go grammar: added initial (small) testing files and utf8.lm

7 files changed, 1656 insertions, 18 deletions

diff --git a/grammar/go/.gitignore b/grammar/go/.gitignore
index 80d228d8..7f1dc001 100644
--- a/grammar/go/.gitignore
+++ b/grammar/go/.gitignore
@@ -1,5 +1,2 @@
-/utf8.lm
 /go.c
 /go
-/insert-semi.c
-/insert-semi
diff --git a/grammar/go/Makefile b/grammar/go/Makefile
index 8f5d8447..faed933e 100644
--- a/grammar/go/Makefile
+++ b/grammar/go/Makefile
@@ -6,7 +6,7 @@ all: go
 go: go.lm parsego.lm utf8.lm $(COLM)
 	$(COLM) -o go parsego.lm
 
-utf8.lm: Makefile utf8pat.rb UnicodeData.txt
+utf8: UnicodeData.txt
 	echo 'rl unicode_letter /' >utf8.lm
 	ruby utf8pat.rb 'L[ultmo]' 1 UnicodeData.txt >> utf8.lm
 	echo '/' >>utf8.lm
diff --git a/grammar/go/go.lm b/grammar/go/go.lm
index 0ab8f7fd..a1523a26 100644
--- a/grammar/go/go.lm
+++ b/grammar/go/go.lm
@@ -101,7 +101,6 @@ lex
 	# Rune literals
 	#
 
-
 	rl escaped_char
 		/ '\\' ( 'a' | 'b' | 'f' | 'n' | 'r' | 't' | 'v' | '\\' | "'" | '"' ) /
 
diff --git a/grammar/go/input.til b/grammar/go/input.til
deleted file mode 100644
index 008182b8..00000000
--- a/grammar/go/input.til
+++ /dev/null
@@ -1,13 +0,0 @@
-var a
-a := 1;
-
-head:
-
-a := ( a + 1 )
-c := d;
-
-if a = 10 then
-	goto head
-end
-
-hi := there; friend := yes
diff --git a/grammar/go/test/input.go b/grammar/go/test/input.go
new file mode 100644
index 00000000..e9e89a51
--- /dev/null
+++ b/grammar/go/test/input.go
@@ -0,0 +1,41 @@
+a b c
+d e f
+treeß
+		foo then /**/
+	bar  //ladkf
+	baz /*
+*/end
+42
+4_2
+0600
+0_600
+0o600
+0O600       // second character is capital letter 'O'
+0xBadFace
+0xBad_Face
+0x_67_7a_2f_cc_40_c6
+170141183460469231731687303715884105727
+170_141183_460469_231731_687303_715884_105727
+_42         // an identifier, not an integer literal
+'a'
+'ä'
+'本'
+'\t'
+'\000'
+'\007'
+'\377'
+'\x07'
+'\xff'
+'\u12e4'
+'\U00101234'
+'\''         // rune literal containing single quote character
+`abc`                // same as "abc"
+`\n
+\n`                  // same as "\\n\n\\n"
+"\n"
+"\""                 // same as `"`
+"Hello, world!\n"
+"日本語"
+"\u65e5本\U00008a9e"
+"\xff\u00FF"
+
diff --git a/grammar/go/test/invalid.go b/grammar/go/test/invalid.go
new file mode 100644
index 00000000..5de15057
--- /dev/null
+++ b/grammar/go/test/invalid.go
@@ -0,0 +1,10 @@
+42_         // invalid: _ must separate successive digits
+4__2        // invalid: only one _ at a time
+0_xBadFace  // invalid: _ must separate successive digits
+'aa'         // illegal: too many characters
+'\xa'        // illegal: too few hexadecimal digits
+'\0'         // illegal: too few octal digits
+'\uDFFF'     // illegal: surrogate half
+'\U00110000' // illegal: invalid Unicode code point
+"\uD800"             // illegal: surrogate half
+"\U00110000"         // illegal: invalid Unicode code point
diff --git a/grammar/go/utf8.lm b/grammar/go/utf8.lm
new file mode 100644
index 00000000..667572f2
