diff options
| author | Gurusamy Sarathy <gsar@cpan.org> | 2000-05-28 10:01:48 +0000 |
|---|---|---|
| committer | Gurusamy Sarathy <gsar@cpan.org> | 2000-05-28 10:01:48 +0000 |
| commit | 7c6f5cd212ae61ef572b511249a68eba23258ef2 (patch) | |
| tree | 073edaa98da549b2d1c8f72259c1b0072d4ab40e /lib | |
| parent | 0b931be4791f7d772317758feb118bd4d693a29e (diff) | |
| download | perl-7c6f5cd212ae61ef572b511249a68eba23258ef2.tar.gz | |
revise mktables.PL for bugs and newness in Unicode 3.0
(from James Bence <jbence@amgen.com>)
p4raw-id: //depot/perl@6139
Diffstat (limited to 'lib')
49 files changed, 1892 insertions, 73 deletions
diff --git a/lib/unicode/Is/BidiAL.pl b/lib/unicode/Is/BidiAL.pl new file mode 100644 index 0000000000..e04f2f562d --- /dev/null +++ b/lib/unicode/Is/BidiAL.pl @@ -0,0 +1,25 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +061b +061f +0621 063a +0640 064a +066d +0671 06d5 +06e5 06e6 +06fa 06fe +0700 070d +0710 +0712 072c +0780 07a5 +fb50 fbb1 +fbd3 fd3d +fd50 fd8f +fd92 fdc7 +fdf0 fdfb +fe70 fe72 +fe74 +fe76 fefc +END diff --git a/lib/unicode/Is/BidiBN.pl b/lib/unicode/Is/BidiBN.pl new file mode 100644 index 0000000000..795a4a9f40 --- /dev/null +++ b/lib/unicode/Is/BidiBN.pl @@ -0,0 +1,15 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +0000 0008 +000e 001b +007f 0084 +0086 009f +070f +180b 180e +200b 200d +206a 206f +feff +fff9 fffb +END diff --git a/lib/unicode/Is/BidiLRE.pl b/lib/unicode/Is/BidiLRE.pl new file mode 100644 index 0000000000..ef2a6e462f --- /dev/null +++ b/lib/unicode/Is/BidiLRE.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +202a +END diff --git a/lib/unicode/Is/BidiLRO.pl b/lib/unicode/Is/BidiLRO.pl new file mode 100644 index 0000000000..e9958c4b81 --- /dev/null +++ b/lib/unicode/Is/BidiLRO.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +202d +END diff --git a/lib/unicode/Is/BidiNSM.pl b/lib/unicode/Is/BidiNSM.pl new file mode 100644 index 0000000000..191bc052a9 --- /dev/null +++ b/lib/unicode/Is/BidiNSM.pl @@ -0,0 +1,97 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +0300 034e +0360 0362 +0483 0486 +0488 0489 +0591 05a1 +05a3 05b9 +05bb 05bd +05bf +05c1 05c2 +05c4 +064b 0655 +0670 +06d6 06e4 +06e7 06e8 +06ea 06ed +0711 +0730 074a +07a6 07b0 +0901 0902 +093c +0941 0948 +094d +0951 0954 +0962 0963 +0981 +09bc +09c1 09c4 +09cd +09e2 09e3 +0a02 +0a3c +0a41 0a42 +0a47 0a48 +0a4b 0a4d +0a70 0a71 +0a81 0a82 +0abc +0ac1 0ac5 +0ac7 0ac8 +0acd +0b01 +0b3c +0b3f +0b41 0b43 +0b4d +0b56 +0b82 +0bc0 +0bcd +0c3e 0c40 +0c46 0c48 +0c4a 0c4d +0c55 0c56 +0cbf +0cc6 +0ccc 0ccd +0d41 0d43 +0d4d +0dca +0dd2 0dd4 +0dd6 +0e31 +0e34 0e3a +0e47 0e4e +0eb1 +0eb4 0eb9 +0ebb 0ebc +0ec8 0ecd +0f18 0f19 +0f35 +0f37 +0f39 +0f71 0f7e +0f80 0f84 +0f86 0f87 +0f90 0f97 +0f99 0fbc +0fc6 +102d 1030 +1032 +1036 1037 +1039 +1058 1059 +17b7 17bd +17c6 +17c9 17d3 +18a9 +20d0 20e3 +302a 302f +3099 309a +fb1e +fe20 fe23 +END diff --git a/lib/unicode/Is/BidiPDF.pl b/lib/unicode/Is/BidiPDF.pl new file mode 100644 index 0000000000..4a3eedd564 --- /dev/null +++ b/lib/unicode/Is/BidiPDF.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +202c +END diff --git a/lib/unicode/Is/BidiRLE.pl b/lib/unicode/Is/BidiRLE.pl new file mode 100644 index 0000000000..d789246ddb --- /dev/null +++ b/lib/unicode/Is/BidiRLE.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +202b +END diff --git a/lib/unicode/Is/BidiRLO.pl b/lib/unicode/Is/BidiRLO.pl new file mode 100644 index 0000000000..fcb81acc93 --- /dev/null +++ b/lib/unicode/Is/BidiRLO.