OSDN Git Service

バイナリファイルを入力したときに落ちる問題の修正
[nkf/nkf.git] / nkf_test.pl
index 6a6ef94..da84c1b 100644 (file)
@@ -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 1.9 2004/12/01 01:59:28 naruse Exp $
+#
 #    Shinji KONO <kono@ie.u-ryukyu.ac.jp>
 # Sun Aug 18 12:25:40 JST 1996
 # Sun Nov  8 00:16:06 JST 1998
@@ -40,13 +43,47 @@ 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
 
-$example{'utf'} = unpack('u',<<'eofeof');
+$example{'utf8'} = unpack('u',<<'eofeof');
+M[[N_1FER<W0@4W1A9V4@XX"%Z9FBY;^<YK.5YKJ`Z(65(%-E8V]N9"!3=&%G
+M92#DN+SI@:4@2&ER86=A;F$@XX&"XX&$XX&&XX&(XX&*XX*/XX*2XX*3"DMA
+M=&%K86YA(.."HN."I.."IN."J.."JN.#K^.#LN.#LR!+:6=O=2#C@[OBB)[O
+1O)#OO*'.L<^)T)'0K^*5@@H`
+eofeof
+
+$example{'utf8N'} = unpack('u',<<'eofeof');
 M1FER<W0@4W1A9V4@XX"%Z9FBY;^<YK.5YKJ`Z(65(%-E8V]N9"!3=&%G92#D
 MN+SI@:4@2&ER86=A;F$@XX&"XX&$XX&&XX&(XX&*XX*/XX*2XX*3"DMA=&%K
 M86YA(.."HN."I.."IN."J.."JN.#K^.#LN.#LR!+:6=O=2#C@[OBB)[OO)#O
 .O*'.L<^)T)'0K^*5@@H`
 eofeof
 
+$example{'u16L'} = unpack('u',<<'eofeof');
+M__Y&`&D`<@!S`'0`(`!3`'0`80!G`&4`(``%,&*6W%_5;(!N58$@`%,`90!C
+M`&\`;@!D`"``4P!T`&$`9P!E`"``/$YED"``2`!I`'(`80!G`&$`;@!A`"``
+M0C!$,$8P2#!*,(\PDC"3,`H`2P!A`'0`80!K`&$`;@!A`"``HC"D,*8PJ#"J
+I,.\P\C#S,"``2P!I`&<`;P!U`"``^S`>(A#_(?^Q`\D#$00O!$(E"@``
+eofeof
+
+$example{'u16L0'} = unpack('u',<<'eofeof');
+M1@!I`'(`<P!T`"``4P!T`&$`9P!E`"``!3!BEMQ?U6R`;E6!(`!3`&4`8P!O
+M`&X`9``@`%,`=`!A`&<`90`@`#Q.99`@`$@`:0!R`&$`9P!A`&X`80`@`$(P
+M1#!&,$@P2C"/,)(PDS`*`$L`80!T`&$`:P!A`&X`80`@`*(PI#"F,*@PJC#O
+G,/(P\S`@`$L`:0!G`&\`=0`@`/LP'B(0_R'_L0/)`Q$$+P1")0H`
+eofeof
+
+$example{'u16B'} = unpack('u',<<'eofeof');
+M_O\`1@!I`'(`<P!T`"``4P!T`&$`9P!E`"`P!99B7]QLU6Z`@54`(`!3`&4`
+M8P!O`&X`9``@`%,`=`!A`&<`90`@3CR090`@`$@`:0!R`&$`9P!A`&X`80`@
+M,$(P1#!&,$@P2C"/,)(PDP`*`$L`80!T`&$`:P!A`&X`80`@,*(PI#"F,*@P
+IJC#O,/(P\P`@`$L`:0!G`&\`=0`@,/LB'O\0_R$#L0/)!!$$+R5"``H`
+eofeof
+
+$example{'u16B0'} = unpack('u',<<'eofeof');
+M`$8`:0!R`',`=``@`%,`=`!A`&<`90`@,`668E_<;-5N@(%5`"``4P!E`&,`
+M;P!N`&0`(`!3`'0`80!G`&4`($X\D&4`(`!(`&D`<@!A`&<`80!N`&$`(#!"
+M,$0P1C!(,$HPCS"2,),`"@!+`&$`=`!A`&L`80!N`&$`(#"B,*0PIC"H,*HP
+G[S#R,/,`(`!+`&D`9P!O`'4`(##[(A[_$/\A`[$#R001!"\E0@`*
+eofeof
 
