@Test
public void testTarget() throws Exception {
Map<String, Object> mb =
ImmutableMap.<String, Object>builder()
.put("inbound.column-names", ImmutableList.of("a", "e", "c"))
.put("inbound.categorical-columns", ImmutableList.of("a", "c"))
.put("inbound.target-column", "c")
.build();
Config conf = overlayConfigOnDefault(mb);
InboundSettings settings = InboundSettings.create(conf);
assertEquals(2, settings.getTargetColumn().intValue());
}
@Override
protected void doPost(HttpServletRequest request, HttpServletResponse response)
throws IOException {
RDFGenerationManager generationManager = getGenerationManager();
Generation generation = generationManager.getCurrentGeneration();
if (generation == null) {
response.sendError(HttpServletResponse.SC_SERVICE_UNAVAILABLE);
return;
}
InboundSettings inboundSettings = getInboundSettings();
TreeBasedClassifier forest = generation.getForest();
Map<Integer, BiMap<String, Integer>> columnToCategoryNameToIDMapping =
generation.getColumnToCategoryNameToIDMapping();
int totalColumns = getTotalColumns();
for (CharSequence line : CharStreams.readLines(request.getReader())) {
generationManager.append(line);
String[] tokens = DelimitedDataUtils.decode(line);
if (tokens.length != totalColumns) {
response.sendError(HttpServletResponse.SC_BAD_REQUEST, "Wrong column count");
return;
}
Feature target = null;
Feature[] features = new Feature[totalColumns]; // Too big by 1 but makes math easier
try {
for (int col = 0; col < features.length; col++) {
if (col == inboundSettings.getTargetColumn()) {
target = buildFeature(col, tokens[col], columnToCategoryNameToIDMapping);
features[col] = IgnoredFeature.INSTANCE;
} else {
features[col] = buildFeature(col, tokens[col], columnToCategoryNameToIDMapping);
}
}
} catch (IllegalArgumentException iae) {
response.sendError(HttpServletResponse.SC_BAD_REQUEST, "Bad input line");
return;
}
Preconditions.checkNotNull(target);
Example example = new Example(target, features);
forest.update(example);
}
}
private Feature buildFeature(
int columnNumber,
String token,
Map<Integer, BiMap<String, Integer>> columnToCategoryNameToIDMapping) {
InboundSettings inboundSettings = getInboundSettings();
if (inboundSettings.isNumeric(columnNumber)) {
return NumericFeature.forValue(Float.parseFloat(token));
}
if (inboundSettings.isCategorical(columnNumber)) {
return CategoricalFeature.forValue(
ClassifyServlet.categoricalFromString(
columnNumber, token, columnToCategoryNameToIDMapping));
}
return IgnoredFeature.INSTANCE;
}
@Test
public void testContinuousSettings() throws Exception {
Map<String, Object> mb =
ImmutableMap.<String, Object>builder()
.put("inbound.column-names", ImmutableList.of("a", "b", "c", "d", "e"))
.put("inbound.id-columns", ImmutableList.of("a"))
.put("inbound.numeric-columns", ImmutableList.of(1, "e"))
.put("inbound.ignored-columns", ImmutableList.of("c"))
.build();
Config conf = overlayConfigOnDefault(mb);
InboundSettings settings = InboundSettings.create(conf);
assertEquals(ImmutableList.of(0), ImmutableList.copyOf(settings.getIdColumns()));
assertEquals(ImmutableList.of(3), ImmutableList.copyOf(settings.getCategoricalColumns()));
assertEquals(ImmutableList.of(1, 4), ImmutableList.copyOf(settings.getNumericColumns()));
assertEquals(ImmutableList.of(2), ImmutableList.copyOf(settings.getIgnoredColumns()));
assertNull(settings.getTargetColumn());
/*
Spec spec = settings.getSpec();
Assert.assertEquals(5, spec.size());
Assert.assertEquals(DataType.STRING, spec.getField(0).spec().getDataType());
Assert.assertEquals(DataType.DOUBLE, spec.getField(1).spec().getDataType());
Assert.assertEquals(DataType.STRING, spec.getField(2).spec().getDataType());
*/
}
@Test
public void testDefaults() throws Exception {
Map<String, Object> mb =
ImmutableMap.<String, Object>builder()
.put("inbound.categorical-columns", ImmutableList.of())
.build();
Config conf = overlayConfigOnDefault(mb);
InboundSettings settings = InboundSettings.create(conf);
assertTrue(settings.getIdColumns().isEmpty());
assertTrue(settings.getIgnoredColumns().isEmpty());
// assertNull(settings.getSpec());
assertTrue(settings.getCategoricalColumns().isEmpty());
assertTrue(settings.getNumericColumns().isEmpty());
assertEquals(Integer.valueOf(1729), settings.getLookupFunction().apply(1729));
assertNull(settings.getTargetColumn());
}
@Test(expected = IllegalArgumentException.class)
public void testUnknownColumn() {
InboundSettings settings = InboundSettings.create(ConfigUtils.getDefaultConfig());
settings.getLookupFunction().apply("foo");
}
@Test(expected = IllegalArgumentException.class)
public void testNoSetting() {
InboundSettings.create(ConfigUtils.getDefaultConfig());
}