def __init__(self, main_window, page_data, ocr_engines):
super(ImageReviewer, self).__init__()
self.configuration_manager = ConfigurationManager()
self.main_window = main_window
self.path_to_image = page_data.image_path
self.text_box_fill_color = (94, 156, 235, 150)
self.box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box', self.updatedBoxBounds)
self.selectable_boxes_area.connect('deselected_box',
self.deselectedBoxCb)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.set_position(500)
self.show()
self.ocr_engines = ocr_engines
self.editor = Editor(self.image_pixbuf, self.ocr_engines, self)
self.boxes_dict = {}
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.get_accessible().set_name(
_('Selectable areas'))
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC, gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.pack1(selectable_boxes_scrolled_window, True, False)
self.pack2(self.editor.box_editor, True, False)
self.page = page_data
self.updatePageData(self.page)
selectable_boxes_scrolled_window.connect_after("size-allocate", self.zoomFitCb)
def __init__(self, path_to_image, ocr_engines):
self.path_to_image = path_to_image
self.text_box_fill_color = (94, 156, 235, 150)
self.text_box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.image_box_stroke_color = (94, 156, 235, 250)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box', self.updatedBoxBounds)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.reviewer_area = gtk.HPaned()
self.reviewer_area.set_position(500)
self.reviewer_area.show()
self.boxeditor_notebook = gtk.Notebook()
self.boxeditor_notebook.set_show_tabs(False)
self.boxeditor_notebook.set_show_border(False)
self.boxeditor_notebook.show()
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC, gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.reviewer_area.pack1(selectable_boxes_scrolled_window, True, False)
self.reviewer_area.pack2(self.boxeditor_notebook, True, False)
self.ocr_engines = ocr_engines
self.editor_list = []
self.page = PageData(self.path_to_image)
selectable_boxes_scrolled_window.connect_after("size-allocate", self.zoomFitCb)
def __init__(self, main_window, page_data, ocr_engines):
super(ImageReviewer, self).__init__()
self.configuration_manager = ConfigurationManager()
self.main_window = main_window
self.path_to_image = page_data.image_path
self.text_box_fill_color = (94, 156, 235, 150)
self.box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box',
self.updatedBoxBounds)
self.selectable_boxes_area.connect('deselected_box',
self.deselectedBoxCb)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.set_position(500)
self.show()
self.ocr_engines = ocr_engines
self.editor = Editor(self.image_pixbuf, self.ocr_engines, self)
self.boxes_dict = {}
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.get_accessible().set_name(
_('Selectable areas'))
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC,
gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.pack1(selectable_boxes_scrolled_window, True, False)
self.pack2(self.editor.box_editor, True, False)
self.page = page_data
self.updatePageData(self.page)
selectable_boxes_scrolled_window.connect_after("size-allocate",
self.zoomFitCb)
def __init__(self, path_to_image, ocr_engines):
self.path_to_image = path_to_image
self.text_box_fill_color = (94, 156, 235, 150)
self.text_box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.image_box_stroke_color = (94, 156, 235, 250)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box',
self.updatedBoxBounds)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.reviewer_area = gtk.HPaned()
self.reviewer_area.set_position(500)
self.reviewer_area.show()
self.boxeditor_notebook = gtk.Notebook()
self.boxeditor_notebook.set_show_tabs(False)
self.boxeditor_notebook.set_show_border(False)
self.boxeditor_notebook.show()
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC,
gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.reviewer_area.pack1(selectable_boxes_scrolled_window, True, False)
