fix (core): Improve Tika Test Coverage by comparing with actual full TTL

java/dev/enola/format/tika/BUILD

@@ -49,6 +49,7 @@ junit_tests(
  ":tika",
  "//java/dev/enola/common/io",
  "//java/dev/enola/thing:thing_java",
+ "//java/dev/enola/thing/testlib",
  "//test",
  ],
 )

java/dev/enola/format/tika/TikaThingConverter.java

@@ -80,7 +80,8 @@ public boolean convertInto(ReadableResource resource, ThingsBuilder thingsBuilde
  parser.parse(is, handler, metadata, parseContext);
  var thing = thingsBuilder.getBuilder(resource.uri().toString());
  convertMetadata(metadata, thing);
- thing.set("https://enola.dev/content-as-text", sw.toString());
+ var text = sw.toString().trim();
+ if (!text.isEmpty()) thing.set("https://enola.dev/content-as-text", text);
  return true;
 
  } catch (TikaException | SAXException e) {

java/dev/enola/format/tika/TikaThingConverterTest.java

@@ -23,6 +23,7 @@
 import dev.enola.common.io.resource.EmptyResource;
 import dev.enola.thing.Thing;
 import dev.enola.thing.repo.ThingsBuilder;
+import dev.enola.thing.testlib.ThingsSubject;
 
 import org.junit.Test;
 
@@ -58,14 +59,19 @@ public void png() throws IOException {
  // TODO @Test public void epubEBook() throws IOException {
 
  private void check(String classpath) throws IOException {
+ var name = "classpath:/" + classpath;
+
  var tb = new ThingsBuilder();
  var c = new TikaThingConverter(new ClasspathResource.Provider());
 
- var r = c.convertInto(URI.create("classpath:/" + classpath), tb);
+ var r = c.convertInto(URI.create(name), tb);
  assertThat(r).isTrue();
+
  assertThat(tb.builders()).hasSize(1);
  var thing = tb.builders().iterator().next().build();
  checkThatAllPredicatesAreAbsoluteURIs(thing);
+
+ ThingsSubject.assertThat(tb).isEqualTo(name + ".ttl");
  }
 
  private void checkThatAllPredicatesAreAbsoluteURIs(Thing thing) {

java/dev/enola/thing/testlib/BUILD

@@ -26,6 +26,7 @@ java_library(
  visibility = ["//:__subpackages__"],
  deps = [
  "//java/dev/enola/common/io",
+ "//java/dev/enola/common/io/testlib",
  "//java/dev/enola/rdf/io",
  "//java/dev/enola/thing:thing_java",
  "@maven//:com_google_errorprone_error_prone_annotations",

java/dev/enola/thing/testlib/ThingsSubject.java

@@ -24,16 +24,22 @@
 import com.google.common.truth.Subject;
 
 import dev.enola.common.io.resource.ClasspathResource;
+import dev.enola.common.io.resource.MemoryResource;
 import dev.enola.common.io.resource.ResourceProvider;
+import dev.enola.common.io.testlib.ResourceSubject;
 import dev.enola.rdf.io.JavaThingRdfConverter;
 import dev.enola.rdf.io.RdfReaderConverter;
+import dev.enola.rdf.io.RdfWriterConverter;
 import dev.enola.thing.repo.ThingRepository;
 
 import org.eclipse.rdf4j.model.Model;
 import org.eclipse.rdf4j.model.impl.DynamicModel;
 import org.eclipse.rdf4j.model.impl.LinkedHashModelFactory;
+import org.eclipse.rdf4j.model.util.Models;
 
-final class ThingsSubject extends Subject {
+import java.io.IOException;
+
+public final class ThingsSubject extends Subject {
 
  // TODO add assertThat(Thing actual) - with a SingleThingRepository ?
 
@@ -50,17 +56,22 @@ public static Factory<ThingsSubject, ThingRepository> resources() {
  private final Model actualModel;
  private final ResourceProvider rp = new ClasspathResource.Provider();
  private final RdfReaderConverter rdfReaderConverter = new RdfReaderConverter(rp);
- private final JavaThingRdfConverter javaThingRdfConverter = new JavaThingRdfConverter();
+ private final RdfWriterConverter rdfWriterConverter = new RdfWriterConverter();
 
  public ThingsSubject(FailureMetadata metadata, ThingRepository actual) {
  super(metadata, actual);
+ JavaThingRdfConverter javaThingRdfConverter = new JavaThingRdfConverter();
  actualModel = javaThingRdfConverter.convert(Streams.stream(actual.list()));
  }
 
