├── README.md
└── pytesser.py


/README.md:
--------------------------------------------------------------------------------
 1 | Pytesser
 2 | ========
 3 | 
 4 | Python wrapper for the tesseract OCR engine. The module is based on OpenCV.
 5 | 
 6 | 
 7 | Informations
 8 | ------------
 9 | 
10 | There is already multiples tesseract python modules, but none of them satisfied me. This one is different on the following point:
11 | 
12 | * All the classes are put in the same file and all inessential class are removed
13 | * Use OpenCV instead of PIL (to really an advantage because PIL as far more widespread, but better fit my needs ;))
14 | * Use subprocess.communicate instead of subprocess.wait to avoid any output in the shell or in the programs that use the module.
15 | * Management of the differents languages via the option '-l' because the original pytesser use the default language which is english. By this way the detection of french for instance is totally inacurrate.
16 | * Management of of the pagesegmode, which allow to modify the behavior of tesseract if we want for instance to detect only one character, a word or a line.
17 | * The code is far more straightforward (my opinion)
18 | 
19 | Installation
20 | ------------
21 | 
22 | ```bash
23 | sudo apt-get install tesseract tesseract-ocr-all
24 | sudo pip install opencv-python
25 | ```
26 | 
27 | How to use it ?
28 | ---------------
29 | 
30 | There is to ways to use it. Either you give it a filename, either directly an image. For a filename you can do:
31 | 
32 | ```python
33 | import pytesser
34 | txt = pytesser.image_file_to_string("myimage.jpg")
35 | #By default language is eng, and page seg mode auto
36 | 
37 | #To give specifify parameters:
38 | txt = pytesser.image_to_string("myimage.jpg","fra",pytesser.PSM_SINGLE_WORD) #Analyse image as a single french word
39 | ```
40 | 
41 | Or you can directly give it an OpenCV image like this:
42 | 
43 | ```python
44 | image = cv2.imread("myimage.jpg")
45 | txt = pytesser.image_to_string(image) 
46 | ```


--------------------------------------------------------------------------------
/pytesser.py:
--------------------------------------------------------------------------------
 1 | import sys
 2 | from subprocess import Popen, PIPE
 3 | import os
 4 | import tempfile
 5 | import cv2
 6 | 
 7 | PROG_NAME = 'tesseract'
 8 | TEMP_IMAGE = tempfile.mktemp()+'.bmp'
 9 | TEMP_FILE = tempfile.mktemp()
10 | 
11 | #All the PSM arguments as a variable name (avoid having to know them)
12 | PSM_OSD_ONLY = 0
13 | PSM_SEG_AND_OSD = 1
14 | PSM_SEG_ONLY = 2
15 | PSM_AUTO = 3
16 | PSM_SINGLE_COLUMN = 4
17 | PSM_VERTICAL_ALIGN = 5
18 | PSM_UNIFORM_BLOCK = 6
19 | PSM_SINGLE_LINE = 7
20 | PSM_SINGLE_WORD = 8
21 | PSM_SINGLE_WORD_CIRCLE = 9
22 | PSM_SINGLE_CHAR = 10
23 | 
24 | class TesseractException(Exception): #Raised when tesseract does not return 0
25 |     pass
26 | 
27 | class TesseractNotFound(Exception): #When tesseract is not found in the path
28 |     pass
29 | 
30 | def check_path(): #Check if tesseract is in the path raise TesseractNotFound otherwise
31 |     for path in os.environ.get('PATH', '').split(':'):
32 |         filepath = os.path.join(path, PROG_NAME)
33 |         if os.path.exists(filepath) and not os.path.isdir(filepath):
34 |             return True
35 |     raise TesseractNotFound()
36 | 
37 | def process_request(input_file, output_file, lang=None, psm=None):
38 |     args = [PROG_NAME, input_file, output_file] #Create the arguments
39 |     if lang is not None:
40 |         args.append("-l")
41 |         args.append(lang)
42 |     if psm is not None:
43 |         args.append("-psm")
44 |         args.append(str(psm))
45 |     proc = Popen(args, stdout=PIPE, stderr=PIPE) #Open process
46 |     ret = proc.communicate() #Launch it
47 | 
48 |     code = proc.returncode
49 |     if code != 0:
50 |         if code == 2:
51 |             raise TesseractException("File not found")
52 |         if code == -11:
53 |             raise TesseractException("Language code invalid: "+ret[1])
54 |         else:
55 |             raise TesseractException(ret[1])
56 | 
57 | def image_to_string(im, lang=None, psm=None):
58 |     grayscale_image = cv2.cvtColor(im, cv2.COLOR_BGR2GRAY)
59 |     cv2.imwrite(TEMP_IMAGE, grayscale_image)
60 |     txt = image_file_to_string(TEMP_IMAGE, lang, psm)
61 |     os.remove(TEMP_IMAGE)
62 |     return txt
63 | 
64 | def image_file_to_string(file, lang=None, psm=None):
65 |     check_path() #Check if tesseract available in the path
66 |     grayscale_image = cv2.cvtColor(cv2.imread(file), cv2.COLOR_BGR2GRAY)
67 |     cv2.imwrite(TEMP_IMAGE, grayscale_image)
68 |     #process_request(file, TEMP_FILE, lang, psm)
69 |     process_request(TEMP_IMAGE, TEMP_FILE, lang, psm)
70 |     f = open(TEMP_FILE+".txt", "r") #Open back the file
71 |     txt = f.read()
72 |     f.close()
73 |     os.remove(TEMP_FILE+".txt")
74 |     os.remove(TEMP_IMAGE)
75 |     return txt
76 | 
77 | 
78 | if __name__ =='__main__':
79 |     print(image_file_to_string(sys.argv[2], sys.argv[1], PSM_AUTO))
80 | 


--------------------------------------------------------------------------------