nkf_test.pl

NKF是网络上写新闻或邮件的时候,汉字的编码转换程序

#!/usr/local/bin/perl
#
# nkf test program for nkf 1.7
#    Shinji KONO <[e-mail]>
# Sun Aug 18 12:25:40 JST 1996
# Sun Nov  8 00:16:06 JST 1998
#
# This is useful when you add new patch on nkf.
# Since this test is too strict, faileurs may not mean
# wrong conversion. 
#
# nkf 1.5 differs on MIME decoding
# nkf 1.4 passes Basic Conversion tests
# nkf PDS version passes Basic Conversion tests  using "nkf -iB -oB "
#

$nkf = "./nkf";
# $nkf = "doscmd nkf.exe";

# If you want to see the testing process, set next flag.

$detail_all = 0;

# &library_test0();
sub test  {
#    use NKF;
#    &library_test(@_);
    &command_test(@_);
}

sub command_test {
    local ($nkf,$in,$ans,$detail) = @_;
    local ($result);

    $result = '';
    open(OUT,"> nkf.in");
    print OUT $in;
    close(OUT);
    system("$nkf <nkf.in >nkf.out");   # easy
    open(IN,"< nkf.out");
    print STDERR "\nINPUT:\n",$in  if ($detail || $detail_all);
    print STDERR "\nEXPECT:\n", $ans if ($detail || $detail_all);
    while(<IN>) {
	$result .= $_;
    }
    print STDERR "\nGOT:\n",$result if ($detail || $detail_all);
    print (($result eq $ans)?"Ok\n":"Fail\n");
    return $result;
}

sub library_test {
    local ($nkf,$in,$ans,$detail) = @_;
    local ($result);


    $nkf =~ s/^[^_]* //;
    print "\nINPUT:\n",$in  if ($detail || $detail_all);
    print "\nEXPECT:\n", $ans if ($detail || $detail_all);
    $result = NKF::nkf($nkf,$in);
    print "\nGOT:\n",$result if ($detail || $detail_all);

    print (($result eq $ans)?"Ok\n":"Fail\n");
    return $result;
}

sub librrary_test0 {
BEGIN { $| = 1; print "1..3\n"; }
END {print "not ok 1\n" unless $loaded;}
    use NKF;
    $loaded = 1;
    print "ok 1\n";

    ######################### End of black magic.

    # Insert your test code below (better if it prints "ok 13"
    # (correspondingly "not ok 13") depending on the success of chunk 13
    # of the test code):

    if ( nkf("-m","-e","1") ) {
	print "ok 2\n";
    } else {
	print "no 2\n";
    }

    $output = nkf("\033\$@#1#3#2%B");
    if ($output) {
    # print $output,"\n";
	print "ok 3\n";
    } else {
	print "no 3\n";
    }

    $output = nkf("-Zj","\033\$@#1#3#2%B");
    if ($output) {
    # print $output,"\n";
	print "ok 4\n";
    } else {
	print "no 4\n";
    }

    # conversion makes strings longer
    $output = "\244\306 " x 1024;
	$old =  length($output);
    $output = nkf("-j",$output);
    if ($output) {
    # print $output,"\n";
	print "ok 5 $old ",length($output),"\n";
    } else {
	print "no 5\n";
    }


}

# Basic Conversion
print "\nBasic Conversion test\n\n";

# I gave up simple literal quote because there are big difference
# on perl4 and perl5 on literal quote. Of course we cannot use
# jperl.