- public void isEqualTo(String classpathResourcePath) {
+ public void isEqualTo(String classpathResourcePath) throws IOException {
  var resource = rp.getReadableResource(classpathResourcePath);
  if (resource == null) throw new IllegalArgumentException(classpathResourcePath);
  var expectedModel = rdfReaderConverter.convert(resource).orElse(EMPTY_MODEL);
- ModelSubject.assertThat(actualModel).isEqualTo(expectedModel);
+ if (!Models.isomorphic(actualModel, expectedModel)) {
+ var actualResource = new MemoryResource(resource.mediaType());
+ rdfWriterConverter.convertInto(actualModel, actualResource);
+ ResourceSubject.assertThat(actualResource).containsCharsOf(resource);
+ }
  }
 }

java/dev/enola/thing/testlib/ThingsSubjectTest.java

@@ -22,16 +22,18 @@
 
 import org.junit.Test;
 
+import java.io.IOException;
+
 public class ThingsSubjectTest {
 
  @Test
- public void empty() {
+ public void empty() throws IOException {
  ThingsRepository r = new ThingsBuilder();
  ThingsSubject.assertThat(r).isEqualTo("classpath:/empty.yaml");
  }
 
  @Test
- public void ttl() {
+ public void ttl() throws IOException {
  ThingsBuilder r = new ThingsBuilder();
  r.getBuilder("https://example.org/greeting1")
  .set("https://example.org/message", "hello, world");

test-cli.bash

@@ -20,7 +20,6 @@ set -euox pipefail
 # This script tests Enola CLI invocations.
 # See also EnolaCLITest
 
-./enola get --load test/test.html enola:/inline
-./enola -v rosetta --in test/test.html --out="fd:2?mediaType=text/turtle"
+# ...
 
 # PS: Update tika.md with anything (of interest) added here

test/test.html

@@ -6,11 +6,6 @@
  <meta name="author" content="The Enola <https://enola.dev> Authors" />
  <meta name="description" content="This HTML is used to test e.g. the TikaThingConverter." />
 </head>
-<body>
- <p>hello, world</p>
-
- <a href="test.html">Recursive Self Link</a>
-
- <a href="https://www.vorburger.ch">Vorburger Link</a>
+<body><p>hello, world</p><a href="test.html">Recursive Self Link</a><a href="https://www.vorburger.ch">Vorburger</a>
 </body>
 </html>

test/test.html.ttl

@@ -0,0 +1,8 @@
+@base <memory:1> .
+
+<classpath:/test.html> <http://purl.org/dc/elements/1.1/title> "Test HTML Title";
+ <http://purl.org/dc/elements/1.1/language> "en";
+ <http://purl.org/dc/elements/1.1/description> "This HTML is used to test e.g. the TikaThingConverter.";
+ <https://enola.dev/mediaType> "text/html; charset=UTF-8";
+ <http://purl.org/dc/elements/1.1/creator> "The Enola <https://enola.dev> Authors";
+ <https://enola.dev/content-as-text> "hello, world\nRecursive Self LinkVorburger".

test/test.png.ttl

@@ -0,0 +1,23 @@
+<classpath:/test.png> <https://enola.dev/tika/Chroma%20ColorSpaceType> "RGB";
+ <https://enola.dev/tika/Compression%20Lossless> "true";
+ <https://enola.dev/tika/Data%20SampleFormat> "UnsignedIntegral";
+ <https://enola.dev/tika/Chroma%20NumChannels> "4";
+ <https://enola.dev/tika/Transparency%20Alpha> "nonpremultipled";
+ <https://enola.dev/tika/IHDR> "width=32, height=32, bitDepth=16, colorType=RGBAlpha, compressionMethod=deflate, filterMethod=adaptive, interlaceMethod=none";
+ <https://enola.dev/tika/imagereader:NumImages> "1";
+ <https://enola.dev/tika/Dimension%20PixelAspectRatio> "1.0";
+ <https://enola.dev/tika/Dimension%20ImageOrientation> "Normal";
+ <https://enola.dev/tika/Data%20PlanarConfiguration> "PixelInterleaved";
+ <https://enola.dev/tika/width> "32";
+ <https://enola.dev/tika/Chroma%20Gamma> "1.0";
+ <https://enola.dev/tika/gAMA> "100000";
+ <https://enola.dev/mediaType> "image/png";
+ <https://enola.dev/tika/Compression%20NumProgressiveScans> "1";
+ <https://enola.dev/tika/Content-Type-Parser-Override> "image/ocr-png";
+ <https://enola.dev/tika/Data%20BitsPerSample> "16 16 16 16";
+ <http://ns.adobe.com/tiff/1.0/ImageWidth> "32";
+ <http://ns.adobe.com/tiff/1.0/ImageLength> "32";
+ <https://enola.dev/tika/height> "32";
+ <https://enola.dev/tika/Chroma%20BlackIsZero> "true";
+ <https://enola.dev/tika/Compression%20CompressionTypeName> "deflate";
+ <http://ns.adobe.com/tiff/1.0/BitsPerSample> "16 16 16 16".