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +202e +END diff --git a/lib/unicode/Is/Cf.pl b/lib/unicode/Is/Cf.pl new file mode 100644 index 0000000000..896c3e6cd6 --- /dev/null +++ b/lib/unicode/Is/Cf.pl @@ -0,0 +1,12 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +070f +180b 180e +200c 200f +202a 202e +206a 206f +feff +fff9 fffb +END diff --git a/lib/unicode/Is/Cn.pl b/lib/unicode/Is/Cn.pl index ec287c456a..3c686154c1 100644 --- a/lib/unicode/Is/Cn.pl +++ b/lib/unicode/Is/Cn.pl @@ -2,4 +2,358 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +0220 0221 +0234 024f +02ae 02af +02ef 02ff +034f 035f +0363 0373 +0376 0379 +037b 037d +037f 0383 +038b +038d +03a2 +03cf +03d8 03d9 +03f4 03ff +0487 +048a 048b +04c5 04c6 +04c9 04ca +04cd 04cf +04f6 04f7 +04fa 0530 +0557 0558 +0560 +0588 +058b 0590 +05a2 +05ba +05c5 05cf +05eb 05ef +05f5 060b +060d 061a +061c 061e +0620 +063b 063f +0656 065f +066e 066f +06ee 06ef +06ff +070e +072d 072f +074b 077f +07b1 0900 +0904 +093a 093b +094e 094f +0955 0957 +0971 0980 +0984 +098d 098e +0991 0992 +09a9 +09b1 +09b3 09b5 +09ba 09bb +09bd +09c5 09c6 +09c9 09ca +09ce 09d6 +09d8 09db +09de +09e4 09e5 +09fb 0a01 +0a03 0a04 +0a0b 0a0e +0a11 0a12 +0a29 +0a31 +0a34 +0a37 +0a3a 0a3b +0a3d +0a43 0a46 +0a49 0a4a +0a4e 0a58 +0a5d +0a5f 0a65 +0a75 0a80 +0a84 +0a8c +0a8e +0a92 +0aa9 +0ab1 +0ab4 +0aba 0abb +0ac6 +0aca +0ace 0acf +0ad1 0adf +0ae1 0ae5 +0af0 0b00 +0b04 +0b0d 0b0e +0b11 0b12 +0b29 +0b31 +0b34 0b35 +0b3a 0b3b +0b44 0b46 +0b49 0b4a +0b4e 0b55 +0b58 0b5b +0b5e +0b62 0b65 +0b71 0b81 +0b84 +0b8b 0b8d +0b91 +0b96 0b98 +0b9b +0b9d +0ba0 0ba2 +0ba5 0ba7 +0bab 0bad +0bb6 +0bba 0bbd +0bc3 0bc5 +0bc9 +0bce 0bd6 +0bd8 0be6 +0bf3 0c00 +0c04 +0c0d +0c11 +0c29 +0c34 +0c3a 0c3d +0c45 +0c49 +0c4e 0c54 +0c57 0c5f +0c62 0c65 +0c70 0c81 +0c84 +0c8d +0c91 +0ca9 +0cb4 +0cba 0cbd +0cc5 +0cc9 +0cce 0cd4 +0cd7 0cdd +0cdf +0ce2 0ce5 +0cf0 0d01 +0d04 +0d0d +0d11 +0d29 +0d3a 0d3d +0d44 0d45 +0d49 +0d4e 0d56 +0d58 0d5f +0d62 0d65 +0d70 0d81 +0d84 +0d97 0d99 +0db2 +0dbc +0dbe 0dbf +0dc7 0dc9 +0dcb 0dce +0dd5 +0dd7 +0de0 0df1 +0df5 0e00 +0e3b 0e3e +0e5c 0e80 +0e83 +0e85 0e86 +0e89 +0e8b 0e8c +0e8e 0e93 +0e98 +0ea0 +0ea4 +0ea6 +0ea8 0ea9 +0eac +0eba +0ebe 0ebf +0ec5 +0ec7 +0ece 0ecf +0eda 0edb +0ede 0eff +0f48 +0f6b 0f70 +0f8c 0f8f +0f98 +0fbd +0fcd 0fce +0fd0 0fff +1022 +1028 +102b +1033 1035 +103a 103f +105a 109f +10c6 10cf +10f7 10fa +10fc 10ff +115a 115e +11a3 11a7 +11fa 11ff +1207 +1247 +1249 +124e 124f +1257 +1259 +125e 125f +1287 +1289 +128e 128f +12af +12b1 +12b6 12b7 +12bf +12c1 +12c6 12c7 +12cf +12d7 +12ef +130f +1311 +1316 1317 +131f +1347 +135b 1360 +137d 139f +13f5 1400 +1677 167f +169d 169f +16f1 177f +17dd 17df +17ea 17ff +180f +181a 181f +1878 187f +18aa 1dff +1e9c 1e9f +1efa 1eff +1f16 1f17 +1f1e 1f1f +1f46 1f47 +1f4e 1f4f +1f58 +1f5a +1f5c +1f5e +1f7e 1f7f +1fb5 +1fc5 +1fd4 1fd5 +1fdc +1ff0 1ff1 +1ff5 +1fff +2047 +204e 2069 +2071 2073 +208f 209f +20b0 20cf +20e4 20ff +213b 2152 +2184 218f +21f4 21ff +22f2 22ff +237c +239b 23ff +2427 243f +244b 245f +24eb 24ff +2596 259f +25f8 25ff +2614 2618 +2672 2700 +2705 +270a 270b +2728 +274c +274e +2753 2755 +2757 +275f 2760 +2768 2775 +2795 2797 +27b0 +27bf 27ff +2900 2e7f +2e9a +2ef4 2eff +2fd6 2fef +2ffc 2fff +303b 303d +3040 +3095 3098 +309f 30a0 +30ff 3104 +312d 3130 +318f +31b8 31ff +321d 321f +3244 325f +327c 327e +32b1 32bf +32cc 32cf +32ff +3377 337a +33de 33df +33ff +4db6 4dff +9fa6 9fff +a48d a48f +a4a2 a4a3 +a4b4 +a4c1 +a4c5 +a4c7 abff +d7a4 d7ff +fa2e faff +fb07 fb12 +fb18 fb1c +fb37 +fb3d +fb3f +fb42 +fb45 +fbb2 fbd2 +fd40 fd4f +fd90 fd91 +fdc8 fdef +fdfc fe1f +fe24 fe2f +fe45 fe48 +fe53 +fe67 +fe6c fe6f +fe73 +fe75 +fefd fefe +ff00 +ff5f ff60 +ffbf ffc1 +ffc8 ffc9 +ffd0 ffd1 +ffd8 ffd9 +ffdd ffdf +ffe7 +ffef fff8 END diff --git a/lib/unicode/Is/Cs.pl b/lib/unicode/Is/Cs.pl new file mode 100644 index 0000000000..8888fb5f3c --- /dev/null +++ b/lib/unicode/Is/Cs.pl @@ -0,0 +1,8 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +d800 db7f +db80 dbff +dc00 dfff +END diff --git a/lib/unicode/Is/DCfraction.pl b/lib/unicode/Is/DCfraction.pl new file mode 100644 index 0000000000..fc2dd6755d --- /dev/null +++ b/lib/unicode/Is/DCfraction.pl @@ -0,0 +1,7 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +00bc 00be +2153 215f +END diff --git a/lib/unicode/Is/Graph.pl b/lib/unicode/Is/Graph.pl index 9c94bb722c..156f1711af 100644 --- a/lib/unicode/Is/Graph.pl +++ b/lib/unicode/Is/Graph.pl @@ -265,7 +265,8 @@ return <<'END'; 1fdd 1fef 1ff2 1ff4 1ff6 1ffe -2000 200b +2000 2008 +200b 2010 2029 202f 2046 2048 204d diff --git a/lib/unicode/Is/Me.pl b/lib/unicode/Is/Me.pl new file mode 100644 index 0000000000..00f446d87d --- /dev/null +++ b/lib/unicode/Is/Me.pl @@ -0,0 +1,9 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +0488 0489 +06dd 06de +20dd 20e0 +20e2 20e3 +END diff --git a/lib/unicode/Is/Nl.pl b/lib/unicode/Is/Nl.pl new file mode 100644 index 0000000000..8f1af469bb --- /dev/null +++ b/lib/unicode/Is/Nl.pl @@ -0,0 +1,9 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +2160 2183 +3007 +3021 3029 +3038 303a +END diff --git a/lib/unicode/Is/Pc.pl b/lib/unicode/Is/Pc.pl new file mode 100644 index 0000000000..342efac344 --- /dev/null +++ b/lib/unicode/Is/Pc.pl @@ -0,0 +1,12 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +005f +203f 2040 +30fb +fe33 fe34 +fe4d fe4f +ff3f +ff65 +END diff --git a/lib/unicode/Is/Pf.pl b/lib/unicode/Is/Pf.pl new file mode 100644 index 0000000000..166c64bbb6 --- /dev/null +++ b/lib/unicode/Is/Pf.pl @@ -0,0 +1,9 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +00bb +2019 +201d +203a +END diff --git a/lib/unicode/Is/Pi.pl b/lib/unicode/Is/Pi.pl new file mode 100644 index 0000000000..7f2243d5d8 --- /dev/null +++ b/lib/unicode/Is/Pi.pl @@ -0,0 +1,10 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +00ab +2018 +201b 201c +201f +2039 +END diff --git a/lib/unicode/Is/Punct.pl b/lib/unicode/Is/Punct.pl index 8fd1e8e183..9e8684d6fc 100644 --- a/lib/unicode/Is/Punct.pl +++ b/lib/unicode/Is/Punct.pl @@ -8,45 +8,45 @@ return <<'END'; 003a 003b 003f 0040 005b 005d -005f -007b -007d -00a1 -00ab -00ad -00b7 -00bb -00bf -037e -0387 +005f +007b +007d +00a1 +00ab +00ad +00b7 +00bb +00bf +037e +0387 055a 055f 0589 058a -05be -05c0 -05c3 +05be +05c0 +05c3 05f3 05f4 -060c -061b -061f +060c +061b +061f 066a 066d -06d4 +06d4 0700 070d 0964 0965 -0970 -0df4 -0e4f +0970 +0df4 +0e4f 0e5a 0e5b 0f04 0f12 0f3a 0f3d -0f85 +0f85 104a 104f -10fb +10fb 1361 1368 166d 166e 169b 169c 16eb 16ed 17d4 17da -17dc +17dc 1800 180a 2010 2027 2030 2043 @@ -58,14 +58,14 @@ return <<'END'; 3001 3003 3008 3011 3014 301f -3030 -30fb +3030 +30fb fd3e fd3f fe30 fe44 fe49 fe52 fe54 fe61 -fe63 -fe68 +fe63 +fe68 fe6a fe6b ff01 ff03 ff05 ff0a @@ -73,8 +73,8 @@ ff0c ff0f ff1a ff1b ff1f ff20 ff3b ff3d -ff3f -ff5b -ff5d +ff3f +ff5b +ff5d ff61 ff65 END diff --git a/lib/unicode/Is/Sk.pl b/lib/unicode/Is/Sk.pl new file mode 100644 index 0000000000..b5f6e591a7 --- /dev/null +++ b/lib/unicode/Is/Sk.pl @@ -0,0 +1,27 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +005e +0060 +00a8 +00af +00b4 +00b8 +02b9 02ba +02c2 02cf +02d2 02df +02e5 02ed +0374 0375 +0384 0385 +1fbd +1fbf 1fc1 +1fcd 1fcf +1fdd 1fdf +1fed 1fef +1ffd 1ffe +309b 309c +ff3e +ff40 +ffe3 +END diff --git a/lib/unicode/Is/Space.pl b/lib/unicode/Is/Space.pl index 4121ef49b8..701329ff82 100644 --- a/lib/unicode/Is/Space.pl +++ b/lib/unicode/Is/Space.pl @@ -2,13 +2,13 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; -0009 000a -000c 000d -0020 -00a0 -1680 +0009 000d +0020 +0085 +00a0 +1680 2000 200b 2028 2029 -202f -3000 +202f +3000 END diff --git a/lib/unicode/Is/SylA.pl b/lib/unicode/Is/SylA.pl index ec287c456a..be1107822d 100644 --- a/lib/unicode/Is/SylA.pl +++ b/lib/unicode/Is/SylA.pl @@ -2,4 +2,157 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1203 +120b +1213 +121b +1223 +122b +1233 +123b +1243 +1253 +1263 +126b +1273 +127b +1283 +1293 +129b +12a3 +12ab +12bb +12cb +12d3 +12db +12e3 +12eb +12f3 +12fb +1303 +130b +131b +1323 +132b +1333 +133b +1343 +134b +1353 +13a0 +13a6 13a7 +13ad +13b3 +13b9 +13be 13bf +13c6 +13cc +13d3 13d4 +13dc 13dd +13e3 +13e9 +13ef +140a +1438 +1455 +146a +1472 +1490 +14aa +14c7 +14da +14f4 +1515 +152d +154b +154d +1559 +1566 +156e +1573 +1579 +1583 +1589 +158d +1593 +159a +159e +15a4 +15ac +15b3 +15b7 +15bb +15bf +15c3 +15c9 +15cf +15d5 +15e1 +15e7 +15ed +15f4 +15fa +1600 +1607 +160d +1613 +161b +1621 +1627 +162d +1633 +1639 +163f +1645 +164d +1653 +1659 +1660 +1666 +166c +1675 +30a1 30a2 +30ab 30ac +30b5 30b6 +30bf 30c0 +30ca +30cf 30d1 +30de +30e3 30e4 +30e9 +30ee 30ef +30f5 +30f7 +32d0 +32d5 +32da +32df +32e4 +32e9 +32ee +32f3 +32f6 +32fb +ff67 +ff6c +ff71 +ff76 +ff7b +ff80 +ff85 +ff8a +ff8f +ff94 +ff97 +ff9c +3041 3042 +304b 304c +3055 3056 +305f 3060 +306a +306f 3071 +307e +3083 3084 +3089 +308e 308f END diff --git a/lib/unicode/Is/SylAA.pl b/lib/unicode/Is/SylAA.pl new file mode 100644 index 0000000000..45d6692de7 --- /dev/null +++ b/lib/unicode/Is/SylAA.pl @@ -0,0 +1,25 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +140b +1439 +1456 +1473 +1491 +14ab +14c8 +14db +14f5 +1516 +152e +154c +155a +1567 +157a +1584 +1594 +15a5 +15ad +1676 +END diff --git a/lib/unicode/Is/SylAAI.pl b/lib/unicode/Is/SylAAI.pl new file mode 100644 index 0000000000..a8b03d4c6c --- /dev/null +++ b/lib/unicode/Is/SylAAI.pl @@ -0,0 +1,19 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1402 +1430 +144d +146c +148a +14a4 +14c1 +14d4 +14ee +1527 +1545 +1554 +157e +158e +END diff --git a/lib/unicode/Is/SylAI.pl b/lib/unicode/Is/SylAI.pl new file mode 100644 index 0000000000..b70d793bc6 --- /dev/null +++ b/lib/unicode/Is/SylAI.pl @@ -0,0 +1,7 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +141c +166f 1670 +END diff --git a/lib/unicode/Is/SylC.pl b/lib/unicode/Is/SylC.pl index ec287c456a..e2a1601dd3 100644 --- a/lib/unicode/Is/SylC.pl +++ b/lib/unicode/Is/SylC.pl @@ -2,4 +2,69 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1205 +120d +1215 +121d +1225 +122d +1235 +123d +1245 +1255 +1265 +126d +1275 +127d +1285 +1295 +129d +12a5 +12ad +12bd +12cd +12d5 +12dd +12e5 +12ed +12f5 +12fd +1305 +130d +131d +1325 +132d +1335 +133d +1345 +134d +1355 +13c0 +13cd +141d +142b 142e +1449 144b +1466 +1483 +1485 1488 +14a1 +14bb 14bf +14d0 14d2 +14ea 14ec +1505 1506 +1508 150b +1525 +153e 1540 +1550 1552 +155d +156a +156f +157b 157d +1585 +1595 1596 +159f +15a6 +15ae 15af +30f3 +ff9d END diff --git a/lib/unicode/Is/SylE.pl b/lib/unicode/Is/SylE.pl index ec287c456a..b3c3e60437 100644 --- a/lib/unicode/Is/SylE.pl +++ b/lib/unicode/Is/SylE.pl @@ -2,4 +2,146 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1204 +120c +1214 +121c +1224 +122c +1234 +123c +1244 +1254 +1264 +126c +1274 +127c +1284 +1294 +129c +12a4 +12ac +12bc +12cc +12d4 +12dc +12e4 +12ec +12f4 +12fc +1304 +130c +131c +1324 +132c +1334 +133c +1344 +134c +1354 +13a1 +13a8 +13ae +13b4 +13ba +13c1 +13c7 +13ce +13d5 13d6 +13de +13e4 +13ea +13f0 +1401 +142f +144c +1467 +146b +1489 +14a3 +14c0 +14d3 +14ed +1510 +1526 +1542 1544 +1553 +155e 155f +156b +1570 +1574 +1586 +158a +1597 +159b +15a7 +15b0 +15b4 +15b8 +15bc +15c0 +15c6 +15cc +15d2 +15de +15e4 +15ea +15f1 +15f7 +15fd +1604 +160a +1610 +1617 +161e +1624 +162a +1630 +1636 +163c +1642 +164a +1650 +1656 +165d +1663 +1669 +30a7 30a8 +30b1 30b2 +30bb 30bc +30c6 30c7 +30cd +30d8 30da +30e1 +30ec +30f1 +30f6 +30f9 +32d3 +32d8 +32dd +32e2 +32e7 +32ec +32f1 +32f9 +32fd +ff6a +ff74 +ff79 +ff7e +ff83 +ff88 +ff8d +ff92 +ff9a +3047 3048 +3051 3052 +305b 305c +3066 3067 +306d +3078 307a +3081 +308c +3091 END diff --git a/lib/unicode/Is/SylEE.pl b/lib/unicode/Is/SylEE.pl new file mode 100644 index 0000000000..0a22f78f65 --- /dev/null +++ b/lib/unicode/Is/SylEE.pl @@ -0,0 +1,34 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1408 +1436 +1453 +15c7 +15cd +15d3 +15df +15e5 +15eb +15f2 +15f8 +15fe +1605 +160b +1611 +1618 +161f +1625 +162b +1631 +1637 +163d +1643 +164b +1651 +1657 +165e +1664 +166a +END diff --git a/lib/unicode/Is/SylI.pl b/lib/unicode/Is/SylI.pl index ec287c456a..f80790ce44 100644 --- a/lib/unicode/Is/SylI.pl +++ b/lib/unicode/Is/SylI.pl @@ -2,4 +2,153 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1202 +120a +1212 +121a +1222 +122a +1232 +123a +1242 +1252 +1262 +126a +1272 +127a +1282 +1292 +129a +12a2 +12aa +12ba +12ca +12d2 +12da +12e2 +12ea +12f2 +12fa +1302 +130a +131a +1322 +132a +1332 +133a +1342 +134a +1352 +13a2 +13a9 +13af +13b5 +13bb +13c2 +13c8 +13cf +13d7 13d8 +13df +13e5 +13eb +13f1 +1403 +1409 +1431 +1437 +144e +1454 +1468 +146d +148b +14a5 +14c2 +14d5 +14ef +1511 +1528 +1541 +1546 +1555 +1560 1561 +156c +1571 +1575 +157f +1587 +158b +158f +1598 +159c +15a0 +15a8 +15b1 +15b5 +15b9 +15bd +15c1 +15c8 +15ce +15d4 +15e0 +15e6 +15ec +15f3 +15f9 +15ff +1606 +160c +1612 +1619 161a +1620 +1626 +162c +1632 +1638 +163e +1644 +164c +1652 +1658 +165f +1665 +166b +1671 +30a3 30a4 +30ad 30ae +30b7 30b8 +30c1 30c2 +30cb +30d2 30d4 +30df +30ea +30f0 +30f8 +32d1 +32d6 +32db +32e0 +32e5 +32ea +32ef +32f7 +32fc +ff68 +ff72 +ff77 +ff7c +ff81 +ff86 +ff8b +ff90 +ff98 +3043 3044 +304d 304e +3057 3058 +3061 3062 +306b +3072 3074 +307f +308a +3090 END diff --git a/lib/unicode/Is/SylII.pl b/lib/unicode/Is/SylII.pl new file mode 100644 index 0000000000..4516d7a32a --- /dev/null +++ b/lib/unicode/Is/SylII.pl @@ -0,0 +1,25 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1404 +1432 +144f +146e +148c +14a6 +14c3 +14d6 +14f0 +1512 +1529 +1547 +1556 +1562 1563 +1576 +1580 +1590 +15a1 +15a9 +1672 +END diff --git a/lib/unicode/Is/SylN.pl b/lib/unicode/Is/SylN.pl new file mode 100644 index 0000000000..215463fb7f --- /dev/null +++ b/lib/unicode/Is/SylN.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +3093 +END diff --git a/lib/unicode/Is/SylO.pl b/lib/unicode/Is/SylO.pl index ec287c456a..a0a6f7dd01 100644 --- a/lib/unicode/Is/SylO.pl +++ b/lib/unicode/Is/SylO.pl @@ -2,4 +2,156 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1206 +120e +1216 +121e +1226 +122e +1236 +123e +1246 +1256 +1266 +126e +1276 +127e +1286 +1296 +129e +12a6 +12ae +12be +12ce +12d6 +12de +12e6 +12ee +12f6 +12fe +1306 +130e +131e +1326 +132e +1336 +133e +1346 +134e +1356 +13a3 +13aa +13b0 +13b6 +13bc +13c3 +13c9 +13d0 +13d9 +13e0 +13e6 +13ec +13f2 +1405 +1433 +1450 +1469 +146f +148d +14a7 +14c4 +14d7 +14f1 +1513 +152a +1548 +154a +1557 +1564 +156d +1572 +1577 +1581 +1588 +158c +1591 +1599 +159d +15a2 +15aa +15b2 +15b6 +15ba +15be +15c2 +15c5 +15cb +15d1 +15dd +15e3 +15e9 +15f0 +15f6 +15fc +1603 +1609 +160f +1616 +161d +1623 +1629 +162f +1635 +163b +1641 +1649 +164f +1655 +165c +1662 +1668 +1673 +30a9 30aa +30b3 30b4 +30bd 30be +30c8 30c9 +30ce +30db 30dd +30e2 +30e7 30e8 +30ed +30f2 +30fa +32d4 +32d9 +32de +32e3 +32e8 +32ed +32f2 +32f5 +32fa +32fe +ff66 +ff6b +ff6e +ff75 +ff7a +ff7f +ff84 +ff89 +ff8e +ff93 +ff96 +ff9b +3049 304a +3053 3054 +305d 305e +3068 3069 +306e +307b 307d +3082 +3087 3088 +308d +3092 END diff --git a/lib/unicode/Is/SylOO.pl b/lib/unicode/Is/SylOO.pl new file mode 100644 index 0000000000..12280534b1 --- /dev/null +++ b/lib/unicode/Is/SylOO.pl @@ -0,0 +1,25 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1406 1407 +1434 1435 +1451 1452 +1470 1471 +148e 148f +14a8 14a9 +14c5 14c6 +14d8 14d9 +14f2 14f3 +1514 +152b 152c +1549 +1558 +1565 +1578 +1582 +1592 +15a3 +15ab +1674 +END diff --git a/lib/unicode/Is/SylU.pl b/lib/unicode/Is/SylU.pl index ec287c456a..c458382f25 100644 --- a/lib/unicode/Is/SylU.pl +++ b/lib/unicode/Is/SylU.pl @@ -2,4 +2,121 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1201 +1209 +1211 +1219 +1221 +1229 +1231 +1239 +1241 +1251 +1261 +1269 +1271 +1279 +1281 +1291 +1299 +12a1 +12a9 +12b9 +12c9 +12d1 +12d9 +12e1 +12e9 +12f1 +12f9 +1301 +1309 +1319 +1321 +1329 +1331 +1339 +1341 +1349 +1351 +13a4 +13ab +13b1 +13b7 +13bd +13c4 +13ca +13d1 +13da +13e1 +13e7 +13ed +13f3 +15c4 +15ca +15d0 +15dc +15e2 +15e8 +15ef +15f5 +15fb +1602 +1608 +160e +1614 1615 +161c +1622 +1628 +162e +1634 +163a +1640 +1648 +164e +1654 +165b +1661 +1667 +30a5 30a6 +30af 30b0 +30b9 30ba +30c3 30c5 +30cc +30d5 30d7 +30e0 +30e5 30e6 +30eb +30f4 +32d2 +32d7 +32dc +32e1 +32e6 +32eb +32f0 +32f4 +32f8 +ff69 +ff6d +ff6f +ff73 +ff78 +ff7d +ff82 +ff87 +ff8c +ff91 +ff95 +ff99 +3045 3046 +304f 3050 +3059 305a +3063 3065 +306c +3075 3077 +3080 +3085 3086 +308b +3094 END