$example{'jis'} = unpack('u',<<'eofeof');
M1FER<W0@4W1A9V4@&R1"(3DQ(3%^2R%+?D]3&RA"(%-E8V]N9"!3=&%G92`;
M)$)0)TU:&RA"($AI<F%G86YA(!LD0B0B)"0D)B0H)"HD;R1R)',;*$(*2V%T
M86MA;F$@&R1")2(E)"4F)2@E*B5O)7(E<QLH0B!+:6=O=2`;)$(A)B%G(S`C
/029!)E@G(B=!*$`;*$(*
eofeof

$example{'sjis'} = unpack('u',<<'eofeof');
M1FER<W0@4W1A9V4@@5B)0(F>ED"6GIAR(%-E8V]N9"!3=&%G92"8I9=Y($AI
M<F%G86YA((*@@J*"I(*F@JB"[8+P@O$*2V%T86MA;F$@@T&#0X-%@T>#28./
>@Y*#DR!+:6=O=2"!18&'@D^"8(._@]:$081@A+X*
eofeof

$example{'euc'} = unpack('u',<<'eofeof');
M1FER<W0@4W1A9V4@H;FQH;'^RZ'+_L_3(%-E8V]N9"!3=&%G92#0I\W:($AI
M<F%G86YA(*2BI*2DIJ2HI*JD[Z3RI/,*2V%T86MA;F$@I:*EI*6FI:BEJJ7O
>I?*E\R!+:6=O=2"AIJ'GH["CP:;!IMBGHJ?!J,`*
eofeof


# From JIS

print "JIS  to JIS ... ";&test("$nkf   ",$example{'jis'},$example{'jis'});
print "JIS  to SJIS... ";&test("$nkf -s",$example{'jis'},$example{'sjis'});
print "JIS  to EUC ... ";&test("$nkf -e",$example{'jis'},$example{'euc'});

# From SJIS

print "SJIS to JIS ... ";&test("$nkf -j",$example{'sjis'},$example{'jis'});
print "SJIS to SJIS... ";&test("$nkf -s",$example{'sjis'},$example{'sjis'});
print "SJIS to EUC ... ";&test("$nkf -e",$example{'sjis'},$example{'euc'});

# From EUC

print "EUC  to JIS ... ";&test("$nkf   ",$example{'euc'},$example{'jis'});
print "EUC  to SJIS... ";&test("$nkf -s",$example{'euc'},$example{'sjis'});
print "EUC  to EUC ... ";&test("$nkf -e",$example{'euc'},$example{'euc'});

# Ambigous Case

$example{'amb'} = unpack('u',<<'eofeof');
MI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&E
MPK"QI<*PL:7"L+&EPK"QI<(*I<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*P
ML:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<(*I<*PL:7"L+&E
MPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"
ML+&EPK"QI<(*I<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"Q
MI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<(*I<*PL:7"L+&EPK"QI<*PL:7"
ML+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<*PL:7"L+&EPK"QI<(*
eofeof

$example{'amb.euc'} = unpack('u',<<'eofeof');
M&R1")4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25"
M,#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*&R1")4(P,25",#$E0C`Q)4(P,25"
M,#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;
M*$(*&R1")4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P
M,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*&R1")4(P,25",#$E0C`Q)4(P
M,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q
M)4(;*$(*&R1")4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q
>)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*
eofeof

$example{'amb.sjis'} = unpack('u',<<'eofeof');
M&RA))4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25"
M,#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*&RA))4(P,25",#$E0C`Q)4(P,25"
M,#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;
M*$(*&RA))4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P
M,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*&RA))4(P,25",#$E0C`Q)4(P
M,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q
M)4(;*$(*&RA))4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q
>)4(P,25",#$E0C`Q)4(P,25",#$E0C`Q)4(;*$(*
eofeof

print "Ambiguous Case. ";
    &test("$nkf",$example{'amb'},$example{'amb.euc'});

# Input assumption

print "SJIS  Input assumption ";
    &test("$nkf -Sx",$example{'amb'},$example{'amb.sjis'});

# X0201 $B2>L>(B
# X0201->X0208 conversion
# X0208 aphabet -> ASCII
# X0201 $BAj8_JQ49(B

print "\nX0201 test\n\n";

$example{'x0201.sjis'} = unpack('u',<<'eofeof');
MD5.*<(-*@TR#3H-0@U*#2X--@T^#48-3"I%3B7""8()A@F*"8X)D@F6"9H*!
M@H*"@X*$@H6"AH*'"I%3BTR-AH%)@9>!E(&0@9.!3X&5@9:!:8%J@7R!>X&!
M@6V!;H%O@7"!CPJ4O(IPMK>X/;FZMMZWWKC>N=ZZWH+&"I2\BG#*W\O?S-_-
MW\[?M]^QW@K*W\O?S`IH86YK86MU(,K?R]_,I`K*W\O?S-VA"I2\BG""S(SC
!"@!"
eofeof

