def __init__(self):
'''
Constructor
'''
self.faker = Faker()
self.dateformat = ListGenWithProb(["%H:%M(:%S)?", "%I:%M(:%S)[ ]?%p"],
[0.5, 0.5])
def gen_compare(sameheight, dsttxt, comparewith):
dstimg = cv2.imread(comparewith)
dstimg = np.mean(dstimg, axis=2).astype(np.uint8)
dstimg_cl = gray2heatmap(dstimg)
cv2.imshow('dst', resizeToHeight(dstimg, sameheight))
cv2.imshow('dst-heatmap', resizeToHeight(dstimg_cl, sameheight))
pipe_name = CMNDPipeName()
hotengen = HoTenGen(os.path.join(RESOURCE_PATH, 'hovaten.csv'))
trashgen = Trash(os.path.join(RESOURCE_PATH, 'hovaten.csv'),0.5)
quanhuyengen = QuanHuyenGen(os.path.join(RESOURCE_PATH, 'tinhtp.csv'), number=2)
filecontent = codecs.open(TEMPORARY_PATH + '4/aligned.txt', encoding='utf8').read()
jsonStr = filecontent.replace('\'', '"')
jsonObj = json.loads(jsonStr)
namegen = ListGenWithProb([hotengen, quanhuyengen, trashgen], [0.4,0.3,0.3])
for i in range(1, 3000):
print i, '-----------------------'
if dsttxt is not None:
pipe_name.txt = dsttxt
else:
pipe_name.txt = namegen.gen()
pipe_name.p.reset(jsonObj)
rs, _, txt = pipe_name.gen()
if rs is None: continue
txt = txt.strip()
print '@@@'+txt+'@@@'
rs_cl = gray2heatmap(rs)
cv2.imshow('gen', resizeToHeight(rs, sameheight))
cv2.imshow('gen-heatmap', resizeToHeight(rs_cl, sameheight))
cv2.waitKey(-1)
def to_weinman(root):
pipe_name = CMNDPipeName()
hotengen = HoTenGen(os.path.join(RESOURCE_PATH, 'hovaten.csv'))
quanhuyengen = QuanHuyenGen(os.path.join(RESOURCE_PATH, 'tinhtp.csv'), number=2)
trashgen = Trash(os.path.join(RESOURCE_PATH, 'hovaten.csv'),0.5)
filecontent = codecs.open(TEMPORARY_PATH + '4/aligned.txt', encoding='utf8').read()
jsonStr = filecontent.replace('\'', '"')
jsonObj = json.loads(jsonStr)
namegen = ListGenWithProb([hotengen, quanhuyengen, trashgen], [0.4,0.3,0.3])
unicodeutil = UnicodeUtil(RESOURCE_PATH + 'diacritics2.csv')
with codecs.open(root + 'anno-train.txt', 'a', encoding='utf8') as annotation_train:
with codecs.open(root + 'anno-test.txt', 'a', encoding='utf8') as annotation_test:
for i in range(1, 3000):
print i, '-----------------------'
p = np.random.rand()
txt = namegen.gen()
txt = re.sub('\d','',txt)
pipe_name.txt = txt
pipe_name.p.reset(jsonObj)
rs, _, txt = pipe_name.gen()
if rs is None: continue
txt = txt.strip()
txt = unicodeutil.to_vni(txt)
newwidth = rs.shape[1] * 32.0 / rs.shape[0]
rs = cv2.resize(rs, (int(newwidth), 32))
cv2.imwrite(root + str(i) + '.jpg', rs)
print '@@@'+txt+'@@@'
cv2.imshow('hihi', rs)
cv2.waitKey(-1)
continue
if i < 2900:
annotation_train.write('./' + str(i) + '.jpg ' + txt + '\n')
else:
annotation_test.write('./' + str(i) + '.jpg ' + txt + '\n')
def __init__(self, afonts):
self.fonts = []
probs = []
for p, fontname, w2h, s2h in afonts:
font = AFont(FONT_PATH+fontname, p, w2h, s2h)
self.allfonts[FONT_PATH+fontname] = font
self.fonts.append(font)
probs.append(p)
self.fontgen = ListGenWithProb(self.fonts, probs)
def __init__(self, afonts=[]):
pygame.init()
self.renderfont = RenderFont(DATA_PATH)
self.height = 100
self.pad_for_cut = 10
self.rel_line_spacing = 1.0
self.loitgfonts = []
self.fonts = []
probs = []
for p, fontname, w2h, s2h in afonts:
self.fonts.append(AFont(FONT_PATH+fontname, p, w2h, s2h))
probs.append(p)
if len(self.fonts) > 0: self.fontgen = ListGenWithProb(self.fonts, probs)
def __init__(self, fromdate=None, todate=None, dateformat=None):
'''
Constructor
'''
self.faker = Faker('en_US')
self.dateformat = ListGenWithProb(*dateformat)
if fromdate is not None:
self.fromdate = datetime.strptime(fromdate, "%Y-%m-%d")
else:
self.fromdate = datetime.now()
if todate is not None:
self.todate = datetime.strptime(todate, "%Y-%m-%d")
else:
self.todate = datetime.now()
class Fonts(object):
allfonts = {}
def __init__(self, afonts):
self.fonts = []
probs = []
for p, fontname, w2h, s2h in afonts:
font = AFont(FONT_PATH+fontname, p, w2h, s2h)
self.allfonts[FONT_PATH+fontname] = font
self.fonts.append(font)
probs.append(p)
self.fontgen = ListGenWithProb(self.fonts, probs)
def genRandom(self):
return self.fontgen.gen()
def genByName(self, query):
rs = []
