Implement getImageItemList

git-svn-id: https://svn.erp5.org/repos/public/erp5/trunk/utils@41090 20353a03-c40f-0410-a6d1-a30d3c3de9de

CHANGES.txt

 1.0.10 (unreleased)
 ===================
+  - Add getImageItemList for OOGranulate
   - Add OdfDocument
   - Add granulate interface.
 
@@ -20,4 +21,3 @@
 =====
   - Remove entry points, treat those as ordinary files.
   - Search all script files using pkg_resources.
-

cloudooo/granulate/oogranulate.py

@@ -27,6 +27,7 @@
 ##############################################################################
 
 from zope.interface import implements
+from cloudooo.document import OdfDocument
 from cloudooo.interfaces.granulate import ITableGranulator, \
                                           IImageGranulator, \
                                           ITextGranulator
@@ -38,6 +39,36 @@ class OOGranulate(object):
 
   implements(ITableGranulator, IImageGranulator, ITextGranulator)
 
+  def __init__(self, file, source_format):
+    self.document = OdfDocument(file, source_format)
+
+  def _getElementsByTagName(self, xml_element, tag):
+    """Returns a list with the xml elements of the given tag
+
+        tag -- tag name with the namespace (e.g. namespace:tag_name)"""
+    return xml_element.xpath('.//%s' % tag, namespaces=xml_element.nsmap)
+
+  def _hasAncestor(self, xml_element, required_ancestor):
+    """Verifies if xml_element have an ancestor tag at a maximum level.
+
+        required_ancestor -- tag name without the namespace"""
+    for ancestor in xml_element.iterancestors():
+      actual_ancestor = ancestor.tag.split('}')[-1]
+      if actual_ancestor == required_ancestor:
+        return True
+    return False
+
+  def _getImageTitle(self, xml_element):
+    """Returns, if exists, the title of the given xml image element"""
+    if self._hasAncestor(xml_element, 'text-box'):
+      draw_frame = xml_element.getparent()
+      text_p = draw_frame.getparent()
+      title = ''
+      for word in text_p.itertext():
+        title += word
+      return title
+    return ''
+
   def getTableItemList(self, file):
     """Returns the list of table IDs in the form of (id, title)."""
     raise NotImplementedError
@@ -50,9 +81,16 @@ class OOGranulate(object):
     """Returns the lines of a given table as (key, value) pairs."""
     raise NotImplementedError
 
-  def getImageItemList(self, file):
-    """Return the list of images in the form of (id, title)."""
-    raise NotImplementedError
+  def getImageItemList(self):
+    """Return a list of tuples with the id and title of image files"""
+    xml_images = self._getElementsByTagName(self.document.parsed_content,
+                                            'draw:image')
+    image_list = []
+    for image in xml_images:
+      title = self._getImageTitle(image)
+      id = image.values()[0].split('/')[-1]
+      image_list.append((id, title))
+    return image_list
 
   def getImage(self, file, image_id, format=None, resolution=None, **kw):
     """Return the given image."""

cloudooo/interfaces/granulate.py

@@ -45,10 +45,10 @@ class ITableGranulator(Interface):
 class IImageGranulator(Interface):
   """Provides methods to granulate a document into images."""
 
-  def getImageItemList(file):
+  def getImageItemList():
     """Return the list of images in the form of (id, title)."""
 
-  def getImage(file, image_id, format=None, resolution=None, **kw):
+  def getImage(image_id, format=None, resolution=None, **kw):
     """Return the given image."""
 
 

cloudooo/tests/testOOGranulate.py

@@ -34,7 +34,33 @@ from cloudooo.granulate.oogranulate import OOGranulate
 class TestOOGranulate(cloudoooTestCase):
 
   def setUp(self):
-    self.oogranulate = OOGranulate()
+    data = open('./data/granulate_test.odt').read()
+    self.oogranulate = OOGranulate(data, 'odt')
+
+  def testGetElementsByTagName(self):
+    """Test if _getElementsByTagName() returns right elements list"""
+    elements = self.oogranulate._getElementsByTagName(
+                                      self.oogranulate.document.parsed_content,
+                                      'draw:image')
+    self.assertEquals(len(elements), 5)
+    for element in elements:
+      self.assertTrue(element.tag.endswith('image'))
+
+  def testHasAncertor(self):
+    images = self.oogranulate._getElementsByTagName(
+                                      self.oogranulate.document.parsed_content,
+                                      'draw:image')
+    self.assertFalse(self.oogranulate._hasAncestor(images[0], 'text-box'))
+    self.assertTrue(self.oogranulate._hasAncestor(images[0], 'frame'))
+    self.assertTrue(self.oogranulate._hasAncestor(images[2], 'text-box'))
+
+  def testGetImageTitle(self):
+    images = self.oogranulate._getElementsByTagName(
+                                      self.oogranulate.document.parsed_content,
+                                      'draw:image')
+    self.assertEquals(self.oogranulate._getImageTitle(images[0]), '')
+    self.assertEquals(self.oogranulate._getImageTitle(images[2]),
+                                                'Illustration 1: TioLive Logo')
 
   def testgetTableItemList(self):
     """Test if getTableItemList() returns the right tables list"""
@@ -55,8 +81,16 @@ class TestOOGranulate(cloudoooTestCase):
 
   def testGetImageItemList(self):
     """Test if getImageItemList() returns the right images list"""
-    self.assertRaises(NotImplementedError, self.oogranulate.getImageItemList,
-                                           'file')
+    image_list = self.oogranulate.getImageItemList()
+    self.assertEquals([
+      ('10000000000000C80000009C38276C51.jpg', ''),
+      ('10000201000000C80000004E7B947D46.png', ''),
+      ('10000201000000C80000004E7B947D46.png', 'Illustration 1: TioLive Logo'),
+      # XXX The svg image are stored into odf as svm
+      ('2000004F00004233000013707E7DE37A.svm', 'Figure 1: Python Logo'),
+      ('10000201000000C80000004E7B947D46.png',
+        'Illustration 2: Again TioLive Logo'),
+      ], image_list)
 
   def testGetImage(self):
     """Test if getImage() returns the right image file"""