Taxonomy: added rejection of name types

Bibliome · Aug 17, 2021 · f9b16d9 · f9b16d9
1 parent 850f494
commit f9b16d9
Show file tree

Hide file tree

Showing 9 changed files with 114 additions and 11 deletions.
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/dict/BuildDictionary.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/dict/BuildDictionary.java
@@ -39,7 +39,7 @@
 import fr.inra.maiage.bibliome.util.taxonomy.Taxon;
 import fr.inra.maiage.bibliome.util.taxonomy.reject.RejectDisjunction;
 import fr.inra.maiage.bibliome.util.taxonomy.reject.RejectName;
-import fr.inra.maiage.bibliome.util.taxonomy.reject.RejectNone;
+import fr.inra.maiage.bibliome.util.taxonomy.reject.RejectNameType;
 import fr.inra.maiage.bibliome.util.taxonomy.saturate.Saturate;
 
 /**
@@ -50,6 +50,7 @@
 public class BuildDictionary extends CLIOParser {
 	private final Collection<File> nodesFiles = new ArrayList<File>();
 	private final Collection<File> namesFiles = new ArrayList<File>();
+	private final Collection<String> rejectedNameTypes = new ArrayList<String>();
 	private File saturationFile;
 	private File rejectionFile;
 	private final List<TaxonNamePattern> pattern = new ArrayList<TaxonNamePattern>(Arrays.asList(
@@ -73,6 +74,11 @@ protected boolean processArgument(String arg) throws CLIOException {
 		return false;
 	}
 
+	@CLIOption("-rejectNameType")
+	public void addRejectedNameType(String nameType) {
+		this.rejectedNameTypes.add(nameType);
+	}
+
 	@CLIOption("-prefix")
 	public void setIdPrefix(String idPrefix) {
 		this.idPrefix = idPrefix;
@@ -226,9 +232,14 @@ public static void main(String[] args) throws CLIOException, IOException, Invali
 		}
 
 		/* Name filter and synonym generation */
-		RejectName reject = RejectNone.INSTANCE;
-		if (inst.rejectionFile != null)
-			reject = new RejectDisjunction(taxonomy.readReject(logger, inst.rejectionFile));
+		RejectDisjunction rejects = new RejectDisjunction();
+		if (inst.rejectionFile != null) {
+			rejects.add(taxonomy.readReject(logger, inst.rejectionFile));
+		}
+		if (!inst.rejectedNameTypes.isEmpty()) {
+			rejects.add(new RejectNameType(inst.rejectedNameTypes));
+		}
+		RejectName reject = rejects.simplify();
 		for (File f : inst.namesFiles) {
 			taxonomy.readNames(logger, f, reject);
 		}

diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectConjunction.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectConjunction.java
@@ -38,4 +38,17 @@ public RejectConjunction(RejectName first, RejectName second) {
 	public boolean reject(String taxid, Name name) {
 		return first.reject(taxid, name) && second.reject(taxid, name);
 	}
+
+	@Override
+	public RejectName simplify() {
+		RejectName first = this.first.simplify();
+		RejectName second = this.second.simplify();
+		if (first.equals(RejectNone.INSTANCE)) {
+			return RejectNone.INSTANCE;
+		}
+		if (second.equals(RejectNone.INSTANCE)) {
+			return RejectNone.INSTANCE;
+		}
+		return new RejectConjunction(first, second);
+	}
 }
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectDisjunction.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectDisjunction.java
@@ -16,13 +16,19 @@
 
 package fr.inra.maiage.bibliome.util.taxonomy.reject;
 
