def test_ocr_image_Whitelist(self):
test_text = "ABABABA"
img = create_img(test_text)
tr = tesserwrap.Tesseract()
tr.set_variable("tessedit_char_whitelist", "A")
out_text = tr.ocr_image(img).strip()
assert out_text != test_text, "%r == %r" % (out_text, test_text)
def test_mean_confidence(self):
tr = tesserwrap.Tesseract()
img = create_img("Hello World")
tr.set_image(img)
tr.get_text() # run recognizer to get all data set
ok_(tr.get_mean_confidence() >= 0,
"Confidence should be positve integer")
def test_set_rectangle(self):
test_text = "A BBB"
img = create_img("A BBB CCC")
tr = tesserwrap.Tesseract()
tr.set_image(img)
tr.set_rectangle(0, 0, 100, 40)
out_text = tr.get_text().decode().strip()
eq_(out_text, test_text, "%s is not %s" % (out_text, test_text))
def test_get_textlines(self):
tr = tesserwrap.Tesseract()
test_text = 'This is a line'
img = create_img(test_text)
tr.set_image(img)
tr.get_text()
res = tr.get_textlines()
result_text = ''.join([l.value.strip() for l in res])
eq_(result_text, test_text, "%s is not %s" % (result_text, test_text))
def test_get_symbols(self):
tr = tesserwrap.Tesseract()
test_text = 'ABCD'
img = create_img(test_text)
tr.set_image(img)
tr.get_text()
res = tr.get_symbols()
result_text = ''.join([l.value for l in res])
eq_(result_text, test_text, "%s is not %s" % (result_text, test_text))
def test_get_rectangle(self):
test_text = "A BBB"
img = create_img("A BBB CCC")
tr = tesserwrap.Tesseract()
tr.set_image(img)
tr.get_text() # run recognizer to get all data set
(l, t), (w, h) = tr.get_rectangle()
eq_(l, 0, "Left attribute incorrect")
eq_(t, 0, "Top attribute incorrect")
eq_(w, 710, "Width attribute incorrect")
eq_(h, 40, "Height attribute incorrect")
def test_get_words(self):
tr = tesserwrap.Tesseract()
img = create_img()
tr.set_image(img)
tr.get_text()
res = tr.get_words()
eq_(len(res), 3, "Each word should have one item in result")
item = res[0]
eq_(item.value, 'Quick', "%s is not %s" % (item.value, 'Quick'))
eq_(len(item.box), 4, 'Box does not contain 4 items')
def test_word_confidences(self):
tr = tesserwrap.Tesseract()
img = create_img()
tr.set_image(img)
tr.get_text() # run recognizer to get all data set
res = tr.get_all_word_confidences()
eq_(len(res), 3, "Each word should have one item in result")
eq_(tr.get_mean_confidence(),
sum(res) / len(res), "Mean confidence incorrect")
# Empty image
img = create_img("")
tr.set_image(img)
res2 = tr.get_all_word_confidences()
eq_([], res2, "Should be empty result and no crash")
import sys
import io
import requests
import PIL.Image
import tesserwrap
#: https://github.com/gregjurman/tesserwrap
tesseract = tesserwrap.Tesseract()
tesseract.set_variable("tessedit_char_whitelist", "abcdefghijklmnopqrstuvwxyz")
def distinguish_captcha(image_url, show_origin_image=True):
#: preprocess
image_bytes = requests.get(image_url).content
origin_image = PIL.Image.open(io.BytesIO(image_bytes))
image = origin_image.point(lambda p: p * 1.5)\
.point(lambda p: 255 if p > 200 else 0)\
.convert("1")
#: distinguish the text
text = tesseract.ocr_image(image)
#: show the origin image
#if show_origin_image:
#origin_image.show()
text.replace(' ', '')
return text.strip()
def decode_captcha(origin_image):
image = origin_image.point(lambda p: p * 1.5)\
.point(lambda p: 255 if p > 200 else 0)\
def test_clear(self):
tr = tesserwrap.Tesseract()
img = create_img("A BBB CCC")
tr.set_image(img)
tr.clear()
def test_bad_handle(self):
tr = tesserwrap.Tesseract()
del tr.handle
del tr
def test_ocr_image_RGB(self):
test_text = "ABABABA"
img = create_img(test_text, "RGB")
tr = tesserwrap.Tesseract()
out_text = tr.ocr_image(img).strip()
eq_(out_text, test_text, "%s is not %s" % (out_text, test_text))
from flask import jsonify, Flask, request
from PIL import Image
import tesserwrap
import traceback
import sys
import io
# Init the tesseract
tr = tesserwrap.Tesseract(lang="por")
# Init web Server
app = Flask(__name__)
app.config['JSON_AS_ASCII'] = False
@app.route('/recognize', methods=["POST"])
def image():
try:
imagefile = request.files.get('imagefile', '')
img = Image.open(imagefile.stream)
text = tr.ocr_image(img)
return jsonify(result="ok", data=text)
except Exception as err:
traceback.print_exc()
print(err)
return jsonify(result="sorry :/")