X-Git-Url: http://git.sourceforge.jp/view?p=nkf%2Fnkf.git;a=blobdiff_plain;f=nkf_test.pl;h=4e79f6f68e3d575bc33ea72785bb579c0dd7ff49;hp=6a6ef9469a01b1e8d78a3ca970a72fae9733692e;hb=31d5827dbe41292855b3ebbee95dfce1e11d5d53;hpb=454835f76a73534739cd4c7732b85e76d0649465 diff --git a/nkf_test.pl b/nkf_test.pl index 6a6ef94..4e79f6f 100644 --- a/nkf_test.pl +++ b/nkf_test.pl @@ -1,6 +1,9 @@ #!/usr/local/bin/perl # -# nkf test program for nkf 1.7 +# nkf test program for nkf-2 +# +# $Id: nkf_test.pl,v 1.29 2008/02/07 19:25:29 naruse Exp $ +# # Shinji KONO # Sun Aug 18 12:25:40 JST 1996 # Sun Nov 8 00:16:06 JST 1998 @@ -40,13 +43,47 @@ MI?*E\R!+:6=O=2"AIJ'GH["CP:;!IMBGHJ?!J,`* eofeof -$example{'utf'} = unpack('u',<<'eofeof'); +$example{'utf8'} = unpack('u',<<'eofeof'); +M[[N_1FER(A#_(?^Q`\D#$00O!$(E"@`` +eofeof + +$example{'u16L0'} = unpack('u',<<'eofeof'); +M1@!I`'(`)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(* eofeof -print "Ambiguous Case. "; +printf "%-40s", "Ambiguous Case."; &test("$nkf -j",$example{'amb'},$example{'amb.euc'}); # Input assumption -print "SJIS Input assumption "; +printf "%-40s", "SJIS Input assumption"; &test("$nkf -jSx",$example{'amb'},$example{'amb.sjis'}); # Broken JIS -print "Broken JIS "; +printf "%-40s", "Broken JIS"; $input = $example{'jis'}; $input =~ s/\033//g; &test("$nkf -Be",$input,$example{'euc'}); -print "Broken JIS is safe on Normal JIS? "; +printf "%-40s", "Broken JIS is safe on Normal JIS?"; $input = $example{'jis'}; &test("$nkf -Be",$input,$example{'euc'}); +# test_data/cp932 + +$example{'test_data/cp932'} = unpack('u',<<'eofeof'); +%^D`@_$L` +eofeof + +$example{'test_data/cp932.ans'} = unpack('u',<<'eofeof'); +%_/$@_.X` +eofeof + +printf "%-40s", "test_data/cp932"; + &test("$nkf -eS",$example{'test_data/cp932'},$example{'test_data/cp932.ans'}); +printf "%-40s", "test_data/cp51932"; + &test("$nkf --ic=cp932 --oc=cp51932",$example{'test_data/cp932'},$example{'test_data/cp932.ans'}); + +# test_data/cp932inv +printf "%-40s", "test_data/cp932inv"; + &test("$nkf -sE --cp932",$example{'test_data/cp932.ans'},$example{'test_data/cp932'}); + +# test_data/no-cp932inv + +$example{'test_data/no-cp932inv.ans'} = unpack('u',<<'eofeof'); +%[N\@[NP` +eofeof + +printf "%-40s", "test_data/no-cp932inv"; + &test("$nkf -sE --no-cp932",$example{'test_data/cp932.ans'},$example{'test_data/no-cp932inv.ans'}); +# JIS X 0212 +$example{'jisx0212_euc'} = "\x8F\xA2\xAF\x8F\xED\xE3"; +$example{'jisx0212_jis'} = "\x1b\x24\x28\x44\x22\x2f\x6d\x63\x1b\x28\x42"; + +printf "%-40s", "ISO-2022-JP-1 to EUC-JP"; + &test("$nkf --ic=ISO-2022-JP-1 --oc=EUC-JP",$example{'jisx0212_jis'},$example{'jisx0212_euc'}); + +printf "%-40s", "EUC-JP to ISO-2022-JP-1"; + &test("$nkf --ic=EUC-JP --oc=ISO-2022-JP-1",$example{'jisx0212_euc'},$example{'jisx0212_jis'}); +# JIS X 0213 + +$example{'jisx0213_sjis'} = unpack('u',<<'eofeof'); +0@:V(G9ATF)WJIN_W\$#\]``` +eofeof + +$example{'jisx0213_euc'} = unpack('u',<<'eofeof'); +2HJ^O_<_5S_WTJ/[YCZ&AC_[V +eofeof + +$example{'jisx0213_jis2000'} = unpack('u',<<'eofeof'); +;&R0H3R(O+WU/54]]="A^>1LD*%`A(7YV&RA" +eofeof +$example{'jisx0213_jis2004'} = unpack('u',<<'eofeof'); +;&R0H42(O+WU/54]]="A^>1LD*%`A(7YV&RA" +eofeof + +printf "%-40s", "Shift_JISX0213 to EUC-JISX0213"; + &test("$nkf --ic=Shift_JISX0213 --oc=EUC-JISX0213",$example{'jisx0213_sjis'},$example{'jisx0213_euc'}); + +printf "%-40s", "EUC-JISX0213 to Shift_JISX0213"; + &test("$nkf --ic=EUC-JISX0213 --oc=Shift_JISX0213",$example{'jisx0213_euc'},$example{'jisx0213_sjis'}); + +printf "%-40s", "ISO-2022-JP-3 to EUC-JISX0213"; + &test("$nkf --ic=ISO-2022-JP-3 --oc=EUC-JISX0213",$example{'jisx0213_jis2000'},$example{'jisx0213_euc'}); + +printf "%-40s", "ISO-2022-JP-2004 to EUC-JISX0213"; + &test("$nkf --ic=ISO-2022-JP-2004 --oc=EUC-JISX0213",$example{'jisx0213_jis2004'},$example{'jisx0213_euc'}); + +printf "%-40s", "EUC-JISX0213 to ISO-2022-JP-2004"; + &test("$nkf --ic=EUC-JISX0213 --oc=ISO-2022-JP-2004",$example{'jisx0213_euc'},$example{'jisx0213_jis2004'}); +# test_data/no_best_fit_chars + +$example{'test_data/no_best_fit_chars'} = unpack('u',<<'eofeof'); +;XH"5XHBE[[R-[[^@[[^A[[^B[[^C[[^D[[^E +eofeof + +printf "%-40s", "no_best_fit_chars (eucJP-ascii)"; + &test("$nkf -W --oc=eucJP-ascii --no-best-fit-chars",$example{'test_data/no_best_fit_chars'},''); + +$example{'test_data/no_best_fit_chars_ms'} = unpack('u',<<'eofeof'); +9PJ+"H\*EPJ;"K.*`E.*`EN*`ON*(DN.`G``` +eofeof + +printf "%-40s", "no_best_fit_chars (eucJP-ms)"; + &test("$nkf -W --oc=eucJP-ms --no-best-fit-chars",$example{'test_data/no_best_fit_chars_ms'},''); + +$example{'test_data/no_best_fit_chars_cp932'} = unpack('u',<<'eofeof'); +MPJ'"HL*CPJ7"IL*IPJK"J\*LPJW"KL*OPK+"L\*UPK?"N,*YPKK"N\.`PX'# +M@L.#PX3#A<.&PX?#B,.)PXK#B\.,PXW#CL./PY##D<.2PY/#E,.5PY;#F,.9 +MPYK#F\.X0208 conversion # X0208 aphabet -> ASCII @@ -219,11 +420,11 @@ $example{'x0201.utf'} = unpack('u',<<'eofeof'); MY86HZ*>2XX*KXX*MXX*OXX*QXX*SXX*LXX*NXX*PXX*RXX*T"N6%J.B+L>^\ MH>^\HN^\H^^\I.^\I>^\IN^\I^^]@>^]@N^]@^^]A.^]A>^]AN^]APKEA:CH MJ)CEC[?OO('OO*#OO(/OO(3OO(7OO+[OO(;OO(KOO(COO(GBB)+OO(OOO)WO -MO+OOO+WOO9OOO9WOOZ4*Y8V*Z*>2[[VV[[VW[[VX/>^]N>^]NN^]MN^^GN^] -MM^^^GN^]N.^^GN^]N>^^GN^]NN^^GN.!J`KEC8KHIY+OOHKOOI_OOHOOOI_O -MOHSOOI_OOHWOOI_OOH[OOI_OO;?OOI_OO;'OOIX*[[Z*[[Z?[[Z+[[Z?[[Z, -M"FAA;FMA:W4@[[Z*[[Z?[[Z+[[Z?