def pdf2text(pdf_filename):

    tool = pyocr.get_available_tools()[0]
    lang = tool.get_available_languages()[1]

    req_image = []
    final_text = []

    image_pdf = Image(filename=pdf_filename, resolution=300)
    image_jpeg = image_pdf.convert('jpeg')

    for img in image_jpeg.sequence:
        img_page = Image(image=img)
        req_image.append(img_page.make_blob('jpeg'))

    for img in req_image:

      txt = tool.image_to_string(
          PI.open(io.BytesIO(img)),
          lang=lang,
          builder=pyocr.builders.TextBuilder()
      )

      final_text.append(txt)

    return final_text

def build_config_info():
	'''Builds configuration information about installed OCR software'''

	tools = pyocr.get_available_tools()
	infos = [{'name': tool.get_name(), 'langs': tool.get_available_languages()} for tool in tools]

	return infos

	def readFile(self, pdfFile, mode = 'top-right' ):
		mode = 0 if mode == 'top-right' else 1
		self.__img_crop = self.__img_crop_modes[mode]
		tool = pyocr.get_available_tools()[0]
		lang = tool.get_available_languages()[1]

		# this could've been done better, but for now will do
		reCaseSubmitted = re.compile(ur'The case is submitted\.\s*\(Whereupon,', re.UNICODE)
		reCaseSubmitted2 = re.compile(ur'Case is submitted\.\s*\(Whereupon,', re.UNICODE)
		reCaseSubmitted3 = re.compile(ur'\(?Whereupon, at \w\w:\w\w', re.UNICODE) # the most-relaxed

		# get file name
		helper = FileHelper()
		outfp = open(self.__outputDir + helper.GetFileName(pdfFile) + ".plain", 'w')

		with Image(filename=pdfFile, resolution=self.__dpi) as image_pdf:
			image_pngs = image_pdf.convert('png')
			idx = 0
			output_text = ''
			for img in image_pngs.sequence:
				if self.__debug:
					print "Parsing Page: " + str(idx + 1)
				cloneImg = img[self.__img_crop[0] : self.__img_crop[2], self.__img_crop[1] : self.__img_crop[3] ]
				cloneImg.alpha_channel = False
#				cloneImg.save(filename = './img_{}.png'.format(idx))
				self.evaluate( cloneImg, 'threshold', self.__threshold)

				txt = tool.image_to_string( PI.open(io.BytesIO(cloneImg.make_blob('png'))), lang=lang, builder=pyocr.builders.TextBuilder())+ "\n"
				output_text = output_text + self._clean_text(txt)
				if reCaseSubmitted.search(txt) != None or reCaseSubmitted2.search(txt) != None or reCaseSubmitted3.search(txt) != None:
					break
				idx += 1

		outfp.write(output_text)

	def testPdf(self, pdfFile, border):
		reCaseSubmitted = re.compile(ur'The case is submitted\.\s*\(Whereupon,', re.UNICODE)
		reCaseSubmitted2 = re.compile(ur'Case is submitted\.\s*\(Whereupon,', re.UNICODE)
		reCaseSubmitted3 = re.compile(ur'\(?Whereupon, at \w\w:\w\w', re.UNICODE)  # the most-relaxed
		tool = pyocr.get_available_tools()[0]
		lang = tool.get_available_languages()[1]

		req_image = []
		final_text = []
		# this could've been done better, but for now will do
		idx = 0
		with Image(filename = pdfFile, resolution=self.__dpi) as img:
#			if self.__debug:
#				print "Parsing Page: " + str(idx + 1)
			cloneImg =img

			cloneImg.alpha_channel = False
			self.evaluate( cloneImg, 'threshold', self.__threshold)

			txt1 = tool.image_to_string( PI.open(io.BytesIO(cloneImg[:, :border].make_blob('png'))), lang=lang, builder=pyocr.builders.TextBuilder())+ "\n"
			txt2 = tool.image_to_string( PI.open(io.BytesIO(cloneImg[:, border:].make_blob('png'))), lang=lang, builder=pyocr.builders.TextBuilder())+ "\n"
			txt = txt1+txt2
			test = self._clean_text(txt1)
			print(txt1)
			print(txt2)
			if reCaseSubmitted.search(txt) != None or reCaseSubmitted2.search(txt) != None or reCaseSubmitted3.search(txt) != None:
				print( 'TRUE' )
			else:
				print('FALSE')

    def image_to_string(self, filename):
        tools = pyocr.get_available_tools()
        if len(tools) == 0:
            raise PyOCRIntegrationNoOCRFound('No OCR tool has been found on '
                                             'this system. Make sure it\'s on'
                                             'PATH variable of your system')

