first attempt to fix tables on the fly

kermitt2 · Dec 4, 2024 · d95a1ac · d95a1ac
1 parent 37dcec1
commit d95a1ac
Show file tree

Hide file tree

Showing 3 changed files with 216 additions and 40 deletions.
diff --git a/grobid-core/src/main/java/org/grobid/core/data/Table.java b/grobid-core/src/main/java/org/grobid/core/data/Table.java
@@ -43,6 +43,9 @@
 public class Table extends Figure {
 	private List<LayoutToken> contentTokens = new ArrayList<>();
 	private List<LayoutToken> fullDescriptionTokens = new ArrayList<>();
+
+    // Contains the raw layoutTokens from the fulltext model
+    private List<LayoutToken> rawLayoutTokens = new ArrayList<>();
 	private boolean goodTable = true;
 
     private StringBuilder note = null;
@@ -423,4 +426,12 @@ public boolean isGoodTable() {
     public String getTeiId() {
         return "tab_" + this.id;
     }
+
+    public List<LayoutToken> getRawLayoutTokens() {
+        return rawLayoutTokens;
+    }
+
+    public void setRawLayoutTokens(List<LayoutToken> rawLayoutTokens) {
+        this.rawLayoutTokens = rawLayoutTokens;
+    }
 }
diff --git a/grobid-core/src/main/java/org/grobid/core/document/Document.java b/grobid-core/src/main/java/org/grobid/core/document/Document.java
@@ -874,6 +874,7 @@ public static List<GraphicObject> getConnectedGraphics(Block block, Document doc
     public void postProcessTables() {
         for (Table table : tables) {
             if (!table.firstCheck()) {
+                table.setGoodTable(false);
                 continue;
             }
 
@@ -919,7 +920,7 @@ public void postProcessTables() {
             table.getContentTokens().clear();
             table.getContentTokens().addAll(contentResult);
 
-            table.secondCheck();
+            table.setGoodTable(table.secondCheck());
         }
     }