|
| 1 | +package de.qaware.chronix.importer; |
| 2 | + |
| 3 | +import org.apache.commons.io.FileUtils; |
| 4 | +import org.apache.commons.io.IOUtils; |
| 5 | +import org.apache.commons.lang3.StringUtils; |
| 6 | +import org.slf4j.Logger; |
| 7 | +import org.slf4j.LoggerFactory; |
| 8 | + |
| 9 | +import java.io.*; |
| 10 | +import java.text.DecimalFormat; |
| 11 | +import java.text.Normalizer; |
| 12 | +import java.text.NumberFormat; |
| 13 | +import java.text.SimpleDateFormat; |
| 14 | +import java.time.Instant; |
| 15 | +import java.util.*; |
| 16 | +import java.util.concurrent.atomic.AtomicInteger; |
| 17 | +import java.util.function.BiConsumer; |
| 18 | +import java.util.zip.GZIPInputStream; |
| 19 | + |
| 20 | +/** |
| 21 | + * Created by f.lautenschlager on 10.06.2015. |
| 22 | + */ |
| 23 | +public class FileImporter { |
| 24 | + |
| 25 | + private static final Logger LOGGER = LoggerFactory.getLogger(FileImporter.class); |
| 26 | + |
| 27 | + private static final String METRICS_FILE_PATH = "metrics.csv"; |
| 28 | + |
| 29 | + private FileImporter() { |
| 30 | + |
| 31 | + } |
| 32 | + |
| 33 | + @SafeVarargs |
| 34 | + public static final Pair<Integer, Integer> importPoints(Map<Metadata, Pair<Instant, Instant>> points, File folder, BiConsumer<List<ImportPoint>, Metadata>... databases) { |
| 35 | + |
| 36 | + final AtomicInteger pointCounter = new AtomicInteger(0); |
| 37 | + final AtomicInteger tsCounter = new AtomicInteger(0); |
| 38 | + final File metricsFile = new File(METRICS_FILE_PATH); |
| 39 | + try { |
| 40 | + final FileWriter metricsFileWriter = new FileWriter(metricsFile); |
| 41 | + |
| 42 | + |
| 43 | + Collection<File> files = new ArrayList<>(); |
| 44 | + if (folder.isFile()) { |
| 45 | + files.add(folder); |
| 46 | + } else { |
| 47 | + files.addAll(FileUtils.listFiles(folder, new String[]{"gz", "csv"}, true)); |
| 48 | + } |
| 49 | + |
| 50 | + AtomicInteger counter = new AtomicInteger(0); |
| 51 | + |
| 52 | + files.parallelStream().forEach(file -> { |
| 53 | + boolean onlyMinusOne = true; |
| 54 | + SimpleDateFormat sdf = new SimpleDateFormat("dd.MM.yyyy HH:mm:ss.SSS"); |
| 55 | + |
| 56 | + |
| 57 | + NumberFormat nf = DecimalFormat.getInstance(Locale.ENGLISH); |
| 58 | + InputStream inputStream = null; |
| 59 | + BufferedReader reader = null; |
| 60 | + try { |
| 61 | + inputStream = new FileInputStream(file); |
| 62 | + |
| 63 | + if (file.getName().endsWith("gz")) { |
| 64 | + inputStream = new GZIPInputStream(inputStream); |
| 65 | + } |
| 66 | + reader = new BufferedReader(new InputStreamReader(inputStream)); |
| 67 | + |
| 68 | + //Read the first line |
| 69 | + String headerLine = reader.readLine(); |
| 70 | + |
| 71 | + if (headerLine == null || headerLine.isEmpty()) { |
| 72 | + boolean deleted = deleteFile(file, inputStream, reader); |
| 73 | + LOGGER.debug("File is empty {}. File {} removed {}", file.getName(), deleted); |
| 74 | + return; |
| 75 | + } |
| 76 | + |
| 77 | + |
| 78 | + //host _ process _ metricGroup |
| 79 | + String[] fileNameMetaData = file.getName().split("_"); |
| 80 | + //build meta data object |
| 81 | + String host = fileNameMetaData[0]; |
