Fix infinite loop in table detection

veraPDF · May 2, 2024 · ff06d95 · ff06d95
1 parent 3726528
commit ff06d95
Show file tree

Hide file tree

Showing 3 changed files with 34 additions and 1 deletion.
diff --git a/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/tables/TableClusterGap.java b/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/tables/TableClusterGap.java
@@ -1,5 +1,7 @@
 package org.verapdf.wcag.algorithms.semanticalgorithms.tables;
 
+import java.util.Objects;
+
 public class TableClusterGap {
     private TableCluster link;
     private double gap;
@@ -24,4 +26,22 @@ public void setGap(double gap) {
     public double getGap() {
         return gap;
     }
+
+    @Override
+    public int hashCode() {
+        return Objects.hash(link.hashCode(), gap);
+    }
+
+    @Override
+    public boolean equals(Object o) {
+        if (this == o) {
+            return true;
+        }
+        if (!super.equals(o)) {
+            return false;
+        }
+        TableClusterGap that = (TableClusterGap) o;
+        return Objects.equals(gap, that.gap) &&
+                Objects.equals(link.getId(), that.getLink().getId());
+    }
 }
diff --git a/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/tables/TableRecognizer.java b/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/tables/TableRecognizer.java
@@ -256,6 +256,7 @@ private List<TableCluster> mergeInitialClusters(List<List<TableCluster>> cluster
                 if (j < clusterRow.size() - 1) {
                     TableCluster nextCluster = clusterRow.get(j + 1);
 
+                    //TODO: this gap could be negative, for example, if clusters intersect
                     double gap = nextCluster.getLeftX() - cluster.getRightX();
                     cluster.getFirstRow().setRightGap(new TableClusterGap(nextCluster, gap));
                     nextCluster.getFirstRow().setLeftGap(new TableClusterGap(cluster, gap));

diff --git a/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/utils/TableUtils.java b/src/main/java/org/verapdf/wcag/algorithms/semanticalgorithms/utils/TableUtils.java
@@ -101,15 +101,27 @@ public static boolean isWeakCluster(TableCluster cluster, List<TableCluster> hea
         if (cluster.getHeader() != null) {
             return false;
         }
-
+        Set<TableClusterGap> visitedGaps = new HashSet<>();
         TableClusterGap gap = cluster.getMinLeftGap();
         while (gap != null && gap.getLink().getHeader() == null) {
+            if (visitedGaps.contains(gap)) {
+                gap = null;
+                break;
+            }
+            visitedGaps.add(gap);
             gap = gap.getLink().getMinLeftGap();
         }
         TableCluster leftHeader = (gap == null) ? null : gap.getLink().getHeader();
 
+        visitedGaps.clear();
+
         gap = cluster.getMinRightGap();
         while (gap != null && gap.getLink().getHeader() == null) {
+            if (visitedGaps.contains(gap)) {
+                gap = null;
+                break;
+            }
+            visitedGaps.add(gap);
             gap = gap.getLink().getMinRightGap();
         }
         TableCluster rightHeader = (gap == null) ? null : gap.getLink().getHeader();