$example{'x0201.euc'} = unpack('u',<<'eofeof');
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#N.4*
eofeof

$example{'x0201.jis'} = unpack('u',<<'eofeof');
M&R1"030S424K)2TE+R4Q)3,E+"4N)3`E,B4T&RA""ALD0D$T,5$C02-"(T,C
M1"-%(T8C1R-A(V(C8R-D(V4C9B-G&RA""ALD0D$T-2TY9B$J(7<A="%P(7,A
M,"%U(78A2B%+(5TA7"%A(4XA3R%0(5$A;QLH0@H;)$)(/C-1&RA)-C<X&RA"
M/1LH23DZ-EXW7CA>.5XZ7ALD0B1(&RA""ALD0D@^,U$;*$E*7TM?3%]-7TY?
M-U\Q7ALH0@H;*$E*7TM?3!LH0@IH86YK86MU(!LH24I?2U],)!LH0@H;*$E*
97TM?3%TA&RA""ALD0D@^,U$D3CAE&RA""@``
eofeof

$example{'x0201.sosi'} = unpack('u',<<'eofeof');
M&R1"030S424K)2TE+R4Q)3,E+"4N)3`E,B4T&RA*"ALD0D$T,5$C02-"(T,C
M1"-%(T8C1R-A(V(C8R-D(V4C9B-G&RA*"ALD0D$T-2TY9B$J(7<A="%P(7,A
M,"%U(78A2B%+(5TA7"%A(4XA3R%0(5$A;QLH2@H;)$)(/C-1&RA*#C8W.`\;
M*$H]#CDZ-EXW7CA>.5XZ7@\;)$(D2!LH2@H;)$)(/C-1&RA*#DI?2U],7TU?
M3E\W7S%>#PH.2E]+7TP/&RA*"FAA;FMA:W4@#DI?2U],)`\;*$H*#DI?2U],
672$/&RA*"ALD0D@^,U$D3CAE&RA""@``
eofeof

$example{'x0201.x0208'} = unpack('u',<<'eofeof');
M&R1"030S424K)2TE+R4Q)3,E+"4N)3`E,B4T&RA""ALD0D$T,5$;*$)!0D-$
M149'86)C9&5F9PH;)$)!-#4M.68;*$(A0",D)5XF*B@I+2L]6UU[?1LD0B%O
M&RA""ALD0D@^,U$E*R4M)2\;*$(]&R1")3$E,R4L)2XE,"4R)30D2!LH0@H;
M)$)(/C-1)5$E5"57)5HE724M(2PE(B$K&RA""ALD0B51)50E51LH0@IH86YK
M86MU(!LD0B51)50E52$B&RA""ALD0B51)50E525S(2,;*$(*&R1"2#XS421.
&.&4;*$(*
eofeof

# -X is necessary to allow X0201 in SJIS
# -Z convert X0208 alphabet to ASCII
print "X0201 conversion: SJIS ";
    &test("$nkf -XZ",$example{'x0201.sjis'},$example{'x0201.x0208'});
print "X0201 conversion: JIS  ";
    &test("$nkf -Z",$example{'x0201.jis'},$example{'x0201.x0208'});
print "X0201 conversion:SI/SO ";
    &test("$nkf -Z",$example{'x0201.sosi'},$example{'x0201.x0208'});
print "X0201 conversion: EUC  ";
    &test("$nkf -Z",$example{'x0201.euc'},$example{'x0201.x0208'});
# -x means X0201 output
print "X0201 output: SJIS     ";
    &test("$nkf -xs",$example{'x0201.euc'},$example{'x0201.sjis'});
print "X0201 output: JIS      ";
    &test("$nkf -xj",$example{'x0201.sjis'},$example{'x0201.jis'});
print "X0201 output: EUC      ";
    &test("$nkf -xe",$example{'x0201.jis'},$example{'x0201.euc'});

