def addImages(self, image_path_list):
item_list = []
item_list_length = len(image_path_list)
if not self.configuration_manager.deskew_images_after_addition and \
not self.configuration_manager.unpaper_images_after_addition:
for index in range(0, len(image_path_list)):
if not self.__addImage(image_path_list[index], index == 0):
debug('Failed to load image "%s"' % image_path_list[index])
return
dialog = QueuedEventsProgressDialog(self.main_window.window)
for index in range(0, item_list_length):
image_path = image_path_list[index]
item = AsyncItem(
self.__imagePreProcessing, (image_path, ),
self.__imagePreProcessingFinishedCb,
(dialog, index == item_list_length - 1, index == 0))
if item_list_length == 1:
item_info = (_('Preparing image'), _(u'Please wait…'))
else:
item_info = (_('Preparing image %(current_index)s/%(total)s') % \
{'current_index': index + 1,
'total': item_list_length},
_(u'Please wait…'))
item_list.append((item_info, item))
dialog.setItemsList(item_list)
dialog.run()
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
engine = self.getEngineFromXml(xml_file)
if engine:
self.ocr_engines.append((engine, xml_file))
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
if not len(self.ocr_engines):
lib.debug("Warning: no engines found!")
elif not favorite_engine_exists:
self.configuration_manager.favorite_engine = self.ocr_engines[0][
0].name
engines_needing_update = {'auto': [], 'manual': []}
for engine, path in self.ocr_engines:
path = engine.engine_path
default_conf = \
self.configuration_manager.getEngineDefaultConfiguration(path)
if default_conf is None:
continue
if float(engine.version) < float(default_conf['version']):
update_type = 'manual'
for arguments in default_conf['old_arguments']:
if engine.arguments == arguments:
update_type = 'auto'
break
engines_needing_update[update_type].append({
'engine':
engine,
'configuration':
default_conf
})
return engines_needing_update
def getAllBlocks(self):
blocks = self.extendBlocksByBelongingSingles()
blocks = self.unifyBlocks(blocks)
if OCRFEEDER_DEBUG:
for block in blocks:
debug(block)
return blocks
def performOcrForDataBox(self, data_box, engine):
if engine.hasLanguages():
engine.setLanguage(data_box.getLanguage())
pixbuf_width = self.image_pixbuf.get_width()
pixbuf_height = self.image_pixbuf.get_height()
new_pixbuf_width = min(data_box.getWidth(), pixbuf_width)
new_pixbuf_height = min(data_box.getHeight(), pixbuf_height)
subpixbuf = self.image_pixbuf.new_subpixbuf(data_box.getX(),
data_box.getY(),
new_pixbuf_width,
new_pixbuf_height)
subpixbuf.x = data_box.getX()
subpixbuf.y = data_box.getY()
subpixbuf.width = pixbuf_width
image = graphics.convertPixbufToImage(subpixbuf)
layout_analysis = LayoutAnalysis(engine,
clean_text = self.configuration_manager.clean_text)
text = layout_analysis.readImage(image)
data_box.setText(text)
self.main_window.copy_to_clipboard_menu.set_sensitive(True)
self.main_window.spellchecker_menu.set_sensitive(True)
debug('Finished reading')
text_size = layout_analysis.getTextSizeFromImage(image,
self.page.resolution[1])
if text_size:
data_box.setFontSize(text_size)
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
engine = self.getEngineFromXml(xml_file)
if engine:
self.ocr_engines.append((engine, xml_file))
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
if not len(self.ocr_engines):
lib.debug("Warning: no engines found!")
