@Override
protected void loadRecentModel(long mostRecentModelGeneration) throws IOException {
if (mostRecentModelGeneration <= modelGeneration) {
return;
}
if (modelGeneration == NO_GENERATION) {
log.info("Most recent generation {} is the first available one", mostRecentModelGeneration);
} else {
log.info(
"Most recent generation {} is newer than current {}",
mostRecentModelGeneration,
modelGeneration);
}
File modelPMMLFile = File.createTempFile("model-", ".pmml.gz");
modelPMMLFile.deleteOnExit();
IOUtils.delete(modelPMMLFile);
Config config = ConfigUtils.getDefaultConfig();
String instanceDir = config.getString("model.instance-dir");
String generationPrefix =
Namespaces.getInstanceGenerationPrefix(instanceDir, mostRecentModelGeneration);
String modelPMMLKey = generationPrefix + "model.pmml.gz";
Store.get().download(modelPMMLKey, modelPMMLFile);
log.info("Loading model description from {}", modelPMMLKey);
Pair<DecisionForest, Map<Integer, BiMap<String, Integer>>> forestAndCatalog =
DecisionForestPMML.read(modelPMMLFile);
IOUtils.delete(modelPMMLFile);
log.info("Loaded model description");
modelGeneration = mostRecentModelGeneration;
currentModel = new Generation(forestAndCatalog.getFirst(), forestAndCatalog.getSecond());
}
@Override
protected void runSteps() throws IOException {
String instanceDir = getInstanceDir();
int generationID = getGenerationID();
String generationPrefix = Namespaces.getInstanceGenerationPrefix(instanceDir, generationID);
int lastGenerationID = getLastGenerationID();
File currentInputDir = Files.createTempDir();
currentInputDir.deleteOnExit();
File tempOutDir = Files.createTempDir();
tempOutDir.deleteOnExit();
try {
Store store = Store.get();
store.downloadDirectory(generationPrefix + "inbound/", currentInputDir);
if (lastGenerationID >= 0) {
store.downloadDirectory(
Namespaces.getInstanceGenerationPrefix(instanceDir, lastGenerationID) + "input/",
currentInputDir);
}
List<Example> examples = new ArrayList<>();
Map<Integer, BiMap<String, Integer>> columnToCategoryNameToIDMapping = new HashMap<>();
new ReadInputs(currentInputDir, examples, columnToCategoryNameToIDMapping).call();
DecisionForest forest = DecisionForest.fromExamplesWithDefault(examples);
DecisionForestPMML.write(
new File(tempOutDir, "model.pmml.gz"), forest, columnToCategoryNameToIDMapping);
store.uploadDirectory(generationPrefix + "input/", currentInputDir, false);
store.uploadDirectory(generationPrefix, tempOutDir, false);
} finally {
IOUtils.deleteRecursively(currentInputDir);
IOUtils.deleteRecursively(tempOutDir);
}
}
