Update chemical formula parser for processing in parallel

src/main/java/org/grobid/core/data/material/ChemicalComposition.java

@@ -12,6 +12,10 @@ public class ChemicalComposition {
     private String formula;
 
     private String name;
+
+    private Integer code;
+
+    private String message;
 
     public String getFormula() {
         return formula;
@@ -38,7 +42,7 @@ public void setComposition(Map<String, String> composition) {
     }
 
     public boolean isEmpty() {
-        return StringUtils.isBlank(formula) && StringUtils.isBlank(name) && composition.keySet().size() == 0;
+        return StringUtils.isBlank(formula) && StringUtils.isBlank(name) && composition.keySet().isEmpty();
     }
 
     @Override
@@ -62,4 +66,20 @@ public boolean equals(Object o) {
     public int hashCode() {
         return Objects.hash(composition, formula, name);
     }
+
+    public Integer getCode() {
+        return code;
+    }
+
+    public void setCode(Integer code) {
+        this.code = code;
+    }
+
+    public String getMessage() {
+        return message;
+    }
+
+    public void setMessage(String message) {
+        this.message = message;
+    }
 }

src/main/java/org/grobid/core/engines/MaterialParser.java

@@ -7,11 +7,13 @@
 import org.apache.commons.text.StringEscapeUtils;
 import org.grobid.core.GrobidModel;
 import org.grobid.core.analyzers.DeepAnalyzer;
+import org.grobid.core.data.document.Span;
 import org.grobid.core.data.material.ChemicalComposition;
 import org.grobid.core.data.material.Formula;
 import org.grobid.core.data.material.Material;
 import org.grobid.core.engines.label.TaggingLabel;
 import org.grobid.core.exceptions.GrobidException;
+import org.grobid.core.exceptions.GrobidExceptionStatus;
 import org.grobid.core.features.FeaturesVectorMaterial;
 import org.grobid.core.layout.BoundingBox;
 import org.grobid.core.layout.LayoutToken;
@@ -27,11 +29,11 @@
 import org.slf4j.LoggerFactory;
 
 import javax.inject.Inject;
-import java.util.ArrayList;
-import java.util.Arrays;
-import java.util.List;
+import java.util.*;
+import java.util.function.Predicate;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
+import java.util.stream.IntStream;
 
 import static org.apache.commons.collections4.CollectionUtils.isEmpty;
 import static org.apache.commons.collections4.CollectionUtils.isNotEmpty;
@@ -87,6 +89,84 @@ public List<Material> process(String text) {
         return process(SuperconductorsParser.textToLayoutTokens(text));
     }
 
+    public List<List<Material>> processParallel(List<String> texts) {
+
+        List<Integer> emptyIndices = IntStream.range(0, texts.size())
+            .filter(i -> StringUtils.isBlank(texts.get(i)))
+            .boxed()
+            .toList();
+
+        Set<Integer> emptyIndicesSet = new HashSet<>(emptyIndices);
+        List<String> textsCopy = new ArrayList<>();
+
+        for (int is = 0; is < texts.size(); is++) {
+            if (!emptyIndices.contains(is)) {
+                textsCopy.add(texts.get(is));
+            }
+        }
+
+        List<List<LayoutToken>> asLayoutTokens = textsCopy.stream()
+            .map(SuperconductorsParser::textToLayoutTokens)
+            .collect(Collectors.toList());
+
+        List<List<Material>> processed = processParallelLT(asLayoutTokens);
+
+        emptyIndices.forEach(i -> processed.add(i, new ArrayList<>()));
+
+        List<List<Material>> output = new ArrayList<>();
+        for (int is = 0; is < texts.size(); is++) {
+            if (!emptyIndices.contains(is)) {
+                output.add(processed.get(is));
+            } else {
+                output.add(new ArrayList<>());
+            }
+        }
+
+        return output;
+    }
+
+    public List<List<Material>> processParallelLT(List<List<LayoutToken>> layoutTokensBatch) {
+
+        List<List<Material>> entities = new ArrayList<>();
+
+        //Normalisation
+        List<List<LayoutToken>> normalisedTokens = layoutTokensBatch.stream()
+            .map(SuperconductorsParser::normalizeAndRetokenizeLayoutTokens)
+            .toList();
+
+        try {
+            List<String> tokensWithFeatures = normalisedTokens.stream().map(nt -> addFeatures(nt) + "\n").toList();
+
+            String labellingResult = null;
+            try {
+                labellingResult = label(tokensWithFeatures);
+            } catch (Exception e) {
+                throw new GrobidException("CRF labeling for superconductors parsing failed.", e);
+            }
+
+            List<String> resultingBlocks = Arrays.asList(labellingResult.split("\n\n"));
+            List<List<Material>> localEntities = extractParallelResults(normalisedTokens, resultingBlocks);
+
+            entities.addAll(localEntities);
+        } catch (Exception e) {
+            throw new GrobidException("An exception occurred while running Grobid.", e);
+        }
+
+        return entities;
+    }
+
+    public List<List<Material>> extractParallelResults(List<List<LayoutToken>> tokens, List<String> results) {
+        List<List<Material>> spans = new ArrayList<>();
+        if (tokens.size() != results.size()) {
+            throw new GrobidException("One of the text provided is invalid or empty and cannot be tagged. Please provide a clean input.", GrobidExceptionStatus.BAD_INPUT_DATA);
+        }
+        for (int i = 0; i < tokens.size(); i++) {
+            spans.add(extractResults(tokens.get(i), results.get(i)));
+        }
+
+        return spans;
+    }
+
 
