Details
-
Bug
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
Description
If a table cell has the role TH it gets a scope info as PDF tag attribute (Entry A) with the name /Scope and a value Column (default), Row or Both.
This works fine, but if the fo:table-cell has an additional number-columns-spanned attribute, the Scope info is lost.
The attached FO document (scope-with-colspan.fo
) shows two table head cells only one with number-columns-spanned=2.
The screenshot shows the tag property view in Acrobat of the rendered PDF (scope-with-colspan.pdf) for both cells.
One has the ColSpan attribute but no Scope info:
The other with no ColSpan attribute has the Scope info:
The reason for the behavior of FOP is that the scope is written directly as "A"-Entry:
static void addScopeAttribute(PDFStructElem th, Scope scope) {
PDFDictionary scopeAttribute = new PDFDictionary();
scopeAttribute.put("O", Table.NAME);
scopeAttribute.put("Scope", scope.getName());
th.put("A", scopeAttribute);
}
Source: StandardStructureAttributes.java:58
But the colspan/rowspan attributes are collected as a special attribute field of PDFStructElem:
public void setTableAttributeColSpan(int colSpan) {
setTableAttributeRowColumnSpan("ColSpan", colSpan);
}
public void setTableAttributeRowSpan(int rowSpan) {
setTableAttributeRowColumnSpan("RowSpan", rowSpan);
}
private void setTableAttributeRowColumnSpan(String typeSpan, int span) {
PDFDictionary attribute = new PDFDictionary();
attribute.put("O", Table.NAME);
attribute.put(typeSpan, span);
if (attributes == null) {
attributes = new ArrayList<PDFDictionary>(2);
}
attributes.add(attribute);
}
Source: PDFStructElem.java:230
At the end the field attribute overwrites the "A"-Entry:
private void attachAttributes() {
if (attributes != null) {
if (attributes.size() == 1) {
put("A", attributes.get(0));
} else {
PDFArray array = new PDFArray(attributes);
put("A", array);
}
}
}
Source: PDFStructElem.java:174
I will see If I can produce a patch in the next days.