elif not favorite_engine_exists:
self.configuration_manager.favorite_engine = self.ocr_engines[0][0].name
engines_needing_update = {'auto': [],
'manual': []}
for engine, path in self.ocr_engines:
path = engine.engine_path
default_conf = \
self.configuration_manager.getEngineDefaultConfiguration(path)
if default_conf is None:
continue
if float(engine.version) < float(default_conf['version']):
update_type = 'manual'
for arguments in default_conf['old_arguments']:
if engine.arguments == arguments:
update_type = 'auto'
break
engines_needing_update[update_type].append({'engine': engine,
'configuration': default_conf})
return engines_needing_update
def addImages(self, image_path_list):
item_list = []
temp_dir = self.configuration_manager.TEMPORARY_FOLDER
image_path_list = graphics.convertMultiImagesInList(image_path_list,
temp_dir)
item_list_length = len(image_path_list)
if not self.configuration_manager.deskew_images_after_addition and \
not self.configuration_manager.unpaper_images_after_addition:
for index in range(0, len(image_path_list)):
if not self.__addImage(image_path_list[index], index == 0):
debug('Failed to load image "%s"' % image_path_list[index])
return
dialog = QueuedEventsProgressDialog(self.main_window)
for index in range(0, item_list_length):
image_path = image_path_list[index]
item = AsyncItem(self.__imagePreProcessing,
(image_path,),
self.__imagePreProcessingFinishedCb,
(dialog,
index == item_list_length - 1,
index == 0))
if item_list_length == 1:
item_info = (_('Preparing image'), _(u'Please wait…'))
else:
item_info = (_('Preparing image %(current_index)s/%(total)s') % \
{'current_index': index + 1,
'total': item_list_length},
_(u'Please wait…'))
item_list.append((item_info,item))
dialog.setItemsList(item_list)
dialog.run()
def getAllBlocks(self):
blocks = self.extendBlocksByBelongingSingles()
blocks = self.unifyBlocks(blocks)
if OCRFEEDER_DEBUG:
for block in blocks:
debug(block)
return blocks
def updateBackgroundImage(self, image_path):
self.path_to_image = self.page.image_path = image_path
if not os.path.exists(self.path_to_image):
return
try:
self.image_pixbuf = GdkPixbuf.Pixbuf.new_from_file(self.path_to_image)
except Exception, exception:
debug(exception.message)
return
def save(self):
name = self.name
if not name.lower().endswith('.odt'):
name += '.odt'
self.document.save(name)
for image in self.temp_images:
try:
os.unlink(image)
except:
debug('Error removing image: %s' % image)
def save(self):
name = self.name
if not name.lower().endswith('.odt'):
name += '.odt'
self.document.save(name)
for image in self.temp_images:
try:
os.unlink(image)
except:
debug('Error removing image: %s' % image)
def updateBackgroundImage(self, image_path):
self.path_to_image = self.page.image_path = image_path
if not os.path.exists(self.path_to_image):
return
try:
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(
self.path_to_image)
except Exception, exception:
debug(exception.message)
return
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
try:
engine = self.getEngineFromXml(xml_file)
self.ocr_engines.append((engine, xml_file))
except WrongSettingsForEngine, we:
lib.debug("Cannot load engine at %s: %s" %( xml_file, str(we)))
else:
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
try:
engine = self.getEngineFromXml(xml_file)
self.ocr_engines.append((engine, xml_file))
except WrongSettingsForEngine, we:
lib.debug("Cannot load engine at %s: %s" % (xml_file, str(we)))
else:
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
def getEngineFromXml(self, xml_file_name):
document = ET.parse(xml_file_name)
root_node = document.getroot()
arguments = {}
for child in root_node.getchildren():
arg_name = child.tag
arg_value = child.text
arguments[arg_name] = arg_value
try:
engine = Engine(**arguments)
except TypeError, exception:
lib.debug('Error when unserializing engine: %s', exception.message)
engine = None
def getEngineFromXml(self, xml_file_name):
document = ET.parse(xml_file_name)
root_node = document.getroot()
arguments = {}
for child in root_node.getchildren():
arg_name = child.tag
arg_value = child.text
arguments[arg_name] = arg_value
try:
engine = Engine(**arguments)
except TypeError, exception:
lib.debug('Error when unserializing engine: %s', exception.message)
engine = None
def choosePageSize(self):
current_reviewer = self.__getCurrentReviewer()
current_page = current_reviewer.page
page_size_dialog = PageSizeDialog((current_page.width, current_page.height))
response = page_size_dialog.run()
if response == gtk.RESPONSE_ACCEPT:
