def test_ocr_text_color(image, color, expected, region):
frame = load_image(image)
mode = stbt.OcrMode.SINGLE_LINE
assert expected not in stbt.ocr(frame, region, mode)
assert expected == stbt.ocr(frame, region, mode, text_color=color)
assert not stbt.match_text(expected, frame, region, mode)
assert stbt.match_text(expected, frame, region, mode, text_color=color)
def test_that_ligatures_and_ambiguous_punctuation_are_normalised():
frame = load_image('ocr/ambig.png')
text = stbt.ocr(frame)
for bad, good in [
# tesseract 3.02
("horizonta|", "horizontal"),
# tesseract 4.00 with tessdata 590567f
("siIIyness", "sillyness"),
("Iigatures", "ligatures"),
]:
text = text.replace(bad, good)
assert ligature_text == text
assert stbt.match_text("em-dash,", frame)
assert stbt.match_text(u"em\u2014dash,", frame)
def test_that_match_text_gives_tesseract_a_hint():
frame = load_image("ocr/itv-player.png")
if "ITV Player" in stbt.ocr(frame=frame):
raise SkipTest("Tesseract doesn't need a hint")
if "ITV Player" not in stbt.ocr(frame=frame, tesseract_user_words=["ITV"]):
raise SkipTest("Giving tesseract a hint doesn't help")
assert stbt.match_text("ITV Player", frame=frame)
def test_that_text_region_is_correct_even_with_regions_larger_than_frame():
frame = load_image("ocr/menu.png")
text, region, _ = list(iterate_menu())[6]
result = stbt.match_text(
text, frame=frame, region=region.extend(right=+12800))
assert result
assert region.contains(result.region)
def test_ocr_doesnt_leak_python_future_newtypes():
f = load_image("ocr/small.png")
result = stbt.ocr(f)
assert type(result).__name__ in ["str", "unicode"]
result = stbt.match_text("Small", f)
assert type(result.text).__name__ in ["str", "unicode"]
def test_match_text_stringify_result():
frame = load_image("ocr/menu.png")
result = stbt.match_text(u"Onion Bhaji", frame=frame)
assert re.match(
r"TextMatchResult\(time=None, match=True, region=Region\(.*\), "
r"frame=<Image\(filename=u?'ocr/menu.png', "
r"dimensions=1280x720x3\)>, text=u?'Onion Bhaji'\)", str(result))
def test_that_match_text_still_returns_if_region_doesnt_intersect_with_frame(
region):
frame = load_image("ocr/menu.png")
result = stbt.match_text("Onion Bhaji", frame=frame, region=region)
assert result.match is False
assert result.region is None
assert result.text == "Onion Bhaji"
# Avoid future.types.newtypes in return values
assert type(result.text).__name__ in ["str", "unicode"]
def test_ocr_debug():
# So that the output directory name doesn't depend on how many tests
# were run before this one.
ImageLogger._frame_number = itertools.count(1) # pylint:disable=protected-access
f = stbt.load_image("action-panel.png")
r = stbt.Region(0, 370, right=1280, bottom=410)
c = (235, 235, 235)
with scoped_curdir(), scoped_debug_level(2):
stbt.ocr(f)
stbt.ocr(f, region=r)
stbt.ocr(f, region=r, text_color=c)
stbt.match_text("Summary", f) # no match
stbt.match_text("Summary", f, region=r) # no match
stbt.match_text("Summary", f, region=r, text_color=c)
files = subprocess.check_output("find stbt-debug | sort", shell=True) \
.decode("utf-8")
assert files == dedent("""\
stbt-debug
stbt-debug/00001
stbt-debug/00001/index.html
stbt-debug/00001/source.png
stbt-debug/00001/tessinput.png
stbt-debug/00001/upsampled.png
stbt-debug/00002
stbt-debug/00002/index.html
stbt-debug/00002/source.png
stbt-debug/00002/tessinput.png
stbt-debug/00002/upsampled.png
stbt-debug/00003
stbt-debug/00003/index.html
stbt-debug/00003/source.png
stbt-debug/00003/tessinput.png
stbt-debug/00003/text_color_difference.png
stbt-debug/00003/text_color_threshold.png
stbt-debug/00003/upsampled.png
stbt-debug/00004
stbt-debug/00004/index.html
stbt-debug/00004/source.png
stbt-debug/00004/tessinput.png
stbt-debug/00004/upsampled.png
stbt-debug/00005
stbt-debug/00005/index.html
stbt-debug/00005/source.png
stbt-debug/00005/tessinput.png
stbt-debug/00005/upsampled.png
stbt-debug/00006
stbt-debug/00006/index.html
stbt-debug/00006/source.png
stbt-debug/00006/tessinput.png
stbt-debug/00006/text_color_difference.png
stbt-debug/00006/text_color_threshold.png
stbt-debug/00006/upsampled.png
""")
def test_match_text_case_sensitivity():
frame = load_image("ocr/menu.png", cv2.IMREAD_GRAYSCALE)
assert stbt.match_text("ONION BHAJI", frame)
assert stbt.match_text("ONION BHAJI", frame, case_sensitive=False)
assert not stbt.match_text("ONION BHAJI", frame, case_sensitive=True)
def test_match_text_on_single_channel_image():
frame = load_image("ocr/menu.png", cv2.IMREAD_GRAYSCALE)
assert stbt.match_text("Onion Bhaji", frame)
def test_that_match_text_returns_no_match_for_non_matching_text():
frame = load_image("ocr/menu.png")
assert not stbt.match_text(u"Noodle Soup", frame=frame)
def test_match_text_case_sensitivity():
frame = load_image("ocr/menu.png", color_channels=1)
assert stbt.match_text("ONION BHAJI", frame)
assert stbt.match_text("ONION BHAJI", frame, case_sensitive=False)
assert not stbt.match_text("ONION BHAJI", frame, case_sensitive=True)
def test_that_default_language_is_configurable():
f = load_image("ocr/unicode.png")
assert not stbt.match_text(u"Röthlisberger", f) # reads Réthlisberger
with temporary_config({"ocr.lang": "deu"}):
assert stbt.match_text(u"Röthlisberger", f)
assert u"Röthlisberger" in stbt.ocr(f)
def test_that_match_text_accepts_unicode():
f = load_image("ocr/unicode.png")
assert stbt.match_text("David", f, lang='eng+deu') # ascii
assert stbt.match_text("Röthlisberger", f, lang='eng+deu') # unicode
assert stbt.match_text("Röthlisberger".encode('utf-8'), f,
lang='eng+deu') # utf-8 bytes
def match_text():
return stbt.match_text("RED", frame=frame)
def test(text, region, upsample):
result = stbt.match_text(text, frame=frame, upsample=upsample)
assert result
assert region.contains(result.region) # pylint:disable=no-member
def test_that_ocr_region_none_isnt_allowed(region):
f = load_image("ocr/small.png")
with pytest.raises((TypeError, ValueError)):
stbt.ocr(frame=f, region=region)
with pytest.raises((TypeError, ValueError)):
stbt.match_text("Small", frame=f, region=region)
def test_match_text_on_single_channel_image():
frame = load_image("ocr/menu.png", color_channels=1)
assert stbt.match_text("Onion Bhaji", frame)