self.reviewer_area.pack2(self.boxeditor_notebook, True, False)
self.ocr_engines = ocr_engines
self.editor_list = []
self.page = PageData(self.path_to_image)
selectable_boxes_scrolled_window.connect_after("size-allocate",
self.zoomFitCb)
class ImageReviewer:
def __init__(self, path_to_image, ocr_engines):
self.path_to_image = path_to_image
self.text_box_fill_color = (94, 156, 235, 150)
self.text_box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.image_box_stroke_color = (94, 156, 235, 250)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box',
self.updatedBoxBounds)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.reviewer_area = gtk.HPaned()
self.reviewer_area.set_position(500)
self.reviewer_area.show()
self.boxeditor_notebook = gtk.Notebook()
self.boxeditor_notebook.set_show_tabs(False)
self.boxeditor_notebook.set_show_border(False)
self.boxeditor_notebook.show()
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC,
gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.reviewer_area.pack1(selectable_boxes_scrolled_window, True, False)
self.reviewer_area.pack2(self.boxeditor_notebook, True, False)
self.ocr_engines = ocr_engines
self.editor_list = []
self.page = PageData(self.path_to_image)
selectable_boxes_scrolled_window.connect_after("size-allocate",
self.zoomFitCb)
def setTextFillColor(self, color):
self.text_box_fill_color = color
self.selectable_boxes_area.setAreaFillRgba(self.text_box_fill_color)
def setTextStrokeColor(self, color):
self.text_box_stroke_color = color
self.selectable_boxes_area.setAreaStrokeRgba(
self.text_box_stroke_color)
def setImageFillColor(self, color):
self.image_box_fill_color = color
def setImageStrokeColor(self, color):
self.image_box_stroke_color = color
def addBoxEditor(self, box):
editor = Editor(box, self.image_pixbuf, self.ocr_engines, self)
self.editor_list.append(editor)
self.boxeditor_notebook.append_page(editor.box_editor)
return editor
def selectedBox(self, widget, box):
page_num = self.__getPageNumFromBox(box)
if page_num != -1:
self.boxeditor_notebook.set_current_page(page_num)
else:
num_boxes = self.boxeditor_notebook.get_n_pages()
self.addBoxEditor(box)
self.boxeditor_notebook.set_current_page(num_boxes)
def updatedBox(self, widget, box):
for editor in self.editor_list:
if editor.box == box:
editor.update(box)
def updatedBoxBounds(self, widget, box):
for editor in self.editor_list:
if editor.box == box:
editor.updateBounds(box)
def removedBox(self, widget, box):
for i in xrange(len(self.editor_list)):
editor = self.editor_list[i]
if editor.box == box:
page_num = self.boxeditor_notebook.page_num(editor.box_editor)
self.boxeditor_notebook.remove_page(page_num)
del self.editor_list[i]
return True
return False
def __getPageNumFromBox(self, box):
for editor in self.editor_list:
if editor.box == box:
return self.boxeditor_notebook.page_num(editor.box_editor)
return -1
def applyTextColors(self):
self.selectable_boxes_area.fill_color_rgba = self.text_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.text_box_stroke_color
def applyImageColors(self):
self.selectable_boxes_area.fill_color_rgba = self.image_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.image_box_stroke_color
def addNewEditorsToAllBoxes(self):
self.editor_list = []
boxes = self.selectable_boxes_area.getAllAreas()
for box in boxes:
self.addBoxEditor(box)
def performOcrForAllEditors(self, engine=None):
for editor in self.editor_list:
editor.performOcr(engine)
editor.performClassification(engine)
if editor.box_editor.getType() == IMAGE_TYPE:
editor.box_editor.setText('')
def __getAllDataBoxes(self):
boxes = []
for editor in self.editor_list:
editor.setDataBox()
data_box = editor.data_box
boxes.append((data_box.y, data_box))
boxes.sort()
boxes_sorted = []
for y, box in boxes:
boxes_sorted.append(box)
boxes = boxes_sorted
return boxes
def getPageData(self):
self.page.data_boxes = self.__getAllDataBoxes()
return self.page
def updatePageData(self, page_data):
self.page = page_data
for data_box in self.page.data_boxes:
self.addDataBox(data_box)
def addDataBox(self, data_box):