--- /dev/null
+++ b/grammar/go/utf8.lm
@@ -0,0 +1,1604 @@
+rl unicode_letter /
+	0x41 .. 0x5A |
+	0x61 .. 0x7A |
+	0xC2 (
+		0xAA |
+		0xB5 |
+		0xBA
+	) |
+	0xC3 (
+		0x80 .. 0x96 |
+		0x98 .. 0xB6 |
+		0xB8 .. 0xBF
+	) |
+	0xC4 .. 0xCA (
+		0x80 .. 0xBF
+	) |
+	0xCB (
+		0x80 .. 0x81 |
+		0x86 .. 0x91 |
+		0xA0 .. 0xA4 |
+		0xAC |
+		0xAE
+	) |
+	0xCD (
+		0xB0 .. 0xB4 |
+		0xB6 .. 0xB7 |
+		0xBA .. 0xBD |
+		0xBF
+	) |
+	0xCE (
+		0x86 |
+		0x88 .. 0x8A |
+		0x8C |
+		0x8E .. 0xA1 |
+		0xA3 .. 0xBF
+	) |
+	0xCF (
+		0x80 .. 0xB5 |
+		0xB7 .. 0xBF
+	) |
+	0xD0 .. 0xD1 (
+		0x80 .. 0xBF
+	) |
+	0xD2 (
+		0x80 .. 0x81 |
+		0x8A .. 0xBF
+	) |
+	0xD3 (
+		0x80 .. 0xBF
+	) |
+	0xD4 (
+		0x80 .. 0xAF |
+		0xB1 .. 0xBF
+	) |
+	0xD5 (
+		0x80 .. 0x96 |
+		0x99 |
+		0xA0 .. 0xBF
+	) |
+	0xD6 (
+		0x80 .. 0x88
+	) |
+	0xD7 (
+		0x90 .. 0xAA |
+		0xAF .. 0xB2
+	) |
+	0xD8 (
+		0xA0 .. 0xBF
+	) |
+	0xD9 (
+		0x80 .. 0x8A |
+		0xAE .. 0xAF |
+		0xB1 .. 0xBF
+	) |
+	0xDA (
+		0x80 .. 0xBF
+	) |
+	0xDB (
+		0x80 .. 0x93 |
+		0x95 |
+		0xA5 .. 0xA6 |
+		0xAE .. 0xAF |
+		0xBA .. 0xBC |
+		0xBF
+	) |
+	0xDC (
+		0x90 |
+		0x92 .. 0xAF
+	) |
+	0xDD (
+		0x8D .. 0xBF
+	) |
+	0xDE (
+		0x80 .. 0xA5 |
+		0xB1
+	) |
+	0xDF (
+		0x8A .. 0xAA |
+		0xB4 .. 0xB5 |
+		0xBA
+	) |
+	0xE0 (
+		0xA0 (
+			0x80 .. 0x95 |
+			0x9A |
+			0xA4 |
+			0xA8
+		) |
+		0xA1 (
+			0x80 .. 0x98 |
+			0xA0 .. 0xAA
+		) |
+		0xA2 (
+			0xA0 .. 0xB4 |
+			0xB6 .. 0xBD
+		) |
+		0xA4 (
+			0x84 .. 0xB9 |
+			0xBD
+		) |
+		0xA5 (
+			0x90 |
+			0x98 .. 0xA1 |
+			0xB1 .. 0xBF
+		) |
+		0xA6 (
+			0x80 |
+			0x85 .. 0x8C |
+			0x8F .. 0x90 |
+			0x93 .. 0xA8 |
+			0xAA .. 0xB0 |
+			0xB2 |
+			0xB6 .. 0xB9 |
+			0xBD
+		) |
+		0xA7 (
+			0x8E |
+			0x9C .. 0x9D |
+			0x9F .. 0xA1 |
+			0xB0 .. 0xB1 |
+			0xBC
+		) |
+		0xA8 (
+			0x85 .. 0x8A |
+			0x8F .. 0x90 |
+			0x93 .. 0xA8 |
+			0xAA .. 0xB0 |
+			0xB2 .. 0xB3 |
+			0xB5 .. 0xB6 |
+			0xB8 .. 0xB9
+		) |
+		0xA9 (
+			0x99 .. 0x9C |
+			0x9E |
+			0xB2 .. 0xB4
+		) |
+		0xAA (
+			0x85 .. 0x8D |
+			0x8F .. 0x91 |
+			0x93 .. 0xA8 |
+			0xAA .. 0xB0 |
+			0xB2 .. 0xB3 |
+			0xB5 .. 0xB9 |
+			0xBD
+		) |
+		0xAB (
+			0x90 |
+			0xA0 .. 0xA1 |
+			0xB9
+		) |
+		0xAC (
+			0x85 .. 0x8C |
+			0x8F .. 0x90 |
+			0x93 .. 0xA8 |
+			0xAA .. 0xB0 |
+			0xB2 .. 0xB3 |
+			0xB5 .. 0xB9 |
+			0xBD
+		) |
+		0xAD (
+			0x9C .. 0x9D |
+			0x9F .. 0xA1 |
+			0xB1
+		) |
+		0xAE (
+			0x83 |
+			0x85 .. 0x8A |
+			0x8E .. 0x90 |
+			0x92 .. 0x95 |
+			0x99 .. 0x9A |
+			0x9C |
+			0x9E .. 0x9F |
+			0xA3 .. 0xA4 |
+			0xA8 .. 0xAA |
+			0xAE .. 0xB9
+		) |
+		0xAF (
+			0x90
+		) |
+		0xB0 (
+			0x85 .. 0x8C |
+			0x8E .. 0x90 |
+			0x92 .. 0xA8 |
+			0xAA .. 0xB9 |
+			0xBD
+		) |
+		0xB1 (
+			0x98 .. 0x9A |
+			0xA0 .. 0xA1
+		) |
+		0xB2 (
+			0x80 |
+			0x85 .. 0x8C |
+			0x8E .. 0x90 |
+			0x92 .. 0xA8 |
+			0xAA .. 0xB3 |
+			0xB5 .. 0xB9 |
+			0xBD
+		) |
+		0xB3 (
+			0x9E |
