PNG  IHDRxsBIT|d pHYs+tEXtSoftwarewww.inkscape.org<,tEXtComment File Manager

File Manager

Path: /opt/cloudlinux/venv/lib/python3.11/site-packages/numpy/core/tests/

Viewing File: test_unicode.py

import pytest

import numpy as np
from numpy.testing import assert_, assert_equal, assert_array_equal

def buffer_length(arr):
    if isinstance(arr, str):
        if not arr:
            charmax = 0
        else:
            charmax = max([ord(c) for c in arr])
        if charmax < 256:
            size = 1
        elif charmax < 65536:
            size = 2
        else:
            size = 4
        return size * len(arr)
    v = memoryview(arr)
    if v.shape is None:
        return len(v) * v.itemsize
    else:
        return np.prod(v.shape) * v.itemsize


# In both cases below we need to make sure that the byte swapped value (as
# UCS4) is still a valid unicode:
# Value that can be represented in UCS2 interpreters
ucs2_value = '\u0900'
# Value that cannot be represented in UCS2 interpreters (but can in UCS4)
ucs4_value = '\U00100900'


def test_string_cast():
    str_arr = np.array(["1234", "1234\0\0"], dtype='S')
    uni_arr1 = str_arr.astype('>U')
    uni_arr2 = str_arr.astype('<U')

    assert_array_equal(str_arr != uni_arr1, np.ones(2, dtype=bool))
    assert_array_equal(uni_arr1 != str_arr, np.ones(2, dtype=bool))
    assert_array_equal(str_arr == uni_arr1, np.zeros(2, dtype=bool))
    assert_array_equal(uni_arr1 == str_arr, np.zeros(2, dtype=bool))

    assert_array_equal(uni_arr1, uni_arr2)


############################################################
#    Creation tests
############################################################

class CreateZeros:
    """Check the creation of zero-valued arrays"""

    def content_check(self, ua, ua_scalar, nbytes):

        # Check the length of the unicode base type
        assert_(int(ua.dtype.str[2:]) == self.ulen)
        # Check the length of the data buffer
        assert_(buffer_length(ua) == nbytes)
        # Small check that data in array element is ok
        assert_(ua_scalar == '')
        # Encode to ascii and double check
        assert_(ua_scalar.encode('ascii') == b'')
        # Check buffer lengths for scalars
        assert_(buffer_length(ua_scalar) == 0)

    def test_zeros0D(self):
        # Check creation of 0-dimensional objects
        ua = np.zeros((), dtype='U%s' % self.ulen)
        self.content_check(ua, ua[()], 4*self.ulen)

    def test_zerosSD(self):
        # Check creation of single-dimensional objects
        ua = np.zeros((2,), dtype='U%s' % self.ulen)
        self.content_check(ua, ua[0], 4*self.ulen*2)
        self.content_check(ua, ua[1], 4*self.ulen*2)

    def test_zerosMD(self):
        # Check creation of multi-dimensional objects
        ua = np.zeros((2, 3, 4), dtype='U%s' % self.ulen)
        self.content_check(ua, ua[0, 0, 0], 4*self.ulen*2*3*4)
        self.content_check(ua, ua[-1, -1, -1], 4*self.ulen*2*3*4)


class TestCreateZeros_1(CreateZeros):
    """Check the creation of zero-valued arrays (size 1)"""
    ulen = 1


class TestCreateZeros_2(CreateZeros):
    """Check the creation of zero-valued arrays (size 2)"""
    ulen = 2


class TestCreateZeros_1009(CreateZeros):
    """Check the creation of zero-valued arrays (size 1009)"""
    ulen = 1009


class CreateValues:
    """Check the creation of unicode arrays with values"""

    def content_check(self, ua, ua_scalar, nbytes):

        # Check the length of the unicode base type
        assert_(int(ua.dtype.str[2:]) == self.ulen)
        # Check the length of the data buffer
        assert_(buffer_length(ua) == nbytes)
        # Small check that data in array element is ok
        assert_(ua_scalar == self.ucs_value*self.ulen)
        # Encode to UTF-8 and double check
        assert_(ua_scalar.encode('utf-8') ==
                        (self.ucs_value*self.ulen).encode('utf-8'))
        # Check buffer lengths for scalars
        if self.ucs_value == ucs4_value:
            # In UCS2, the \U0010FFFF will be represented using a
            # surrogate *pair*
            assert_(buffer_length(ua_scalar) == 2*2*self.ulen)
        else:
            # In UCS2, the \uFFFF will be represented using a
            # regular 2-byte word
            assert_(buffer_length(ua_scalar) == 2*self.ulen)

    def test_values0D(self):
        # Check creation of 0-dimensional objects with values
        ua = np.array(self.ucs_value*self.ulen, dtype='U%s' % self.ulen)
        self.content_check(ua, ua[()], 4*self.ulen)

    def test_valuesSD(self):
        # Check creation of single-dimensional objects with values
        ua = np.array([self.ucs_value*self.ulen]*2, dtype='U%s' % self.ulen)
        self.content_check(ua, ua[0], 4*self.ulen*2)
        self.content_check(ua, ua[1], 4*self.ulen*2)

    def test_valuesMD(self):
        # Check creation of multi-dimensional objects with values
        ua = np.array([[[self.ucs_value*self.ulen]*2]*3]*4, dtype='U%s' % self.ulen)
        self.content_check(ua, ua[0, 0, 0], 4*self.ulen*2*3*4)
        self.content_check(ua, ua[-1, -1, -1], 4*self.ulen*2*3*4)


class TestCreateValues_1_UCS2(CreateValues):
    """Check the creation of valued arrays (size 1, UCS2 values)"""
    ulen = 1
    ucs_value = ucs2_value


class TestCreateValues_1_UCS4(CreateValues):
    """Check the creation of valued arrays (size 1, UCS4 values)"""
    ulen = 1
    ucs_value = ucs4_value


class TestCreateValues_2_UCS2(CreateValues):
    """Check the creation of valued arrays (size 2, UCS2 values)"""
    ulen = 2
    ucs_value = ucs2_value


class TestCreateValues_2_UCS4(CreateValues):
    """Check the creation of valued arrays (size 2, UCS4 values)"""
    ulen = 2
    ucs_value = ucs4_value


class TestCreateValues_1009_UCS2(CreateValues):
    """Check the creation of valued arrays (size 1009, UCS2 values)"""
    ulen = 1009
    ucs_value = ucs2_value


class TestCreateValues_1009_UCS4(CreateValues):
    """Check the creation of valued arrays (size 1009, UCS4 values)"""
    ulen = 1009
    ucs_value = ucs4_value


############################################################
#    Assignment tests
############################################################

class AssignValues:
    """Check the assignment of unicode arrays with values"""

    def content_check(self, ua, ua_scalar, nbytes):

        # Check the length of the unicode base type
        assert_(int(ua.dtype.str[2:]) == self.ulen)
        # Check the length of the data buffer
        assert_(buffer_length(ua) == nbytes)
        # Small check that data in array element is ok
        assert_(ua_scalar == self.ucs_value*self.ulen)
        # Encode to UTF-8 and double check
        assert_(ua_scalar.encode('utf-8') ==
                        (self.ucs_value*self.ulen).encode('utf-8'))
        # Check buffer lengths for scalars
        if self.ucs_value == ucs4_value:
            # In UCS2, the \U0010FFFF will be represented using a
            # surrogate *pair*
            assert_(buffer_length(ua_scalar) == 2*2*self.ulen)
        else:
            # In UCS2, the \uFFFF will be represented using a
            # regular 2-byte word
            assert_(buffer_length(ua_scalar) == 2*self.ulen)

    def test_values0D(self):
        # Check assignment of 0-dimensional objects with values
        ua = np.zeros((), dtype='U%s' % self.ulen)
        ua[()] = self.ucs_value*self.ulen
        self.content_check(ua, ua[()], 4*self.ulen)

    def test_valuesSD(self):
        # Check assignment of single-dimensional objects with values
        ua = np.zeros((2,), dtype='U%s' % self.ulen)
        ua[0] = self.ucs_value*self.ulen
        self.content_check(ua, ua[0], 4*self.ulen*2)
        ua[1] = self.ucs_value*self.ulen
        self.content_check(ua, ua[1], 4*self.ulen*2)

