390 lines
14 KiB
Python
390 lines
14 KiB
Python
|
import pytest
|
||
|
|
import unittest
|
||
|
|
import kivy.multistroke
|
||
|
|
from kivy.multistroke import Recognizer, MultistrokeGesture
|
||
|
|
from kivy.vector import Vector
|
||
|
|
|
||
|
|
best_score = 0.0
|
||
|
|
counter = 0
|
||
|
|
|
||
|
|
|
||
|
|
def best_score_cb(result):
|
||
|
|
global best_score
|
||
|
|
best_score = result.best['score']
|
||
|
|
|
||
|
|
|
||
|
|
def counter_cb(result):
|
||
|
|
global counter
|
||
|
|
counter += 1
|
||
|
|
|
||
|
|
|
||
|
|
# These are taken from the examples in JavaScript code (but made unistrokes)
|
||
|
|
TGesture = [Vector(30, 7), Vector(103, 7), Vector(66, 7), Vector(66, 87)]
|
||
|
|
NGesture = [Vector(177, 92), Vector(177, 2), Vector(182, 1), Vector(246, 95),
|
||
|
|
Vector(247, 87), Vector(247, 1)]
|
||
|
|
|
||
|
|
|
||
|
|
# dataset that matches N pretty well
|
||
|
|
Ncandidate = [
|
||
|
|
Vector(160, 271), Vector(160, 263), Vector(158, 257), Vector(156, 249),
|
||
|
|
Vector(146, 187), Vector(144, 181), Vector(144, 175), Vector(142, 167),
|
||
|
|
Vector(140, 113), Vector(140, 107), Vector(140, 103), Vector(140, 99),
|
||
|
|
Vector(140, 85), Vector(138, 85), Vector(138, 87), Vector(138, 89),
|
||
|
|
Vector(166, 151), Vector(176, 171), Vector(188, 189), Vector(200, 205),
|
||
|
|
Vector(238, 263), Vector(242, 269), Vector(244, 273), Vector(246, 277),
|
||
|
|
Vector(252, 289), Vector(254, 291), Vector(256, 291), Vector(258, 291),
|
||
|
|
Vector(260, 281), Vector(260, 275), Vector(260, 267), Vector(260, 255),
|
||
|
|
Vector(254, 189), Vector(254, 175), Vector(254, 161), Vector(254, 147),
|
||
|
|
Vector(260, 103), Vector(260, 101), Vector(260, 99), Vector(260, 95),
|
||
|
|
Vector(260, 93), Vector(260, 91), Vector(260, 89)
|
||
|
|
]
|
||
|
|
|
||
|
|
|
||
|
|
class MultistrokeTestCase(unittest.TestCase):
|
||
|
|
|
||
|
|
def setUp(self):
|
||
|
|
global best_score
|
||
|
|
best_score = 0
|
||
|
|
counter = 0
|
||
|
|
|
||
|
|
self.Tinvar = MultistrokeGesture('T', [TGesture],
|
||
|
|
orientation_sensitive=False)
|
||
|
|
self.Tbound = MultistrokeGesture('T', [TGesture],
|
||
|
|
orientation_sensitive=True)
|
||
|
|
self.Ninvar = MultistrokeGesture('N', [NGesture],
|
||
|
|
orientation_sensitive=False)
|
||
|
|
self.Nbound = MultistrokeGesture('N', [NGesture],
|
||
|
|
orientation_sensitive=True)
|
||
|
|
|
||
|
|
@pytest.fixture(autouse=True)
|
||
|
|
def set_clock(self, kivy_clock):
|
||
|
|
self.kivy_clock = kivy_clock
|
||
|
|
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
# Recognizer scheduling
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
def test_immediate(self):
|
||
|
|
gdb = Recognizer(db=[self.Tinvar, self.Ninvar])
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=0)
|
||
|
|
self.assertEqual(r._match_ops, 4)
|
||
|
|
self.assertEqual(r._completed, 2)
|
||
|
|
self.assertEqual(r.progress, 1)
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
def test_scheduling(self):
|
||
|
|
global best_score
|
||
|
|
from kivy.clock import Clock
|
||
|
|
gdb = Recognizer(db=[self.Tinvar, self.Ninvar])
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r.bind(on_complete=best_score_cb)
|
||
|
|
|
||
|
|
# _recognize_tick is scheduled here; compares to Tinvar
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, .5)
|
||
|
|
self.assertEqual(best_score, .0)
|
||
|
|
|
||
|
|
# Now complete the search operation
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 1)
|
||
|
|
self.assertTrue(best_score > 0.94 and best_score < 0.95)
|
||
|
|
|
||
|
|
def test_scheduling_limits(self):
|
||
|
|
global best_score
|
||
|
|
from kivy.clock import Clock
|
||
|
|
gdb = Recognizer(db=[self.Ninvar])
|
||
|
|
tpls = len(self.Ninvar.templates)
|
||
|
|
|
||
|
|
best_score = 0
|
||
|
|
gdb.db.append(self.Ninvar)
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r.bind(on_complete=best_score_cb)
|
||
|
|
self.assertEqual(r.progress, 0)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 0.5)
|
||
|
|
self.assertEqual(best_score, 0)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 1)
|
||
|
|
self.assertTrue(best_score > 0.94 and best_score < 0.95)
|
||
|
|
|
||
|
|
best_score = 0
|
||
|
|
gdb.db.append(self.Ninvar)
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r.bind(on_complete=best_score_cb)
|
||
|
|
self.assertEqual(r.progress, 0)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 1 / 3.)
