use of org.apache.hudi.hadoop.FileStatusWithBootstrapBaseFile in project hudi by apache.
the class HoodieMergeOnReadTableInputFormat method createRealtimeFileStatusUnchecked.
/**
* Creates {@link RealtimeFileStatus} for the file-slice where base file is present
*/
private static RealtimeFileStatus createRealtimeFileStatusUnchecked(HoodieBaseFile baseFile, Stream<HoodieLogFile> logFiles, String basePath, Option<HoodieInstant> latestCompletedInstantOpt, Option<HoodieVirtualKeyInfo> virtualKeyInfoOpt) {
FileStatus baseFileStatus = getFileStatusUnchecked(baseFile);
List<HoodieLogFile> sortedLogFiles = logFiles.sorted(HoodieLogFile.getLogFileComparator()).collect(Collectors.toList());
try {
RealtimeFileStatus rtFileStatus = new RealtimeFileStatus(baseFileStatus, basePath, sortedLogFiles, false, virtualKeyInfoOpt);
if (latestCompletedInstantOpt.isPresent()) {
HoodieInstant latestCompletedInstant = latestCompletedInstantOpt.get();
checkState(latestCompletedInstant.isCompleted());
rtFileStatus.setMaxCommitTime(latestCompletedInstant.getTimestamp());
}
if (baseFileStatus instanceof LocatedFileStatusWithBootstrapBaseFile || baseFileStatus instanceof FileStatusWithBootstrapBaseFile) {
rtFileStatus.setBootStrapFileStatus(baseFileStatus);
}
return rtFileStatus;
} catch (IOException e) {
throw new HoodieIOException(String.format("Failed to init %s", RealtimeFileStatus.class.getSimpleName()), e);
}
}
use of org.apache.hudi.hadoop.FileStatusWithBootstrapBaseFile in project hudi by apache.
the class HoodieMergeOnReadTableInputFormat method collectAllIncrementalFiles.
private static List<FileStatus> collectAllIncrementalFiles(List<HoodieFileGroup> fileGroups, String maxCommitTime, String basePath, Map<String, FileStatus> candidateFileStatus, Option<HoodieVirtualKeyInfo> virtualKeyInfoOpt) {
List<FileStatus> result = new ArrayList<>();
fileGroups.stream().forEach(f -> {
try {
List<FileSlice> baseFiles = f.getAllFileSlices().filter(slice -> slice.getBaseFile().isPresent()).collect(Collectors.toList());
if (!baseFiles.isEmpty()) {
FileStatus baseFileStatus = HoodieInputFormatUtils.getFileStatus(baseFiles.get(0).getBaseFile().get());
String baseFilePath = baseFileStatus.getPath().toUri().toString();
if (!candidateFileStatus.containsKey(baseFilePath)) {
throw new HoodieException("Error obtaining fileStatus for file: " + baseFilePath);
}
List<HoodieLogFile> deltaLogFiles = f.getLatestFileSlice().get().getLogFiles().collect(Collectors.toList());
// We cannot use baseFileStatus.getPath() here, since baseFileStatus.getPath() missing file size information.
// So we use candidateFileStatus.get(baseFileStatus.getPath()) to get a correct path.
RealtimeFileStatus fileStatus = new RealtimeFileStatus(candidateFileStatus.get(baseFilePath), basePath, deltaLogFiles, true, virtualKeyInfoOpt);
fileStatus.setMaxCommitTime(maxCommitTime);
if (baseFileStatus instanceof LocatedFileStatusWithBootstrapBaseFile || baseFileStatus instanceof FileStatusWithBootstrapBaseFile) {
fileStatus.setBootStrapFileStatus(baseFileStatus);
}
result.add(fileStatus);
}
// add file group which has only logs.
if (f.getLatestFileSlice().isPresent() && baseFiles.isEmpty()) {
List<FileStatus> logFileStatus = f.getLatestFileSlice().get().getLogFiles().map(logFile -> logFile.getFileStatus()).collect(Collectors.toList());
if (logFileStatus.size() > 0) {
List<HoodieLogFile> deltaLogFiles = logFileStatus.stream().map(l -> new HoodieLogFile(l.getPath(), l.getLen())).collect(Collectors.toList());
RealtimeFileStatus fileStatus = new RealtimeFileStatus(logFileStatus.get(0), basePath, deltaLogFiles, true, virtualKeyInfoOpt);
fileStatus.setMaxCommitTime(maxCommitTime);
result.add(fileStatus);
}
}
} catch (IOException e) {
throw new HoodieException("Error obtaining data file/log file grouping ", e);
}
});
return result;
}
Aggregations