    def test_valuesMD(self):
        # Check assignment of multi-dimensional objects with values
        ua = np.zeros((2, 3, 4), dtype='U%s' % self.ulen)
        ua[0, 0, 0] = self.ucs_value*self.ulen
        self.content_check(ua, ua[0, 0, 0], 4*self.ulen*2*3*4)
        ua[-1, -1, -1] = self.ucs_value*self.ulen
        self.content_check(ua, ua[-1, -1, -1], 4*self.ulen*2*3*4)


class TestAssignValues_1_UCS2(AssignValues):
    """Check the assignment of valued arrays (size 1, UCS2 values)"""
    ulen = 1
    ucs_value = ucs2_value


class TestAssignValues_1_UCS4(AssignValues):
    """Check the assignment of valued arrays (size 1, UCS4 values)"""
    ulen = 1
    ucs_value = ucs4_value


class TestAssignValues_2_UCS2(AssignValues):
    """Check the assignment of valued arrays (size 2, UCS2 values)"""
    ulen = 2
    ucs_value = ucs2_value


class TestAssignValues_2_UCS4(AssignValues):
    """Check the assignment of valued arrays (size 2, UCS4 values)"""
    ulen = 2
    ucs_value = ucs4_value


class TestAssignValues_1009_UCS2(AssignValues):
    """Check the assignment of valued arrays (size 1009, UCS2 values)"""
    ulen = 1009
    ucs_value = ucs2_value


class TestAssignValues_1009_UCS4(AssignValues):
    """Check the assignment of valued arrays (size 1009, UCS4 values)"""
    ulen = 1009
    ucs_value = ucs4_value


############################################################
#    Byteorder tests
############################################################

class ByteorderValues:
    """Check the byteorder of unicode arrays in round-trip conversions"""

    def test_values0D(self):
        # Check byteorder of 0-dimensional objects
        ua = np.array(self.ucs_value*self.ulen, dtype='U%s' % self.ulen)
        ua2 = ua.newbyteorder()
        # This changes the interpretation of the data region (but not the
        #  actual data), therefore the returned scalars are not
        #  the same (they are byte-swapped versions of each other).
        assert_(ua[()] != ua2[()])
        ua3 = ua2.newbyteorder()
        # Arrays must be equal after the round-trip
        assert_equal(ua, ua3)

    def test_valuesSD(self):
        # Check byteorder of single-dimensional objects
        ua = np.array([self.ucs_value*self.ulen]*2, dtype='U%s' % self.ulen)
        ua2 = ua.newbyteorder()
        assert_((ua != ua2).all())
        assert_(ua[-1] != ua2[-1])
        ua3 = ua2.newbyteorder()
        # Arrays must be equal after the round-trip
        assert_equal(ua, ua3)

    def test_valuesMD(self):
        # Check byteorder of multi-dimensional objects
        ua = np.array([[[self.ucs_value*self.ulen]*2]*3]*4,
                      dtype='U%s' % self.ulen)
        ua2 = ua.newbyteorder()
        assert_((ua != ua2).all())
        assert_(ua[-1, -1, -1] != ua2[-1, -1, -1])
        ua3 = ua2.newbyteorder()
        # Arrays must be equal after the round-trip
        assert_equal(ua, ua3)

    def test_values_cast(self):
        # Check byteorder of when casting the array for a strided and
        # contiguous array:
        test1 = np.array([self.ucs_value*self.ulen]*2, dtype='U%s' % self.ulen)
        test2 = np.repeat(test1, 2)[::2]
        for ua in (test1, test2):
            ua2 = ua.astype(dtype=ua.dtype.newbyteorder())
            assert_((ua == ua2).all())
            assert_(ua[-1] == ua2[-1])
            ua3 = ua2.astype(dtype=ua.dtype)
            # Arrays must be equal after the round-trip
            assert_equal(ua, ua3)