+			0xA0 .. 0xA1 |
+			0xB1 .. 0xB2
+		) |
+		0xB4 (
+			0x85 .. 0x8C |
+			0x8E .. 0x90 |
+			0x92 .. 0xBA |
+			0xBD
+		) |
+		0xB5 (
+			0x8E |
+			0x94 .. 0x96 |
+			0x9F .. 0xA1 |
+			0xBA .. 0xBF
+		) |
+		0xB6 (
+			0x85 .. 0x96 |
+			0x9A .. 0xB1 |
+			0xB3 .. 0xBB |
+			0xBD
+		) |
+		0xB7 (
+			0x80 .. 0x86
+		) |
+		0xB8 (
+			0x81 .. 0xB0 |
+			0xB2 .. 0xB3
+		) |
+		0xB9 (
+			0x80 .. 0x86
+		) |
+		0xBA (
+			0x81 .. 0x82 |
+			0x84 |
+			0x86 .. 0x8A |
+			0x8C .. 0xA3 |
+			0xA5 |
+			0xA7 .. 0xB0 |
+			0xB2 .. 0xB3 |
+			0xBD
+		) |
+		0xBB (
+			0x80 .. 0x84 |
+			0x86 |
+			0x9C .. 0x9F
+		) |
+		0xBC (
+			0x80
+		) |
+		0xBD (
+			0x80 .. 0x87 |
+			0x89 .. 0xAC
+		) |
+		0xBE (
+			0x88 .. 0x8C
+		)
+	) |
+	0xE1 (
+		0x80 (
+			0x80 .. 0xAA |
+			0xBF
+		) |
+		0x81 (
+			0x90 .. 0x95 |
+			0x9A .. 0x9D |
+			0xA1 |
+			0xA5 .. 0xA6 |
+			0xAE .. 0xB0 |
+			0xB5 .. 0xBF
+		) |
+		0x82 (
+			0x80 .. 0x81 |
+			0x8E |
+			0xA0 .. 0xBF
+		) |
+		0x83 (
+			0x80 .. 0x85 |
+			0x87 |
+			0x8D |
+			0x90 .. 0xBA |
+			0xBC .. 0xBF
+		) |
+		0x84 .. 0x88 (
+			0x80 .. 0xBF
+		) |
+		0x89 (
+			0x80 .. 0x88 |
+			0x8A .. 0x8D |
+			0x90 .. 0x96 |
+			0x98 |
+			0x9A .. 0x9D |
+			0xA0 .. 0xBF
+		) |
+		0x8A (
+			0x80 .. 0x88 |
+			0x8A .. 0x8D |
+			0x90 .. 0xB0 |
+			0xB2 .. 0xB5 |
+			0xB8 .. 0xBE
+		) |
+		0x8B (
+			0x80 |
+			0x82 .. 0x85 |
+			0x88 .. 0x96 |
+			0x98 .. 0xBF
+		) |
+		0x8C (
+			0x80 .. 0x90 |
+			0x92 .. 0x95 |
+			0x98 .. 0xBF
+		) |
+		0x8D (
+			0x80 .. 0x9A
+		) |
+		0x8E (
+			0x80 .. 0x8F |
+			0xA0 .. 0xBF
+		) |
+		0x8F (
+			0x80 .. 0xB5 |
+			0xB8 .. 0xBD
+		) |
+		0x90 (
+			0x81 .. 0xBF
+		) |
+		0x91 .. 0x98 (
+			0x80 .. 0xBF
+		) |
+		0x99 (
+			0x80 .. 0xAC |
+			0xAF .. 0xBF
+		) |
+		0x9A (
+			0x81 .. 0x9A |
+			0xA0 .. 0xBF
+		) |
+		0x9B (
+			0x80 .. 0xAA |
+			0xB1 .. 0xB8
+		) |
+		0x9C (
+			0x80 .. 0x8C |
+			0x8E .. 0x91 |
+			0xA0 .. 0xB1
+		) |
+		0x9D (
+			0x80 .. 0x91 |
+			0xA0 .. 0xAC |
+			0xAE .. 0xB0
+		) |
+		0x9E (
+			0x80 .. 0xB3
+		) |
+		0x9F (
+			0x97 |
+			0x9C
+		) |
+		0xA0 (
+			0xA0 .. 0xBF
+		) |
+		0xA1 (
+			0x80 .. 0xB8
+		) |
+		0xA2 (
+			0x80 .. 0x84 |
+			0x87 .. 0xA8 |
+			0xAA |
+			0xB0 .. 0xBF
+		) |
+		0xA3 (
+			0x80 .. 0xB5
+		) |
+		0xA4 (
+			0x80 .. 0x9E
+		) |
+		0xA5 (
+			0x90 .. 0xAD |
+			0xB0 .. 0xB4
+		) |
+		0xA6 (
+			0x80 .. 0xAB |
+			0xB0 .. 0xBF
+		) |
+		0xA7 (
+			0x80 .. 0x89
+		) |
+		0xA8 (
+			0x80 .. 0x96 |
+			0xA0 .. 0xBF
+		) |
+		0xA9 (
+			0x80 .. 0x94
+		) |
+		0xAA (
+			0xA7
+		) |
+		0xAC (
+			0x85 .. 0xB3
+		) |
+		0xAD (
+			0x85 .. 0x8B
+		) |
+		0xAE (
+			0x83 .. 0xA0 |
+			0xAE .. 0xAF |
+			0xBA .. 0xBF
+		) |
+		0xAF (
+			0x80 .. 0xA5
+		) |
+		0xB0 (
+			0x80 .. 0xA3
+		) |
+		0xB1 (
+			0x8D .. 0x8F |
+			0x9A .. 0xBD
+		) |
+		0xB2 (
+			0x80 .. 0x88 |
+			0x90 .. 0xBA |
+			0xBD .. 0xBF
+		) |
+		0xB3 (
+			0xA9 .. 0xAC |
+			0xAE .. 0xB3 |
