Changeset 155

Timestamp:

05/09/06 06:55:01 (2 years ago)

Author:

jordi

Message:

The PDF document now can actually search for a text in a page and return the rectangles with the position of the found text. The rectangle is kept in a new class called DocumentRectangle? that just holds the rectangle corners' coordinates.

The DocumentLink? class has been also updated to use the new DocumentRectangle? class instead of holding the rectangles coordinates by itself.

Location:

trunk

Files:

• src/DocumentLink.cxx (modified) (3 diffs)
• src/DocumentLink.h (modified) (1 diff)
• src/DocumentRectangle.cxx (added)
• src/DocumentRectangle.h (added)
• src/Makefile.am (modified) (1 diff)
• src/PDFDocument.cxx (modified) (2 diffs)
• src/epdfview.h (modified) (1 diff)
• tests/Makefile.am (modified) (2 diffs)
• tests/PDFDocumentTest.cxx (modified) (2 diffs)
• tests/PDFDocumentTest.h (modified) (2 diffs)

trunk/src/DocumentLink.cxx

@@ -34 +34 @@
 {
     m_DestinationPage = destinationPage;
-    m_X1 = x1;
-    m_X2 = x2;
-    m_Y1 = y1;
-    m_Y2 = y2;
+    m_Rect = new DocumentRectangle (x1, y1, x2, y2);
 }
 
@@ -45 +42 @@
 DocumentLink::~DocumentLink ()
 {
+    delete m_Rect;
 }
 
@@ -72 +70 @@
 DocumentLink::positionIsOver (gint x, gint y)
 {
-    return ( (gint)m_X1 <= x && (gint)m_Y1 <= y &&
-             (gint)m_X2 >= x && (gint)m_Y2 >= y );
+    return ( (gint)m_Rect->getX1 () <= x && (gint)m_Rect->getY1 () <= y &&
+             (gint)m_Rect->getX2 () >= x && (gint)m_Rect->getY2 () >= y );
 }

trunk/src/DocumentLink.h

@@ -42 +42 @@
             /// The number of the link's destination page.
             gint m_DestinationPage;
-            /// The X coordinate of the link's top-left corner.
-            gdouble m_X1;
-            /// The X coordinate of the link's bottom-right corner.
-            gdouble m_X2;
-            /// The Y coordinate of the link's top-left corner.
-            gdouble m_Y1;
-            /// The Y coordinate of the link's bottom-right corner.
-            gdouble m_Y2;
+            /// The link rectangle.
+            DocumentRectangle *m_Rect;
     };
 }

trunk/src/Makefile.am

@@ -13 +13 @@
     DocumentPage.h      \
     DocumentPage.cxx    \
+    DocumentRectangle.cxx   \
+    DocumentRectangle.h \
     epdfview.h          \
     FindPter.cxx    \

trunk/src/PDFDocument.cxx

@@ -57 +57 @@
 PDFDocument::findTextInPage (gint pageNum, const gchar *textToFind)
 {
-    return NULL;
+    GList *results = NULL;
+    
+    if ( NULL == m_Document )
+    {    
+        return results;
+    }
+
+    PopplerPage *page = poppler_document_get_page (m_Document, pageNum - 1);
+    if ( NULL != page )
+    {
+        gdouble height = 1.0;
+        poppler_page_get_size (page, NULL, &height);
+        gdouble scale = getZoom ();        
+        GList *matches = poppler_page_find_text (page, textToFind);
+        for ( GList *match = g_list_first (matches) ;
+              NULL != match ;
+              match = g_list_next (match) )
+        {
+            PopplerRectangle *matchRect = (PopplerRectangle *)match->data;
+            DocumentRectangle *rect = 
+                new DocumentRectangle (matchRect->x1 * scale,
+                                       (height - matchRect->y2) * scale,
+                                       matchRect->x2 * scale,
+                                       (height - matchRect->y1) * scale);
+            results = g_list_prepend (results, rect);
+        }
+    }
+
+    return g_list_reverse (results);
 }
 
@@ -167 +195 @@
 {
     g_assert (NULL != m_Document && "The document has not been loaded.");
-
 
     gchar *author = NULL;

trunk/src/epdfview.h

@@ -24 +24 @@
 #include <Config.h>
 
+#include <DocumentRectangle.h>
 #include <DocumentLink.h>
 #include <DocumentOutline.h>

trunk/tests/Makefile.am

@@ -25 +25 @@
     FindPterTest.h  \
     main.cxx                \
+    PDFDocumentTest.cxx \
+    PDFDocumentTest.h   \
     Utils.cxx               \
     Utils.h
@@ -32 +34 @@
 #   PagePterTest.cxx
 #   PagePterTest.h
-#   PDFDocumentTest.cxx
-#   PDFDocumentTest.h
 
 test_epdfview_CXXFLAGS =                        \

trunk/tests/PDFDocumentTest.cxx

@@ -621 +621 @@
     gchar *testFile = getTestFile ("test1.pdf");
     m_Document->load (testFile, NULL);
+    g_free (testFile);
     while ( !m_Observer->loadFinished () ) { }
     CPPUNIT_ASSERT (m_Observer->notifiedLoaded ());
@@ -669 +670 @@
     CPPUNIT_ASSERT_EQUAL (5, m_Document->getCurrentPageNum ());
 }
+
+///
+/// @brief Tests finding text in a page.
+///
+void
+PDFDocumentTest::pageFindText ()
+{
+    gchar *testFile = getTestFile ("test1.pdf");
+    m_Document->load (testFile, NULL);
+    g_free (testFile);
+    while ( !m_Observer->loadFinished () ) { }
+
+    // If a text doesn't exist on a page the list of found text is
+    // empty.
+    CPPUNIT_ASSERT (NULL == m_Document->findTextInPage (4, "nothing"));
+    // On the other hand, if something is found the list is filled with
+    // the rectangles the text is found.
+    GList *results = m_Document->findTextInPage (4, "first");
+    CPPUNIT_ASSERT (NULL != results);
+    // In this case it should have found 2 results.
+    CPPUNIT_ASSERT_EQUAL ((guint)2, g_list_length (results));
+
+    GList *firstResult = g_list_first (results);
+    {
+        DocumentRectangle *rect = (DocumentRectangle *)firstResult->data;
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (82.00, rect->getX1 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (137.11, rect->getY1 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (100.34, rect->getX2 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (146.11, rect->getY2 (), 0.0001);
+    }
+    GList *secondResult = g_list_next (firstResult);
+    {
+        DocumentRectangle *rect = (DocumentRectangle *)secondResult->data;
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (96.00, rect->getX1 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (148.11, rect->getY1 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (114.34, rect->getX2 (), 0.0001);
+        CPPUNIT_ASSERT_DOUBLES_EQUAL (157.11, rect->getY2 (), 0.0001);
+    }
+    g_list_free (results);
+}

trunk/tests/PDFDocumentTest.h

@@ -37 +37 @@
         CPPUNIT_TEST (pageRender);
         CPPUNIT_TEST (pageLinks);
+        CPPUNIT_TEST (pageFindText);
         CPPUNIT_TEST_SUITE_END ();
 
@@ -54 +55 @@
             void pageRender (void);
             void pageLinks (void);
+            void pageFindText (void);
             
         private: