def testNoPunct(self):
test_string = [u"abc", u"a;m".encode("utf-8")]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.HAS_NO_PUNCT_OR_SYMBOL)
self.assertAllEqual(shapes, [True, False])
def testTrailingPunct(self):
test_string = [u"abc", u";b", u"b;", u";,\u0f08".encode("utf-8")]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.ENDS_WITH_PUNCT_OR_SYMBOL)
self.assertAllEqual(shapes, [False, False, True, True])
def testNoDigits(self):
test_string = [u"abc", u"a\u06f3m".encode("utf-8")]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.HAS_NO_DIGITS)
self.assertAllEqual(shapes, [True, False])
def testNumericValue(self):
test_string = [u"98.6", u"-0.3", u"2.783E4", u"e4", u"1e10"]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.IS_NUMERIC_VALUE)
self.assertAllEqual(shapes, [True, True, True, False, True])
def testEllipsis(self):
test_string = [u"abc", u"abc...", u"...abc", u"abc\u2026".encode("utf-8")]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.ENDS_WITH_ELLIPSIS)
self.assertAllEqual(shapes, [False, True, False, True])
def testNonShapePassedToShapeArg(self):
test_string = [u"abc", u"ABc", u"ABC"]
with self.assertRaises(TypeError):
wordshape_ops.wordshape(test_string, "This is not a Shape")
def testMathSymbol(self):
test_string = [u"''", u"\u003c", u"\uff07".encode("utf-8")]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.HAS_MATH_SYMBOL)
self.assertAllEqual(shapes, [False, True, False])
def testIsEmoticon(self):
test_string = [u"abc", u":-)", u"O:)", u"8)x", u":\u3063C", u"abc:-)"]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.IS_EMOTICON)
self.assertAllEqual(shapes, [False, True, False, False, True, False])
def testMixedCase(self):
test_string = [u"abc", u"ABc", u"ABC", u"abC"]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.HAS_MIXED_CASE)
self.assertAllEqual(shapes, [False, True, False, True])
def testMixedCaseLetters(self):
test_string = [u"abc", u"ABc", u"ABC", u"abC", u"abC."]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.IS_MIXED_CASE_LETTERS)
self.assertAllEqual(shapes, [False, True, False, True, False])
def testAllLowercase(self):
test_string = [u"abc", u"ABc", u"ABC"]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.IS_LOWERCASE)
self.assertAllEqual(shapes, [True, False, False])
def testExtendedEmojis(self):
test_string = [
"‼",
"⁉",
"ℹ",
"↘",
"↩",
"⌚",
"⌛",
"⏳",
"⌨",
"⏏",
"⏩",
"⏺",
"⏰",
"⏱",
"⏲",
"🕰",
"Ⓜ",
"▪",
"⬛",
"⬜",
"✂",
"✅",
"✈",
"✉",
"✊",
"✊🏿",
"✋",
"✌",
"🤘🏾",
"🤞🏿",
"✍",
"✏",
"✒",
"✔",
"✝",
"✡",
"✨",
"✳",
"✴",
"❄",
"❇",
"❌",
"❎",
"❓",
"❔",
"❗",
"❕",
"❣",
"❤",
"➕",
"➖",
"➗",
"⤴",
"⤵",
"⬅",
"⭐",
"⭕",
"〰",
"〽",
"㊗",
"🀄",
"🃏",
"🅰",
"🅱",
"🅾",
"🅿",
"🆎",
"🆑",
"🆒",
"🆔",
"🆗",
"🆘",
"🆙",
"🆚",
"🈁",
"🈂",
"🈚",
"🈯",
"🈴",
"🈳",
"🈺",
"🉐",
"🉑",
"🌍",
"🏔",
"🍾",
"🐯",
"🐆",
"🦇",
"🦅",
"🐝",
"🦖",
"🐉",
"🦠",
"🔎",
"⚗",
"🕯",
"💡",
"📽",
"📡",
"🧮",
"🔋",
"📲",
"☎",
"🥁",
"🎧",
"🎼",
"🔊",
"💍",
"👗",
"🕶",
"🎭",
"🔮",
"🧬",
"🔬",
"🤹",
"🚵",
"🧗",
"🧗🏼♀️",
"🧗🏿♂️",
"🥋",
"🎳",
"🏈",
"🏅",
"🎑",
"🎉",
"🎄",
"🌊",
"⚡",
"🌖",
"🚀",
"🚠",
"🛩",
"🛴",
"🏎",
"🚅",
"🌆",
"🕌",
"🕍",
"⛪",
"🗽",
"🏘",
"🍵",
"🍫",
"🦑",
"🍱",
"🥦",
"🥑",
"🌴",
"🌼",
"🦂",
"🐬",
"🥀",
"🧖🏾",
"🧕🏿",
"🧔🏼",
"🧒🏾",
"🧛",
"🧝🏻",
"🧞",
"🧟",
"🧙🏾",
"🧚🏻",
"💃🏽",
"👯",
"🧘",
"🦱",
"👪",
"👩👩👧👦",
"👨🏿🤝👨🏻",
"🕵️♀️",
"🧑🚀",
"👩✈️",
"🧑🏿⚕️",
"🧑🏾⚖️",
"🧠",
"👁️🗨️",
"🙉",
"🤗",
"👏",
"💏",
"🧯",
"🛒",
"🧺",
"🧷",
"💊",
"🧲",
"⛓",
"⚖",
"🛡",
"🏹",
"🎣",
"⚔",
"🔨",
"📌",
"📊",
"📈",
"💹",
"💸",
"💵",
"📜",
"📚",
"📆",
"💼",
"📝",
"📬",
"🔏",
"🔓",
"🔑",
"🗃",
"🚿",
"🛏",
"🗿",
"🏧",
"🚮",
"🚰",
"♿",
"🚻",
"🚾",
"🛄",
"⚠",
"🚸",
"⛔",
"🚭",
"☣",
"🔃",
"🔚",
"🔚",
"⚛",
"♈",
"🔆",
"🎦",
"⚕",
"♻",
"⚜",
"💠",
"🏁",
"🚩",
"🎌",
"🏴☠️",
"🇺🇸",
"🇨🇭",
"🇺🇦",
"🇿🇼",
"🇦🇴",
"🇦🇨",
"🇦🇶",
"🇺🇳",
"🇪🇺",
"🇧🇿",
"🇵🇲",
"🇮🇴",
"🇻🇮",
"🇨🇽",
"🏴",
"🇧🇱",
u"\U0001fa70".encode("utf-8"), # ballet shoes.
u"\U0001fa7a".encode("utf-8"), # stethoscope.
u"\U0001fa80".encode("utf-8"), # yo-yo.
u"\U0001fa82".encode("utf-8"), # parachute.
u"\U0001fa86".encode("utf-8"), # nesting dolls.
u"\U0001fa90".encode("utf-8"), # ringed planet.
u"\U0001fa97".encode("utf-8"), # accordion.
u"\U0001fa99".encode("utf-8"), # coin.
u"\U0001fa9c".encode("utf-8"), # ladder.
u"\U0001fa9f".encode("utf-8"), # window.
u"\U0001faa1".encode("utf-8"), # sewing needle.
u"\U0001faa8".encode("utf-8"), # rock.
u"\U0001fab0".encode("utf-8"), # fly.
u"\U0001fab4".encode("utf-8"), # potted plant.
u"\U0001fab6".encode("utf-8"), # feather.
u"\U0001fac0".encode("utf-8"), # anatomical heart.
u"\U0001fac2".encode("utf-8"), # people hugging.
u"\U0001fad0".encode("utf-8"), # blueberries.
u"\U0001fad2".encode("utf-8"), # olive.
u"\U0001fad6".encode("utf-8"), # teapot.
]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.HAS_EMOJI)
self.assertAllEqual(shapes, [True] * len(test_string))
def testOnlyDigits(self):
test_string = [u"abc", u"a9b".encode("utf-8"), u"90\u06f3".encode("utf-8")]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.HAS_ONLY_DIGITS)
self.assertAllEqual(shapes, [False, False, True])
def testSentenceTerminal(self):
test_string = [u"abc", u".b", u"b.", u"b,", u"b!!!", u"abc?!"]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.ENDS_WITH_SENTENCE_TERMINAL)
self.assertAllEqual(shapes, [False, False, True, False, True, True])
def testCurrencySymbol(self):
test_string = [u"''", u"ABc$", u"$\uff07".encode("utf-8")]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.HAS_CURRENCY_SYMBOL)
self.assertAllEqual(shapes, [False, True, True])
def testMultipleTerminalPunct(self):
test_string = [u"abc", u".b", u"b.", u"b,,", u"b!!!", u"abc?!"]
shapes = wordshape_ops.wordshape(
test_string,
wordshape_ops.WordShape.ENDS_WITH_MULTIPLE_TERMINAL_PUNCT)
self.assertAllEqual(shapes, [False, False, False, True, True, True])
def testCurrencySymbolAtBeginning(self):
test_string = [u"''", u"ABc$", u"$ABc", u"A$Bc"]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.HAS_CURRENCY_SYMBOL)
self.assertAllEqual(shapes, [False, True, True, True])
def testAcronym(self):
test_string = [u"abc", u"A.B.", u"A.B.C.)", u"ABC"]
shapes = wordshape_ops.wordshape(
test_string, wordshape_ops.WordShape.IS_ACRONYM_WITH_PERIODS)
self.assertAllEqual(shapes, [False, True, False, False])
def testAllPunct(self):
test_string = [u"abc", u"a;b".encode("utf-8"), u";,\u0f08".encode("utf-8")]
shapes = wordshape_ops.wordshape(test_string,
wordshape_ops.WordShape.IS_PUNCT_OR_SYMBOL)
self.assertAllEqual(shapes, [False, False, True])