+			0xB5 .. 0xB6 |
+			0xBA
+		) |
+		0xB4 .. 0xB6 (
+			0x80 .. 0xBF
+		) |
+		0xB8 .. 0xBB (
+			0x80 .. 0xBF
+		) |
+		0xBC (
+			0x80 .. 0x95 |
+			0x98 .. 0x9D |
+			0xA0 .. 0xBF
+		) |
+		0xBD (
+			0x80 .. 0x85 |
+			0x88 .. 0x8D |
+			0x90 .. 0x97 |
+			0x99 |
+			0x9B |
+			0x9D |
+			0x9F .. 0xBD
+		) |
+		0xBE (
+			0x80 .. 0xB4 |
+			0xB6 .. 0xBC |
+			0xBE
+		) |
+		0xBF (
+			0x82 .. 0x84 |
+			0x86 .. 0x8C |
+			0x90 .. 0x93 |
+			0x96 .. 0x9B |
+			0xA0 .. 0xAC |
+			0xB2 .. 0xB4 |
+			0xB6 .. 0xBC
+		)
+	) |
+	0xE2 (
+		0x81 (
+			0xB1 |
+			0xBF
+		) |
+		0x82 (
+			0x90 .. 0x9C
+		) |
+		0x84 (
+			0x82 |
+			0x87 |
+			0x8A .. 0x93 |
+			0x95 |
+			0x99 .. 0x9D |
+			0xA4 |
+			0xA6 |
+			0xA8 |
+			0xAA .. 0xAD |
+			0xAF .. 0xB9 |
+			0xBC .. 0xBF
+		) |
+		0x85 (
+			0x85 .. 0x89 |
+			0x8E
+		) |
+		0x86 (
+			0x83 .. 0x84
+		) |
+		0xB0 (
+			0x80 .. 0xAE |
+			0xB0 .. 0xBF
+		) |
+		0xB1 (
+			0x80 .. 0x9E |
+			0xA0 .. 0xBF
+		) |
+		0xB2 (
+			0x80 .. 0xBF
+		) |
+		0xB3 (
+			0x80 .. 0xA4 |
+			0xAB .. 0xAE |
+			0xB2 .. 0xB3
+		) |
+		0xB4 (
+			0x80 .. 0xA5 |
+			0xA7 |
+			0xAD |
+			0xB0 .. 0xBF
+		) |
+		0xB5 (
+			0x80 .. 0xA7 |
+			0xAF
+		) |
+		0xB6 (
+			0x80 .. 0x96 |
+			0xA0 .. 0xA6 |
+			0xA8 .. 0xAE |
+			0xB0 .. 0xB6 |
+			0xB8 .. 0xBE
+		) |
+		0xB7 (
+			0x80 .. 0x86 |
+			0x88 .. 0x8E |
+			0x90 .. 0x96 |
+			0x98 .. 0x9E
+		) |
+		0xB8 (
+			0xAF
+		)
+	) |
+	0xE3 (
+		0x80 (
+			0x85 .. 0x86 |
+			0xB1 .. 0xB5 |
+			0xBB .. 0xBC
+		) |
+		0x81 (
+			0x81 .. 0xBF
+		) |
+		0x82 (
+			0x80 .. 0x96 |
+			0x9D .. 0x9F |
+			0xA1 .. 0xBF
+		) |
+		0x83 (
+			0x80 .. 0xBA |
+			0xBC .. 0xBF
+		) |
+		0x84 (
+			0x85 .. 0xAF |
+			0xB1 .. 0xBF
+		) |
+		0x85 (
+			0x80 .. 0xBF
+		) |
+		0x86 (
+			0x80 .. 0x8E |
+			0xA0 .. 0xBA
+		) |
+		0x87 (
+			0xB0 .. 0xBF
+		) |
+		0x90 (
+			0x80
+		)
+	) |
+	0xE4 (
+		0xB6 (
+			0xB5
+		) |
+		0xB8 (
+			0x80
+		)
+	) |
+	0xE9 (
+		0xBF (
+			0xAF
+		)
+	) |
+	0xEA (
+		0x80 .. 0x91 (
+			0x80 .. 0xBF
+		) |
+		0x92 (
+			0x80 .. 0x8C
+		) |
+		0x93 (
+			0x90 .. 0xBD
+		) |
+		0x94 .. 0x97 (
+			0x80 .. 0xBF
+		) |
+		0x98 (
+			0x80 .. 0x8C |
+			0x90 .. 0x9F |
+			0xAA .. 0xAB
+		) |
+		0x99 (
+			0x80 .. 0xAE |
+			0xBF
+		) |
+		0x9A (
+			0x80 .. 0x9D |
+			0xA0 .. 0xBF
+		) |
+		0x9B (
+			0x80 .. 0xA5
+		) |
+		0x9C (
+			0x97 .. 0x9F |
+			0xA2 .. 0xBF
+		) |
+		0x9D (
+			0x80 .. 0xBF
+		) |
+		0x9E (
+			0x80 .. 0x88 |
+			0x8B .. 0xBF
+		) |
+		0x9F (
+			0x82 .. 0x86 |
+			0xB7 .. 0xBF
+		) |
+		0xA0 (
+			0x80 .. 0x81 |
+			0x83 .. 0x85 |
+			0x87 .. 0x8A |
+			0x8C .. 0xA2
+		) |
+		0xA1 (
+			0x80 .. 0xB3
+		) |
+		0xA2 (
+			0x82 .. 0xB3
+		) |
+		0xA3 (
+			0xB2 .. 0xB7 |
+			0xBB |
+			0xBD .. 0xBE
+		) |
+		0xA4 (
+			0x8A .. 0xA5 |
+			0xB0 .. 0xBF
+		) |
+		0xA5 (
+			0x80 .. 0x86 |
+			0xA0 .. 0xBC
+		) |
+		0xA6 (
+			0x84 .. 0xB2
+		) |
+		0xA7 (
+			0x8F |
+			0xA0 .. 0xA4 |
+			0xA6 .. 0xAF |
+			0xBA .. 0xBE
+		) |
+		0xA8 (
+			0x80 .. 0xA8
+		) |
+		0xA9 (
+			0x80 .. 0x82 |
+			0x84 .. 0x8B |
+			0xA0 .. 0xB6 |
+			0xBA |
+			0xBE .. 0xBF
+		) |
+		0xAA (
+			0x80 .. 0xAF |
+			0xB1 |
+			0xB5 .. 0xB6 |
+			0xB9 .. 0xBD
+		) |
+		0xAB (
+			0x80 |
+			0x82 |
+			0x9B .. 0x9D |
+			0xA0 .. 0xAA |
+			0xB2 .. 0xB4
+		) |
+		0xAC (
+			0x81 .. 0x86 |
+			0x89 .. 0x8E |
+			0x91 .. 0x96 |
+			0xA0 .. 0xA6 |
+			0xA8 .. 0xAE |
+			0xB0 .. 0xBF
+		) |
+		0xAD (
+			0x80 .. 0x9A |
+			0x9C .. 0xA7 |
+			0xB0 .. 0xBF
+		) |
+		0xAE (
+			0x80 .. 0xBF
+		) |
+		0xAF (
+			0x80 .. 0xA2
+		) |
+		0xB0 (
+			0x80
+		)
+	) |
+	0xED (
+		0x9E (
+			0xA3 |
+			0xB0 .. 0xBF
+		) |
+		0x9F (
+			0x80 .. 0x86 |
+			0x8B .. 0xBB
+		)
+	) |
+	0xEF (
+		0xA4 .. 0xA8 (
+			0x80 .. 0xBF
+		) |
+		0xA9 (
+			0x80 .. 0xAD |
+			0xB0 .. 0xBF
+		) |
+		0xAA (
+			0x80 .. 0xBF
+		) |
+		0xAB (
+			0x80 .. 0x99
+		) |
+		0xAC (
+			0x80 .. 0x86 |
+			0x93 .. 0x97 |
+			0x9D |
+			0x9F .. 0xA8 |
+			0xAA .. 0xB6 |
+			0xB8 .. 0xBC |
+			0xBE
+		) |
+		0xAD (
+			0x80 .. 0x81 |
+			0x83 .. 0x84 |
+			0x86 .. 0xBF
+		) |
+		0xAE (
+			0x80 .. 0xB1
+		) |
+		0xAF (
+			0x93 .. 0xBF
+		) |
+		0xB0 .. 0xB3 (
+			0x80 .. 0xBF
+		) |
+		0xB4 (
+			0x80 .. 0xBD
+		) |
+		0xB5 (
+			0x90 .. 0xBF
+		) |
+		0xB6 (
+			0x80 .. 0x8F |
+			0x92 .. 0xBF
+		) |
+		0xB7 (
+			0x80 .. 0x87 |
+			0xB0 .. 0xBB
+		) |
+		0xB9 (
+			0xB0 .. 0xB4 |
+			0xB6 .. 0xBF
+		) |
+		0xBA (
+			0x80 .. 0xBF
+		) |
+		0xBB (
+			0x80 .. 0xBC
+		) |
+		0xBC (
+			0xA1 .. 0xBA
+		) |
+		0xBD (
+			0x81 .. 0x9A |
+			0xA6 .. 0xBF
+		) |
+		0xBE (
+			0x80 .. 0xBE
+		) |
+		0xBF (
+			0x82 .. 0x87 |
+			0x8A .. 0x8F |
+			0x92 .. 0x97 |
+			0x9A .. 0x9C
+		)
+	) |
+	0xF0 (
+		0x90 (
+			0x80 (
+				0x80 .. 0x8B |
+				0x8D .. 0xA6 |
+				0xA8 .. 0xBA |
+				0xBC .. 0xBD |
+				0xBF
+			) |
+			0x81 (
+				0x80 .. 0x8D |
+				0x90 .. 0x9D
+			) |
+			0x82 (
+				0x80 .. 0xBF
+			) |
+			0x83 (
+				0x80 .. 0xBA
+			) |
+			0x8A (
+				0x80 .. 0x9C |
+				0xA0 .. 0xBF
+			) |
+			0x8B (
+				0x80 .. 0x90
+			) |
+			0x8C (
+				0x80 .. 0x9F |
+				0xAD .. 0xBF
+			) |
+			0x8D (
+				0x80 |
+				0x82 .. 0x89 |
+				0x90 .. 0xB5
+			) |
+			0x8E (
+				0x80 .. 0x9D |
+				0xA0 .. 0xBF
+			) |
+			0x8F (
+				0x80 .. 0x83 |
+				0x88 .. 0x8F
+			) |
+			0x90 .. 0x91 (
+				0x80 .. 0xBF
+			) |
+			0x92 (
+				0x80 .. 0x9D |
+				0xB0 .. 0xBF
+			) |
+			0x93 (
+				0x80 .. 0x93 |
+				0x98 .. 0xBB
+			) |
+			0x94 (
+				0x80 .. 0xA7 |
+				0xB0 .. 0xBF
+			) |
+			0x95 (
+				0x80 .. 0xA3
+			) |
+			0x98 .. 0x9B (
+				0x80 .. 0xBF
+			) |
+			0x9C (
+				0x80 .. 0xB6
+			) |
+			0x9D (
+				0x80 .. 0x95 |
+				0xA0 .. 0xA7
+			) |
+			0xA0 (
+				0x80 .. 0x85 |
+				0x88 |
+				0x8A .. 0xB5 |
+				0xB7 .. 0xB8 |
+				0xBC |
+				0xBF
+			) |
+			0xA1 (
+				0x80 .. 0x95 |
+				0xA0 .. 0xB6
+			) |
+			0xA2 (
