Merge branch 'master' into JoelGardes-patch-1

Author: ChristopheMaldivi

Makefile

@@ -4,12 +4,12 @@ VERSION=`git describe`
 build-notest:
 (cd simdoc/core/java/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install -DskipTests=true)
 (cd simdoc/apps && mvn -Dmaven.repo.local=${HOME}/.m2/repository install -DskipTests=true)
-(cd simdoc/simdoc-server/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install -DskipTests=true)
+(cd simdoc/clustering-server/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install -DskipTests=true)
 
 build-tests:
 (cd simdoc/core/java/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install)
 (cd simdoc/apps && mvn -Dmaven.repo.local=${HOME}/.m2/repository install)
-(cd simdoc/simdoc-server/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install)
+(cd simdoc/clustering-server/ && mvn -Dmaven.repo.local=${HOME}/.m2/repository install)
 
 # Used to build debian package
 # here we consider that artifacts are already built (to speed up continuous integration)
@@ -27,7 +27,7 @@ deb:
 # Debian package install
 install:
 rm -rf usr && mkdir -p usr/share/java
-cp simdoc/simdoc-server/target/simdoc-server-*.jar usr/share/java/simdoc-server.jar
+cp simdoc/clustering-server/target/clustering-server-*.jar usr/share/java/clustering-server.jar
 rm -rf etc && mkdir -p etc/init.d && cp debian/server-init-d etc/init.d/simdoc-server
 mkdir etc/rc2.d && (cd etc/rc2.d && ln -s ../init.d/simdoc-server S08simdoc-server)
 

README.md

@@ -2,7 +2,7 @@
 [![License](https://img.shields.io/aur/license/yaourt.svg?maxAge=2592000)](https://github.com/Orange-OpenSource/documentare-simdoc/blob/master/LICENSE)
 
 # Documentare, SimDoc library & tools
-Library and tools for similarity measurement, classification and clustering of digital content and segmentation images from digitized document
+Library and tools for similarity measurement, classification and clustering of digital content and segmentation images from digitized document.
 
 ## Build
 

simdoc/apps/clustering-remote/src/main/java/com/orange/documentare/app/clusteringremote/ClusteringRemoteApp.java

@@ -14,16 +14,17 @@
 import org.apache.commons.cli.ParseException;
 
 import java.io.IOException;
+import java.util.Currency;
 
 @Slf4j
 public class ClusteringRemoteApp {
 
- private static ClusteringRequest clusteringRequest;
+ private static ClusteringRequest remoteClusteringRequest;
 
  public static void main(String[] args) throws IllegalAccessException, IOException, ParseException {
  System.out.println("\n[ClusteringRemote - Start]");
  try {
- clusteringRequest = (new CommandLineOptions(args)).clusteringRequest();
+ remoteClusteringRequest = (new CommandLineOptions(args)).clusteringRequest();
  } catch (Exception e) {
  CommandLineOptions.showHelp(e);
  return;
@@ -37,8 +38,49 @@ public static void main(String[] args) throws IllegalAccessException, IOExceptio
  }
 
  private static void doTheJob() throws IOException {
- System.out.println("Clustering parameters, " + clusteringRequest);
+ System.out.println("Clustering parameters, " + remoteClusteringRequest);
  RemoteClustering remoteClustering = new RemoteClustering();
- remoteClustering.request("http://localhost:8080", clusteringRequest);
+ String url = remoteClusteringRequest.url;
+ ClusteringRequest clusteringRequest = buildClusteringRequest();
+
+ remoteClustering.request(url, clusteringRequest);
+ }
+
+ private static ClusteringRequest buildClusteringRequest() {
+ ClusteringRequest.ClusteringRequestBuilder clusteringRequestBuilder = ClusteringRequest.builder()
+ .bytesData(remoteClusteringRequest.bytesData)
+ .outputDirectory(remoteClusteringRequest.outputDirectory);
+
+ if (remoteClusteringRequest.acutSdFactor != null) {
+ clusteringRequestBuilder.acut(remoteClusteringRequest.acutSdFactor);
+ }
+
+ if (remoteClusteringRequest.ccutPercentile != null) {
+ clusteringRequestBuilder.ccut(remoteClusteringRequest.ccutPercentile);
+ }
+
+ if (remoteClusteringRequest.qcutSdFactor != null) {
+ clusteringRequestBuilder.qcut(remoteClusteringRequest.qcutSdFactor);
+ }
+
+ if (remoteClusteringRequest.scutSdFactor != null) {
+ clusteringRequestBuilder.scut(remoteClusteringRequest.scutSdFactor);
+ }
+
+
+ if ((remoteClusteringRequest.wcut != null) && (remoteClusteringRequest.wcut == true)) {
+ clusteringRequestBuilder.wcut();
+ }
+
+ if ((remoteClusteringRequest.debug != null) && (remoteClusteringRequest.debug == true)) {
+ clusteringRequestBuilder.debug();
+ }
+
+ if ((remoteClusteringRequest.sloop != null) && (remoteClusteringRequest.sloop == true)) {
+ clusteringRequestBuilder.sloop();
+ }
+
+ return clusteringRequestBuilder
+ .build();
  }
 }

simdoc/apps/clustering-remote/src/main/java/com/orange/documentare/app/clusteringremote/ClusteringRequest.java

