Refactoring classes related to a pdf page.

ZaqueuCavalcante · jazzido · commit 008c395b9af4 · 2020-11-24T17:59:14.000-03:00
diff --git a/src/main/java/technology/tabula/ObjectExtractor.java b/src/main/java/technology/tabula/ObjectExtractor.java
@@ -13,55 +13,50 @@ public ObjectExtractor(PDDocument pdfDocument) {
         this.pdfDocument = pdfDocument;
     }
 
+    // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - //
     protected Page extractPage(Integer pageNumber) throws IOException {
-
-        if (pageNumber > this.pdfDocument.getNumberOfPages() || pageNumber < 1) {
-            throw new java.lang.IndexOutOfBoundsException(
-                    "Page number does not exist");
+        if (pageNumber > pdfDocument.getNumberOfPages() || pageNumber < 1) {
+            throw new java.lang.IndexOutOfBoundsException("Page number does not exist.");
         }
+        PDPage page = pdfDocument.getPage(pageNumber - 1);
 
-        PDPage p = this.pdfDocument.getPage(pageNumber - 1);
-
-        ObjectExtractorStreamEngine se = new ObjectExtractorStreamEngine(p);
-        se.processPage(p);
-
-
-        TextStripper pdfTextStripper = new TextStripper(this.pdfDocument, pageNumber);
+        ObjectExtractorStreamEngine streamEngine = new ObjectExtractorStreamEngine(page);
+        streamEngine.processPage(page);
 
-        pdfTextStripper.process();
+        TextStripper textStripper = new TextStripper(pdfDocument, pageNumber);
+        textStripper.process();
 
-        Utils.sort(pdfTextStripper.textElements, Rectangle.ILL_DEFINED_ORDER);
+        Utils.sort(textStripper.textElements, Rectangle.ILL_DEFINED_ORDER);
 
-        float w, h;
-        int pageRotation = p.getRotation();
-        if (Math.abs(pageRotation) == 90 || Math.abs(pageRotation) == 270) {
-            w = p.getCropBox().getHeight();
-            h = p.getCropBox().getWidth();
+        float width, height;
+        int rotation = page.getRotation();
+        if (Math.abs(rotation) == 90 || Math.abs(rotation) == 270) {
+            width = page.getCropBox().getHeight();
+            height = page.getCropBox().getWidth();
         } else {
-            w = p.getCropBox().getWidth();
-            h = p.getCropBox().getHeight();
+            width = page.getCropBox().getWidth();
+            height = page.getCropBox().getHeight();
         }
 
-        return new Page(0, 0, w, h, pageRotation, pageNumber, p, this.pdfDocument, pdfTextStripper.textElements,
-                se.rulings, pdfTextStripper.minCharWidth, pdfTextStripper.minCharHeight, pdfTextStripper.spatialIndex);
+        return new Page(0, 0, width, height, rotation, pageNumber, page, pdfDocument, streamEngine, textStripper);
     }
 
+    // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - //
     public PageIterator extract(Iterable<Integer> pages) {
         return new PageIterator(this, pages);
     }
 
     public PageIterator extract() {
-        return extract(Utils.range(1, this.pdfDocument.getNumberOfPages() + 1));
+        return extract(Utils.range(1, pdfDocument.getNumberOfPages() + 1));
     }
 
     public Page extract(int pageNumber) {
         return extract(Utils.range(pageNumber, pageNumber + 1)).next();
     }
 
+    // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - //
     public void close() throws IOException {
-        this.pdfDocument.close();
+        pdfDocument.close();
     }
-
-
-
+    
 }
diff --git a/src/main/java/technology/tabula/Page.java b/src/main/java/technology/tabula/Page.java
@@ -39,6 +39,13 @@ public Page(float top, float left, float width, float height, int rotation, int
     this.rulings = rulings;
   }
 
+  public Page(float top, float left, float width, float height, int rotation, int number, PDPage pdPage, PDDocument doc,
+              ObjectExtractorStreamEngine streamEngine, TextStripper textStripper) {
+    this(top, left, width, height, rotation, number, pdPage, doc, textStripper.textElements, streamEngine.rulings);
+    this.minCharWidth = textStripper.minCharWidth;
+    this.minCharHeight = textStripper.minCharHeight;
+    this.spatial_index = textStripper.spatialIndex;
+  }
 
   public Page(float top, float left, float width, float height, int rotation, int page_number, PDPage pdPage, PDDocument doc,
               List<TextElement> characters, List<Ruling> rulings,
diff --git a/src/main/java/technology/tabula/PageIterator.java b/src/main/java/technology/tabula/PageIterator.java
@@ -5,39 +5,39 @@
 
 public class PageIterator implements Iterator<Page> {
 
-    private ObjectExtractor oe;
+    private ObjectExtractor objectExtractor;
     private Iterator<Integer> pageIndexIterator;
-    
-    public PageIterator(ObjectExtractor oe, Iterable<Integer> pages) {
+
+    public PageIterator(ObjectExtractor objectExtractor, Iterable<Integer> pages) {
         super();
-        this.oe = oe;
+        this.objectExtractor = objectExtractor;
         this.pageIndexIterator = pages.iterator();
     }
 
+    // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - //
     @Override
     public boolean hasNext() {
-        return this.pageIndexIterator.hasNext();
+        return pageIndexIterator.hasNext();
     }
 
     @Override
     public Page next() {
-        Page page = null;
+        Page nextPage = null;
         if (!this.hasNext()) {
             throw new IllegalStateException();
         }
         try {
-            page = oe.extractPage(this.pageIndexIterator.next());
+            nextPage = objectExtractor.extractPage(pageIndexIterator.next());
         } catch (IOException e) {
-            // TODO Auto-generated catch block
             e.printStackTrace();
         }
-        return page;
+        return nextPage;
     }
 
+    // - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - //
     @Override
     public void remove() {
         throw new UnsupportedOperationException();
-
     }
 
-}
+}
