Kali pertama

.gitignore

+*egg-info
+__pycache__

CHANGES.txt

+0.1 2023-04-02
+----------
+- Kali pertama

MANIFEST.in

+include *.txt *.rst *.py
+recursive-include ocr_ktp *.py *.traineddata

README.rst

+OCR KTP
+=======
+
+Optical Character Recognition adalah membaca foto yang berisi huruf, angka,
+serta karakter lainnya. Paket ini memilah hasil pembacaan foto KTP
+menjadi bentuk baku seperti NIK, Nama, Alamat, dst. 
+
+
+Pemasangan Tesseract
+--------------------
+
+Paket ini membutuhkan `Tesseract <https://github.com/tesseract-ocr/tesseract>`_.
+Untuk Debian dan turunannya bisa dengan cara berikut::
+
+    $ sudo add-apt-repository ppa:alex-p/tesseract-ocr-devel
+    $ sudo apt update
+    $ sudo apt install -y tesseract-ocr
+
+Setelah paket ini dipasang fotolah sebuah KTP dan coba::
+
+    $ cd contrib
+    $ ~/env/bin/python uji-ktp.py ktp.jpg
+
+Jika NIK tidak sesuai - misalnya tidak 16 digit - maka tidak akan ditampilkan.
+Hal lainnya yang tidak baku juga tidak akan tampak misalnya tanggal yang tidak
+benar, RT / RW bukan angka, dst. Cobalah foto ulang, pastikan alas meja tidak
+terlihat.
+
+Semoga berhasil.

contrib/uji-ktp.py

+import sys
+from argparse import ArgumentParser
+from pprint import pprint
+from ocr_ktp.ktp import ocr
+
+
+lang_dir = '../ocr_ktp/data'
+help_lang_dir = 'default ' + lang_dir
+pars = ArgumentParser()
+pars.add_argument('img_file')
+pars.add_argument('--lang-dir', default=lang_dir, help=help_lang_dir)
+option = pars.parse_args(sys.argv[1:])
+d = ocr(option.img_file, option.lang_dir)
+pprint(d)

ocr_ktp/__init__.py

+from cv2 import (
+    imread,
+    cvtColor,
+    COLOR_BGR2GRAY,
+    GaussianBlur,
+    threshold,
+    THRESH_BINARY,
+    )
+from pytesseract import image_to_string
+
+
+def ocr(image, lang='eng', tess_opt='--psm 6'):
+    if isinstance(image, str):
+        image = imread(image)  # Color
+    gray = cvtColor(image, COLOR_BGR2GRAY)  # Grayscale
+    blurred = GaussianBlur(gray, (3, 3), 0)  # Smoothing
+    ret, thresh = threshold(blurred, 127, 255, THRESH_BINARY)  # Black & white
+    height, width, _ = image.shape
+    width = int(width*7/10)
+    cropped = thresh[0:height, 0:width]  # Buang tanda tangan
+    return image_to_string((cropped), lang=lang, config=tess_opt)