diff --git a/lib/unicode/Is/SylV.pl b/lib/unicode/Is/SylV.pl index ec287c456a..b6e76f81b9 100644 --- a/lib/unicode/Is/SylV.pl +++ b/lib/unicode/Is/SylV.pl @@ -2,4 +2,53 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1200 +1208 +1210 +1218 +1220 +1228 +1230 +1238 +1240 +1250 +1260 +1268 +1270 +1278 +1280 +1290 +1298 +12a0 +12a8 +12b8 +12c8 +12d0 +12d8 +12e0 +12e8 +12f0 +12f8 +1300 +1308 +1318 +1320 +1328 +1330 +1338 +1340 +1348 +1350 +13a5 +13ac +13b2 +13b8 +13c5 +13cb +13d2 +13db +13e2 +13e8 +13ee +13f4 END diff --git a/lib/unicode/Is/SylWA.pl b/lib/unicode/Is/SylWA.pl index ec287c456a..9bb529ed01 100644 --- a/lib/unicode/Is/SylWA.pl +++ b/lib/unicode/Is/SylWA.pl @@ -2,4 +2,48 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +120f +1217 +121f +1227 +122f +1237 +123f +124b +125b +1267 +126f +1277 +127f +128b +1297 +129f +12a7 +12b3 +12c3 +12df +12e7 +12f7 +12ff +1307 +1313 +1327 +132f +1337 +133f +134f +1357 +1417 1418 +1444 1445 +1461 1462 +147e 147f +149c 149d +14b6 14b7 +14cb 14cc +14e6 14e7 +1500 1501 +150c 150f +1521 1522 +1539 153a +15db END diff --git a/lib/unicode/Is/SylWAA.pl b/lib/unicode/Is/SylWAA.pl new file mode 100644 index 0000000000..5f3b784d0c --- /dev/null +++ b/lib/unicode/Is/SylWAA.pl @@ -0,0 +1,19 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1419 141b +1446 1448 +1463 1465 +1480 1482 +149e 14a0 +14b8 14ba +14cd 14cf +14e8 14e9 +1502 1504 +1523 1524 +153b 153d +154e 154f +155b 155c +1568 1569 +END diff --git a/lib/unicode/Is/SylWC.pl b/lib/unicode/Is/SylWC.pl index ec287c456a..3ad968c505 100644 --- a/lib/unicode/Is/SylWC.pl +++ b/lib/unicode/Is/SylWC.pl @@ -2,4 +2,12 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +124d +125d +128d +12b5 +12c5 +1315 +1484 +1507 END diff --git a/lib/unicode/Is/SylWE.pl b/lib/unicode/Is/SylWE.pl index ec287c456a..9e32c0e602 100644 --- a/lib/unicode/Is/SylWE.pl +++ b/lib/unicode/Is/SylWE.pl @@ -2,4 +2,22 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +124c +125c +128c +12b4 +12c4 +1314 +140c 140d +143a 143b +1457 1458 +1474 1475 +1492 1493 +14ac 14ad +14c9 14ca +14dc 14dd +14f6 14f7 +1517 1518 +152f 1530 +15d8 END diff --git a/lib/unicode/Is/SylWEE.pl b/lib/unicode/Is/SylWEE.pl new file mode 100644 index 0000000000..c4bccb5240 --- /dev/null +++ b/lib/unicode/Is/SylWEE.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +15d9 +END diff --git a/lib/unicode/Is/SylWI.pl b/lib/unicode/Is/SylWI.pl index ec287c456a..4cd6c6789c 100644 --- a/lib/unicode/Is/SylWI.pl +++ b/lib/unicode/Is/SylWI.pl @@ -2,4 +2,21 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +124a +125a +128a +12b2 +12c2 +1312 +140e 140f +143c 143d +1459 145a +1476 1477 +1494 1495 +14ae 14af +14de 14df +14f8 14f9 +1519 151a +1531 1532 +15da END diff --git a/lib/unicode/Is/SylWII.pl b/lib/unicode/Is/SylWII.pl new file mode 100644 index 0000000000..bd68aeadf5 --- /dev/null +++ b/lib/unicode/Is/SylWII.pl @@ -0,0 +1,15 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1410 1411 +143e 143f +145b 145c +1478 1479 +1496 1497 +14b0 14b1 +14e0 14e1 +14fa 14fb +151b 151c +1533 1534 +END diff --git a/lib/unicode/Is/SylWO.pl b/lib/unicode/Is/SylWO.pl new file mode 100644 index 0000000000..7676564130 --- /dev/null +++ b/lib/unicode/Is/SylWO.pl @@ -0,0 +1,16 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1412 1413 +1440 1441 +145d 145e +147a 147b +1498 1499 +14b2 14b3 +14e2 14e3 +14fc 14fd +151d 151e +1535 1536 +15d7 +END diff --git a/lib/unicode/Is/SylWOO.pl b/lib/unicode/Is/SylWOO.pl new file mode 100644 index 0000000000..0ab766a553 --- /dev/null +++ b/lib/unicode/Is/SylWOO.pl @@ -0,0 +1,15 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +1414 1416 +1442 1443 +145f 1460 +147c 147d +149a 149b +14b4 14b5 +14e4 14e5 +14fe 14ff +151f 1520 +1537 1538 +END diff --git a/lib/unicode/Is/SylWU.pl b/lib/unicode/Is/SylWU.pl new file mode 100644 index 0000000000..76af7aefad --- /dev/null +++ b/lib/unicode/Is/SylWU.pl @@ -0,0 +1,6 @@ +# !!!!!!! DO NOT EDIT THIS FILE !!!!!!! +# This file is built by mktables.PL from e.g. Unicode.300. +# Any changes made here will be lost! +return <<'END'; +15d6 +END diff --git a/lib/unicode/Is/SylWV.pl b/lib/unicode/Is/SylWV.pl index ec287c456a..8bd8849042 100644 --- a/lib/unicode/Is/SylWV.pl +++ b/lib/unicode/Is/SylWV.pl @@ -2,4 +2,10 @@ # This file is built by mktables.PL from e.g. Unicode.300. # Any changes made here will be lost! return <<'END'; +1248 +1258 +1288 +12b0 +12c0 +1310 END diff --git a/lib/unicode/Is/Upper.pl b/lib/unicode/Is/Upper.pl index 8dde2742d0..4fda655dc4 100644 --- a/lib/unicode/Is/Upper.pl +++ b/lib/unicode/Is/Upper.pl @@ -86,9 +86,9 @@ return <<'END'; 01b5 01b7 01b8 01bc -01c4 -01c7 -01ca +01c4 01c5 +01c7 01c8 +01ca 01cb 01cd 01cf 01d1 @@ -106,7 +106,7 @@ return <<'END'; 01ea 01ec 01ee -01f1 +01f1 01f2 01f4 01f6 01f8 01fa @@ -355,11 +355,14 @@ return <<'END'; 1f5d 1f5f 1f68 1f6f -1fb8 1fbb -1fc8 1fcb +1f88 1f8f +1f98 1f9f +1fa8 1faf +1fb8 1fbc +1fc8 1fcc 1fd8 1fdb 1fe8 1fec -1ff8 1ffb +1ff8 1ffc 2102 2107 210b 210d diff --git a/lib/unicode/Makefile b/lib/unicode/Makefile index c68fa3af00..af5e77b47b 100644 --- a/lib/unicode/Makefile +++ b/lib/unicode/Makefile @@ -1,6 +1,5 @@ all: - ./mktables.PL - ./MakeEthiopicSyllables.PL + ../../miniperl -I../../lib ./mktables.PL clean: rm -f *.pl */*.pl diff --git a/lib/unicode/mktables.PL b/lib/unicode/mktables.PL index 4f705a4016..241d2e6bb3 100755 --- a/lib/unicode/mktables.PL +++ b/lib/unicode/mktables.PL @@ -1,6 +1,11 @@ #!../../miniperl +use bytes; + $UnicodeData = "Unicode.300"; +$SyllableData = "syllables.txt"; +$PropData = "Props.txt"; + # Note: we try to keep filenames unique within first 8 chars. Using # subdirectories for the following helps. @@ -14,16 +19,15 @@ mkdir "To", 0777; ['IsWord', '$cat =~ /^L[ulot]|^Nd/ or $code eq "005F"', ''], ['IsAlnum', '$cat =~ /^L[ulot]|^Nd/', ''], ['IsAlpha', '$cat =~ /^L[ulot]/', ''], - # XXX broken: recursive definition (/\s/ will look up IsSpace in future) - ['IsSpace', '$cat =~ /^Z/ or $code lt "0020" and chr(hex $code) =~ /^\s/', ''], + ['IsSpace', 'White space', $PropData], ['IsDigit', '$cat =~ /^Nd$/', ''], - ['IsUpper', '$cat =~ /^Lu$/', ''], + ['IsUpper', '$cat =~ /^L[ut]$/', ''], ['IsLower', '$cat =~ /^Ll$/', ''], ['IsASCII', 'hex $code <= 127', ''], ['IsCntrl', '$cat =~ /^C/', ''], - ['IsGraph', '$cat =~ /^[^C]/ and $code ne "0020"', ''], + ['IsGraph', '$cat =~ /^[^C]/ and ($cat !~ /^Z/ and $code ne "0020" or chr(hex $code) !~ /^\s/)', ''], ['IsPrint', '$cat =~ /^[^C]/', ''], - ['IsPunct', '$cat =~ /^P/', ''], + ['IsPunct', 'Punctuation', $PropData], ['IsXDigit', '$code =~ /^00(3[0-9]|[46][1-6])$/', ''], ['ToUpper', '$up', '$up'], ['ToLower', '$down', '$down'], @@ -43,12 +47,14 @@ mkdir "To", 0777; ['IsM', '$cat =~ /^M/', ''], # Mark ['IsMn', '$cat eq "Mn"', ''], # Mark, Non-Spacing ['IsMc', '$cat eq "Mc"', ''], # Mark, Combining + ['IsMe', '$cat eq "Me"', ''], # Mark, Enclosing ['IsN', '$cat =~ /^N/', ''], # Number ['IsNd', '$cat eq "Nd"', ''], # Number, Decimal Digit ['IsNo', '$cat eq "No"', ''], # Number, Other + ['IsNl', '$cat eq "Nl"', ''], # Number, Letter - ['IsZ', '$cat =~ /^Z/', ''], # Zeparator + ['IsZ', '$cat =~ /^Z/', ''], # Separator ['IsZs', '$cat eq "Zs"', ''], # Separator, Space ['IsZl', '$cat eq "Zl"', ''], # Separator, Line ['IsZp', '$cat eq "Zp"', ''], # Separator, Paragraph @@ -57,6 +63,9 @@ mkdir "To", 0777; ['IsCc', '$cat eq "Cc"', ''], # Other, Control or Format ['IsCo', '$cat eq "Co"', ''], # Other, Private Use ['IsCn', '$cat eq "Cn"', ''], # Other, Not Assigned + ['IsCf', '$cat eq "Cf"', ''], # Other, Format + ['IsCs', '$cat eq "Cs"', ''], # Other, Surrogate + ['IsCn', 'Unassigned Code Value',$PropData], # Other, Not Assigned # Informative @@ -72,9 +81,13 @@ mkdir "To", 0777; ['IsPs', '$cat eq "Ps"', ''], # Punctuation, Open ['IsPe', '$cat eq "Pe"', ''], # Punctuation, Close ['IsPo', '$cat eq "Po"', ''], # Punctuation, Other + ['IsPc', '$cat eq "Pc"', ''], # Punctuation, Connector + ['IsPi', '$cat eq "Pi"', ''], # Punctuation, Initial quote + ['IsPf', '$cat eq "Pf"', ''], # Punctuation, Final quote ['IsS', '$cat =~ /^S/', ''], # Symbol ['IsSm', '$cat eq "Sm"', ''], # Symbol, Math + ['IsSk', '$cat eq "Sk"', ''], # Symbol, Modifier ['IsSc', '$cat eq "Sc"', ''], # Symbol, Currency ['IsSo', '$cat eq "So"', ''], # Symbol, Other @@ -95,6 +108,15 @@ mkdir "To", 0777; # and punctuation specific to # those scripts + ['IsBidiLRE', '$bid eq "LRE"', ''], # Left-to-Right Embedding + ['IsBidiLRO', '$bid eq "LRO"', ''], # Left-to-Right Override + ['IsBidiAL', '$bid eq "AL"', ''], # Right-to-Left Arabic + ['IsBidiRLE', '$bid eq "RLE"', ''], # Right-to-Left Embedding + ['IsBidiRLO', '$bid eq "RLO"', ''], # Right-to-Left Override + ['IsBidiPDF', '$bid eq "PDF"', ''], # Pop Directional Format + ['IsBidiNSM', '$bid eq "NSM"', ''], # Non-Spacing Mark + ['IsBidiBN', '$bid eq "BN"', ''], # Boundary Neutral + # Weak types: ['IsBidiEN','$bid eq "EN"', ''], # European Number @@ -134,6 +156,7 @@ mkdir "To", 0777; ['IsDCnarrow', '$decomp =~ /^<narrow>/', ''], ['IsDCsmall', '$decomp =~ /^<small>/', ''], ['IsDCsquare', '$decomp =~ /^<square>/', ''], + ['IsDCfraction', '$decomp =~ /^<fraction>/', ''], ['IsDCcompat', '$decomp =~ /^<compat>/', ''], # Number @@ -155,19 +178,8 @@ mkdir "To", 0777; # Syllables - ['IsSylV', '$syl eq "V"', ''], - ['IsSylU', '$syl eq "U"', ''], - ['IsSylI', '$syl eq "I"', ''], - ['IsSylA', '$syl eq "A"', ''], - ['IsSylE', '$syl eq "E"', ''], - ['IsSylC', '$syl eq "C"', ''], - ['IsSylO', '$syl eq "O"', ''], - ['IsSylWV', '$syl eq "V"', ''], - ['IsSylWI', '$syl eq "I"', ''], - ['IsSylWA', '$syl eq "A"', ''], - ['IsSylWE', '$syl eq "E"', ''], - ['IsSylWC', '$syl eq "C"', ''], - + syllable_defs(), + # Line break properties - Normative ['IsLbrkBK','$brk eq "BK"', ''], # Mandatory Break @@ -232,8 +244,8 @@ END exit if @ARGV and not grep { $_ eq Block } @ARGV; print "Block\n"; -open(UD, 'Blocks.txt') or die "Can't open blocks.txt: $!\n"; -open(OUT, ">Block.pl") or die "Can't create $table.pl: $!\n"; +open(UD, 'Blocks.txt') or die "Can't open Blocks.txt: $!\n"; +open(OUT, ">Block.pl") or die "Can't create Block.pl: $!\n"; print OUT <<EOH; # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! # This file is built by $0 from e.g. $UnicodeData. @@ -277,6 +289,8 @@ sub proplist { my $out; my $split; + return listFromPropFile($wanted) if $val eq $PropData; + if ($table =~ /^Arab/) { open(UD, "ArabShap.txt") or warn "Can't open $table: $!"; @@ -288,7 +302,7 @@ sub proplist { $split = '($code, $short, $name) = split(/; */); $code =~ s/^U\+//;'; } elsif ($table =~ /^IsSyl/) { - open(UD, "syllables.txt") or warn "Can't open $table: $!"; + open(UD, $SyllableData) or warn "Can't open $table: $!"; $split = '($code, $short, $syl) = split(/; */); $code =~ s/^U\+//;'; } @@ -308,8 +322,8 @@ sub proplist { eval <<"END"; while (<UD>) { next if /^#/; - next if /^\s/; - chop; + next if /^\\s/; + s/\\s+\$//; $split if ($wanted) { push(\@wanted, [hex \$code, hex $val, \$name =~ /, First>\$/]); @@ -343,7 +357,7 @@ END eval <<"END"; while (<UD>) { next if /^#/; - next if /^\s*\$/; + next if /^\\s*\$/; chop; $split if ($wanted) { @@ -376,4 +390,44 @@ END $out; } +sub listFromPropFile { + my ($wanted) = @_; + my $out; + + open (UD, $PropData) or die "Can't open $PropData: $!\n"; + local($/) = "\n" . '*' x 43 . "\n\nProperty dump for:"; # not 42? + + <UD>; + while (<UD>) { + chomp; + if (s/0x[\d\w]+\s+\((.*?)\)// and $wanted eq $1) { + s/\(\d+ chars\)//g; + s/^\s+//mg; + s/\s+$//mg; + s/\.\./\t/g; + $out = lc $_; + last; + } + } + close (UD); + "$out\n"; +} + +sub syllable_defs { + my @defs; + my %seen; + + open (SD, $SyllableData) or die "Can't open $SyllableData: $!\n"; + while (<SD>) { + next if /^\s*(#|$)/; + s/\s+$//; + ($code, $name, $syl) = split /; */; + next unless $syl; + push (@defs, ["IsSyl$syl", qq{\$syl eq "$syl"}, '']) + unless $seen{$syl}++; + } + close (SD); + return (@defs); +} + # eof |