german translation

monarch-initiative · May 11, 2024 · 3134826 · 3134826
1 parent df015bd
commit 3134826
Show file tree

Hide file tree

Showing 9 changed files with 659 additions and 5 deletions.
diff --git a/src/main/java/org/monarchinitiative/phenopacket2prompt/cmd/GbtTranslateBatchCommand.java b/src/main/java/org/monarchinitiative/phenopacket2prompt/cmd/GbtTranslateBatchCommand.java
@@ -65,6 +65,8 @@ public Integer call() throws Exception {
         // output all non-English languages here
         PromptGenerator spanish = PromptGenerator.spanish(hpo, internationalMap.get("es"));
         outputPromptsInternational(ppktFiles, hpo, "es", spanish);
+        PromptGenerator german = PromptGenerator.german(hpo, internationalMap.get("de"));
+        outputPromptsInternational(ppktFiles, hpo, "de", german);
         // output file with correct diagnosis list
         outputCorrectResults(correctResultList);
         return 0;
@@ -97,7 +99,7 @@ private void outputPromptsInternational(List<File> ppktFiles, Ontology hpo, Stri
             PpktIndividual individual = new PpktIndividual(f);
             List<PhenopacketDisease> diseaseList = individual.getDiseases();
             if (diseaseList.size() != 1) {
-                System.err.println(String.format("[ERROR] Got %d diseases for %s.\n", diseaseList.size(), individual.getPhenopacketId()));
+                System.err.printf("[ERROR] Got %d diseases for %s.\n", diseaseList.size(), individual.getPhenopacketId());
                 continue;
             }
             PhenopacketDisease pdisease = diseaseList.get(0);
@@ -123,7 +125,7 @@ private List<CorrectResult> outputPromptsEnglish(List<File> ppktFiles, Ontology
             PpktIndividual individual = new PpktIndividual(f);
             List<PhenopacketDisease> diseaseList = individual.getDiseases();
             if (diseaseList.size() != 1) {
-                System.err.println(String.format("[ERROR] Got %d diseases for %s.\n", diseaseList.size(), individual.getPhenopacketId()));
+                System.err.printf("[ERROR] Got %d diseases for %s.\n", diseaseList.size(), individual.getPhenopacketId());
                 continue;
             }
             PhenopacketDisease pdisease = diseaseList.get(0);

diff --git a/src/main/java/org/monarchinitiative/phenopacket2prompt/cmd/GptTranslateCommand.java b/src/main/java/org/monarchinitiative/phenopacket2prompt/cmd/GptTranslateCommand.java
@@ -62,6 +62,10 @@ public Integer call() throws Exception {
         PromptGenerator spanish = PromptGenerator.spanish(hpo, internationalMap.get("es"));
         prompt = spanish.createPrompt(individual);
         System.out.println(prompt);
+        System.out.println("GERMAN");
+        PromptGenerator german = PromptGenerator.spanish(hpo, internationalMap.get("de"));
+        prompt = german.createPrompt(individual);
+        System.out.println(prompt);
 
         return 0;
     }

diff --git a/...java/org/monarchinitiative/phenopacket2prompt/international/HpInternationalOboParser.java b/...java/org/monarchinitiative/phenopacket2prompt/international/HpInternationalOboParser.java
@@ -53,7 +53,7 @@ public Optional<String> getTranslation(String annots) {
     public HpInternationalOboParser(File file) {
         languageToInternationalMap = new HashMap<>();
         String pattern = "id: (HP:\\d{7,7})";
-        Set<String> acronyms = Set.of("cs", "en", "es", "fr", "ja", "nl", "nna", "tr", "tw", "zh");
+        Set<String> acronyms = Set.of("cs", "en", "de", "it", "es", "fr", "ja", "nl", "nna", "tr", "tw", "zh");
         for (String acronym : acronyms) {
             languageToInternationalMap.put(acronym, new HpInternational(acronym));
         }