        filename_split, fileextension_split = os.path.splitext(filename)

        grayscale_filename = filename_split + '_gray' + fileextension_split
        with WandImage(filename=filename) as img:
            img.type = 'grayscale'
            img.save(filename=grayscale_filename)

        adaptive_thresh_filename = filename_split + '_adt' + fileextension_split
        OpenCVIntegration.adaptive_threshold(filename, adaptive_thresh_filename)

        processes = []
        for tool in tools:
            if tool.get_name() == "Tesseract":

                thread_t = self._OCRProcessingThread(tool, self.lang, filename)
                thread_t.start()
                processes.append(thread_t)

            else:
                thread_c_raw = self._OCRProcessingThread(tool, self.lang,
                                                         filename)
                thread_c_raw.start()
                processes.append(thread_c_raw)

                thread_c_gs = self._OCRProcessingThread(tool, self.lang,
                                                        grayscale_filename)
                thread_c_gs.start()
                processes.append(thread_c_gs)

                thread_c_prd = self._OCRProcessingThread(tool, self.lang,
                                                         adaptive_thresh_filename)
                thread_c_prd.start()
                processes.append(thread_c_prd)

        # Wait this all threads finish processing
        result = []
        threads_running = True
        while threads_running:
            found_thread_alive = False
            for p in processes:
                if p.is_alive():
                    found_thread_alive = True

            if not found_thread_alive:
                threads_running = False
                for p in processes:
                    result.append(p.return_value)

        # Removing generated files
        self._cleanup(grayscale_filename)
        self._cleanup(adaptive_thresh_filename)

        return result

def pdf2ocr(pdffile):
    """
    Optical Character Recognition on PDF files using Python
    see https://pythontips.com/2016/02/25/ocr-on-pdf-files-using-python/
    :param pdffile: pdffile to be OCR'd
    :return:
    """
    from wand.image import Image
    from PIL import Image as PI
    import pyocr
    import pyocr.builders
    import io

    tool = pyocr.get_available_tools()[0]
    lang = tool.get_available_languages()[0]  # [0] for english
    req_image = []
    final_text = []
    print "Reading {0}".format(pdffile)
    image_pdf = Image(filename=pdffile, resolution=300)
    image_jpeg = image_pdf.convert("jpeg")
    for img in image_jpeg.sequence:
        img_page = Image(image=img)
        print ("appending image")
        req_image.append(img_page.make_blob("jpeg"))
    print "Generating text"
    for img in req_image:
        txt = tool.image_to_string(PI.open(io.BytesIO(img)), lang=lang, builder=pyocr.builders.TextBuilder())
        final_text.append(txt)
    return final_text

def extract_text(filename, output_filename):
    tools = pyocr.get_available_tools()
    if len(tools) == 0:
        print("No OCR tool found")
        sys.exit(1)
    # The tools are returned in the recommended order of usage
    tool = tools[0]
    print("Will use tool '%s'" % (tool.get_name()))

    langs = tool.get_available_languages()
    print("Available languages: %s" % ", ".join(langs))
    lang = langs[0]
    print("Will use lang '%s'" % (lang))

    txt = tool.image_to_string(
        Image.open(filename),
        lang=lang,
        builder=pyocr.builders.TextBuilder()
    )


    with open(output_filename, 'w') as txtfile:
        txtfile.write(txt.encode("UTF-8"))

    print "Text contents saved as '{}'".format(output_filename)

def test(image_name):
    """
    只能识别验证码在正中间的情况
    :param image_name:
    :return:
    """
    with Image.open(image_name) as image:
        # 把彩色图像转化为灰度图像。彩色图像转化为灰度图像的方法很多，这里采用RBG转化到HSI彩色空间，采用L分量。
        image = image.convert("L")

        # 需要把图像中的噪声去除掉。这里的图像比较简单，直接阈值化就行了。我们把大于阈值threshold的像素置为1，其他的置为0。对此，先生成一张查找表，映射过程让库函数帮我们做。
        image = cut_noise(image)

        tools = pyocr.get_available_tools()
        if len(tools) == 0:
            print("No OCR tool found")
            sys.exit(1)
        tool = tools[0]
        image.save("test.jpg")

        txt = tool.image_to_string(image, lang="eng", builder=pyocr.builders.TextBuilder())
        # Digits - Only Tesseract
        digits = tool.image_to_string(image, lang="eng", builder=pyocr.tesseract.DigitBuilder())
        print(txt)
        print(digits)