@@ -15,10 +15,13 @@
 import lombok.RequiredArgsConstructor;
 import lombok.ToString;
 
+import java.io.File;
+
 @ToString
 @EqualsAndHashCode
 @RequiredArgsConstructor
 public class ClusteringRequest {
+ public final String url;
  public final String inputDirectory;
  public final BytesData[] bytesData;
  public final String outputDirectory;
@@ -36,6 +39,7 @@ public static ClusteringRequestBuilder builder() {
  }
 
  public static class ClusteringRequestBuilder {
+ public String url;
  private String inputDirectory;
  private BytesData[] bytesData;
  private String outputDirectory;
@@ -52,12 +56,17 @@ private ClusteringRequestBuilder() {
  }
 
  public ClusteringRequest build() {
- return new ClusteringRequest(inputDirectory, bytesData, outputDirectory, debug, acutSdFactor, qcutSdFactor, scutSdFactor, ccutPercentile, wcut, kNearestNeighboursThreshold, sloop);
+ return new ClusteringRequest(url, inputDirectory, bytesData, outputDirectory, debug, acutSdFactor, qcutSdFactor, scutSdFactor, ccutPercentile, wcut, kNearestNeighboursThreshold, sloop);
  }
 
+ public ClusteringRequestBuilder url(String url) {
+ this.url = url;
+ return this;
+ }
 
  public ClusteringRequestBuilder inputDirectory(String inputDirectory) {
- this.inputDirectory = inputDirectory;
+ BytesData[] bytesData = BytesData.loadFromDirectory(new File(inputDirectory), File::getName);
+ this.bytesData(bytesData);
  return this;
  }
 

simdoc/apps/clustering-remote/src/main/java/com/orange/documentare/app/clusteringremote/cmdline/CommandLineOptions.java

@@ -22,6 +22,7 @@
 import static com.orange.documentare.core.comp.clustering.graph.ClusteringParameters.*;
 
 public class CommandLineOptions {
+ private static final String URL_REMOTE_SERVER = "url";
  private static final String HELP = "h";
  private static final String INPUT_DIRECTORY = "din";
  private static final String OUTPUT_DIRECTORY = "dout";
@@ -58,12 +59,13 @@ private void init(String[] args) throws ParseException, IOException {
  }
 
  private void initOptions(CommandLine commandLine) throws IOException {
+ boolean dUrlOption = commandLine.hasOption(URL_REMOTE_SERVER);
  boolean dInOption = commandLine.hasOption(INPUT_DIRECTORY);
  boolean dOutOption = commandLine.hasOption(OUTPUT_DIRECTORY);
  boolean bytesDataOption = commandLine.hasOption(BYTE_DATA_JSON);
 
- if (!dInOption && !bytesDataOption && !dOutOption) {
- throw new CommandLineException("input directory or bytes data json or output directory argument is missing\n");
+ if (!dUrlOption && !dInOption && !bytesDataOption && !dOutOption) {
+ throw new CommandLineException("remote server url or input directory or bytes data json or output directory argument is missing\n");
  } else {
  if (commandLine.hasOption(ACUT)) {
  builder.acut(Float.parseFloat(commandLine.getOptionValue(ACUT, String.valueOf(A_DEFAULT_SD_FACTOR))));
@@ -87,6 +89,10 @@ private void initOptions(CommandLine commandLine) throws IOException {
  builder.kNearestNeighboursThreshold(Integer.parseInt(commandLine.getOptionValue(KNN)));
  }
 
+ if (dUrlOption) {
+ builder.url(commandLine.getOptionValue(URL_REMOTE_SERVER));
+ }
+
  if (dInOption) {
  builder.inputDirectory(commandLine.getOptionValue(INPUT_DIRECTORY));
  }
@@ -110,6 +116,7 @@ private BytesData[] loadBytesDataJson(String filePath) throws IOException {
  }
 
  private CommandLine getCommandLineFromArgs(String[] args) throws ParseException {
+ Option url = OptionBuilder.withArgName("url for remote server").hasArg().withDescription("url for remote server").create(URL_REMOTE_SERVER);
  Option help = new Option(HELP, "print this message");
 
  Option din = OptionBuilder.withArgName("input directory").hasArg().withDescription("input directory").create(INPUT_DIRECTORY);
@@ -151,6 +158,7 @@ private CommandLine getCommandLineFromArgs(String[] args) throws ParseException
  .desc("kNearestNeighboursThreshold")
  .build();
 