| 82 | + String process = fileNameMetaData[1]; |
| 83 | + String metricGroup = fileNameMetaData[2]; |
| 84 | + String measurement = file.getParentFile().getName(); |
| 85 | + |
| 86 | + |
| 87 | + String[] metrics = headerLine.split(";"); |
| 88 | + |
| 89 | + Map<Integer, Metadata> metadatas = new HashMap<>(metrics.length); |
| 90 | + |
| 91 | + |
| 92 | + for (int i = 1; i < metrics.length; i++) { |
| 93 | + String metric = metrics[i]; |
| 94 | + String metricOnlyAscii = Normalizer.normalize(metric, Normalizer.Form.NFD); |
| 95 | + metricOnlyAscii = metric.replaceAll("[^\\x00-\\x7F]", ""); |
| 96 | + Metadata metadata = new Metadata(host, process, metricGroup, metricOnlyAscii, measurement); |
| 97 | + |
| 98 | + //Check if meta data is completely set |
| 99 | + if (isEmpty(metadata)) { |
| 100 | + boolean deleted = deleteFile(file, inputStream, reader); |
| 101 | + LOGGER.info("Metadata contains empty values {}. File {} deleted {}", metadata, file.getName(), deleted); |
| 102 | + continue; |
| 103 | + } |
| 104 | + |
| 105 | + if (metadata.getMetric().equals(".*")) { |
| 106 | + boolean deleted = deleteFile(file, inputStream, reader); |
| 107 | + LOGGER.info("Metadata metric{}. File {} deleted {}", metadata.getMetric(), file.getName(), deleted); |
| 108 | + continue; |
| 109 | + } |
| 110 | + metadatas.put(i, metadata); |
| 111 | + tsCounter.incrementAndGet(); |
| 112 | + |
| 113 | + } |
| 114 | + |
| 115 | + Map<Integer, List<ImportPoint>> dataPoints = new HashMap<>(); |
| 116 | + |
| 117 | + String line; |
| 118 | + boolean instantdate = true; |
| 119 | + boolean onlyOnce = true; |
| 120 | + while ((line = reader.readLine()) != null) { |
| 121 | + String[] splits = line.split(";"); |
| 122 | + String date = splits[0]; |
| 123 | + |
| 124 | + if (onlyOnce) { |
| 125 | + try { |
| 126 | + Instant.parse(date); |
| 127 | + } catch (Exception e) { |
| 128 | + instantdate = false; |
| 129 | + } |
| 130 | + onlyOnce = false; |
| 131 | + } |
| 132 | + Instant dateObject; |
| 133 | + if (instantdate) { |
| 134 | + dateObject = Instant.parse(date); |
| 135 | + } else { |
| 136 | + dateObject = sdf.parse(date).toInstant(); |
| 137 | + } |
| 138 | + |
| 139 | + |
| 140 | + String[] values = splits; |
| 141 | + |
| 142 | + for (int column = 1; column < values.length; column++) { |
| 143 | + |
| 144 | + String value = values[column]; |
| 145 | + double numericValue = nf.parse(value).doubleValue(); |
| 146 | + |
| 147 | + ImportPoint point; |
| 148 | + if (instantdate) { |
| 149 | + point = new ImportPoint(dateObject, numericValue); |
| 150 | + } else { |
| 151 | + point = new ImportPoint(dateObject, numericValue); |
| 152 | + |
| 153 | + } |
| 154 | + |
| 155 | + if (!dataPoints.containsKey(column)) { |
| 156 | + dataPoints.put(column, new ArrayList<>()); |
| 157 | + } |
| 158 | + dataPoints.get(column).add(point); |
| 159 | + pointCounter.incrementAndGet(); |
| 160 | + } |
| 161 | + |
| 162 | + } |
| 163 | +/* |
| 164 | + if (onlyMinusOne) { |
| 165 | + pointCounter.addAndGet(-dataPoints.size()); |
| 166 | + tsCounter.decrementAndGet(); |
| 167 | +
|
| 168 | + //close all streams |