[[Z,[[VD"N^^BN^^G^^^B^^^G^^^C.^^ -2G>^]H0KEC8KHIY+C@:[EOHP* +MO+OOO+WOO9OOO9W"I0KEC8KHIY+OO;;OO;?OO;@][[VY[[VZ[[VV[[Z>[[VW +M[[Z>[[VX[[Z>[[VY[[Z>[[VZ[[Z>XX&H"N6-BNBGDN^^BN^^G^^^B^^^G^^^ +MC.^^G^^^C>^^G^^^CN^^G^^]M^^^G^^]L>^^G@KOOHKOOI_OOHOOOI_OOHP* +M:&%N:V%K=2#OOHKOOI_OOHOOOI_OOHSOO:0*[[Z*[[Z?[[Z+[[Z?[[Z,[[Z= +1[[VA"N6-BNBGDN.!KN6^C`H` eofeof $example{'x0201.jis'} = unpack('u',<<'eofeof'); @@ -255,24 +456,24 @@ eofeof # -X is necessary to allow X0201 in SJIS # -Z convert X0208 alphabet to ASCII -print "X0201 conversion: SJIS "; +printf "%-40s", "X0201 conversion: SJIS"; &test("$nkf -jXZ",$example{'x0201.sjis'},$example{'x0201.x0208'}); -print "X0201 conversion: JIS "; +printf "%-40s", "X0201 conversion: JIS"; &test("$nkf -jZ",$example{'x0201.jis'},$example{'x0201.x0208'}); -print "X0201 conversion:SI/SO "; +printf "%-40s", "X0201 conversion: SI/SO"; &test("$nkf -jZ",$example{'x0201.sosi'},$example{'x0201.x0208'}); -print "X0201 conversion: EUC "; +printf "%-40s", "X0201 conversion: EUC"; &test("$nkf -jZ",$example{'x0201.euc'},$example{'x0201.x0208'}); -print "X0201 conversion: UTF8 "; +printf "%-40s", "X0201 conversion: UTF8"; &test("$nkf -jZ",$example{'x0201.utf'},$example{'x0201.x0208'}); # -x means X0201 output -print "X0201 output: SJIS "; +printf "%-40s", "X0201 output: SJIS"; &test("$nkf -xs",$example{'x0201.euc'},$example{'x0201.sjis'}); -print "X0201 output: JIS "; +printf "%-40s", "X0201 output: JIS"; &test("$nkf -xj",$example{'x0201.sjis'},$example{'x0201.jis'}); -print "X0201 output: EUC "; +printf "%-40s", "X0201 output: EUC"; &test("$nkf -xe",$example{'x0201.jis'},$example{'x0201.euc'}); -print "X0201 output: UTF8 "; +printf "%-40s", "X0201 output: UTF8"; &test("$nkf -xw",$example{'x0201.jis'},$example{'x0201.utf'}); # MIME decode @@ -340,8 +541,8 @@ M(TB1./RD[=ALH0@I"tmp1");print OUT pack('u',$tmp);close(OUT); + # open(OUT,">tmp1");printf "%-40s", OUT pack('u',$tmp);close(OUT); # unbuf mode implies more pessimistic decode -print "MIME decode (unbuf) "; +printf "%-40s", "MIME decode (unbuf)"; $tmp = &test("$nkf -jmNu",$example{'mime.iso2022'},$example{'mime.unbuf'},$example{'mime.unbuf.alt'}); - # open(OUT,">tmp2");print OUT pack('u',$tmp);close(OUT); -print "MIME decode (base64) "; - &test("$nkf -jTmB",$example{'mime.base64'},$example{'mime.base64.ans'}); + # open(OUT,">tmp2");printf "%-40s", OUT pack('u',$tmp);close(OUT); +printf "%-40s", "MIME decode (base64)"; + &test("$nkf -jmB",$example{'mime.base64'},$example{'mime.base64.ans'}); +#MIME BASE64 must be LF? # MIME ISO-8859-1 @@ -386,8 +588,8 @@ eofeof # Without -l, ISO-8859-1 was handled as X0201. -print "MIME ISO-8859-1 (Q) "; - &test("$nkf -ml",$example{'mime.is8859'},$example{'mime.is8859.ans'}); +printf "%-40s", "MIME ISO-8859-1 (Q)"; + &test("$nkf -jml",$example{'mime.is8859'},$example{'mime.is8859.ans'}); # test for -f is not so simple. @@ -403,7 +605,7 @@ $example{'test_data/cr.ans'} = unpack('u',<<'eofeof'); 7&R1")$8D.21(&RA""G1E5)#2D-):TI#46U*0V=K2VE1 +M5%X2D1-:TY343-*1&MB2T5)/3\]"B`]/TE33RTR,#(R+4I0/T(_ +M1WE20TI$5)#2D-):TI#46U*0V=K2VE1 -M5%X2D1-:TY343-*1&MK3WAS;U%G/3T_/2`*"3T_25-/+3(P,C(M -M2E`_0C]'>5)#2D0P:U!Y4D)*15%K4FE224I%;VM3>5)-2D4P:U1I4E!*1DEK -M5E-264=Y:$,_/2`*"3T_25-/+3(P,C(M2E`_0C]'>5)#2D9S:UAI4F9*1T%K -M65-2:4I'46M*0U)M2D-G:V%"5%X2D1-:TY343-'>6A#/ST*(#T_25-/+3(P,C(M2E`_0C]'>5)# +M2D1K:T]Y43E*1#AK45-214I%66M30U)+2D5S:U1#4DY*131K5'E24TI&56M7 +M0U)B1WEH0S\]"B`]/TE33RTR,#(R+4I0/T(_1WE20TI&-&M8>5)G2D=%:UEI +M4FM*0U%K6FE1;TI'9V)+14D]/ST*+2TM+0I3=6)J96-T.B!A86%A(&%A86$@ +M86%A82!A86%A(&%A86$@86%A82!A86%A"B`]/TE33RTR,#(R+4I0/T(_1WE2 +M0TI#26M*0U%M2D-G:TMH5)#2D-):TI#46U*0V=K2VE1 +M5%X2D1-:TY343-*1&MK3WE1.4=Y:$,_/0H@/3])4T\M,C`R,BU* +M4#]"/T=Y4D-*1#AK45-214I%66M30U)+2D5S:U1#4DY*131K5'E24TI&56M7 +M0U)B2D8T:UAY4F='>6A#/ST*(#T_25-/+3(P,C(M2E`_0C]'>5)#2D=%:UEI +M4FM*0U%K6FE1;TI'9V)+14D]/ST*+2TM+0I3=6)J96-T.B!A86%A(&%A86$@ +M86%A82!A86%A(&%A86$@86%A82!A86%A(#T_25-/+3(P,C(M2E`_0C]'>5)# +M2D-):TI"5)#2D-9:TM#47%' +@>6A#/ST@86%A82!A86%A(&%A86$@86%A80HM+2TM"@H` +eofeof + +$example{'test_data/mime_out.ans.alt3'} = unpack('u',<<'eofeof'); +M"BTM+2T*4W5B:F5C=#H@86%A82!A86%A(&%A86$@86%A82!A86%A(&%A86$@ +M86%A82!A86%A(&%A86$@86%A82!A86%A(&%A86$@86%A80H@86%A80HM+2TM +M"E-U8FIE8W0Z(#T_25-/+3(P,C(M2E`_0C]'>5)#2D-):TI#46U*0V=K2VE1 +M5%X2D1-:TY343-*1&MK3WE1.4=Y:$,_/0H@/3])4T\M,C`R,BU* +M4#]"/T=Y4D-*1#AK45-214I%66M30U)+2D5S:U1#4DY*131K5'E24TI&56M7 +M0U)B2D8T:UAX5)#2D=!:UE3 +M4FE*1U%K2D-2;4I#9VMA0G-O46<]/3\]"BTM+2T*4W5B:F5C=#H@86%A82!A +M86%A(&%A86$@86%A82!A86%A(&%A86$@86%A82`]/TE33RTR,#(R+4I0/T(_ +M1WE20TI#26M*0G-O46<]/3\]"B`]/TE33RTR,#(R+4I0/T(_1WE20TI#66M+ +D0U%Q1WEH0S\](&%A86$@86%A82!A86%A(&%A86$*+2TM+0H*1 eofeof -print "test_data/mime_out "; - &test("$nkf -jM",$example{'test_data/mime_out'},$example{'test_data/mime_out.ans'}); +printf "%-40s", "test_data/mime_out"; + &test("$nkf -jM",$example{'test_data/mime_out'},$example{'test_data/mime_out.ans'},$example{'test_data/mime_out.ans.alt'},$example{'test_data/mime_out.ans.alt2'},$example{'test_data/mime_out.ans.alt3'}); +# test_data/mime_out3 + +$example{'test_data/mime_out3'} = "\x82\xD9\x82\xB0 A"; + +$example{'test_data/mime_out3.ans'} = "=?ISO-2022-JP?B?GyRCJFskMhsoQg==?