+import java.util.ArrayList;
 import java.util.Collection;
+import java.util.List;
 
 import fr.inra.maiage.bibliome.util.taxonomy.Name;
 
 public class RejectDisjunction implements RejectName {
 	private final Collection<RejectName> rejects;
-
+
+	public RejectDisjunction() {
+		this(new ArrayList<RejectName>());
+	}
+
 	public RejectDisjunction(Collection<RejectName> rejects) {
 		super();
 		this.rejects = rejects;
@@ -35,4 +41,30 @@ public boolean reject(String taxid, Name name) {
 				return true;
 		return false;
 	}
+
+	public void add(RejectName reject) {
+		rejects.add(reject);
+	}
+
+	public void add(Collection<RejectName> rejects) {
+		this.rejects.addAll(rejects);
+	}
+
+	@Override
+	public RejectName simplify() {
+		List<RejectName> rejects = new ArrayList<RejectName>(this.rejects.size());
+		for (RejectName rn : this.rejects) {
+			RejectName srn = rn.simplify();
+			if (!srn.equals(RejectNone.INSTANCE)) {
+				rejects.add(srn);
+			}
+		}
+		if (rejects.isEmpty()) {
+			return RejectNone.INSTANCE;
+		}
+		if (rejects.size() == 1) {
+			return rejects.get(0);
+		}
+		return new RejectDisjunction(rejects);
+	}
 }
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectName.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectName.java
@@ -30,4 +30,6 @@ public interface RejectName {
 	 * @param name
 	 */
 	boolean reject(String taxid, Name name);
+
+	RejectName simplify();
 }
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNamePattern.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNamePattern.java
@@ -41,4 +41,9 @@ public RejectNamePattern(Pattern pattern) {
 	public boolean reject(String taxid, Name name) {
 		return pattern.matcher(name.name).matches();
 	}
+
+	@Override
+	public RejectName simplify() {
+		return this;
+	}
 }
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNameType.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNameType.java
@@ -0,0 +1,32 @@
+package fr.inra.maiage.bibliome.util.taxonomy.reject;
+
+import java.util.Arrays;
+import java.util.Collection;
+import java.util.HashSet;
+
+import fr.inra.maiage.bibliome.util.taxonomy.Name;
+
+public class RejectNameType implements RejectName {
+	private final Collection<String> nameTypes = new HashSet<String>();
+
+	public RejectNameType(String... nameTypes) {
+		this(Arrays.asList(nameTypes));
+	}
+
+	public RejectNameType(Collection<String> nameTypes) {
+		this.nameTypes.addAll(nameTypes);
+	}
+
+	@Override
+	public boolean reject(String taxid, Name name) {
+		return nameTypes.contains(name.type);
+	}
+
+	@Override
+	public RejectName simplify() {
+		if (nameTypes.isEmpty()) {
+			return RejectNone.INSTANCE;
+		}
+		return this;
+	}
+}
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNone.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectNone.java
@@ -18,15 +18,16 @@
 
 import fr.inra.maiage.bibliome.util.taxonomy.Name;
 
-public class RejectNone implements RejectName {
-	public static final RejectNone INSTANCE = new RejectNone();
-
-	private RejectNone() {
-		super();
-	}
+public enum RejectNone implements RejectName {
+	INSTANCE;
 
 	@Override
 	public boolean reject(String taxid, Name name) {
 		return false;
 	}
+
+	@Override
+	public RejectName simplify() {
+		return this;
+	}
 }
diff --git a/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectTaxid.java b/src/main/java/fr/inra/maiage/bibliome/util/taxonomy/reject/RejectTaxid.java
@@ -39,4 +39,9 @@ public RejectTaxid(String taxid) {
 	public boolean reject(String taxid, Name name) {
 		return this.taxid.equals(taxid);
 	}
+
+	@Override
+	public RejectName simplify() {
+		return this;
+	}
 }
diff --git a/src/main/resources/fr/inra/maiage/bibliome/util/taxonomy/dict/BuildDictionaryHelp.properties b/src/main/resources/fr/inra/maiage/bibliome/util/taxonomy/dict/BuildDictionaryHelp.properties
@@ -27,6 +27,8 @@ options = Options:
 -saturationFile.help = path to the saturation patterns file
 -rejectionFile.args = FILE
 -rejectionFile.help = path to rejected names file
+-rejectNameType.args = TYPE
+-rejectNameType.help = reject synonyms TYPE
 -pathSeparator.args = STRING
 -pathSeparator.help = separator between nodes in node path output (default: '/')
 -pattern.args = PATTERN