+				0x80 .. 0x9E
+			) |
+			0xA3 (
+				0xA0 .. 0xB2 |
+				0xB4 .. 0xB5
+			) |
+			0xA4 (
+				0x80 .. 0x95 |
+				0xA0 .. 0xB9
+			) |
+			0xA6 (
+				0x80 .. 0xB7 |
+				0xBE .. 0xBF
+			) |
+			0xA8 (
+				0x80 |
+				0x90 .. 0x93 |
+				0x95 .. 0x97 |
+				0x99 .. 0xB5
+			) |
+			0xA9 (
+				0xA0 .. 0xBC
+			) |
+			0xAA (
+				0x80 .. 0x9C
+			) |
+			0xAB (
+				0x80 .. 0x87 |
+				0x89 .. 0xA4
+			) |
+			0xAC (
+				0x80 .. 0xB5
+			) |
+			0xAD (
+				0x80 .. 0x95 |
+				0xA0 .. 0xB2
+			) |
+			0xAE (
+				0x80 .. 0x91
+			) |
+			0xB0 (
+				0x80 .. 0xBF
+			) |
+			0xB1 (
+				0x80 .. 0x88
+			) |
+			0xB2 .. 0xB3 (
+				0x80 .. 0xB2
+			) |
+			0xB4 (
+				0x80 .. 0xA3
+			) |
+			0xBC (
+				0x80 .. 0x9C |
+				0xA7 |
+				0xB0 .. 0xBF
+			) |
+			0xBD (
+				0x80 .. 0x85
+			) |
+			0xBF (
+				0xA0 .. 0xB6
+			)
+		) |
+		0x91 (
+			0x80 (
+				0x83 .. 0xB7
+			) |
+			0x82 (
+				0x83 .. 0xAF
+			) |
+			0x83 (
+				0x90 .. 0xA8
+			) |
+			0x84 (
+				0x83 .. 0xA6
+			) |
+			0x85 (
+				0x84 |
+				0x90 .. 0xB2 |
+				0xB6
+			) |
+			0x86 (
+				0x83 .. 0xB2
+			) |
+			0x87 (
+				0x81 .. 0x84 |
+				0x9A |
+				0x9C
+			) |
+			0x88 (
+				0x80 .. 0x91 |
+				0x93 .. 0xAB
+			) |
+			0x8A (
+				0x80 .. 0x86 |
+				0x88 |
+				0x8A .. 0x8D |
+				0x8F .. 0x9D |
+				0x9F .. 0xA8 |
+				0xB0 .. 0xBF
+			) |
+			0x8B (
+				0x80 .. 0x9E
+			) |
+			0x8C (
+				0x85 .. 0x8C |
+				0x8F .. 0x90 |
+				0x93 .. 0xA8 |
+				0xAA .. 0xB0 |
+				0xB2 .. 0xB3 |
+				0xB5 .. 0xB9 |
+				0xBD
+			) |
+			0x8D (
+				0x90 |
+				0x9D .. 0xA1
+			) |
+			0x90 (
+				0x80 .. 0xB4
+			) |
+			0x91 (
+				0x87 .. 0x8A |
+				0x9F
+			) |
+			0x92 (
+				0x80 .. 0xAF
+			) |
+			0x93 (
+				0x84 .. 0x85 |
+				0x87
+			) |
+			0x96 (
+				0x80 .. 0xAE
+			) |
+			0x97 (
+				0x98 .. 0x9B
+			) |
+			0x98 (
+				0x80 .. 0xAF
+			) |
+			0x99 (
+				0x84
+			) |
+			0x9A (
+				0x80 .. 0xAA |
+				0xB8
+			) |
+			0x9C (
+				0x80 .. 0x9A
+			) |
+			0xA0 (
+				0x80 .. 0xAB
+			) |
+			0xA2 (
+				0xA0 .. 0xBF
+			) |
+			0xA3 (
+				0x80 .. 0x9F |
+				0xBF
+			) |
+			0xA6 (
+				0xA0 .. 0xA7 |
+				0xAA .. 0xBF
+			) |
+			0xA7 (
+				0x80 .. 0x90 |
+				0xA1 |
+				0xA3
+			) |
+			0xA8 (
+				0x80 |
+				0x8B .. 0xB2 |
+				0xBA
+			) |
+			0xA9 (
+				0x90 |
+				0x9C .. 0xBF
+			) |
+			0xAA (
+				0x80 .. 0x89 |
+				0x9D
+			) |
+			0xAB (
+				0x80 .. 0xB8
+			) |
+			0xB0 (
+				0x80 .. 0x88 |
+				0x8A .. 0xAE
+			) |
+			0xB1 (
+				0x80 |
+				0xB2 .. 0xBF
+			) |
+			0xB2 (
+				0x80 .. 0x8F
+			) |
+			0xB4 (
+				0x80 .. 0x86 |
+				0x88 .. 0x89 |
+				0x8B .. 0xB0
+			) |
+			0xB5 (
+				0x86 |
+				0xA0 .. 0xA5 |
+				0xA7 .. 0xA8 |
+				0xAA .. 0xBF
+			) |
+			0xB6 (
+				0x80 .. 0x89 |
+				0x98
+			) |
+			0xBB (
+				0xA0 .. 0xB2
+			)
+		) |
+		0x92 (
+			0x80 .. 0x8D (
+				0x80 .. 0xBF
+			) |
+			0x8E (
+				0x80 .. 0x99
+			) |
+			0x92 .. 0x94 (