dimensions = (int(data_box.x), int(data_box.y), int(data_box.width),
int(data_box.height))
box = self.selectable_boxes_area.addArea(dimensions)
editor = self.addBoxEditor(box)
editor.box = box
editor.updateDataBox(data_box)
def updateBackgroundImage(self, image_path):
self.path_to_image = image_path
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.selectable_boxes_area.setBackgroundImage(self.path_to_image)
def updateBoxesColors(self):
for editor in self.editor_list:
editor.updateBoxColor()
def zoomFitCb(self, widget, data):
self.zoomFit()
widget.disconnect_by_func(self.zoomFitCb)
def zoomFit(self):
parent = self.selectable_boxes_area.get_parent()
parent_height, parent_width = parent.allocation.height, parent.allocation.width
image_height, image_width = self.selectable_boxes_area.getImageSize()
changed = False
if image_height > parent_height:
image_height = parent_height / image_height
changed = True
if image_width > parent_width:
image_width = parent_width / image_width
changed = True
if changed:
self.selectable_boxes_area.zoom(min(image_height, image_width),
False)
class ImageReviewer(gtk.HPaned):
def __init__(self, main_window, page_data, ocr_engines):
super(ImageReviewer, self).__init__()
self.configuration_manager = ConfigurationManager()
self.main_window = main_window
self.path_to_image = page_data.image_path
self.text_box_fill_color = (94, 156, 235, 150)
self.box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box',
self.updatedBoxBounds)
self.selectable_boxes_area.connect('deselected_box',
self.deselectedBoxCb)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.set_position(500)
self.show()
self.ocr_engines = ocr_engines
self.editor = Editor(self.image_pixbuf, self.ocr_engines, self)
self.boxes_dict = {}
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.get_accessible().set_name(
_('Selectable areas'))
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC,
gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.pack1(selectable_boxes_scrolled_window, True, False)
self.pack2(self.editor.box_editor, True, False)
self.page = page_data
self.updatePageData(self.page)
selectable_boxes_scrolled_window.connect_after("size-allocate",
self.zoomFitCb)
def setTextFillColor(self, color):
self.text_box_fill_color = color
self.selectable_boxes_area.setAreaFillRgba(self.text_box_fill_color)
def setBoxesStrokeColor(self, color):
self.box_stroke_color = color
self.selectable_boxes_area.setAreaStrokeRgba(self.box_stroke_color)
def setImageFillColor(self, color):
self.image_box_fill_color = color
def addBoxEditor(self, box, data_box=None):
if not data_box:
data_box = DataBox()
data_box.updateBoundsFromBox(box)
self.boxes_dict[box] = data_box
self.editor.setBoxes(box, data_box)
def selectedBox(self, widget, box):
data_box = self.boxes_dict.get(box)
if data_box:
self.editor.setBoxes(box, data_box)
else:
self.addBoxEditor(box)
self.updateMainWindow()
def deselectedBoxCb(self, widget, box):
self.updateMainWindow()
def updatedBox(self, widget, box):
if self.editor.box == box:
self.editor.update(box)
def updatedBoxBounds(self, widget, box):
if self.editor.box == box:
self.editor.data_box.updateBoundsFromBox(box)
def removedBox(self, widget, box):
self.updateMainWindow()
if not self.boxes_dict.has_key(box):
return False
del self.boxes_dict[box]
if self.editor.box == box:
self.editor.updateDataBox(None)
self.editor.box = None
self.selectable_boxes_area.selectNextArea()
return False
def clear(self):
self.selectable_boxes_area.clearAreas()
self.editor.updateDataBox(None)
self.editor.box = None
self.boxes_dict.clear()
def applyTextColors(self):
self.selectable_boxes_area.fill_color_rgba = self.text_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.box_stroke_color
def applyImageColors(self):
self.selectable_boxes_area.fill_color_rgba = self.image_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.box_stroke_color
def addNewEditorsToAllBoxes(self):
self.editor_list = []
boxes = self.selectable_boxes_area.getAllAreas()
for box in boxes:
self.addBoxEditor(box)
def performOcrForSelectedBoxes(self):