diff --git a/src/main/java/org/monarchinitiative/phenopacket2prompt/output/PromptGenerator.java b/src/main/java/org/monarchinitiative/phenopacket2prompt/output/PromptGenerator.java
@@ -7,6 +7,8 @@
 import org.monarchinitiative.phenopacket2prompt.model.PhenopacketSex;
 import org.monarchinitiative.phenopacket2prompt.model.PpktIndividual;
 import org.monarchinitiative.phenopacket2prompt.output.impl.english.EnglishPromptGenerator;
+import org.monarchinitiative.phenopacket2prompt.output.impl.german.GermanPromptGenerator;
+import org.monarchinitiative.phenopacket2prompt.output.impl.german.PpktPhenotypicfeatureGerman;
 import org.monarchinitiative.phenopacket2prompt.output.impl.spanish.*;
 
 import java.util.List;
@@ -33,6 +35,11 @@ static PromptGenerator spanish(Ontology hpo, HpInternational international) {
         return new SpanishPromptGenerator(hpo, pfgen);
     }
 
+    static PromptGenerator german(Ontology hpo, HpInternational international) {
+        PpktPhenotypicFeatureGenerator pfgen = new PpktPhenotypicfeatureGerman(international);
+        return new GermanPromptGenerator(hpo, pfgen);
+    }
+
     /**
      * The following structure should work for most other languages, but the function
      * can be overridden if necessary.

diff --git a/...va/org/monarchinitiative/phenopacket2prompt/output/impl/german/GermanPromptGenerator.java b/...va/org/monarchinitiative/phenopacket2prompt/output/impl/german/GermanPromptGenerator.java
@@ -0,0 +1,63 @@
+package org.monarchinitiative.phenopacket2prompt.output.impl.german;
+
+import org.monarchinitiative.phenol.ontology.data.Ontology;
+import org.monarchinitiative.phenopacket2prompt.model.OntologyTerm;
+import org.monarchinitiative.phenopacket2prompt.model.PhenopacketAge;
+import org.monarchinitiative.phenopacket2prompt.model.PhenopacketSex;
+import org.monarchinitiative.phenopacket2prompt.model.PpktIndividual;
+import org.monarchinitiative.phenopacket2prompt.output.PhenopacketIndividualInformationGenerator;
+import org.monarchinitiative.phenopacket2prompt.output.PhenopacketTextGenerator;
+import org.monarchinitiative.phenopacket2prompt.output.PpktPhenotypicFeatureGenerator;
+import org.monarchinitiative.phenopacket2prompt.output.PromptGenerator;
+
+import java.util.List;
+
+public class GermanPromptGenerator implements PromptGenerator {
+
+    private final Ontology hpo;
+
+
+    private final PhenopacketIndividualInformationGenerator ppktAgeSexGenerator;
+
+    private final PhenopacketTextGenerator ppktTextGenerator;
+
+    private final PpktPhenotypicFeatureGenerator ppktPhenotypicFeatureGenerator;
+
+
+
+    public GermanPromptGenerator(Ontology hpo, PpktPhenotypicFeatureGenerator pfgen) {
+        this.hpo = hpo;
+        ppktAgeSexGenerator = new PpktIndividualGerman();
+        ppktTextGenerator = new PpktTextGerman();
+        this.ppktPhenotypicFeatureGenerator = pfgen;
+    }
+
+    @Override
+    public String queryHeader() {
+        return ppktTextGenerator.QUERY_HEADER();
+    }
+
+    @Override
+    public String getIndividualInformation(PpktIndividual ppktIndividual) {
+        return this.ppktAgeSexGenerator.getIndividualDescription(ppktIndividual);
+    }
+
+    @Override
+    public String formatFeatures(List<OntologyTerm> ontologyTerms) {
+        return ppktPhenotypicFeatureGenerator.formatFeatures(ontologyTerms);
+    }
+
+    @Override
+    public String getVignetteAtAge(PhenopacketAge page, PhenopacketSex psex, List<OntologyTerm> terms) {
+        String ageString = this.ppktAgeSexGenerator.atAge(page);
+        String features = formatFeatures(terms);
+        return String.format("%s, %s presentó %s", ageString, ppktAgeSexGenerator.heSheIndividual(psex), features);
+    }
+
+
+
+
+
+
+
+}