|
||
|
|
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 2 / 3.)
|
||
|
|
self.assertEqual(best_score, 0)
|
||
|
|
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(r.progress, 1)
|
||
|
|
self.assertTrue(best_score > 0.94 and best_score < 0.95)
|
||
|
|
|
||
|
|
def test_parallel_recognize(self):
|
||
|
|
global counter
|
||
|
|
from kivy.clock import Clock
|
||
|
|
|
||
|
|
counter = 0
|
||
|
|
gdb = Recognizer()
|
||
|
|
for i in range(9):
|
||
|
|
gdb.add_gesture('T', [TGesture], priority=50)
|
||
|
|
gdb.add_gesture('N', [NGesture])
|
||
|
|
|
||
|
|
r1 = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r1.bind(on_complete=counter_cb)
|
||
|
|
Clock.tick() # first run scheduled here; 9 left
|
||
|
|
|
||
|
|
r2 = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r2.bind(on_complete=counter_cb)
|
||
|
|
Clock.tick() # 8 left
|
||
|
|
|
||
|
|
r3 = gdb.recognize([Ncandidate], max_gpf=1)
|
||
|
|
r3.bind(on_complete=counter_cb)
|
||
|
|
Clock.tick() # 7 left
|
||
|
|
|
||
|
|
# run some immediate searches, should not interfere.
|
||
|
|
for i in range(5):
|
||
|
|
n = gdb.recognize([TGesture], max_gpf=0)
|
||
|
|
self.assertEqual(n.best['name'], 'T')
|
||
|
|
self.assertTrue(round(n.best['score'], 1) == 1.0)
|
||
|
|
|
||
|
|
for i in range(6):
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(counter, 0)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(counter, 1)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(counter, 2)
|
||
|
|
Clock.tick()
|
||
|
|
self.assertEqual(counter, 3)
|
||
|
|
|
||
|
|
def test_timeout_case_1(self):
|
||
|
|
global best_score
|
||
|
|
from kivy.clock import Clock
|
||
|
|
from time import sleep
|
||
|
|
|
||
|
|
best_score = 0
|
||
|
|
gdb = Recognizer(db=[self.Tbound, self.Ninvar])
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=1, timeout=0.4)
|
||
|
|
Clock.tick() # matches Tbound in this tick
|
||
|
|
self.assertEqual(best_score, 0)
|
||
|
|
sleep(0.4)
|
||
|
|
Clock.tick() # should match Ninv, but times out (got T)
|
||
|
|
self.assertEqual(r.status, 'timeout')
|
||
|
|
self.assertEqual(r.progress, .5)
|
||
|
|
self.assertTrue(r.best['name'] == 'T')
|
||
|
|
self.assertTrue(r.best['score'] < 0.5)
|
||
|
|
|
||
|
|
def test_timeout_case_2(self):
|
||
|
|
global best_score
|
||
|
|
from kivy.clock import Clock
|
||
|
|
from time import sleep
|
||
|
|
|
||
|
|
best_score = 0
|
||
|
|
gdb = Recognizer(db=[self.Tbound, self.Ninvar, self.Tinvar])
|
||
|
|
r = gdb.recognize([Ncandidate], max_gpf=1, timeout=0.8)
|
||
|
|
|
||
|
|
Clock.tick() # matches Tbound in this tick
|
||
|
|
self.assertEqual(best_score, 0)
|
||
|
|
sleep(0.4)
|
||
|
|
Clock.tick() # matches Ninvar in this tick
|
||
|
|
sleep(0.4)
|
||
|
|
Clock.tick() # should match Tinvar, but times out
|
||
|
|
self.assertEqual(r.status, 'timeout')
|
||
|
|
self.assertEqual(r.progress, 2 / 3.)