size = page_size_dialog.getSize()
if page_size_dialog.all_pages_radio.get_active():
for page in self.source_images_selector_widget.getAllPages():
page.setSize(size)
else:
current_reviewer.page.setSize(size)
debug('Page size: ', size)
page_size_dialog.destroy()
self.__updateStatusBar(current_reviewer)
def choosePageSize(self):
current_reviewer = self.__getCurrentReviewer()
current_page = current_reviewer.page
page_size_dialog = PageSizeDialog(
(current_page.width, current_page.height))
response = page_size_dialog.run()
if response == gtk.RESPONSE_ACCEPT:
size = page_size_dialog.getSize()
if page_size_dialog.all_pages_radio.get_active():
for page in self.source_images_selector_widget.getAllPages():
page.setSize(size)
else:
current_reviewer.page.setSize(size)
debug('Page size: ', size)
page_size_dialog.destroy()
self.__updateStatusBar(current_reviewer)
def addText(self, data_box):
text = data_box.getText()
frame_style = Style(name='FrameStyle', family = 'graphic')
debug('Angle: ', data_box.text_data.angle)
angle = data_box.text_data.angle
if angle:
frame_style = Style(name='FrameStyleRotated', family = 'graphic')
x, y, width, height = data_box.getBoundsPrintSize(self.current_page_resolution)
frame = Frame(stylename = frame_style, width = str(width) + 'in', height = str(height) + 'in', x = str(x) + 'in', y = str(y) + 'in', anchortype = 'paragraph')
if angle:
frame.addAttribute('transform', 'rotate (%s) translate (%scm %scm)' % (abs(math.radians(angle)), x, y))
self.current_page.addElement(frame)
textbox = TextBox()
frame.addElement(textbox)
for line in text.split('\n'):
textbox.addElement(P(stylename = self.__handleFrameStyle(data_box.text_data), text = line))
class OcrEnginesManager:
def __init__(self, configuration_manager):
self.ocr_engines = []
self.configuration_manager = configuration_manager
def getEnginesNames(self):
return [engine.name for engine, path in self.ocr_engines]
def getEnginePath(self, engine):
for eng, path in self.ocr_engines:
if eng == engine:
return path
return None
def replaceEngine(self, engine, new_engine):
for i in xrange(len(self.ocr_engines)):
eng, path = self.ocr_engines[i]
if eng == engine:
new_path = self.engineToXml(new_engine, path)
self.ocr_engines[i] = new_engine, path
return True
return False
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
try:
engine = self.getEngineFromXml(xml_file)
self.ocr_engines.append((engine, xml_file))
except WrongSettingsForEngine, we:
lib.debug("Cannot load engine at %s: %s" % (xml_file, str(we)))
else:
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
if not len(self.ocr_engines):
lib.debug("Warning: no engines found!")
elif not favorite_engine_exists:
self.configuration_manager.favorite_engine = self.ocr_engines[0][
0].name
engines_needing_update = {'auto': [], 'manual': []}
for engine, path in self.ocr_engines:
path = engine.engine_path
default_conf = \
self.configuration_manager.getEngineDefaultConfiguration(path)
if default_conf is None:
continue
if float(engine.version) < float(default_conf['version']):
update_type = 'manual'
for arguments in default_conf['old_arguments']:
if engine.arguments == arguments:
update_type = 'auto'
break
engines_needing_update[update_type].append({
'engine':
engine,
'configuration':
default_conf
})
return engines_needing_update
def addText(self, data_box):
text = data_box.getText()
frame_style = Style(name='FrameStyle', family = 'graphic')
debug('Angle: ', data_box.text_data.angle)
angle = data_box.text_data.angle
if angle:
frame_style = Style(name='FrameStyleRotated', family = 'graphic')
x, y, width, height = data_box.getBoundsPrintSize(self.current_page_resolution)
frame = Frame(stylename = frame_style, width = str(width) + 'in', height = str(height) + 'in', x = str(x) + 'in', y = str(y) + 'in', anchortype = 'paragraph')
if angle:
frame.addAttribute('transform', 'rotate (%s) translate (%scm %scm)' % (abs(math.radians(angle)), x, y))
self.current_page.addElement(frame)
textbox = TextBox()
frame.addElement(textbox)
for line in text.split('\n'):
textbox.addElement(P(stylename = self.__handleFrameStyle(data_box.text_data), text = line))
def choosePageSize(self):
current_reviewer = self.__getCurrentReviewer()
current_page = current_reviewer.page
page_size_dialog = PageSizeDialog(self.main_window,
(current_page.width, current_page.height))
response = page_size_dialog.run()
if response == Gtk.ResponseType.ACCEPT:
size = page_size_dialog.getSize()
if page_size_dialog.all_pages_radio.get_active():
for page in self.pages_icon_view.getAllPages():