selected_boxes = self.selectable_boxes_area.getSelectedAreas()
configured_engine = None
for engine, path in self.ocr_engines:
if engine.name == self.configuration_manager.favorite_engine:
configured_engine = engine
break
if not configured_engine:
return
for box in selected_boxes:
data_box = self.boxes_dict[box]
self.performOcrForDataBox(data_box, engine)
self.editor.updateDataBox(self.editor.data_box)
self.updateMainWindow()
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 getTextFromBoxes(self, boxes):
text = ''
#@todo: Implement a proper way to get the boxes
# in reading order
boxes.reverse()
self.editor.saveDataBox()
if boxes:
text = self.page.getTextFromBoxes(
[self.boxes_dict.get(box) for box in boxes])
else:
if self.editor.box_editor.getType() == TEXT_TYPE:
text = self.box_editor.getText()
return text
def copyTextToClipboard(self):
selected_boxes = self.selectable_boxes_area.getSelectedAreas()
text = self.getTextFromBoxes(selected_boxes)
gtk.Clipboard().set_text(text)
def getAllText(self):
boxes = self.selectable_boxes_area.getAllAreas()
return self.getTextFromBoxes(boxes)
def __getAllDataBoxes(self):
boxes = []
for data_box in self.boxes_dict.values():
boxes.append((data_box.y, data_box))
boxes.sort()
boxes_sorted = []
for y, box in boxes:
boxes_sorted.append(box)
boxes = boxes_sorted
return boxes
def savePageData(self):
self.editor.saveDataBox()
self.page.data_boxes = self.__getAllDataBoxes()
return self.page
def updatePageData(self, page_data):
self.page = page_data
for data_box in self.page.data_boxes:
self.addDataBox(data_box)
def addDataBox(self, data_box):
dimensions = (int(data_box.x), int(data_box.y), int(data_box.width),
int(data_box.height))
box = self.selectable_boxes_area.addArea(dimensions)
self.addBoxEditor(box, data_box)
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
self.selectable_boxes_area.setBackgroundImage(self.path_to_image)
class ImageReviewer(gtk.HPaned):
def __init__(self, main_window, page_data, ocr_engines):
super(ImageReviewer, self).__init__()
self.configuration_manager = ConfigurationManager()
self.main_window = main_window
self.path_to_image = page_data.image_path
self.text_box_fill_color = (94, 156, 235, 150)
self.box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box', self.updatedBoxBounds)
self.selectable_boxes_area.connect('deselected_box',
self.deselectedBoxCb)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.set_position(500)
self.show()
self.ocr_engines = ocr_engines
self.editor = Editor(self.image_pixbuf, self.ocr_engines, self)
self.boxes_dict = {}
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.get_accessible().set_name(
_('Selectable areas'))
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC, gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.pack1(selectable_boxes_scrolled_window, True, False)
self.pack2(self.editor.box_editor, True, False)
self.page = page_data
self.updatePageData(self.page)
selectable_boxes_scrolled_window.connect_after("size-allocate", self.zoomFitCb)
def setTextFillColor(self, color):
self.text_box_fill_color = color
self.selectable_boxes_area.setAreaFillRgba(self.text_box_fill_color)
def setBoxesStrokeColor(self, color):
self.box_stroke_color = color
self.selectable_boxes_area.setAreaStrokeRgba(self.box_stroke_color)
def setImageFillColor(self, color):
self.image_box_fill_color = color
def addBoxEditor(self, box, data_box = None):
if not data_box:
data_box = DataBox()
data_box.updateBoundsFromBox(box)
self.boxes_dict[box] = data_box
self.editor.setBoxes(box, data_box)
def selectedBox(self, widget, box):
data_box = self.boxes_dict.get(box)
if data_box:
self.editor.setBoxes(box, data_box)
else:
self.addBoxEditor(box)
self.updateMainWindow()
def deselectedBoxCb(self, widget, box):
self.updateMainWindow()
def updatedBox(self, widget, box):
if self.editor.box == box:
self.editor.update(box)
def updatedBoxBounds(self, widget, box):
if self.editor.box == box:
self.editor.data_box.updateBoundsFromBox(box)
def removedBox(self, widget, box):
self.updateMainWindow()