|
||
|
|
self.assertTrue(r.best['score'] >= .94 and r.best['score'] <= .95)
|
||
|
|
|
||
|
|
def test_priority_sorting(self):
|
||
|
|
gdb = Recognizer()
|
||
|
|
gdb.add_gesture('N', [NGesture], priority=10)
|
||
|
|
gdb.add_gesture('T', [TGesture], priority=5)
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], goodscore=0.01, max_gpf=0,
|
||
|
|
force_priority_sort=True)
|
||
|
|
self.assertEqual(r.best['name'], 'T')
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], goodscore=0.01,
|
||
|
|
force_priority_sort=False, max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], goodscore=0.01, max_gpf=0,
|
||
|
|
priority=10)
|
||
|
|
self.assertEqual(r.best['name'], 'T')
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], goodscore=0.01, max_gpf=0,
|
||
|
|
priority=4)
|
||
|
|
self.assertEqual(r.best['name'], None)
|
||
|
|
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
# Recognizer - filter tests
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
def test_name_filter(self):
|
||
|
|
gdb = Recognizer(db=[self.Ninvar, self.Nbound])
|
||
|
|
n = gdb.filter()
|
||
|
|
self.assertEqual(len(n), 2)
|
||
|
|
n = gdb.filter(name='X')
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
|
||
|
|
def test_numpoints_filter(self):
|
||
|
|
gdb = Recognizer(db=[self.Ninvar, self.Nbound])
|
||
|
|
n = gdb.filter(numpoints=100)
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
|
||
|
|
gdb.add_gesture('T', [TGesture], numpoints=100)
|
||
|
|
n = gdb.filter(numpoints=100)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(numpoints=[100, 16])
|
||
|
|
self.assertEqual(len(n), 3)
|
||
|
|
|
||
|
|
def test_numstrokes_filter(self):
|
||
|
|
gdb = Recognizer(db=[self.Ninvar, self.Nbound])
|
||
|
|
n = gdb.filter(numstrokes=2)
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
|
||
|
|
gdb.add_gesture('T', [TGesture, TGesture])
|
||
|
|
n = gdb.filter(numstrokes=2)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(numstrokes=[1, 2])
|
||
|
|
self.assertEqual(len(n), 3)
|
||
|
|
|
||
|
|
def test_priority_filter(self):
|
||
|
|
gdb = Recognizer(db=[self.Ninvar, self.Nbound])
|
||
|
|
n = gdb.filter(priority=50)
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
|
||
|
|
gdb.add_gesture('T', [TGesture], priority=51)
|
||
|
|
n = gdb.filter(priority=50)
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
n = gdb.filter(priority=51)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
|
||
|
|
gdb.add_gesture('T', [TGesture], priority=52)
|
||
|
|
n = gdb.filter(priority=[0, 51])
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(priority=[0, 52])
|
||
|
|
self.assertEqual(len(n), 2)
|
||
|
|
n = gdb.filter(priority=[51, 52])
|
||
|
|
self.assertEqual(len(n), 2)
|
||
|
|
n = gdb.filter(priority=[52, 53])
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(priority=[53, 54])
|
||
|
|
self.assertEqual(len(n), 0)
|
||
|
|
|
||
|
|
def test_orientation_filter(self):
|
||
|
|
gdb = Recognizer(db=[self.Ninvar, self.Nbound])
|
||
|
|
n = gdb.filter(orientation_sensitive=True)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(orientation_sensitive=False)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(orientation_sensitive=None)
|
||
|
|
self.assertEqual(len(n), 2)
|
||
|
|
|
||
|
|
gdb.db.append(self.Tinvar)
|
||
|
|
n = gdb.filter(orientation_sensitive=True)
|
||
|
|
self.assertEqual(len(n), 1)
|
||
|
|
n = gdb.filter(orientation_sensitive=False)
|
||
|
|
self.assertEqual(len(n), 2)
|
||
|
|