 $example{'jis1'} = unpack('u',<<'eofeof');
 M&R1";3%Q<$$L&RA""ALD0F4Z3F\;*$(*&R1"<FT;*$()&R1"/F5.3D]+&RA"
@@ -86,28 +123,40 @@ eofeof
 print "JIS  to JIS ... ";&test("$nkf -j",$example{'jis'},$example{'jis'});
 print "JIS  to SJIS... ";&test("$nkf -s",$example{'jis'},$example{'sjis'});
 print "JIS  to EUC ... ";&test("$nkf -e",$example{'jis'},$example{'euc'});
-print "JIS  to UTF8... ";&test("$nkf -w",$example{'jis'},$example{'utf'});
+print "JIS  to UTF8... ";&test("$nkf -w",$example{'jis'},$example{'utf8N'});
+print "JIS  to U16L... ";&test("$nkf -w16L",$example{'jis'},$example{'u16L'});
+print "JIS  to U16B... ";&test("$nkf -w16B",$example{'jis'},$example{'u16B'});
 
 # 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'},$example{'euc'});
-print "SJIS to UTF8... ";&test("$nkf -w",$example{'sjis'},$example{'utf'});
+print "SJIS to UTF8... ";&test("$nkf -w",$example{'sjis'},$example{'utf8N'});
+print "SJIS to U16L... ";&test("$nkf -w16L",$example{'sjis'},$example{'u16L'});
+print "SJIS to U16B... ";&test("$nkf -w16B",$example{'sjis'},$example{'u16B'});
 
 # From EUC
 
 print "EUC  to JIS ... ";&test("$nkf -j",$example{'euc'},$example{'jis'});
 print "EUC  to SJIS... ";&test("$nkf -s",$example{'euc'},$example{'sjis'});
 print "EUC  to EUC ... ";&test("$nkf -e",$example{'euc'},$example{'euc'});
-print "EUC  to UTF8... ";&test("$nkf -w",$example{'euc'},$example{'utf'});
+print "EUC  to UTF8... ";&test("$nkf -w",$example{'euc'},$example{'utf8N'});
+print "EUC  to U16L... ";&test("$nkf -w16L",$example{'euc'},$example{'u16L'});
+print "EUC  to U16B... ";&test("$nkf -w16B",$example{'euc'},$example{'u16B'});
 
 # From UTF8
 
-print "UTF8 to JIS ... ";&test("$nkf -j",$example{'utf'},$example{'jis'});
-print "UTF8 to SJIS... ";&test("$nkf -s",$example{'utf'},$example{'sjis'});
-print "UTF8 to EUC ... ";&test("$nkf -e",$example{'utf'},$example{'euc'});
-print "UTF8 to UTF8... ";&test("$nkf -w",$example{'utf'},$example{'utf'});
+print "UTF8 to JIS ... ";&test("$nkf -j",      $example{'utf8N'},$example{'jis'});
+print "UTF8 to SJIS... ";&test("$nkf -s",      $example{'utf8N'},$example{'sjis'});
+print "UTF8 to EUC ... ";&test("$nkf -e",      $example{'utf8N'},$example{'euc'});
+print "UTF8 to UTF8N.. ";&test("$nkf -w",      $example{'utf8N'},$example{'utf8N'});
+print "UTF8 to UTF8... ";&test("$nkf -w8",     $example{'utf8N'},$example{'utf8'});
+print "UTF8 to UTF8N.. ";&test("$nkf -w80",    $example{'utf8N'},$example{'utf8N'});
+print "UTF8 to U16L... ";&test("$nkf -w16L",   $example{'utf8N'},$example{'u16L'});
+print "UTF8 to U16L0.. ";&test("$nkf -w16L0",  $example{'utf8N'},$example{'u16L0'});
+print "UTF8 to U16B... ";&test("$nkf -w16B",   $example{'utf8N'},$example{'u16B'});
+print "UTF8 to U16B0.. ";&test("$nkf -w16B0",  $example{'utf8N'},$example{'u16B0'});
 
 
 
@@ -191,6 +240,62 @@ print "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
+
+print "test_data/cp932    ";
+    &test("$nkf -eS",$example{'test_data/cp932'},$example{'test_data/cp932.ans'});
+
+# test_data/cp932inv
+print "test_data/cp932inv    ";
+    &test("$nkf -sE --cp932inv",$example{'test_data/cp932.ans'},$example{'test_data/cp932'});
+
+# test_data/no-cp932inv
+
+$example{'test_data/no-cp932inv.ans'} = unpack('u',<<'eofeof');
+%[N\@[NP`
+eofeof
+
+print "test_data/no-cp932inv    ";
+    &test("$nkf -sE --no-cp932",$example{'test_data/cp932.ans'},$example{'test_data/no-cp932inv.ans'});
+
+# test_data/irv
+
+# $example{'test_data/irv'} = unpack('u',<<'eofeof');
+# %#B`/(!L`
+# eofeof
+# 
+# $example{'test_data/irv.ans'} = unpack('u',<<'eofeof');
+# %#B`/(!L`
+# eofeof
+# 
+# print "test_data/irv    ";
+#    &test("$nkf -wE",$example{'test_data/irv'},$example{'test_data/irv.ans'});
+
+
+# UCS Mapping Test
+print "\n\nUCS Mapping Test\n";
+
+print "Shift_JIS to UTF-16\n";
+$example{'ms_ucs_map_1_sjis'} = "\x81\x60\x81\x61\x81\x7C\x81\x91\x81\x92\x81\xCA";
+$example{'ms_ucs_map_1_utf16'} = "\x30\x1C\x20\x16\x22\x12\x00\xA2\x00\xA3\x00\xAC";
+$example{'ms_ucs_map_1_utf16_ms'} = "\xFF\x5E\x22\x25\xFF\x0D\xFF\xE0\xFF\xE1\xFF\xE2";
+
+print "Normal UCS Mapping : ";
+    &test("$nkf -w16B0 -S",$example{'ms_ucs_map_1_sjis'},$example{'ms_ucs_map_1_utf16'});
+
+print "Microsoft UCS Mapping : ";
+    &test("$nkf -w16B0 -S --ms-ucs-map",$example{'ms_ucs_map_1_sjis'},$example{'ms_ucs_map_1_utf16_ms'});
+
+print"\n";
+
 # X0201 \e$B2>L>\e(B
 # X0201->X0208 conversion
 # X0208 aphabet -> ASCII
@@ -366,7 +471,8 @@ print "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'},$example{'mime.base64.ans'});
+    &test("$nkf -jmB",$example{'mime.base64'},$example{'mime.base64.ans'});
+#MIME BASE64 must be LF?
 
 # MIME ISO-8859-1
 
@@ -439,7 +545,7 @@ M4B`;)$(D3CE4(2,;*$(*"ALD0B0S)#,D3QLH0B!,1B`;)$(D3CE4(2,;*$(*
 eofeof
 
 print "test_data/long-fold-1    ";
-    &test("$nkf -jTF60",$example{'test_data/long-fold-1'},$example{'test_data/long-fold-1.ans'});
+    &test("$nkf -jF60",$example{'test_data/long-fold-1'},$example{'test_data/long-fold-1.ans'});
 # test_data/long-fold
 
 $example{'test_data/long-fold'} = unpack('u',<<'eofeof');
@@ -457,7 +563,7 @@ M)&\D:R0D)#<A(B1()$$D920F)$<D021G)',D+B1L)&LD*R1B)#<D<QLH0@H;
 eofeof
 
 print "test_data/long-fold    ";
-    &test("$nkf -jTf60",$example{'test_data/long-fold'},$example{'test_data/long-fold.ans'});
+    &test("$nkf -jf60",$example{'test_data/long-fold'},$example{'test_data/long-fold.ans'});
 # test_data/mime_out
 
 $example{'test_data/mime_out'} = unpack('u',<<'eofeof');
@@ -471,20 +577,66 @@ eofeof
 
 $example{'test_data/mime_out.ans'} = unpack('u',<<'eofeof');
 M"BTM+2T*4W5B:F5C=#H@86%A82!A86%A(&%A86$@86%A82!A86%A(&%A86$@
-M86%A82!A86%A(&%A86$*(&%A86$@86%A82!A86%A(&%A86$@86%A80HM+2TM
-M"E-U8FIE8W0Z(#T_25-/+3(P,C(M2E`_0C]'>5)#2D-):TI#46U*0V=K2VE1
-M<DI#,&M,>5%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:$,_/2`*"3T_25-/+3(P,C(M2E`_0C]'>5)#2D9S:UAI4F9*1T%K
-M65-2:4I'46M*0U)M2D-G:V%"<V]19ST]/ST@"BTM+2T*4W5B:F5C=#H@86%A
-M82!A86%A(&%A86$@86%A82!A86%A(&%A86$@86%A82`]/TE33RTR,#(R+4I0
-M/T(_1WE20TI#26)+14D]/ST@"@D]/TE33RTR,#(R+4I0/T(_1WE20TI#46M*
-J:5%O2D-O8DM%23T_/2`@86%A80H@86%A82!A86%A(&%A86$*+2TM+0H*
+M86%A82!A86%A(&%A86$@86%A80H@86%A82!A86%A(&%A86$@86%A82!A86%A
+M"BTM+2T*4W5B:F5C=#H@/3])4T\M,C`R,BU*4#]"/T=Y4D-*0TEK2D-1;4I#
+M9VM+:5%R2D,P:TQY47A*1$UK3E-1,T=Y:$,_/0H@/3])4T\M,C`R,BU*4#]"
+M/T=Y4D-*1&MK3WE1.4I$.&M14U)%2D59:U-#4DM*17-K5$-23DI%-&M4>5)3
+M2D95:U=#4F)'>6A#/ST*(#T_25-/+3(P,C(M2E`_0C]'>5)#2D8T:UAY4F=*
+M1T5K66E2:TI#46M::5%O2D=G8DM%23T_/0HM+2TM"E-U8FIE8W0Z(&%A86$@
+M86%A82!A86%A(&%A86$@86%A82!A86%A(&%A86$*(#T_25-/+3(P,C(M2E`_
+M0C]'>5)#2D-):TI#46U*0V=K2VAS;U%G/3T_/2!A86%A(&%A86$@86%A82!A
+086%A"B!A86%A"BTM+2T*"@``
 eofeof
 