| 169 | + boolean deleted = deleteFile(file, inputStream, reader); |
| 170 | + LOGGER.info("{} contains only -1. Deleted {}", file.getName(), deleted); |
| 171 | +
|
| 172 | + return; |
| 173 | + } |
| 174 | + */ |
| 175 | + |
| 176 | + dataPoints.values().forEach(Collections::sort); |
| 177 | + |
| 178 | + IOUtils.closeQuietly(reader); |
| 179 | + IOUtils.closeQuietly(inputStream); |
| 180 | + |
| 181 | + dataPoints.forEach((key, value) -> { |
| 182 | + for (BiConsumer<List<ImportPoint>, Metadata> database : databases) { |
| 183 | + database.accept(value, metadatas.get(key)); |
| 184 | + } |
| 185 | + points.put(metadatas.get(key), Pair.of(value.get(0).getDate(), value.get(value.size() - 1).getDate())); |
| 186 | + //write the stats to the file |
| 187 | + Instant start = value.get(0).getDate(); |
| 188 | + Instant end = value.get(value.size() - 1).getDate(); |
| 189 | + |
| 190 | + try { |
| 191 | + writeStatsLine(metricsFileWriter, metadatas.get(key), start, end); |
| 192 | + } catch (IOException e) { |
| 193 | + LOGGER.error("Could not write stats line", e); |
| 194 | + } |
| 195 | + LOGGER.info("{} of {} time series imported", counter.incrementAndGet(), tsCounter.get()); |
| 196 | + }); |
| 197 | + |
| 198 | + |
| 199 | + } catch (Exception e) { |
| 200 | + LOGGER.info("Exception while reading points.", e); |
| 201 | + } finally { |
| 202 | + //close all streams |
| 203 | + IOUtils.closeQuietly(reader); |
| 204 | + IOUtils.closeQuietly(inputStream); |
| 205 | + } |
| 206 | + |
| 207 | + }); |
| 208 | + } catch (Exception e) { |
| 209 | + LOGGER.error("Exception occurred during reading points."); |
| 210 | + } |
| 211 | + return Pair.of(tsCounter.get(), pointCounter.get()); |
| 212 | + } |
| 213 | + |
| 214 | + private static void writeStatsLine(FileWriter metricsFile, Metadata metadata, Instant start, Instant end) throws IOException { |
| 215 | + //host:process:metric-group:metric:start:end |
| 216 | + StringBuilder line = new StringBuilder(); |
| 217 | + line.append(metadata.getHost()).append(";") |
| 218 | + .append(metadata.getProcess()).append(";") |
| 219 | + .append(metadata.getMetricGroup()).append(";") |
| 220 | + .append(metadata.getMetric()).append(";") |
| 221 | + .append(start).append(";") |
| 222 | + .append(end).append(";") |
| 223 | + .append("\n"); |
| 224 | + |
| 225 | + metricsFile.write(line.toString()); |
| 226 | + metricsFile.flush(); |
| 227 | + |
| 228 | + } |
| 229 | + |
| 230 | + private static boolean deleteFile(File file, InputStream inputStream, BufferedReader reader) { |
| 231 | + IOUtils.closeQuietly(reader); |
| 232 | + IOUtils.closeQuietly(inputStream); |
| 233 | + |
| 234 | + //remove file |
| 235 | + return file.delete(); |
| 236 | + } |
| 237 | + |
| 238 | + private static boolean isEmpty(Metadata metadata) { |
| 239 | + return empty(metadata.getMetric()) || empty(metadata.getMetricGroup()) || empty(metadata.getProcess()) || empty(metadata.getHost()); |
| 240 | + } |
| 241 | + |
| 242 | + private static boolean empty(String metric) { |
| 243 | + return StringUtils.isEmpty(metric); |
| 244 | + } |
| 245 | +} |
0 commit comments