def getTextSizeFromImage(image):
width, height = image.size
colors = image.getcolors(width * height)
background_color = 255
if colors:
colors.sort()
background_color = (list(colors[-1][1])).pop()
text_sizes = [0]
for i in xrange(getTextBeginHeight(image), height):
current_line = image.crop((0, i - 1, width, i))
current_line.convert('L')
blank_line = True
for i in range(0, current_line.size[0], 3):
color = current_line.getpixel((i, 0))
if colorsContrast(list(color).pop(), background_color):
blank_line = False
break
if blank_line:
if text_sizes[-1]:
text_sizes.append(0)
else:
text_sizes[-1] += 1
text_sizes.sort()
text_sizes = [i for i in text_sizes if i != 0]
text_size = 0
if text_sizes:
text_sizes_avg = sum(text_sizes) / len(text_sizes)
for i in text_sizes:
if i > text_sizes_avg:
text_size = math.floor(i)
break
text_size = max(text_sizes)
debug('Text Size: ', text_size)
return text_size
def getTextSizeFromImage(image):
width, height = image.size
colors = image.getcolors(width * height)
background_color = 255
if colors:
colors.sort()
background_color = (list(colors[-1][1])).pop()
text_sizes = [0]
for i in xrange(getTextBeginHeight(image), height):
current_line = image.crop((0, i - 1, width, i))
current_line.convert('L')
blank_line = True
for i in range(0, current_line.size[0], 3):
color = current_line.getpixel((i, 0))
if colorsContrast(list(color).pop(), background_color):
blank_line = False
break
if blank_line:
if text_sizes[-1]:
text_sizes.append(0)
else:
text_sizes[-1] += 1
text_sizes.sort()
text_sizes = [i for i in text_sizes if i != 0]
text_size = 0
if text_sizes:
text_sizes_avg = sum(text_sizes) / len(text_sizes)
for i in text_sizes:
if i > text_sizes_avg:
text_size = math.floor(i)
break
text_size = max(text_sizes)
debug('Text Size: ', text_size)
return text_size
def getAllBlocks(self):
blocks = self.extendBlocksByBelongingSingles()
blocks = self.unifyBlocks(blocks)
if OCRFEEDER_DEBUG:
for block in blocks:
debug(block)
return blocks
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 getAllBlocks(self):
blocks = self.extendBlocksByBelongingSingles()
blocks = self.unifyBlocks(blocks)
if OCRFEEDER_DEBUG:
for block in blocks:
debug(block)
return blocks
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 __init__(self, path_to_image, window_size=None):
self.window_size = window_size
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:
raise ImageManipulationError
else:
raise IOError
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 key, reviewer in self.image_reviewer_dict.items():
reviewer.page.setSize(size)
else:
current_reviewer.page.setSize(size)
debug('Page size: ', size)
page_size_dialog.destroy()
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 key, reviewer in self.image_reviewer_dict.items():
reviewer.page.setSize(size)
else:
current_reviewer.page.setSize(size)
debug('Page size: ', size)
page_size_dialog.destroy()
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 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 OLDgetTextSizeFromImage(image):
width, height = image.size
global count
debug('Image #%s' % count)
image.save('/tmp/img%s.png' % count, 'PNG')
count += 1
text_sizes = [0]
for i in xrange(getTextBeginHeight(image), height):
current_line = image.crop((0, i - 1, width, i))
current_line.convert('L')
blank_line = True
colors = current_line.getcolors()
background_color = 255
if colors:
colors.sort()
background_color = (list(colors[-1][1])).pop()
for color in colors:
if colorsContrast(list(color[1]).pop(), background_color):
blank_line = False
break
if blank_line:
if text_sizes[-1]:
text_sizes.append(0)
else:
text_sizes[-1] += 1
text_sizes.sort()
text_sizes = [i for i in text_sizes if i != 0]
text_size = 0
if text_sizes:
text_sizes_avg = sum(text_sizes) / len(text_sizes)
for i in text_sizes:
if i > text_sizes_avg:
text_size = math.floor(i)
break
text_size = max(text_sizes)
debug('Text Size: ', text_size)
return text_size
def OLDgetTextSizeFromImage(image):
width, height = image.size
global count
debug('Image #%s' % count)
image.save('/tmp/img%s.png' % count , 'PNG')
count += 1
text_sizes = [0]
for i in xrange(getTextBeginHeight(image), height):
current_line = image.crop((0, i - 1, width, i))
current_line.convert('L')
blank_line = True
colors = current_line.getcolors()
background_color = 255
if colors:
colors.sort()
background_color = (list(colors[-1][1])).pop()
for color in colors:
if colorsContrast(list(color[1]).pop(), background_color):
blank_line = False
break
if blank_line:
if text_sizes[-1]:
text_sizes.append(0)
else:
text_sizes[-1] += 1
text_sizes.sort()
text_sizes = [i for i in text_sizes if i != 0]
text_size = 0
if text_sizes:
text_sizes_avg = sum(text_sizes) / len(text_sizes)
for i in text_sizes:
if i > text_sizes_avg:
text_size = math.floor(i)
break
text_size = max(text_sizes)
debug('Text Size: ', text_size)
return text_size
def performOcr(self, engine_name = None):
selected_engine_index = self.box_editor.getSelectedOcrEngine()
if engine_name:
for i in xrange(len(self.ocr_engines)):
if self.ocr_engines[i][0].name == engine_name:
selected_engine_index = i
break
self.box_editor.selectOcrEngine(selected_engine_index)
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = self.box_editor.getAngle()
if angle:
image = graphics.getImageRotated(image, angle)
if selected_engine_index != None:
engine = self.ocr_engines[selected_engine_index][0]
engine.setImage(image)
text = engine.read()
self.box_editor.setText(text)
debug('Finished reading')
text_size = graphics.getTextSizeFromImage(image)
if text_size:
y_resolution = float(self.reviewer.page.resolution[1])
text_size /= y_resolution
text_size *= 72.0
self.box_editor.setFontSize(math.floor(text_size))
def performOcr(self, engine_name=None):
selected_engine_index = self.box_editor.getSelectedOcrEngine()
if engine_name:
for i in xrange(len(self.ocr_engines)):
if self.ocr_engines[i][0].name == engine_name:
selected_engine_index = i
break
self.box_editor.selectOcrEngine(selected_engine_index)
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = self.box_editor.getAngle()
if angle:
image = graphics.getImageRotated(image, angle)
if selected_engine_index != None:
engine = self.ocr_engines[selected_engine_index][0]
engine.setImage(image)
text = engine.read()
self.box_editor.setText(text)
debug('Finished reading')
text_size = graphics.getTextSizeFromImage(image)
if text_size:
y_resolution = float(self.reviewer.page.resolution[1])
text_size /= y_resolution
text_size *= 72.0
self.box_editor.setFontSize(math.floor(text_size))
def __pressedAngleDetectionButton(self, widget):
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = graphics.getHorizontalAngleForText(image)
debug('ANGLE: ', angle)
self.box_editor.setAngle(angle)
def __pressedAngleDetectionButton(self, widget):
image = graphics.convertPixbufToImage(self.box_editor.getImage())
angle = graphics.getHorizontalAngleForText(image)
debug('ANGLE: ', angle)
self.box_editor.setAngle(angle)