for name, font in self.allfonts.iteritems():
name = name.split('/')[-1]
if query.lower() in name.lower():
rs.append(font)
if len(rs) == 1:
return rs[0]
else:
return None
class DateGen(Gen):
'''
Date
'''
def __init__(self, fromdate=None, todate=None, dateformat=None):
'''
Constructor
'''
self.faker = Faker('en_US')
self.dateformat = ListGenWithProb(*dateformat)
if fromdate is not None:
self.fromdate = datetime.strptime(fromdate, "%Y-%m-%d")
else:
self.fromdate = datetime.now()
if todate is not None:
self.todate = datetime.strptime(todate, "%Y-%m-%d")
else:
self.todate = datetime.now()
def gen(self):
# d = self.faker.date_between(start_date="-365d", end_date="-60d")
d = self.faker.date_time_between_dates(datetime_start=self.fromdate,
datetime_end=self.todate,
tzinfo=None).date()
fm = self.dateformat.gen()
d = d.strftime(fm)
d = replaceZeroWithProb(d, 0.15)
d = changeCaseWithProb(d, upper=0.5, lower=0.0)
return d
class TimeGen(Gen):
'''
Time
'''
def __init__(self):
'''
Constructor
'''
self.faker = Faker()
self.dateformat = ListGenWithProb(["%H:%M(:%S)?", "%I:%M(:%S)[ ]?%p"],
[0.5, 0.5])
def gen(self):
fm = self.dateformat.gen()
t = self.faker.time(pattern=fm, end_datetime=None)
d = replaceZeroWithProb(t)
d = changeCaseWithProb(d, upper=0.5, lower=0.1)
return d
class RenderText(object):
def __init__(self, afonts=[]):
pygame.init()
self.renderfont = RenderFont(DATA_PATH)
self.height = 100
self.pad_for_cut = 10
self.rel_line_spacing = 1.0
self.loitgfonts = []
self.fonts = []
probs = []
for p, fontname, w2h, s2h in afonts:
self.fonts.append(AFont(FONT_PATH+fontname, p, w2h, s2h))
probs.append(p)
if len(self.fonts) > 0: self.fontgen = ListGenWithProb(self.fonts, probs)
def init_font(self, fontpath, spacing):
font = freetype.Font(FONT_PATH + fontpath, size=self.height)
font.underline = False
font.strong = False
font.oblique = False
font.strength = False
font.antialiased = True
font.origin = True
self.spacing = spacing
return font
def genFont(self):
if len(self.fonts) > 0:
return self.fontgen.gen()
else:
return None
def toMask2(self, afont, txt):
###
s2h = random.uniform(afont.s2h[0], afont.s2h[1])
w2h = random.uniform(afont.w2h[0], afont.w2h[1])
###
txt_arr, _, bbs = self.renderfont.render_singleline(afont.font, txt , w2h, s2h)
newwidth = int(txt_arr.shape[1]*w2h)
txt_arr = cv2.resize(txt_arr,(newwidth, txt_arr.shape[0]))
return txt_arr, txt
def toMask(self, loitgfont, txt, otherlines=False):
if loitgfont.fontpath == '/home/loitg/Downloads/fonts/fontss/receipts/westgate/PKMN-Mystery-Dungeon.ttf' and any(c in txt for c in ['<','>',';','!','$','#','&','/','`','~','@','%','^','*','.']):
return None, None
if loitgfont.fontpath == '/home/loitg/Downloads/fonts/fontss/receipts/general_fairprice/LEFFC2.TTF' and any(c.isdigit() for c in txt):
return None, None
if (loitgfont.fontpath == '/home/loitg/Downloads/fonts/fontss/receipts/general_fairprice/PRINTF Regular.ttf')\
or (loitgfont.fontpath == '/home/loitg/Downloads/fonts/fontss/receipts/dotted/fake receipt.ttf'):
txt = txt.upper()
above = rstr.rstr('ABC0123456789abcdef ', len(txt))
below = rstr.rstr('ABC0123456789abcdef ', len(txt))
multilines = above + '\n' + txt + '\n' + below
if otherlines:
txt_arr, _, bbs = self.renderfont.render_multiline(loitgfont.font, multilines , 0.3, 0.2, 1)
# angle = np.random.rand() * 3 + 3
# if np.random.rand() < 0.5: angle = -angle
# txt_arr= rotate_bound(txt_arr,angle)
else:
txt_arr, _, bbs = self.renderfont.render_multiline(loitgfont.font, multilines , 0.05, 0.5, 1)
# angle = np.random.randn() * 2
# if np.random.rand() < 0.5 and abs(angle) > 1 and abs(angle) < 10:
# txt_arr= rotate_bound(txt_arr,angle)
angle = np.random.randn() * 3
if np.random.rand() < 0.4:
txt_arr= rotate_bound(txt_arr,angle)
newwidth = int(txt_arr.shape[1]*loitgfont.getRatio())
if np.random.rand() < 0.4:
found = re.search( r'\d\.\d\d', txt)
if found:
newwidth *= 2
elif any(c in txt for c in receiptgenline.TOTAL_KEYS) or \
any(c.upper() in txt for c in receiptgenline.TOTAL_KEYS):
newwidth *= 2
txt_arr = cv2.resize(txt_arr,(newwidth, txt_arr.shape[0]))
return txt_arr, txt