proper affiliation coordinates

kermitt2 · Dec 13, 2023 · d7b2ff1 · d7b2ff1
1 parent 8ade3c0
commit d7b2ff1
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 8 deletions.
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
@@ -5,6 +5,8 @@
 import org.grobid.core.layout.LayoutToken;
 import org.grobid.core.utilities.OffsetPosition;
 import org.grobid.core.engines.label.TaggingLabel;
+import org.grobid.core.engines.config.GrobidAnalysisConfig;
+import org.grobid.core.utilities.LayoutTokensUtil;
 
 import java.util.*;
 
@@ -355,7 +357,7 @@ public void setLayoutTokens(List<LayoutToken> tokens) {
 
     public void appendLayoutTokens(List<LayoutToken> tokens) {
         if (this.layoutTokens == null)
-            layoutTokens = new ArrayList<>();
+            this.layoutTokens = new ArrayList<>();
         this.layoutTokens.addAll(tokens);
     }
 
@@ -572,11 +574,29 @@ public String toTEI() {
     }*/
 
     public static String toTEI(Affiliation aff, int nbTag) {
+        return toTEI(aff, nbTag, null);
+    }
+
+    public static String toTEI(Affiliation aff, int nbTag, GrobidAnalysisConfig config) {
         StringBuffer tei = new StringBuffer();
         TextUtilities.appendN(tei, '\t', nbTag + 1);
+
+        boolean withAffCoords = (config != null) && 
+                                (config.getGenerateTeiCoordinates() != null) && 
+                                (config.getGenerateTeiCoordinates().contains("affiliation"));
+        boolean orgNameCoords = (config != null) && 
+                                (config.getGenerateTeiCoordinates() != null) && 
+                                (config.getGenerateTeiCoordinates().contains("orgName"));
+
         tei.append("<affiliation");
         if (aff.getKey() != null)
             tei.append(" key=\"").append(aff.getKey()).append("\"");
+        if (withAffCoords) {
+            String coords = LayoutTokensUtil.getCoordsString(aff.getLayoutTokens());
+            if (coords != null && coords.length()>0) {
+                tei.append(" coord=\"" + coords + "\"");
+            }
+        }
         tei.append(">\n");
 
         if (aff.getDepartments() != null) {

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
@@ -3779,17 +3779,20 @@ 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"));
+        boolean affiliationWithCoords = (config != null) && 
+                                        (config.getGenerateTeiCoordinates() != null) && 
+                                        (config.getGenerateTeiCoordinates().contains("affiliation"));
+        boolean orgnameWithCoords = (config != null) && 
+                                    (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);
+            String coords = LayoutTokensUtil.getCoordsString(aff.getLayoutTokens());
             if (coords != null && coords.length()>0) {
                 tei.append(" coord=\"" + coords + "\"");
             }
@@ -3824,8 +3827,8 @@ private void appendAffiliation(
                 int q = 1;
                 for (String depa : aff.getDepartments()) {
                     TextUtilities.appendN(tei, '\t', nbTag + 1);
-                    tei.append("<orgName type=\"department\" key=\"dep" + q + "\">" +
-                            TextUtilities.HTMLEncode(depa) + "</orgName>\n");
+                    tei.append("<orgName type=\"department\" key=\"dep" + q + "\"");
+                    tei.append(">" +TextUtilities.HTMLEncode(depa) + "</orgName>\n");
                     q++;
                 }
             }

diff --git a/grobid-core/src/main/java/org/grobid/core/engines/AffiliationAddressParser.java b/grobid-core/src/main/java/org/grobid/core/engines/AffiliationAddressParser.java
@@ -327,6 +327,10 @@ protected List<Affiliation> resultExtractionLayoutTokens(String result,
                 }
                 affiliation.addLabeledResult(TaggingLabels.AFFILIATION_ADDRESSLINE, tokens);
             } 
+
+            if (!clusterLabel.equals(TaggingLabels.OTHER) && affiliation.isNotNull()) {
+                affiliation.appendLayoutTokens(tokens);
+            }
         }
 
         // last affiliation

diff --git a/grobid-core/src/main/java/org/grobid/core/engines/label/TaggingLabels.java b/grobid-core/src/main/java/org/grobid/core/engines/label/TaggingLabels.java
@@ -295,7 +295,8 @@ public class TaggingLabels {
     public static final TaggingLabel AFFILIATION_REGION = new TaggingLabelImpl(GrobidModels.AFFILIATION_ADDRESS, REGION_LABEL);
     public static final TaggingLabel AFFILIATION_SETTLEMENT = new TaggingLabelImpl(GrobidModels.AFFILIATION_ADDRESS, SETTLEMENT_LABEL);
     public static final TaggingLabel AFFILIATION_ADDRESSLINE = new TaggingLabelImpl(GrobidModels.AFFILIATION_ADDRESS, ADDRESSLINE_LABEL);
-
+    public static final TaggingLabel AFFILIATION_OTHER = new TaggingLabelImpl(GrobidModels.AFFILIATION_ADDRESS, OTHER_LABEL);
+
     protected static void register(TaggingLabel label) {
         cache.putIfAbsent(new Pair<>(label.getGrobidModel(), label.getLabel()), label);
     }
@@ -455,6 +456,7 @@ protected static void register(TaggingLabel label) {
         register(AFFILIATION_REGION);
         register(AFFILIATION_SETTLEMENT);
         register(AFFILIATION_ADDRESSLINE);
+        register(AFFILIATION_OTHER);
     }
 
     protected TaggingLabels() {