 def picture2text(self, picture):
     tools = pyocr.get_available_tools()
     tool = tools[0]
     self.txt = tool.image_to_string(
         picture,
         lang="jpn",
         builder=pyocr.builders.TextBuilder(tesseract_layout=6)
     )

 def GetLanguages(cls):        
     lang = ''
     for tool in pyocr.get_available_tools():
         if tool.get_name() == 'Tesseract (sh)':
             ocr = tool
             break
     if ocr:
         lang = '+'.join(ocr.get_available_languages())
     return lang

def get_named_ocr_tool(toolname):
	'''For a given tool name, this function will return an ocr tool'''
	tools = pyocr.get_available_tools()

	for tool in tools:
		if tool.get_name() == toolname:
			return tool

	return None

	def __init__(self):
		#初始化浏览器和ocr tool
		self.driver = webdriver.Firefox(executable_path='./geckodriver')
		tools = pyocr.get_available_tools()
		if len(tools) == 0:
			sys.exit(1)
		self.tool = tools[0]
		self.im = None
		self.path = None
		self.crop_xy = [(35, 300, 770, 610), (35, 550, 770, 720), (35, 730, 770, 830), (35, 840, 770, 940)]

def ocr_recipe_name(coords, img):
    img = img.crop(coords)
    tool = pyocr.get_available_tools()[0]
    lang = tool.get_available_languages()[2]
    recipe_name = tool.image_to_string(
        img, lang=lang, builder=pyocr.builders.TextBuilder()
    )
    recipe_name = recipe_name.replace('"', '') \
        .replace(' ', '_').replace('/', '_').replace('(', '').replace(')', '').replace('!', '').lower()
    return recipe_name

def recognition_phone(phone_img_file):
    tools = pyocr.get_available_tools()
    tool = tools[0]
    langs = tool.get_available_languages()
    lang = langs[1]
    phone_text = tool.image_to_string(Image.open(phone_img_file), lang=lang, builder=pyocr.builders.TextBuilder())

    print phone_text

    return phone_text

def ptoi(name):
	with open(name,'r') as f:
		img = Image.open(name)
	img = img.convert('1')
	img.save('tmp.bmp')
	tools = pyocr.get_available_tools()
	if len(tools) == 0:
		print('No OCR tool found!')
		sys.exit(1)
	print("Using '%s'" % (tools[0].get_name()))
	return tools[0].image_to_string(img)

    def __init__(self):
        self.ocr = None
        self.lang = None

        for tool in pyocr.get_available_tools():
            if tool.get_name() == 'Tesseract (sh)':
                self.ocr = tool
                break
        
        if self.ocr:
            self.lang = '+'.join(self.ocr.get_available_languages())

def image_to_string(args):
    img, lang = args
    ocr = pyocr.get_available_tools()[0]
    with Image.open(os.path.join(Consumer.SCRATCH, img)) as f:
        if ocr.can_detect_orientation():
            try:
                orientation = ocr.detect_orientation(f, lang=lang)
                f = f.rotate(orientation["angle"], expand=1)
            except TesseractError:
                pass
        return ocr.image_to_string(f, lang=lang)

 def check_required_software():
     logger = logging.getLogger(__name__)
     tools = pyocr.get_available_tools()
     if len(tools) == 0:
         raise PyOCRIntegrationNoOCRFound('No OCR tool has been found on '
                                          'this system. Make sure it\'s on')
     elif len(tools) == 1:
         logger.info("I've found only one ocr tool [%s]. This is not exactly "
                     "an error but you should get better results if you have "
                     "both Tesseract and Cuneiform installed"
                     % tools[0].get_name())
     else:
         logger.info("I've found all required software. We're good to go =)")

 def ocr(self, img, angles=None):
     """
     Returns immediately.
     Listen for the signal ocr-done to get the result
     """
     if (not self.__config['ocr_enabled'].value or
             len(pyocr.get_available_tools()) == 0):
         angles = 0
     elif angles is None:
         angles = 4
     img.load()
     job = self.factories['ocr'].make(img, angles)
     self.schedulers['ocr'].schedule(job)
     return job

def get_default_ocr_lang():
    # Try to guess based on the system locale what would be
    # the best OCR language

    ocr_tools = pyocr.get_available_tools()
    if len(ocr_tools) == 0:
        return DEFAULT_OCR_LANG
    ocr_langs = ocr_tools[0].get_available_languages()

    lang = find_language()
    if hasattr(lang, 'iso639_3_code') and lang.iso639_3_code in ocr_langs:
        return lang.iso639_3_code
    if hasattr(lang, 'terminology') and lang.terminology in ocr_langs:
        return lang.terminology
    return DEFAULT_OCR_LANG