+				0x80 .. 0xBF
+			) |
+			0x95 (
+				0x80 .. 0x83
+			)
+		) |
+		0x93 (
+			0x80 .. 0x8F (
+				0x80 .. 0xBF
+			) |
+			0x90 (
+				0x80 .. 0xAE
+			)
+		) |
+		0x94 (
+			0x90 .. 0x98 (
+				0x80 .. 0xBF
+			) |
+			0x99 (
+				0x80 .. 0x86
+			)
+		) |
+		0x96 (
+			0xA0 .. 0xA7 (
+				0x80 .. 0xBF
+			) |
+			0xA8 (
+				0x80 .. 0xB8
+			) |
+			0xA9 (
+				0x80 .. 0x9E
+			) |
+			0xAB (
+				0x90 .. 0xAD
+			) |
+			0xAC (
+				0x80 .. 0xAF
+			) |
+			0xAD (
+				0x80 .. 0x83 |
+				0xA3 .. 0xB7 |
+				0xBD .. 0xBF
+			) |
+			0xAE (
+				0x80 .. 0x8F
+			) |
+			0xB9 (
+				0x80 .. 0xBF
+			) |
+			0xBC (
+				0x80 .. 0xBF
+			) |
+			0xBD (
+				0x80 .. 0x8A |
+				0x90
+			) |
+			0xBE (
+				0x93 .. 0x9F
+			) |
+			0xBF (
+				0xA0 .. 0xA1 |
+				0xA3
+			)
+		) |
+		0x97 (
+			0x80 (
+				0x80
+			)
+		) |
+		0x98 (
+			0x9F (
+				0xB7
+			) |
+			0xA0 .. 0xAA (
+				0x80 .. 0xBF
+			) |
+			0xAB (
+				0x80 .. 0xB2
+			)
+		) |
+		0x9B (
+			0x80 .. 0x83 (
+				0x80 .. 0xBF
+			) |
+			0x84 (
+				0x80 .. 0x9E
+			) |
+			0x85 (
+				0x90 .. 0x92 |
+				0xA4 .. 0xA7 |
+				0xB0 .. 0xBF
+			) |
+			0x86 .. 0x8A (
+				0x80 .. 0xBF
+			) |
+			0x8B (
+				0x80 .. 0xBB
+			) |
+			0xB0 (
+				0x80 .. 0xBF
+			) |
+			0xB1 (
+				0x80 .. 0xAA |
+				0xB0 .. 0xBC
+			) |
+			0xB2 (
+				0x80 .. 0x88 |
+				0x90 .. 0x99
+			)
+		) |
+		0x9D (
+			0x90 (
+				0x80 .. 0xBF
+			) |
+			0x91 (
+				0x80 .. 0x94 |
+				0x96 .. 0xBF
+			) |
+			0x92 (
+				0x80 .. 0x9C |
+				0x9E .. 0x9F |
+				0xA2 |
+				0xA5 .. 0xA6 |
+				0xA9 .. 0xAC |
+				0xAE .. 0xB9 |
+				0xBB |
+				0xBD .. 0xBF
+			) |
+			0x93 (
+				0x80 .. 0x83 |
+				0x85 .. 0xBF
+			) |
+			0x94 (
+				0x80 .. 0x85 |
+				0x87 .. 0x8A |
+				0x8D .. 0x94 |
+				0x96 .. 0x9C |
+				0x9E .. 0xB9 |
+				0xBB .. 0xBE
+			) |
+			0x95 (
+				0x80 .. 0x84 |
+				0x86 |
+				0x8A .. 0x90 |
+				0x92 .. 0xBF
+			) |
+			0x96 .. 0x99 (
+				0x80 .. 0xBF
+			) |
+			0x9A (
+				0x80 .. 0xA5 |
+				0xA8 .. 0xBF
+			) |
+			0x9B (
+				0x80 |
+				0x82 .. 0x9A |
+				0x9C .. 0xBA |
+				0xBC .. 0xBF
+			) |
+			0x9C (
+				0x80 .. 0x94 |
+				0x96 .. 0xB4 |
+				0xB6 .. 0xBF
+			) |
+			0x9D (
+				0x80 .. 0x8E |
+				0x90 .. 0xAE |
+				0xB0 .. 0xBF
+			) |
+			0x9E (
+				0x80 .. 0x88 |
+				0x8A .. 0xA8 |
+				0xAA .. 0xBF
+			) |
+			0x9F (
+				0x80 .. 0x82 |
+				0x84 .. 0x8B
+			)
+		) |
+		0x9E (
+			0x84 (
+				0x80 .. 0xAC |
+				0xB7 .. 0xBD
+			) |
+			0x85 (
+				0x8E
+			) |
+			0x8B (
+				0x80 .. 0xAB
+			) |
+			0xA0 .. 0xA2 (
+				0x80 .. 0xBF
+			) |
+			0xA3 (
+				0x80 .. 0x84
+			) |
+			0xA4 (
+				0x80 .. 0xBF
+			) |
+			0xA5 (
+				0x80 .. 0x83 |
+				0x8B
+			) |
+			0xB8 (
+				0x80 .. 0x83 |
+				0x85 .. 0x9F |
+				0xA1 .. 0xA2 |
+				0xA4 |
+				0xA7 |
+				0xA9 .. 0xB2 |
+				0xB4 .. 0xB7 |
+				0xB9 |
+				0xBB
+			) |
+			0xB9 (
+				0x82 |
+				0x87 |
+				0x89 |
+				0x8B |
+				0x8D .. 0x8F |