    def test_values_updowncast(self):
        # Check byteorder of when casting the array to a longer and shorter
        # string length for strided and contiguous arrays
        test1 = np.array([self.ucs_value*self.ulen]*2, dtype='U%s' % self.ulen)
        test2 = np.repeat(test1, 2)[::2]
        for ua in (test1, test2):
            # Cast to a longer type with zero padding
            longer_type = np.dtype('U%s' % (self.ulen+1)).newbyteorder()
            ua2 = ua.astype(dtype=longer_type)
            assert_((ua == ua2).all())
            assert_(ua[-1] == ua2[-1])
            # Cast back again with truncating:
            ua3 = ua2.astype(dtype=ua.dtype)
            # Arrays must be equal after the round-trip
            assert_equal(ua, ua3)


class TestByteorder_1_UCS2(ByteorderValues):
    """Check the byteorder in unicode (size 1, UCS2 values)"""
    ulen = 1
    ucs_value = ucs2_value


class TestByteorder_1_UCS4(ByteorderValues):
    """Check the byteorder in unicode (size 1, UCS4 values)"""
    ulen = 1
    ucs_value = ucs4_value


class TestByteorder_2_UCS2(ByteorderValues):
    """Check the byteorder in unicode (size 2, UCS2 values)"""
    ulen = 2
    ucs_value = ucs2_value


class TestByteorder_2_UCS4(ByteorderValues):
    """Check the byteorder in unicode (size 2, UCS4 values)"""
    ulen = 2
    ucs_value = ucs4_value


class TestByteorder_1009_UCS2(ByteorderValues):
    """Check the byteorder in unicode (size 1009, UCS2 values)"""
    ulen = 1009
    ucs_value = ucs2_value


