doc; prepare coord in affiliations

kermitt2 · Nov 18, 2023 · 169dcfd · 169dcfd
1 parent 6f4bcdb
commit 169dcfd
Show file tree

Hide file tree

Showing 3 changed files with 28 additions and 4 deletions.
diff --git a/doc/Grobid-docker.md b/doc/Grobid-docker.md
@@ -67,7 +67,9 @@ By default, this image runs Deep Learning models for:
 
 - segmentation of the bibliographical reference section into individual references,
 
-- parsing of the header metadata.
+- parsing of the header metadata,
+
+- parsing of funding and acknowledgement sections.
 
 With a GPU (at least 4GB GPU memory required), the processing runtime is similar as with the CRF-only image with CPU only. 
 

diff --git a/grobid-core/src/main/java/org/grobid/core/data/Affiliation.java b/grobid-core/src/main/java/org/grobid/core/data/Affiliation.java
@@ -4,9 +4,9 @@
 import org.grobid.core.lexicon.Lexicon;
 import org.grobid.core.layout.LayoutToken;
 import org.grobid.core.utilities.OffsetPosition;
+import org.grobid.core.engines.label.TaggingLabel;
 
-import java.util.ArrayList;
-import java.util.List;
+import java.util.*;
 
 /**
  * Class for representing and exchanging affiliation information.
@@ -37,6 +37,9 @@ public class Affiliation {
 
     private List<LayoutToken> layoutTokens = null;
 
+    // map of model labels to LayoutToken
+    private Map<String, List<LayoutToken>> labeledTokens;
+
     // an identifier for the affiliation independent from the marker, present in the TEI result
     private String key = null;
 
@@ -672,5 +675,13 @@ public String toString() {
                 ", failAffiliation=" + failAffiliation +
                 '}';
     }
-
+
+    public void addLabeledResult(TaggingLabel label, List<LayoutToken> tokenizations) {
+        if (labeledTokens == null)
+            labeledTokens = new TreeMap<>();
+
+        List<LayoutToken> theTokenList = tokenizations == null ? new ArrayList<>() : tokenizations;
+        labeledTokens.put(label.getLabel(), theTokenList);
+    }
+
 }
diff --git a/grobid-core/src/main/java/org/grobid/core/data/BiblioItem.java b/grobid-core/src/main/java/org/grobid/core/data/BiblioItem.java
@@ -3779,10 +3779,21 @@ private void appendAffiliation(
         GrobidAnalysisConfig config,
         Lexicon lexicon
     ) {
+        boolean affiliationWithCoords = (config.getGenerateTeiCoordinates() != null) && (config.getGenerateTeiCoordinates().contains("affiliation"));
+        boolean orgnameWithCoords = (config.getGenerateTeiCoordinates() != null) && (config.getGenerateTeiCoordinates().contains("orgName"));
+
         TextUtilities.appendN(tei, '\t', nbTag);
         tei.append("<affiliation");
         if (aff.getKey() != null)
             tei.append(" key=\"").append(aff.getKey()).append("\"");
+        if (affiliationWithCoords) {
+            // we serialize the coordinates for the whole affiliation block
+            List<LayoutToken> affTokens = aff.getLayoutTokens();
+            String coords = LayoutTokensUtil.getCoordsString(affTokens);
+            if (coords != null && coords.length()>0) {
+                tei.append(" coord=\"" + coords + "\"");
+            }
+        }
         tei.append(">\n");
 
         if (