Page.java: TextExtractor.java: Add page number to XML output.

CrossRef · Jun 10, 2010 · bc595f6 · bc595f6
1 parent 53ee8ee
commit bc595f6
Show file tree

Hide file tree

Showing 2 changed files with 13 additions and 5 deletions.
diff --git a/src/org/crossref/pdf2xml/TextExtractor.java b/src/org/crossref/pdf2xml/TextExtractor.java
@@ -44,19 +44,20 @@
  */
 public class TextExtractor extends PDFTextStripper {
 
-	private ArrayList<Page> previousPages;
+	private ArrayList<Page> previousPages = new ArrayList<Page>();
 
-	private Page currentPage;
+	private Page currentPage = null;
+
+	private int pageCount = 0;
 
 	public TextExtractor() throws IOException {
 		super();
-		previousPages = new ArrayList<Page>();
 	}
 
 	@Override
 	public void processStream(PDPage aPage, PDResources resources,
 			COSStream cosStream) throws IOException {
-		currentPage = new Page(aPage.findCropBox());
+		currentPage = new Page(aPage.findCropBox(), ++pageCount);
 
 		super.processStream(aPage, resources, cosStream);
 		coalesceRows(currentPage);
@@ -172,6 +173,7 @@ public String toXml() {
 				pageEle.setAttribute("left", String.valueOf(cb.getLowerLeftX()));
 				pageEle.setAttribute("width", String.valueOf(cb.getWidth()));
 				pageEle.setAttribute("height", String.valueOf(cb.getHeight()));
+				pageEle.setAttribute("number", String.valueOf(page.getNumber()));
 				pdf2xml.appendChild(pageEle);
 
 				for (Text t : page.getText()) {

diff --git a/src/org/crossref/pdf2xml/data/Page.java b/src/org/crossref/pdf2xml/data/Page.java
@@ -13,11 +13,13 @@ public class Page {
 	private ArrayList<Text> texts;
 	private HashMap<Float, ArrayList<Text>> yPosMap;
 	private PDRectangle clipBox;
+	private int number;
 
-	public Page(PDRectangle newClipBox) {
+	public Page(PDRectangle newClipBox, int newNumber) {
 		texts = new ArrayList<Text>();
 		yPosMap = new HashMap<Float, ArrayList<Text>>();
 		clipBox = newClipBox;
+		number = newNumber;
 	}
 
 	public void addText(Text t) {
@@ -49,6 +51,10 @@ public List<Text> getText() {
 		return texts;
 	}
 
+	public int getNumber() {
+	    return number;
+	}
+
 	public List<Text> getTextAtY(float y) {
 		Float fObj = new Float(y);
 		if (yPosMap.containsKey(fObj)) {