     public List<Material> process(List<LayoutToken> tokens) {
 

src/main/java/org/grobid/core/utilities/client/ChemicalMaterialParserClient.java

@@ -66,10 +66,16 @@ public ChemicalComposition convertNameToFormula(String name) {
             try (CloseableHttpResponse response = httpClient.execute(request)) {
                 int statusCode = response.getStatusLine().getStatusCode();
 
-                if (statusCode == HttpURLConnection.HTTP_OK) {
-                    outputFormula = fromJsonToChemicalComposition(response.getEntity().getContent());
+                if (statusCode != HttpURLConnection.HTTP_OK) {
+                    LOGGER.debug("Not OK answer. Input: " + name + ". Status code: " + response.getStatusLine().getStatusCode());
                 } else {
-                    LOGGER.debug("Not OK answer. Input: " + name + ", status code: " + statusCode);
+                    outputFormula = fromJsonToChemicalComposition(response.getEntity().getContent());
+                    if (outputFormula != null && outputFormula.getCode() != HttpURLConnection.HTTP_OK) {
+                        LOGGER.debug("Not OK answer. Input: " + name + ". " +
+                            "Status code: " + outputFormula.getCode() +
+                            "Message: " + outputFormula.getMessage());
+                        outputFormula = new ChemicalComposition();
+                    }
                 }
             }
 
@@ -101,6 +107,11 @@ public ChemicalComposition convertFormulaToComposition(String formula) {
                     LOGGER.debug("Not OK answer. Input: " + formula + ". Status code: " + response.getStatusLine().getStatusCode());
                 } else {
                     outputComposition = fromJsonToChemicalComposition(response.getEntity().getContent());
+                    if (outputComposition != null && outputComposition.getCode() != HttpURLConnection.HTTP_OK) {
+                        LOGGER.debug("Not OK answer. Input: " + formula + ". Status code: " + outputComposition.getCode() +
+                            "Message: " + outputComposition.getMessage());
+                        outputComposition = new ChemicalComposition();
+                    }
                 }
             }
 

src/main/java/org/grobid/service/controller/MaterialController.java

@@ -12,6 +12,7 @@
 import javax.ws.rs.Path;
 import javax.ws.rs.Produces;
 import javax.ws.rs.core.MediaType;
+import java.util.Arrays;
 import java.util.List;
 
 @Singleton
@@ -42,10 +43,32 @@ public List<Material> processTextSuperconductorsGet(@FormDataParam("text") Strin
         return parseMaterial(text);
     }
 
+
+    @Path("multiparse")
+    @Produces(MediaType.APPLICATION_JSON)
+    @POST
+    public List<List<Material>> processTextSuperconductorsPost2(@FormDataParam("texts") String texts) {
+        return parseMaterials(texts);
+    }
+
+    @Path("multiparse")
+    @Produces(MediaType.APPLICATION_JSON)
+    @GET
+    public List<List<Material>> processTextSuperconductorsGet2(@FormDataParam("texts") String texts) {
+        return parseMaterials(texts);
+    }
+
     private List<Material> parseMaterial(@FormDataParam("text") String text) {
         String textPreprocessed = text.replace("\r\n", "\n");
 
         return materialParser.process(textPreprocessed);
     }
 
+    private List<List<Material>> parseMaterials(@FormDataParam("text") String text) {
+        String textPreprocessed = text.replace("\r\n", "\n");
+
+        List<String> list = Arrays.asList(textPreprocessed.split("\n"));
+        return materialParser.processParallel(list);
+    }
+
 }