# MIME decode

print "\nMIME test\n\n";

# MIME ISO-2022-JP

$example{'mime.iso2022'} = unpack('u',<<'eofeof');
M/3])4T\M,C`R,BU*4#]"/T=Y4D%.144W96E23TI566Q/4U9)1WEH2S\]"CT_
M:7-O+3(P,C(M2E`_0C]'>5)!3D5%-V5I4D]*55EL3U-624=Y:$L_/0H]/VES
M;RTR,#(R+4I0/U$_/3%")$(D1B11/3%"*$)?96YD/ST*&R1`)#TD)B0K)$H;
M*$H@/3])4T\M,C`R,BU*4#]"/T=Y4D%.144W96E23U!Y:S=D:'-O4V<]/3\]
M(&5N9"!O9B!L:6YE"CT_25-/+3(P,C(M2E`_0C]'>5)!3D5%-V5I4D]0>6LW
M9&AS;U-G/3T_/2`]/TE33RTR,#(R+4I0/T(_1WE204Y%13=E:5)/4'EK-V1H
M<V]39ST]/ST*0G)O:V5N(&-A<V4*/3])4T\M,C`R,BU*4#]"/T=Y4D%.144W
M96E23U!Y:S=D"FAS;U-G/3T_/2`]/TE33RTR,`HR,BU*4#]"/T=Y4D%.144W
M96E23U!Y:S=D:'-O4V<]/3\]"CT_25-/+3(P,C(M2E`_0C]'>5)!3D5%-V5I
44D]*55EL3QM;2U-624=Y:$L_/0H_
eofeof

$example{'mime.ans.strict'} = unpack('u',<<'eofeof');
M&R1"-$$[>B1.)48E.25(&RA""ALD0C1!.WHD3B5&)3DE2!LH0@H;)$(D1B11
M&RA"(&5N9`H;)$(D/20F)"LD2ALH0B`;)$(T03MZ)$X_*3MV&RA"96YD(&]F
M(&QI;F4*&R1"-$$[>B1./RD[=C1!.WHD3C\I.W8;*$(*0G)O:V5N(&-A<V4*
M/3])4T\M,C`R,BU*4#]"/T=Y4D%.144W96E23U!Y:S=D"FAS;U-G/3T_/2`]
M/TE33RTR,`HR,BU*4#]"/T=Y4D%.144W96E23U!Y:S=D:'-O4V<]/3\]"CT_
L25-/+3(P,C(M2E`_0C]'>5)!3D5%-V5I4D]*55EL3QM;2U-624=Y:$L_/0H_
eofeof

$example{'mime.unbuf.strict'} = unpack('u',<<'eofeof');
M&R1"-$$[>B1.)48E.25(&RA""ALD0C1!.WHD3B5&)3DE2!LH0@H;)$(D1B11
M&RA"(&5N9`H;)$(D/20F)"LD2ALH0B`;)$(T03MZ)$X_*3MV&RA"96YD(&]F
M(&QI;F4*&R1"-$$[>B1./RD[=C1!.WHD3C\I.W8;*$(*0G)O:V5N(&-A<V4*
M&R1"-$$[>B1./RD;*$)H<V]39ST]/ST@/3])4T\M,C`*,C(M2E`_0C]'>5)!
M3D5%-V5I4D]0>6LW9&AS;U-G/3T_/0H;)$(T03MZ)$XE1ALH0EM+4U9)1WEH
$2S\]"F5I
eofeof

$example{'mime.ans'} = unpack('u',<<'eofeof');
M&R1"-$$[>B1.)48E.25(&RA""ALD0C1!.WHD3B5&)3DE2!LH0@H;)$(D1B11
M&RA"(&5N9`H;)$(D/20F)"LD2ALH0B`;)$(T03MZ)$X_*3MV&RA"96YD(&]F
M(&QI;F4*&R1"-$$[>B1./RD[=C1!.WHD3C\I.W8;*$(*0G)O:V5N(&-A<V4*
M&R1"-$$[>B1./RD;*$)H<V]39ST]/ST@&R1"-$$[>B1./RD[=ALH0@H;)$(T
603MZ)$XE1ALH0EM+4U9)1WEH2S\]"@`*
eofeof