page.setSize(size)
else:
current_reviewer.page.setSize(size)
debug('Page size: ', size)
page_size_dialog.destroy()
self.__updateStatusBar(current_reviewer)
def performOcrForDataBox(self, data_box, engine):
pixbuf_width = self.image_pixbuf.get_width()
pixbuf_height = self.image_pixbuf.get_height()
subpixbuf = self.image_pixbuf.subpixbuf(
data_box.getX(), data_box.getY(),
min(data_box.getWidth(), pixbuf_width),
min(data_box.getHeight(), pixbuf_height))
image = graphics.convertPixbufToImage(subpixbuf)
layout_analysis = LayoutAnalysis(
engine, clean_text=self.configuration_manager.clean_text)
text = layout_analysis.readImage(image)
data_box.setText(text)
self.main_window.copy_to_clipboard_menu.set_sensitive(True)
self.main_window.spellchecker_menu.set_sensitive(True)
debug('Finished reading')
text_size = layout_analysis.getTextSizeFromImage(
image, self.page.resolution[1])
if text_size:
data_box.setFontSize(text_size)
def addText(self, box):
x, y, width, height = box.getBoundsPrintSize(self.page_data.resolution)
text = self.canvas.beginText()
# Make the text transparent if we are not
# creating a PDF from scratch
if not self._from_scratch:
text.setTextRenderMode(3)
text.setTextOrigin(x * units.inch,
(self.page_data.height - y) * units.inch)
text.setCharSpace(box.text_data.letter_space)
text.setLeading(box.text_data.line_space + box.text_data.size)
text.moveCursor(0, box.text_data.size)
try:
self.setFont(box.text_data.face,
box.text_data.size)
except:
debug('Error setting font %s' % box.text_data.face)
self.canvas.setFontSize(box.text_data.size)
text.textLines(box.text)
self.canvas.drawText(text)
def loadConfiguration(self, folder = None):
folder = folder or self.configuration_dir
project_xml = os.path.join(folder, 'project.xml')
if not project_xml:
return None
document = minidom.parse(project_xml)
root_node = document.documentElement
images_node = document.getElementsByTagName('image')
images = self.__getImagesInfo(images_node)
page_data_nodes = document.getElementsByTagName('PageData')
pages = []
for page_data in self.__getPageDataInfo(page_data_nodes):
debug('Page Data:', page_data)
data_boxes = []
for data_box in page_data['data_boxes']:
args = []
# text variable is to avoid problems with
# escaping characters
text = ''
for var_name, value in data_box.items():
if var_name == 'text':
text = value
continue
real_value = '"""%s"""' % re.escape(value)
try:
real_value = int(value)
except ValueError:
pass
args.append('%s = %s' % (var_name, real_value))
exec('box = DataBox(%s)' % ', '.join(args))
box.text = text
data_boxes.append(box)
image_path = page_data['image_path']
if not os.path.exists(image_path):
image_path = os.path.join(self.configuration_dir, 'images', images[image_path])
page = PageData(image_path, data_boxes)
pages.append(page)
return pages
def loadConfiguration(self, folder = None):
folder = folder or self.configuration_dir
project_xml = os.path.join(folder, 'project.xml')
if not project_xml:
return None
document = minidom.parse(project_xml)
root_node = document.documentElement
images_node = document.getElementsByTagName('image')
images = self.__getImagesInfo(images_node)
page_data_nodes = document.getElementsByTagName('PageData')
pages = []
for page_data in self.__getPageDataInfo(page_data_nodes):
debug('Page Data:', page_data)
data_boxes = []
for data_box in page_data['data_boxes']:
args = []
# text variable is to avoid problems with
# escaping characters
text = ''
for var_name, value in data_box.items():
if var_name == 'text':
text = value
continue
real_value = '"""%s"""' % re.escape(value)
try:
real_value = int(value)
except ValueError:
pass
args.append('%s = %s' % (var_name, real_value))
exec('box = DataBox(%s)' % ', '.join(args))
box.text = text
data_boxes.append(box)
image_path = page_data['image_path']
if not os.path.exists(image_path):
image_path = os.path.join(self.configuration_dir, 'images', images[image_path])
page = PageData(image_path, data_boxes)
pages.append(page)
return pages
def __init__(self, path_to_image,
window_size = None, contrast_tolerance = 120):
self.window_size = window_size
self.contrast_tolerance = contrast_tolerance
error_message = _("A problem occurred while trying to open the image:\n %s\n"
"Ensure the image exists or try converting it to another format.") % path_to_image
if os.path.isfile(path_to_image):
try:
self.original_image = Image.open(path_to_image)
self.black_n_white_image = self.original_image.convert('L')
if not self.window_size:
self.window_size = self.original_image.size[1] / 60.