if not self.boxes_dict.has_key(box):
return False
del self.boxes_dict[box]
if self.editor.box == box:
self.editor.updateDataBox(None)
self.editor.box = None
self.selectable_boxes_area.selectNextArea()
return False
def clear(self):
self.selectable_boxes_area.clearAreas()
self.editor.updateDataBox(None)
self.editor.box = None
self.boxes_dict.clear()
def applyTextColors(self):
self.selectable_boxes_area.fill_color_rgba = self.text_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.box_stroke_color
def applyImageColors(self):
self.selectable_boxes_area.fill_color_rgba = self.image_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.box_stroke_color
def addNewEditorsToAllBoxes(self):
self.editor_list = []
boxes = self.selectable_boxes_area.getAllAreas()
for box in boxes:
self.addBoxEditor(box)
def performOcrForSelectedBoxes(self):
selected_boxes = self.selectable_boxes_area.getSelectedAreas()
configured_engine = None
for engine, path in self.ocr_engines:
if engine.name == self.configuration_manager.favorite_engine:
configured_engine = engine
break
if not configured_engine:
return
for box in selected_boxes:
data_box = self.boxes_dict[box]
self.performOcrForDataBox(data_box, engine)
self.editor.updateDataBox(self.editor.data_box)
self.updateMainWindow()
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 getTextFromBoxes(self, boxes):
text = ''
boxes.reverse()
self.editor.saveDataBox()
if boxes:
text = getTextFromBoxes([self.boxes_dict.get(box) for box in boxes])
else:
if self.editor.box_editor.getType() == TEXT_TYPE:
text = self.box_editor.getText()
return text
def copyTextToClipboard(self):
selected_boxes = self.selectable_boxes_area.getSelectedAreas()
text = self.getTextFromBoxes(selected_boxes)
gtk.Clipboard().set_text(text)
def getAllText(self):
boxes = self.selectable_boxes_area.getAllAreas()
return self.getTextFromBoxes(boxes)
def __getAllDataBoxes(self):
boxes = []
for data_box in self.boxes_dict.values():
boxes.append((data_box.y, data_box))
boxes.sort()
boxes_sorted = []
for y, box in boxes:
boxes_sorted.append(box)
boxes = boxes_sorted
return boxes
def savePageData(self):
self.editor.saveDataBox()
self.page.data_boxes = self.__getAllDataBoxes()
return self.page
def updatePageData(self, page_data):
self.page = page_data
for data_box in self.page.data_boxes:
self.addDataBox(data_box)
def addDataBox(self, data_box):
dimensions = (int(data_box.x), int(data_box.y), int(data_box.width), int(data_box.height))
box = self.selectable_boxes_area.addArea(dimensions)
self.addBoxEditor(box, data_box)
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
self.selectable_boxes_area.setBackgroundImage(self.path_to_image)
class ImageReviewer:
def __init__(self, path_to_image, ocr_engines):
self.path_to_image = path_to_image
self.text_box_fill_color = (94, 156, 235, 150)
self.text_box_stroke_color = (94, 156, 235, 250)
self.image_box_fill_color = (0, 183, 0, 150)
self.image_box_stroke_color = (94, 156, 235, 250)
self.selectable_boxes_area = SelectableBoxesArea(self.path_to_image)
self.selectable_boxes_area.connect('selected_box', self.selectedBox)
self.selectable_boxes_area.connect('removed_box', self.removedBox)
self.selectable_boxes_area.connect('updated_box', self.updatedBox)
self.selectable_boxes_area.connect('dragged_box', self.updatedBoxBounds)
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.reviewer_area = gtk.HPaned()
self.reviewer_area.set_position(500)
self.reviewer_area.show()
self.boxeditor_notebook = gtk.Notebook()
self.boxeditor_notebook.set_show_tabs(False)
self.boxeditor_notebook.set_show_border(False)
self.boxeditor_notebook.show()
selectable_boxes_scrolled_window = gtk.ScrolledWindow()
selectable_boxes_scrolled_window.set_policy(gtk.POLICY_AUTOMATIC, gtk.POLICY_AUTOMATIC)
selectable_boxes_scrolled_window.add(self.selectable_boxes_area)
self.selectable_boxes_area.show()
selectable_boxes_scrolled_window.show()
self.reviewer_area.pack1(selectable_boxes_scrolled_window, True, False)
self.reviewer_area.pack2(self.boxeditor_notebook, True, False)