 print "test_data/mime_out    ";
     &test("$nkf -jM",$example{'test_data/mime_out'},$example{'test_data/mime_out.ans'});
-# test_data/multi-line
+# test_data/mime_out2
+
+$example{'test_data/mime_out2'} = unpack('u',<<'eofeof');
+M5&AI<R!M96UO(&1E<V-R:6)E<R!S:6UI;&%R('1E8VAN:7%U97,@=&\@86QL
+M;W<@=&AE(&5N8V]D:6YG(&]F(&YO;BU!4T-)22!T97AT(&EN('9A<FEO=7,@
+M<&]R=&EO;G,@;V8@82!21D,@.#(R(%LR72!M97-S86=E(&AE861E<BP@:6X@
+M82!M86YN97(@=VAI8V@@:7,@=6YL:6ME;'D@=&\@8V]N9G5S92!E>&ES=&EN
+M9R!M97-S86=E(&AA;F1L:6YG('-O9G1W87)E+@H*4W5B:F5C=#H@=&5S=#$@
+M=&5S=#(@@L2"MX+&@J<@=&5S=#,@@L2"MX+&@O$@=&5S=#0*"E-U8FIE8W0Z
+M('1E<W0Q("!T97-T,B""Q"""MR""QB""IR!T97-T,R`@@L2"MX+&@O$@('1E
+M<W0T"@I!4T-)22"3^I9[C.H@05-#24D@05-#24D@D_J6>XSJ()/ZEGN,ZB!!
+M4T-)22!!4T-)29/ZEGN,ZB!!4T-)20H*@J`@@J(@@J0@@J8@@J@@@JD@@JL@
+M@JT@@J\@@K$@@K,@@K4@@K<@@KD@@KL@@KT@@K\@@L(@@L0@@L8@@L@@@LD@
+8@LH@@LL@@LP*"@H*"@H*"@H*"@H*"@H*
+eofeof
+
+$example{'test_data/mime_out2.ans'} = unpack('u',<<'eofeof');
+M5&AI<R!M96UO(&1E<V-R:6)E<R!S:6UI;&%R('1E8VAN:7%U97,@=&\@86QL
+M;W<@=&AE(&5N8V]D:6YG(&5N8V]D:6YG"B!O9B!N;VXM05-#24D@=&5X="!I
+M;B!V87)I;W5S('!O<G1I;VYS(&]F(&$@80H@4D9#(#@R,B!;,ET@;65S<V%G
+M92!H96%D97(L(&EN(&$@;6%N;F5R('=H:6-H(&ES('5N;&EK96QY('5N;&EK
+M96QY"B!T;R!C;VYF=7-E(&5X:7-T:6YG(&UE<W-A9V4@:&%N9&QI;F<@<V]F
+M='=A<F4N"@I3=6)J96-T.B!T97-T,2!T97-T,B`]/TE33RTR,#(R+4I0/T(_
+M1WE20TI%66M/4U))2D-K8DM%23T_/2!T97-T,PH@/3])4T\M,C`R,BU*4#]"
+M/T=Y4D-*15EK3U-224I(36)+14D]/ST@=&5S=#0*"E-U8FIE8W0Z('1E<W0Q
+M("!T97-T,B`]/TE33RTR,#(R+4I0/T(_1WE20TI%66)+14EG1WE20TI$:V)+
+M14EG1WE20TI%9V)+14D]/ST*(#T_25-/+3(P,C(M2E`_0C]'>5)#1WEH0TE"
+M<VM1:5%P1WEH0S\]('1E<W0S(`H@/3])4T\M,C`R,BU*4#]"/T=Y4D-*15EK
+M3U-224I(36)+14D]/ST@('1E<W0T"@I!4T-)22`]/TE33RTR,#(R+4I0/T(_
+M1WE20U)N>$Q81&AS1WEH0S\]($%30TE)($%30TE)"B`]/TE33RTR,#(R+4I0
+M/T(_1WE20U)N>$Q81&AS1WEH0TE"<VM1:UHX4S%W-&)"<V]19ST]/ST@05-#
+M24D*(#T_25-/+3(P,C(M2E`_0C]15DY$4U5K8DI%2D=F171C3T=W8DM%23T_
+M/2!!4T-)20H*/3])4T\M,C`R,BU*4#]"/T=Y4D-*0TEB2T5)9T=Y4D-*0U%B
+M2T5)9T=Y4D-*0UEB2T5)9T=Y4D-*0V=B2T5)/3\]"B`]/TE33RTR,#(R+4I0
+M/T(_24)S:U%I47%'>6A#24)S:U%I47)'>6A#24)S:U%I471'>6A#24)S:U%I
+M479'>6A#/ST*(#T_25-/+3(P,C(M2E`_0C])0G-K46E1>$=Y:$-)0G-K46E1
+M>D=Y:$-)0G-K46E1,4=Y:$-)0G-K46E1,T=Y:$,_/0H@/3])4T\M,C`R,BU*
+M4#]"/TE"<VM1:5$U1WEH0TE"<VM1:5$W1WEH0TE"<VM1:5$Y1WEH0TE"<VM1
+M:5$O1WEH0S\]"B`]/TE33RTR,#(R+4I0/T(_24)S:U%I4D)'>6A#24)S:U%I
+M4D5'>6A#24)S:U%I4D='>6A#24)S:U%I4DE'>6A#/ST*(#T_25-/+3(P,C(M
+M2E`_0C])0G-K46E22T=Y:$-)0G-K46E23$=Y:$-)0G-K46E234=Y:$-)0G-K
+M46E23D=Y:$,_/0H@/3])4T\M,C`R,BU*4#]"/TE"<VM1:5)/1WEH0S\]"@H*
+-"@H*"@H*"@H*"@H*"@``
+eofeof
+
+print "test_data/mime_out2    ";
+    &test("$nkf -jM",$example{'test_data/mime_out2'},$example{'test_data/mime_out2.ans'});# test_data/multi-line
 
 $example{'test_data/multi-line'} = unpack('u',<<'eofeof');
 MI,JDK*2DI,JDK*2DI,JDK*'!I*2DKJ3GI*:DK*2BI.JDWJ2WI,:AH@"DLZ3L
@@ -552,7 +704,7 @@ M&R1")$8D)"0_)$`D)"1&)%XD.2$C&RA"#0H-"ALD0CMD)$\[?B$Y)6PE.21+
 eofeof
 
 print "test_data/non-strict-mime    ";
-    &test("$nkf -jTmN",$example{'test_data/non-strict-mime'},$example{'test_data/non-strict-mime.ans'});
+    &test("$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');
@@ -564,7 +716,7 @@ $example{'test_data/q-encode-softrap.ans'} = unpack('u',<<'eofeof');
 eofeof
 
 print "test_data/q-encode-softrap    ";
-    &test("$nkf -jTmQ",$example{'test_data/q-encode-softrap'},$example{'test_data/q-encode-softrap.ans'});
+    &test("$nkf -jmQ",$example{'test_data/q-encode-softrap'},$example{'test_data/q-encode-softrap.ans'});
 # test_data/rot13
 
 $example{'test_data/rot13'} = unpack('u',<<'eofeof');
@@ -634,4 +786,5 @@ eofeof
 print "test_data/z1space-2    ";
     &test("$nkf -e -Z2",$example{'test_data/z1space-2'},$example{'test_data/z1space-2.ans'});
 
+
 # end