debug('Window Size: ', self.window_size)
except:
debug(sys.exc_info())
raise ImageManipulationError(error_message)
else:
debug(sys.exc_info())
raise ImageManipulationError(error_message)
self.bg_color = 255
def __pressedAngleDetectionButton(self, widget):
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = graphics.getHorizontalAngleForText(image)
debug('ANGLE: ', angle)
self.box_editor.setAngle(angle)
class OcrEnginesManager:
def __init__(self, configuration_manager):
self.ocr_engines = []
self.configuration_manager = configuration_manager
def getEnginesNames(self):
return [engine.name for engine, path in self.ocr_engines]
def getEnginePath(self, engine):
for eng, path in self.ocr_engines:
if eng == engine:
return path
return None
def replaceEngine(self, engine, new_engine):
for i in xrange(len(self.ocr_engines)):
eng, path = self.ocr_engines[i]
if eng == engine:
new_path = self.engineToXml(new_engine, path)
self.ocr_engines[i] = new_engine, path
return True
return False
def makeEnginesFromFolder(self, folder):
self.ocr_engines = []
favorite_engine_exists = False
for xml_file in self.getXmlFilesInFolder(folder):
engine = self.getEngineFromXml(xml_file)
if engine:
self.ocr_engines.append((engine, xml_file))
favorite_engine_exists = favorite_engine_exists or \
self.configuration_manager.favorite_engine == engine.name
if not len(self.ocr_engines):
lib.debug("Warning: no engines found!")
elif not favorite_engine_exists:
self.configuration_manager.favorite_engine = self.ocr_engines[0][
0].name
engines_needing_update = {'auto': [], 'manual': []}
for engine, path in self.ocr_engines:
path = engine.engine_path
default_conf = \
self.configuration_manager.getEngineDefaultConfiguration(path)
if default_conf is None:
continue
if float(engine.version) < float(default_conf['version']):
update_type = 'manual'
for arguments in default_conf['old_arguments']:
if engine.arguments == arguments:
update_type = 'auto'
break
engines_needing_update[update_type].append({
'engine':
engine,
'configuration':
default_conf
})
return engines_needing_update
def migrateEngine(self, engine, configuration, only_version=False):
if not only_version:
engine.arguments = configuration['arguments']
engine.language_argument = configuration['language_argument']
engine.setLanguages(configuration['languages'])
engine.version = configuration['version']
self.replaceEngine(engine, engine)
def getEngineFromXml(self, xml_file_name):
document = ET.parse(xml_file_name)
root_node = document.getroot()
arguments = {}
for child in root_node.getchildren():
arg_name = child.tag
arg_value = child.text
arguments[arg_name] = arg_value
try:
engine = Engine(**arguments)
except TypeError, exception:
lib.debug('Error when unserializing engine: %s', exception.message)
engine = None
except WrongSettingsForEngine, we:
lib.debug("Cannot load engine at %s: %s" %
(xml_file_name, str(we)))
engine = None
def removeTemporaryFolder(self):
try:
shutil.rmtree(self.TEMPORARY_FOLDER)
except:
debug('Error when removing the temporary folder: ' + \
self.TEMPORARY_FOLDER)
def __pressedAngleDetectionButton(self, widget):
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = graphics.getHorizontalAngleForText(image)
debug('ANGLE: ', angle)
self.box_editor.setAngle(angle)
def removeTemporaryFolder(self):
try:
shutil.rmtree(self.TEMPORARY_FOLDER)
except:
debug('Error when removing the temporary folder: ' + \
self.TEMPORARY_FOLDER)