagedJSON(String input_dir )28 public PagedJSON(String input_dir, int verbosity) 28 29 { 29 30 _input_dir = input_dir; 31 _verbosity = verbosity; 30 32 } 31 33 … … 87 89 int ef_page_count = ef_features.getInt("pageCount"); 88 90 91 if (_verbosity >= 1) { 92 System.out.println("Processing: " + json_file_in); 93 System.out.println(" pageCount = " + ef_page_count); 94 } 95 89 96 JSONArray ef_pages = ef_features.getJSONArray("pages"); 90 97 int ef_num_pages = ef_pages.length(); -
other-projects/hathitrust/solr-extracted-features/trunk/src/main/java/org/hathitrust/PrepareForIngest.java
r30941 r30942 47 47 JavaRDD<String> json_list_data = jsc.textFile(_json_list_filename).cache(); 48 48 49 JavaRDD<String> json_ids = json_list_data.flatMap(new PagedJSON(_input_dir ));49 JavaRDD<String> json_ids = json_list_data.flatMap(new PagedJSON(_input_dir,_verbosity)); 50 50 51 51 … … 63 63 System.out.println(""); 64 64 System.out.println("############"); 65 System.out.println("# number of IDS: " + num_ids);65 System.out.println("# Number of page ids: " + num_ids); 66 66 System.out.println("############"); 67 67 System.out.println("");
Note:
See TracChangeset
for help on using the changeset viewer.