self.ocr_engines = ocr_engines
self.editor_list = []
self.page = PageData(self.path_to_image)
selectable_boxes_scrolled_window.connect_after("size-allocate", self.zoomFitCb)
def setTextFillColor(self, color):
self.text_box_fill_color = color
self.selectable_boxes_area.setAreaFillRgba(self.text_box_fill_color)
def setTextStrokeColor(self, color):
self.text_box_stroke_color = color
self.selectable_boxes_area.setAreaStrokeRgba(self.text_box_stroke_color)
def setImageFillColor(self, color):
self.image_box_fill_color = color
def setImageStrokeColor(self, color):
self.image_box_stroke_color = color
def addBoxEditor(self, box):
editor = Editor(box, self.image_pixbuf, self.ocr_engines, self)
self.editor_list.append(editor)
self.boxeditor_notebook.append_page(editor.box_editor)
return editor
def selectedBox(self, widget, box):
page_num = self.__getPageNumFromBox(box)
if page_num != -1:
self.boxeditor_notebook.set_current_page(page_num)
else:
num_boxes = self.boxeditor_notebook.get_n_pages()
self.addBoxEditor(box)
self.boxeditor_notebook.set_current_page(num_boxes)
def updatedBox(self, widget, box):
for editor in self.editor_list:
if editor.box == box:
editor.update(box)
def updatedBoxBounds(self, widget, box):
for editor in self.editor_list:
if editor.box == box:
editor.updateBounds(box)
def removedBox(self, widget, box):
for i in xrange(len(self.editor_list)):
editor = self.editor_list[i]
if editor.box == box:
page_num = self.boxeditor_notebook.page_num(editor.box_editor)
self.boxeditor_notebook.remove_page(page_num)
del self.editor_list[i]
return True
return False
def __getPageNumFromBox(self, box):
for editor in self.editor_list:
if editor.box == box:
return self.boxeditor_notebook.page_num(editor.box_editor)
return -1
def applyTextColors(self):
self.selectable_boxes_area.fill_color_rgba = self.text_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.text_box_stroke_color
def applyImageColors(self):
self.selectable_boxes_area.fill_color_rgba = self.image_box_fill_color
self.selectable_boxes_area.stroke_color_rgba = self.image_box_stroke_color
def addNewEditorsToAllBoxes(self):
self.editor_list = []
boxes = self.selectable_boxes_area.getAllAreas()
for box in boxes:
self.addBoxEditor(box)
def performOcrForAllEditors(self, engine = None):
for editor in self.editor_list:
editor.performOcr(engine)
editor.performClassification(engine)
if editor.box_editor.getType() == IMAGE_TYPE:
editor.box_editor.setText('')
def __getAllDataBoxes(self):
boxes = []
for editor in self.editor_list:
editor.setDataBox()
data_box = editor.data_box
boxes.append((data_box.y, data_box))
boxes.sort()
boxes_sorted = []
for y, box in boxes:
boxes_sorted.append(box)
boxes = boxes_sorted
return boxes
def getPageData(self):
self.page.data_boxes = self.__getAllDataBoxes()
return self.page
def updatePageData(self, page_data):
self.page = page_data
for data_box in self.page.data_boxes:
self.addDataBox(data_box)
def addDataBox(self, data_box):
dimensions = (int(data_box.x), int(data_box.y), int(data_box.width), int(data_box.height))
box = self.selectable_boxes_area.addArea(dimensions)
editor = self.addBoxEditor(box)
editor.box = box
editor.updateDataBox(data_box)
def updateBackgroundImage(self, image_path):
self.path_to_image = image_path
self.image_pixbuf = gtk.gdk.pixbuf_new_from_file(self.path_to_image)
self.selectable_boxes_area.setBackgroundImage(self.path_to_image)
def updateBoxesColors(self):
for editor in self.editor_list:
editor.updateBoxColor()
def zoomFitCb(self, widget, data):
self.zoomFit()
widget.disconnect_by_func(self.zoomFitCb)
def zoomFit(self):
parent = self.selectable_boxes_area.get_parent()
parent_height, parent_width = parent.allocation.height, parent.allocation.width
image_height, image_width = self.selectable_boxes_area.getImageSize()
changed = False
if image_height > parent_height:
image_height = parent_height / image_height
changed = True
if image_width > parent_width:
image_width = parent_width / image_width
changed = True
if changed:
self.selectable_boxes_area.zoom(min(image_height, image_width), False)