$example{'mime.unbuf'} = unpack('u',<<'eofeof');
M&R1"-$$[>B1.)48E.25(&RA""ALD0C1!.WHD3B5&)3DE2!LH0@H;)$(D1B11
M&RA"(&5N9`H;)$(D/20F)"LD2ALH0B`;)$(T03MZ)$X_*3MV&RA"96YD(&]F
M(&QI;F4*&R1"-$$[>B1./RD[=C1!.WHD3C\I.W8;*$(*0G)O:V5N(&-A<V4*
M&R1"-$$[>B1./RD;*$)H<V]39ST]/ST@&R1"-$$[>B1./RD[=ALH0@H;)$(T
603MZ)$XE1ALH0EM+4U9)1WEH2S\]"@`*
eofeof

$example{'mime.base64'} = unpack('u',<<'eofeof');
M9W-M5"])3&YG<FU#>$I+-&=Q=4,S24LS9W%Q0E%:3TUI-39,,S0Q-&=S5T)1
M43!+9VUA1%9O3T@*9S)+1%1O3'=K8C)1;$E+;V=Q2T-X24MG9W5M0W%*3EEG
<<T=#>$E+9V=U;4,X64Q&9W)70S592VMG<6U""F=Q
eofeof

$example{'mime.base64.ans'} = unpack('u',<<'eofeof');
M&R1")$M&?B1I)#LD1D0Z)"TD7B0Y)"PA(D5L-7XV83E9)$<A(ALH0@T*&R1"
M(T<E-R5G)4,E+R1R0C\_="0J)"0D1B0B)&LD*D4Y)$,D1B0B)&LD<R1')#<D
(9R0F)"L;*$(E
eofeof

print "Next test is expeced to Fail.\n";
print "MIME decode (strict)   ";
    $tmp = &test("$nkf -m",$example{'mime.iso2022'},$example{'mime.ans.strict'});
print "MIME decode (nonstrict)";
    $tmp = &test("$nkf -m",$example{'mime.iso2022'},$example{'mime.ans'});
#    open(OUT,">tmp1");print OUT pack('u',$tmp);close(OUT);
# unbuf mode implies more pessimistic decode
print "MIME decode (unbuf)    ";
    &test("$nkf -mu",$example{'mime.iso2022'},$example{'mime.unbuf'});
print "MIME decode (base64)   ";
    &test("$nkf -mB",$example{'mime.base64'},$example{'mime.base64.ans'});

# MIME ISO-8859-1

$example{'mime.is8859'} = unpack('u',<<'eofeof');
M/3])4T\M.#@U.2TQ/U$_*CU#-V%V83\_/2`*4&5E<B!4]G)N9W)E;@I,87-S
M92!(:6QL97+X92!0971E<G-E;B`@7"`B36EN(&MA97!H97-T(&AA<B!F86%E
M="!E="!F;V5L(2(*06%R:'5S(%5N:79E<G-I='DL($1%3DU!4DL@(%P@(DUI
<;B!KYG!H97-T(&AA<B!FY65T(&5T(&;X;"$B"@!K
eofeof

$example{'mime.is8859.ans'} = unpack('u',<<'eofeof');
M*L=A=F$_(`I0965R(%3V<FYG<F5N"DQA<W-E($AI;&QE<OAE(%!E=&5R<V5N
M("!<(")-:6X@:V%E<&AE<W0@:&%R(&9A865T(&5T(&9O96PA(@I!87)H=7,@
M56YI=F5R<VET>2P@1$5.34%22R`@7"`B36EN(&OF<&AE<W0@:&%R(&;E970@
)970@9OAL(2(*
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'});

# test for -f is not so simple.

# end