ocr_ktp/ktp.py

+from datetime import datetime
+from string import (
+    ascii_letters,
+    ascii_uppercase,
+    digits,
+    )
+from cv2 import (
+    imread,
+    cvtColor,
+    COLOR_BGR2GRAY,
+    GaussianBlur,
+    threshold,
+    THRESH_BINARY,
+    )
+from pytesseract import image_to_string
+from jaccard_index.jaccard import jaccard_index
+from . import ocr as base_ocr
+
+
+LABELS = (
+    'NIK', 'Nama', 'Tempat/Tgl Lahir', 'Jenis', 'Alamat', 'RT/RW',
+    'Kel/Desa', 'Kecamatan', 'Agama', 'Status', 'Pekerjaan',
+    'Kewarganegaraan')
+
+VALUES = dict(
+    Jenis=('LAKI-LAKI', 'PEREMPUAN'),
+    Agama=(
+        'ISLAM', 'KRISTEN', 'KATOLIK', 'BUDHA', 'HINDU', 'KONGHUCHU',
+        'KEPERCAYAAN'),
+    Status=('KAWIN', 'BELUM'),
+    Kewarganegaraan=('WNI', 'WNA'))
+
+NUMBERS = ('NIK', 'RT/RW')
+
+digits_letters = digits + ascii_letters
+
+
+def is_date(s):
+    try:
+        d = datetime.strptime(s, '%d-%m-%Y')
+    except ValueError:
+        return
+    return d.date()
+
+
+def clean(s, chars=ascii_letters):
+    r = ''
+    i = -1
+    for ch in s:
+        i += 1
+        if ch in chars:
+            r = s[i:]
+            break
+    while r:
+        if r[-1] in chars:
+            return r
+        r = r[:-1]
+    return r
+
+
+def match_value(values, ref_values):
+    last_j_index = 0
+    found = None
+    for value in values:
+        if not value[2:]:
+            continue
+        for ref_value in ref_values:
+            j_index = jaccard_index(value, ref_value)
+            if j_index < 0.3:
+                continue
+            if last_j_index < j_index:
+                # print(f'Jaccard Index {value} vs {ref_value} = {j_index}')
+                last_j_index = j_index
+                found = ref_value
+    return found
+
+
+def nik(v):
+    d = dict()
+    v = clean(v, digits)
+    if len(v) != 16:
+        return d
+    for ch in v:
+        if ch not in digits:
+            return d
+    return dict(NIK=v)
+
+
+def rt_rw(v):
+    d = dict()
+    t = v.split('/')
+    if t[0]:
+        v = clean(t[0], digits)
+        if v:
+            d['RT'] = v
+    if t[1:] and t[1]:
+        v = clean(t[1], digits)
+        if v:
+            d['RW'] = v
+    return d
+
+
+def tempat_tgl_lahir(v):
+    d = dict()
+    t = v.split()
+    tgl = is_date(t[-1])
+    if tgl:
+        d['Tgl_Lahir'] = tgl
+    tempat = ' '.join(t[:-1])
+    tempat = clean(tempat)
+    if tempat:
+        tempat = tempat.lstrip('Lahir')
+        d['Tempat_Lahir'] = clean(tempat)
+    return d
+
+
+def pekerjaan(v):
+    v = clean(v)
+    return dict(Pekerjaan=v)
+
+
+PARSER = {
+    'RT/RW': rt_rw,
+    'Tempat/Tgl Lahir': tempat_tgl_lahir,
+    'NIK': nik,
+    'Pekerjaan': pekerjaan}
+
+
+def parse(s):
+    d = dict()
+    for line in s.splitlines():
+        line = clean(line, digits_letters)
+        if not line:
+            continue
+        values = line.split()
+        label = match_value(values[:1], LABELS)
+        if not label:
+            continue
+        values = values[1:]
+        if label in VALUES:
+            v = match_value(values, VALUES[label])
+            if v:
+                d[label] = v
+        else:
+            v = ' '.join(values)
+            func = PARSER.get(label)
+            if func:
+                d.update(func(v))
+            else:
+                v = clean(v, ascii_uppercase)
+                d[label] = v
+    return d
+
+
+def ocr(img, lang_dir='data'):
+    tess_opt = '--psm 6 --tessdata-dir ' + lang_dir
+    s = base_ocr(img, 'ind', tess_opt)
+    return parse(s)

setup.py

+import os
+import sys
+import subprocess
+from setuptools import (
+    setup,
+    find_packages,
+    )
+
+
+here = os.path.abspath(os.path.dirname(__file__))
+with open(os.path.join(here, 'README.rst')) as f:
+    README = f.read()
+with open(os.path.join(here, 'CHANGES.txt')) as f:
+    CHANGES = f.read()
+
+line = CHANGES.splitlines()[0]
+version = line.split()[0]
+
+requires = [
+    'PyTesseract',
+    'opencv-python',
+    'jaccard-index',
+    ]
+
+
+setup(
+    name='ocr-ktp',
+    version=version,
+    description='Menerjemahkan foto KTP ke dictionary',
+    long_description=README + '\n\n' + CHANGES,
+    classifiers=[
+        'Programming Language :: Python :: 3.6',
+    ],
+    author='Owo Sugiana',
+    author_email='sugiana@gmail.com',
+    url='https://git.opensipkd.com/sugiana/ocr-ktp',
+    keywords='ocr ktp',
+    packages=find_packages(),
+    include_package_data=True,
+    zip_safe=False,
+    install_requires=requires,
+)