POI-54722 table text in ppt files

git-svn-id: https://svn.apache.org/repos/asf/poi/trunk@1526960 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Tim Allison 2013-09-27 15:45:55 +00:00
parent ad8b09095e
commit 58dd7d7c0c
3 changed files with 42 additions and 0 deletions

View File

@ -252,6 +252,12 @@ public final class PowerPointExtractor extends POIOLE2TextExtractor {
// Slide text
textRunsToText(ret, slide.getTextRuns());
// Table text
for (Shape shape : slide.getShapes()){
if (shape instanceof Table){
extractTableText(ret, (Table)shape);
}
}
// Slide footer, if set
if (hf != null && hf.isFooterVisible() && hf.getFooterText() != null) {
ret.append(hf.getFooterText() + "\n");
@ -306,6 +312,23 @@ public final class PowerPointExtractor extends POIOLE2TextExtractor {
return ret.toString();
}
private void extractTableText(StringBuffer ret, Table table) {
for (int row = 0; row < table.getNumberOfRows(); row++){
for (int col = 0; col < table.getNumberOfColumns(); col++){
TableCell cell = table.getCell(row, col);
//defensive null checks; don't know if they're necessary
if (cell != null){
String txt = cell.getText();
txt = (txt == null) ? "" : txt;
ret.append(txt);
if (col < table.getNumberOfColumns()-1){
ret.append("\t");
}
}
}
ret.append('\n');
}
}
private void textRunsToText(StringBuffer ret, TextRun[] runs) {
if (runs==null) {
return;

View File

@ -367,4 +367,23 @@ public final class TestExtractor extends POITestCase {
assertEquals(expectText, extractor.getText());
}
}
public void testTable() throws Exception{
ppe = new PowerPointExtractor(slTests.openResourceAsStream("54111.ppt"));
String text = ppe.getText();
String target = "TH Cell 1\tTH Cell 2\tTH Cell 3\tTH Cell 4\n"+
"Row 1, Cell 1\tRow 1, Cell 2\tRow 1, Cell 3\tRow 1, Cell 4\n"+
"Row 2, Cell 1\tRow 2, Cell 2\tRow 2, Cell 3\tRow 2, Cell 4\n"+
"Row 3, Cell 1\tRow 3, Cell 2\tRow 3, Cell 3\tRow 3, Cell 4\n"+
"Row 4, Cell 1\tRow 4, Cell 2\tRow 4, Cell 3\tRow 4, Cell 4\n"+
"Row 5, Cell 1\tRow 5, Cell 2\tRow 5, Cell 3\tRow 5, Cell 4\n";
assertTrue(text.contains(target));
ppe = new PowerPointExtractor(slTests.openResourceAsStream("54722.ppt"));
text = ppe.getText();
target = "this\tText\tis\twithin\ta\n"+
"table\t1\t2\t3\t4";
assertTrue(text.contains(target));
}
}

Binary file not shown.