+ options.addOption(url);
  options.addOption(help);
  options.addOption(din);
  options.addOption(bytesDataJson);

simdoc/apps/clustering-remote/src/test/java/com/orange/documentare/app/clusteringremote/RemoteClusteringTest.java

@@ -41,7 +41,7 @@ public void cleanup() {
  }
 
  @Test
- @Ignore // a running simdoc-server is mandatory for this test
+ @Ignore // a running clustering-server on port 6969 is mandatory for this test
  public void remote_build_animals_dna_clustering() throws IOException {
  // Given
  ClusteringRequest req = ClusteringRequest.builder()
@@ -53,14 +53,14 @@ public void remote_build_animals_dna_clustering() throws IOException {
  RemoteClustering remoteClustering = new RemoteClustering();
 
  // When
- ClusteringRequestResult result = remoteClustering.request("http://localhost:8080", req);
+ ClusteringRequestResult result = remoteClustering.request("http://localhost:6969", req);
 
  // Then
- Assertions.assertThat(result).isEqualTo(expectedClusteringResult(false));
+ Assertions.assertThat(result).isEqualTo(expectedClusteringResult());
  }
 
  @Test
- @Ignore // a running simdoc-server is mandatory for this test
+ @Ignore // a running clustering-server on port 6969 is mandatory for this test
  public void remote_build_animals_dna_clustering_with_bytes_data() throws IOException {
  // Given
  ClusteringRequest req = ClusteringRequest.builder()
@@ -71,11 +71,51 @@ public void remote_build_animals_dna_clustering_with_bytes_data() throws IOExcep
 
  RemoteClustering remoteClustering = new RemoteClustering();
 
+ // When
+ ClusteringRequestResult result = remoteClustering.request("http://localhost:6969", req);
+
+ // Then
+ ClusteringRequestResult expected = expectedClusteringResult();
+ Assertions.assertThat(result).isEqualTo(expected);
+ }
+
+
+ @Test
+ @Ignore // a running mediation-server on port 8080 and clustering-server on port 6969 is mandatory for this test
+ public void remote_build_animals_dna_clustering_with_mediation() throws IOException {
+ // Given
+ ClusteringRequest req = ClusteringRequest.builder()
+ .inputDirectory(inputDirectory())
+ .outputDirectory(OUTPUT_DIRECTORY.getAbsolutePath())
+ .debug()
+ .build();
+
+ RemoteClustering remoteClustering = new RemoteClustering();
+
  // When
  ClusteringRequestResult result = remoteClustering.request("http://localhost:8080", req);
 
  // Then
- ClusteringRequestResult expected = expectedClusteringResult(true);
+ Assertions.assertThat(result).isEqualTo(expectedClusteringResultWithMediation());
+ }
+
+ @Test
+ @Ignore // a running mediation-server on port 8080 and clustering-server on port 6969 is mandatory for this test
+ public void remote_build_animals_dna_clustering_with_bytes_data_with_mediation() throws IOException {
+ // Given
+ ClusteringRequest req = ClusteringRequest.builder()
+ .bytesData(bytesData())
+ .outputDirectory(OUTPUT_DIRECTORY.getAbsolutePath())
+ .debug()
+ .build();
+
+ RemoteClustering remoteClustering = new RemoteClustering();
+
+ // When
+ ClusteringRequestResult result = remoteClustering.request("http://localhost:8080", req);
+
+ // Then
+ ClusteringRequestResult expected = expectedClusteringResultWithMediation();
  Assertions.assertThat(result).isEqualTo(expected);
  }
 
@@ -87,9 +127,15 @@ private BytesDataArray bytesDataWithBytes() throws IOException {
  return (BytesDataArray)jsonGenericHandler.getObjectFromJsonFile(BytesDataArray.class, new File(getClass().getResource("/bytes-data-animals-dna.json").getFile()));
  }
 
- private ClusteringRequestResult expectedClusteringResult(boolean bytesDataMode) throws IOException {
- File file = new File(getClass().getResource(
- bytesDataMode ? "/expected-clustering-result-bytes-data.json" : "/expected-clustering-result.json").getFile());
+ private ClusteringRequestResult expectedClusteringResult() throws IOException {
+
+ File file = new File(getClass().getResource("/expected-clustering-result-bytes-data-animals-dna.json").getFile());
+ return (ClusteringRequestResult) jsonGenericHandler.getObjectFromJsonFile(ClusteringRequestResult.class, file);
+ }
+
+ private ClusteringRequestResult expectedClusteringResultWithMediation() throws IOException {
+
+ File file = new File(getClass().getResource("/expected-clustering-result-bytes-data-files-animals-dna.json").getFile());
  return (ClusteringRequestResult) jsonGenericHandler.getObjectFromJsonFile(ClusteringRequestResult.class, file);
  }