class TestByteorder_1009_UCS4(ByteorderValues):
    """Check the byteorder in unicode (size 1009, UCS4 values)"""
    ulen = 1009
    ucs_value = ucs4_value
b IDATxytVսϓ22 A@IR :hCiZ[v*E:WũZA ^dQeQ @ !jZ'>gsV仿$|?g)&x-EIENT ;@xT.i%-X}SvS5.r/UHz^_$-W"w)Ɗ/@Z &IoX P$K}JzX:;` &, ŋui,e6mX ԵrKb1ԗ)DADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADADA݀!I*]R;I2$eZ#ORZSrr6mteffu*((Pu'v{DIߔ4^pIm'77WEEE;vƎ4-$]'RI{\I&G :IHJ DWBB=\WR޽m o$K(V9ABB.}jѢv`^?IOȅ} ڶmG}T#FJ`56$-ھ}FI&v;0(h;Б38CӧOWf!;A i:F_m9s&|q%=#wZprrrla A &P\\СC[A#! {olF} `E2}MK/vV)i{4BffV\|ۭX`b@kɶ@%i$K z5zhmX[IXZ` 'b%$r5M4º/l ԃߖxhʔ)[@=} K6IM}^5k㏷݆z ΗÿO:gdGBmyT/@+Vɶ纽z񕏵l.y޴it뭷zV0[Y^>Wsqs}\/@$(T7f.InݺiR$푔n.~?H))\ZRW'Mo~v Ov6oԃxz! S,&xm/yɞԟ?'uaSѽb,8GלKboi&3t7Y,)JJ c[nzӳdE&KsZLӄ I?@&%ӟ۶mSMMњ0iؐSZ,|J+N ~,0A0!5%Q-YQQa3}$_vVrf9f?S8`zDADADADADADADADADAdqP,تmMmg1V?rSI꒟]u|l RCyEf٢9 jURbztѰ!m5~tGj2DhG*{H9)꒟ר3:(+3\?/;TUݭʴ~S6lڧUJ*i$d(#=Yݺd{,p|3B))q:vN0Y.jkק6;SɶVzHJJЀ-utѹսk>QUU\޲~]fFnK?&ߡ5b=z9)^|u_k-[y%ZNU6 7Mi:]ۦtk[n X(e6Bb."8cۭ|~teuuw|ήI-5"~Uk;ZicEmN/:]M> cQ^uiƞ??Ңpc#TUU3UakNwA`:Y_V-8.KKfRitv޲* 9S6ֿj,ՃNOMߤ]z^fOh|<>@Å5 _/Iu?{SY4hK/2]4%it5q]GGe2%iR| W&f*^]??vq[LgE_3f}Fxu~}qd-ږFxu~I N>\;͗O֊:̗WJ@BhW=y|GgwܷH_NY?)Tdi'?խwhlmQi !SUUsw4kӺe4rfxu-[nHtMFj}H_u~w>)oV}(T'ebʒv3_[+vn@Ȭ\S}ot}w=kHFnxg S 0eޢm~l}uqZfFoZuuEg `zt~? b;t%>WTkķh[2eG8LIWx,^\thrl^Ϊ{=dž<}qV@ ⠨Wy^LF_>0UkDuʫuCs$)Iv:IK;6ֲ4{^6եm+l3>݆uM 9u?>Zc }g~qhKwڭeFMM~pМuqǿz6Tb@8@Y|jx](^]gf}M"tG -w.@vOqh~/HII`S[l.6nØXL9vUcOoB\xoǤ'T&IǍQw_wpv[kmO{w~>#=P1Pɞa-we:iǏlHo׈꒟f9SzH?+shk%Fs:qVhqY`jvO'ρ?PyX3lх]˾uV{ݞ]1,MzYNW~̈́ joYn}ȚF߾׮mS]F z+EDxm/d{F{-W-4wY듏:??_gPf ^3ecg ҵs8R2מz@TANGj)}CNi/R~}c:5{!ZHӋӾ6}T]G]7W6^n 9*,YqOZj:P?Q DFL|?-^.Ɵ7}fFh׶xe2Pscz1&5\cn[=Vn[ĶE鎀uˌd3GII k;lNmشOuuRVfBE]ۣeӶu :X-[(er4~LHi6:Ѻ@ԅrST0trk%$Č0ez" *z"T/X9|8.C5Feg}CQ%͞ˣJvL/?j^h&9xF`њZ(&yF&Iݻfg#W;3^{Wo^4'vV[[K';+mӍִ]AC@W?1^{එyh +^]fm~iԵ]AB@WTk̏t uR?l.OIHiYyԶ]Aˀ7c:q}ힽaf6Z~қm(+sK4{^6}T*UUu]n.:kx{:2 _m=sAߤU@?Z-Vކеz왍Nэ{|5 pڶn b p-@sPg]0G7fy-M{GCF'%{4`=$-Ge\ eU:m+Zt'WjO!OAF@ik&t݆ϥ_ e}=]"Wz_.͜E3leWFih|t-wZۍ-uw=6YN{6|} |*={Ѽn.S.z1zjۻTH]흾 DuDvmvK.`V]yY~sI@t?/ϓ. m&["+P?MzovVЫG3-GRR[(!!\_,^%?v@ҵő m`Y)tem8GMx.))A]Y i`ViW`?