n = gdb.filter(orientation_sensitive=None)
|
||
|
|
self.assertEqual(len(n), 3)
|
||
|
|
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
# misc tests
|
||
|
|
# -----------------------------------------------------------------------------
|
||
|
|
def test_resample(self):
|
||
|
|
r = kivy.multistroke.resample([Vector(0, 0), Vector(1, 1)], 11)
|
||
|
|
self.assertEqual(len(r), 11)
|
||
|
|
self.assertEqual(round(r[9].x, 1), 0.9)
|
||
|
|
|
||
|
|
r = kivy.multistroke.resample(TGesture, 25)
|
||
|
|
self.assertEqual(len(r), 25)
|
||
|
|
self.assertEqual(round(r[12].x), 81)
|
||
|
|
self.assertEqual(r[12].y, 7)
|
||
|
|
self.assertEqual(TGesture[3].x, r[24].x)
|
||
|
|
self.assertEqual(TGesture[3].y, r[24].y)
|
||
|
|
|
||
|
|
def test_rotateby(self):
|
||
|
|
r = kivy.multistroke.rotate_by(NGesture, 24)
|
||
|
|
self.assertEqual(round(r[2].x, 1), 158.59999999999999)
|
||
|
|
self.assertEqual(round(r[2].y, 1), 54.899999999999999)
|
||
|
|
|
||
|
|
def test_transfer(self):
|
||
|
|
gdb1 = Recognizer(db=[self.Ninvar])
|
||
|
|
gdb2 = Recognizer()
|
||
|
|
gdb1.transfer_gesture(gdb2, name='N')
|
||
|
|
|
||
|
|
r = gdb2.recognize([Ncandidate], max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
def test_export_import_case_1(self):
|
||
|
|
gdb1 = Recognizer(db=[self.Ninvar])
|
||
|
|
gdb2 = Recognizer()
|
||
|
|
|
||
|
|
g = gdb1.export_gesture(name='N')
|
||
|
|
gdb2.import_gesture(g)
|
||
|
|
|
||
|
|
r = gdb2.recognize([Ncandidate], max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
def test_export_import_case_2(self):
|
||
|
|
from tempfile import mkstemp
|
||
|
|
import os
|
||
|
|
gdb1 = Recognizer(db=[self.Ninvar, self.Tinvar])
|
||
|
|
gdb2 = Recognizer()
|
||
|
|
fh, fn = mkstemp()
|
||
|
|
os.close(fh)
|
||
|
|
g = gdb1.export_gesture(name='N', filename=fn)
|
||
|
|
|
||
|
|
gdb2.import_gesture(filename=fn)
|
||
|
|
os.unlink(fn)
|
||
|
|
|
||
|
|
self.assertEqual(len(gdb1.db), 2)
|
||
|
|
self.assertEqual(len(gdb2.db), 1)
|
||
|
|
r = gdb2.recognize([Ncandidate], max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
# ------------------------------------------------------------------------
|
||
|
|
# Test protractor
|
||
|
|
# ------------------------------------------------------------------------
|
||
|
|
def test_protractor_invariant(self):
|
||
|
|
gdb = Recognizer(db=[self.Tinvar, self.Ninvar])
|
||
|
|
r = gdb.recognize([NGesture], orientation_sensitive=False,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] == 1.0)
|
||
|
|
|
||
|
|
r = gdb.recognize([NGesture], orientation_sensitive=True,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], None)
|
||
|
|
self.assertEqual(r.best['score'], 0)
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], orientation_sensitive=False,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
def test_protractor_bound(self):
|
||
|
|
gdb = Recognizer(db=[self.Tbound, self.Nbound])
|
||
|
|
r = gdb.recognize([NGesture], orientation_sensitive=True,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] >= 0.99)
|
||
|
|
|
||
|
|
r = gdb.recognize([NGesture], orientation_sensitive=False,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], None)
|
||
|
|
self.assertEqual(r.best['score'], 0)
|
||
|
|
|
||
|
|
r = gdb.recognize([Ncandidate], orientation_sensitive=True,
|
||
|
|
max_gpf=0)
|
||
|
|
self.assertEqual(r.best['name'], 'N')
|
||
|
|
self.assertTrue(r.best['score'] > 0.94 and r.best['score'] < 0.95)
|
||
|
|
|
||
|
|
|
||
|
|
if __name__ == '__main__':
|
||
|
|
unittest.main()
|