+				0x91 .. 0x92 |
+				0x94 |
+				0x97 |
+				0x99 |
+				0x9B |
+				0x9D |
+				0x9F |
+				0xA1 .. 0xA2 |
+				0xA4 |
+				0xA7 .. 0xAA |
+				0xAC .. 0xB2 |
+				0xB4 .. 0xB7 |
+				0xB9 .. 0xBC |
+				0xBE
+			) |
+			0xBA (
+				0x80 .. 0x89 |
+				0x8B .. 0x9B |
+				0xA1 .. 0xA3 |
+				0xA5 .. 0xA9 |
+				0xAB .. 0xBB
+			)
+		) |
+		0xA0 (
+			0x80 (
+				0x80
+			)
+		) |
+		0xAA (
+			0x9B (
+				0x96
+			) |
+			0x9C (
+				0x80
+			)
+		) |
+		0xAB (
+			0x9C (
+				0xB4
+			) |
+			0x9D (
+				0x80
+			) |
+			0xA0 (
+				0x9D |
+				0xA0
+			)
+		) |
+		0xAC (
+			0xBA (
+				0xA1 |
+				0xB0
+			)
+		) |
+		0xAE (
+			0xAF (
+				0xA0
+			)
+		) |
+		0xAF (
+			0xA0 .. 0xA7 (
+				0x80 .. 0xBF
+			) |
+			0xA8 (
+				0x80 .. 0x9D
+			)
+		)
+	)
+/
+rl unicode_digit /
+	0x30 .. 0x39 |
+	0xD9 (
+		0xA0 .. 0xA9
+	) |
+	0xDB (
+		0xB0 .. 0xB9
+	) |
+	0xDF (
+		0x80 .. 0x89
+	) |
+	0xE0 (
+		0xA5 (
+			0xA6 .. 0xAF
+		) |
+		0xA7 (
+			0xA6 .. 0xAF
+		) |
+		0xA9 (
+			0xA6 .. 0xAF
+		) |
+		0xAB (
+			0xA6 .. 0xAF
+		) |
+		0xAD (
+			0xA6 .. 0xAF
+		) |
+		0xAF (
+			0xA6 .. 0xAF
+		) |
+		0xB1 (
+			0xA6 .. 0xAF
+		) |
+		0xB3 (
+			0xA6 .. 0xAF
+		) |
+		0xB5 (
+			0xA6 .. 0xAF
+		) |
+		0xB7 (
+			0xA6 .. 0xAF
+		) |
+		0xB9 (
+			0x90 .. 0x99
+		) |
+		0xBB (
+			0x90 .. 0x99
+		) |
+		0xBC (
+			0xA0 .. 0xA9
+		)
+	) |
+	0xE1 (
+		0x81 (
+			0x80 .. 0x89
+		) |
+		0x82 (
+			0x90 .. 0x99
+		) |
+		0x9F (
+			0xA0 .. 0xA9
+		) |
+		0xA0 (
+			0x90 .. 0x99
+		) |
+		0xA5 (
+			0x86 .. 0x8F
+		) |
+		0xA7 (
+			0x90 .. 0x99
+		) |
+		0xAA (
+			0x80 .. 0x89 |
+			0x90 .. 0x99
+		) |
+		0xAD (
+			0x90 .. 0x99
+		) |
+		0xAE (
+			0xB0 .. 0xB9
+		) |
+		0xB1 (
+			0x80 .. 0x89 |
+			0x90 .. 0x99
+		)
+	) |
+	0xEA (
+		0x98 (
+			0xA0 .. 0xA9
+		) |
+		0xA3 (
+			0x90 .. 0x99
+		) |
+		0xA4 (
+			0x80 .. 0x89
+		) |
+		0xA7 (
+			0x90 .. 0x99 |
+			0xB0 .. 0xB9
+		) |
+		0xA9 (
+			0x90 .. 0x99
+		) |
+		0xAF (
+			0xB0 .. 0xB9
+		)
+	) |
+	0xEF (
+		0xBC (
+			0x90 .. 0x99
+		)
+	) |
+	0xF0 (
+		0x90 (
+			0x92 (
+				0xA0 .. 0xA9
+			) |
+			0xB4 (
+				0xB0 .. 0xB9
+			)
+		) |
+		0x91 (
+			0x81 (
+				0xA6 .. 0xAF
+			) |
+			0x83 (
+				0xB0 .. 0xB9
+			) |
+			0x84 (
+				0xB6 .. 0xBF
+			) |
+			0x87 (
+				0x90 .. 0x99
+			) |
+			0x8B (
+				0xB0 .. 0xB9
+			) |
+			0x91 (
+				0x90 .. 0x99
+			) |
+			0x93 (
+				0x90 .. 0x99
+			) |
+			0x99 (
+				0x90 .. 0x99
+			) |
+			0x9B (
+				0x80 .. 0x89
+			) |
+			0x9C (
+				0xB0 .. 0xB9
+			) |
+			0xA3 (
+				0xA0 .. 0xA9
+			) |
+			0xB1 (
+				0x90 .. 0x99
+			) |
+			0xB5 (
+				0x90 .. 0x99
+			) |
+			0xB6 (
+				0xA0 .. 0xA9
+			)
+		) |
+		0x96 (
+			0xA9 (
+				0xA0 .. 0xA9
+			) |
+			0xAD (
+				0x90 .. 0x99
+			)
+		) |
+		0x9D (
+			0x9F (
+				0x8E .. 0xBF
+			)
+		) |
+		0x9E (
+			0x85 (
+				0x80 .. 0x89
+			) |
+			0x8B (
+				0xB0 .. 0xB9
+			) |
+			0xA5 (
+				0x90 .. 0x99
+			)
+		)
+	)
+/