^~!S#^+ѽGZj?Vģ0.))A꨷lzL*]OXrY`DBBLOj{-MH'ii-ϰ ok7^ )쭡b]UXSְmռY|5*cֽk0B7镹%ڽP#8nȎq}mJr23_>lE5$iwui+ H~F`IjƵ@q \ @#qG0".0" l`„.0! ,AQHN6qzkKJ#o;`Xv2>,tێJJ7Z/*A .@fفjMzkg @TvZH3Zxu6Ra'%O?/dQ5xYkU]Rֽkق@DaS^RSּ5|BeHNN͘p HvcYcC5:y #`οb;z2.!kr}gUWkyZn=f Pvsn3p~;4p˚=ē~NmI] ¾ 0lH[_L hsh_ғߤc_њec)g7VIZ5yrgk̞W#IjӪv>՞y睝M8[|]\շ8M6%|@PZڨI-m>=k='aiRo-x?>Q.}`Ȏ:Wsmu u > .@,&;+!!˱tﭧDQwRW\vF\~Q7>spYw$%A~;~}6¾ g&if_=j,v+UL1(tWake:@Ș>j$Gq2t7S?vL|]u/ .(0E6Mk6hiۺzښOrifޱxm/Gx> Lal%%~{lBsR4*}{0Z/tNIɚpV^#Lf:u@k#RSu =S^ZyuR/.@n&΃z~B=0eg뺆#,Þ[B/?H uUf7y Wy}Bwegל`Wh(||`l`.;Ws?V@"c:iɍL֯PGv6zctM̠':wuW;d=;EveD}9J@B(0iհ bvP1{\P&G7D޴Iy_$-Qjm~Yrr&]CDv%bh|Yzni_ˆR;kg}nJOIIwyuL}{ЌNj}:+3Y?:WJ/N+Rzd=hb;dj͒suݔ@NKMԄ jqzC5@y°hL m;*5ezᕏ=ep XL n?מ:r`۵tŤZ|1v`V뽧_csج'ߤ%oTuumk%%%h)uy]Nk[n 'b2 l.=͜E%gf$[c;s:V-͞WߤWh-j7]4=F-X]>ZLSi[Y*We;Zan(ӇW|e(HNNP5[= r4tP &0<pc#`vTNV GFqvTi*Tyam$ߏWyE*VJKMTfFw>'$-ؽ.Ho.8c"@DADADADADADADADADA~j*֘,N;Pi3599h=goضLgiJ5փy~}&Zd9p֚ e:|hL``b/d9p? fgg+%%hMgXosج, ΩOl0Zh=xdjLmhݻoO[g_l,8a]٭+ӧ0$I]c]:粹:Teꢢ"5a^Kgh,&= =՟^߶“ߢE ܹS J}I%:8 IDAT~,9/ʃPW'Mo}zNƍ쨓zPbNZ~^z=4mswg;5 Y~SVMRXUյڱRf?s:w ;6H:ºi5-maM&O3;1IKeamZh͛7+##v+c ~u~ca]GnF'ټL~PPPbn voC4R,ӟgg %hq}@#M4IÇ Oy^xMZx ) yOw@HkN˖-Sǎmb]X@n+i͖!++K3gd\$mt$^YfJ\8PRF)77Wא!Cl$i:@@_oG I{$# 8磌ŋ91A (Im7֭>}ߴJq7ޗt^ -[ԩSj*}%]&' -ɓ'ꫯVzzvB#;a 7@GxI{j޼ƌ.LÇWBB7`O"I$/@R @eee@۷>}0,ɒ2$53Xs|cS~rpTYYY} kHc %&k.], @ADADADADADADADADA@lT<%''*Lo^={رc5h %$+CnܸQ3fҥK}vUVVs9G R,_{xˇ3o߾;TTTd}馛]uuuG~iԩ@4bnvmvfϞ /Peeeq}}za I~,誫{UWW뮻}_~YƍSMMMYχ֝waw\ďcxꩧtEƍկ_?۷5@u?1kNׯWzz/wy>}zj3 k(ٺuq_Zvf̘:~ ABQ&r|!%KҥKgԞ={<_X-z !CyFUUz~ ABQIIIjݺW$UXXDٳZ~ ABQƍecW$<(~<RSSvZujjjԧOZQu@4 8m&&&jԩg$ď1h ͟?_{768@g =@`)))5o6m3)ѣƌJ;wҿUTT /KZR{~a=@0o<*狔iFɶ[ˎ;T]]OX@?K.ۈxN pppppppppppppppppPfl߾] ,{ァk۶mڿo5BTӦMӴiӴ|r DB2e|An!Dy'tkΝ[A $***t5' "!駟oaDnΝ:t֭[gDШQ06qD;@ x M6v(PiizmZ4ew"@̴ixf [~-Fٱc&IZ2|n!?$@{[HTɏ#@hȎI# _m(F /6Z3z'\r,r!;w2Z3j=~GY7"I$iI.p_"?pN`y DD?: _  Gÿab7J !Bx@0 Bo cG@`1C[@0G @`0C_u V1 aCX>W ` | `!<S `"<. `#c`?cAC4 ?c p#~@0?:08&_MQ1J h#?/`7;I  q 7a wQ A 1 Hp !#<8/#@1Ul7=S=K.4Z?E_$i@!1!E4?`P_  @Bă10#: "aU,xbFY1 [n|n #'vEH:`xb #vD4Y hi.i&EΖv#O H4IŶ}:Ikh @tZRF#(tXҙzZ ?I3l7q@õ|ۍ1,GpuY Ꮿ@hJv#xxk$ v#9 5 }_$c S#=+"K{F*m7`#%H:NRSp6I?sIՖ{Ap$I$I:QRv2$Z @UJ*$]<FO4IENDB`