= A"; + +printf "%-40s", "test_data/mime_out3"; + &test("$nkf -jSM",$example{'test_data/mime_out3'},$example{'test_data/mime_out3.ans'}); # test_data/multi-line $example{'test_data/multi-line'} = unpack('u',<<'eofeof'); @@ -500,8 +749,28 @@ MHJ3(I,&DY:2FI,>DP:3GI/.DKJ3LI.NDJZ3BI+>D\Z3*I*2AHPJDLZ2SI,_# 8NZ2DN=2AHP`*I+.DLZ3/P[NDI+G4H:,* eofeof -print "test_data/multi-line "; +printf "%-40s", "test_data/multi-line"; &test("$nkf -e",$example{'test_data/multi-line'},$example{'test_data/multi-line.ans'}); +# test_data/-Z4 + +$example{'test_data/-Z4'} = unpack('u',<<'eofeof'); +MH:.AUJ'7H:*AIJ&\H:NAK*6AI:*EHZ6DI:6EIJ6GI:BEJ:6JI:NEK*6MI:ZE +MKZ6PI;&ELJ6SI;2EM:6VI;>EN*6YI;JENZ6\I;VEOJ6_I<"EP:7"I<.EQ*7% +MI<:EQZ7(I&EXJ7CI>2EY:7FI>>EZ*7II>JEZZ7LI>VE[Z7RI?.E]``` +eofeof + +$example{'test_data/-Z4.ans'} = unpack('u',<<'eofeof'); +MCJ&.HHZCCJ2.I8ZPCMZ.WXZGCK&.J(ZRCJF.LXZJCK2.JXZUCK:.MH[>CK>. +MMX[>CKB.N([>CKF.N8[>CKJ.NH[>CKN.NX[>CKR.O([>CKV.O8[>CKZ.OH[> +MCK^.OX[>CL".P([>CL&.P8[>CJ^.PH["CMZ.PX[#CMZ.Q([$CMZ.Q8[&CL>. +MR([)CLJ.RH[>CLJ.WX[+CLN.WH[+CM^.S([,CMZ.S([?CLV.S8[>CLV.WX[. +MCLZ.WH[.CM^.SX[0CM&.TH[3CJR.U(ZMCM6.KH[6CM>.V([9CMJ.VX[&RA""@`` eofeof -print "test_data/non-strict-mime "; - &test("$nkf -jTmN",$example{'test_data/non-strict-mime'},$example{'test_data/non-strict-mime.ans'}); +printf "%-40s", "test_data/non-strict-mime"; + &test("$nkf -jmN",$example{'test_data/non-strict-mime'},$example{'test_data/non-strict-mime.ans'}); # test_data/q-encode-softrap $example{'test_data/q-encode-softrap'} = unpack('u',<<'eofeof'); @@ -563,8 +832,8 @@ $example{'test_data/q-encode-softrap.ans'} = unpack('u',<<'eofeof'); >&R1"-$$[>B4S(3PE221.&RA""ALD0DI1-#D;*$(* eofeof -print "test_data/q-encode-softrap "; - &test("$nkf -jTmQ",$example{'test_data/q-encode-softrap'},$example{'test_data/q-encode-softrap.ans'}); +printf "%-40s", "test_data/q-encode-softrap"; + &test("$nkf -jmQ",$example{'test_data/q-encode-softrap'},$example{'test_data/q-encode-softrap.ans'}); # test_data/rot13 $example{'test_data/rot13'} = unpack('u',<<'eofeof'); @@ -583,7 +852,7 @@ M4U-U4U-3>5-34SE355-V4%%?>6%K4WU3.5-54WIY(F-H4V13/5,O4VI31%!2 A&RA""@HE(')P=6(@)W5B='(G('P@87AS("UE"G5B='(* eofeof -print "test_data/rot13 "; +printf "%-40s", "test_data/rot13"; &test("$nkf -jr",$example{'test_data/rot13'},$example{'test_data/rot13.ans'}); # test_data/slash @@ -595,8 +864,8 @@ $example{'test_data/slash.ans'} = unpack('u',<<'eofeof'); 7("`]/U8\5"U5.5=%2RTK.U4J..# +MH>.#O..#J^BAJ.FAC"#HJ:;IJ)/GE*CC@Z'C@[SC@ZOHH:CIH8P@Z*FFZ:B3 +3YY2HXX.AXX.\XX.KZ*&HZ:&,"@`` +eofeof + +$example{'test_data/bugs10904.ans'} = unpack('u',<<'eofeof'); +M4W5B:F5C=#H@/3])4T\M,C`R,BU*4#]"/T=Y4D-/,C0T33`Q4DI716A00U9R +?4U0Q0V%H5)#2E=%:%!#5G)35#%#86AS;U%I06)*14DW +?8FIG>E1616Q94T4X2E=T2E!52G%'>6A#24$]/3\]"@`` +M(#T_25-/+3(P,C(M2E`_0C]'>5)#3S(T-$TP,5)*5T5H4$-6