| 30 | public class Speed { |
| 31 | public static final int TRIALS = 20; |
| 32 | public static void main(String[] args) throws Exception { |
| 33 | String langname = args[0].substring(1); |
| 34 | String testFilename = args[1]; |
| 35 | LangDescriptor language = null; |
| 36 | for (int i = 0; i<languages.length; i++) { |
| 37 | if ( languages[i].name.equals(langname) ) { |
| 38 | language = languages[i]; |
| 39 | break; |
| 40 | } |
| 41 | } |
| 42 | if ( language==null ) { |
| 43 | System.err.println("Language "+langname+" unknown"); |
| 44 | return; |
| 45 | } |
| 46 | |
| 47 | // load all files up front |
| 48 | long load_start = System.nanoTime(); |
| 49 | List<String> allFiles = getFilenames(new File(language.corpusDir), language.fileRegex); |
| 50 | List<InputDocument> documents = Tool.load(allFiles, language); |
| 51 | long load_stop = System.nanoTime(); |
| 52 | long load_time = (load_stop-load_start)/1_000_000; |
| 53 | System.out.printf("Loaded %d files in %dms\n", documents.size(), load_time); |
| 54 | |
| 55 | final String path = new File(testFilename).getAbsolutePath(); |
| 56 | List<InputDocument> others = filter(documents, d -> !d.fileName.equals(path)); |
| 57 | List<InputDocument> excluded = filter(documents, d -> d.fileName.equals(path)); |
| 58 | assert others.size() == documents.size() - 1; |
| 59 | if ( excluded.size()==0 ) { |
| 60 | System.err.println("Doc not in corpus: "+path); |
| 61 | return; |
| 62 | } |
| 63 | InputDocument testDoc = excluded.get(0); |
| 64 | |
| 65 | List<Integer> training = new ArrayList<>(); |
| 66 | List<Integer> formatting = new ArrayList<>(); |
| 67 | for (int i = 1; i<=TRIALS; i++) { |
| 68 | Pair<Integer, Integer> timing = test(language, others, testDoc); |
| 69 | training.add(timing.a); |
| 70 | formatting.add(timing.b); |
| 71 | } |
| 72 | // drop first four |
| 73 | training = training.subList(5,training.size()); |
| 74 | formatting = formatting.subList(5,formatting.size()); |
| 75 | System.out.printf("median of [5:%d] training %dms\n", TRIALS-1, BuffUtils.median(training)); |
| 76 | System.out.printf("median of [5:%d] formatting %dms\n", TRIALS-1, BuffUtils.median(formatting)); |
| 77 | } |
| 78 | |
| 79 | public static Pair<Integer,Integer> test(LangDescriptor language, |
| 80 | List<InputDocument> others, |