有点儿好奇，你们的这些表都是哪儿来的？
在微软的官网上有Windows 10 Sorting Weight Table文件下载，大小约为20MB，开头部分源代码如下：
; Unisort.txt
; This file contains all the NLS sorting data. It is UTF-8 for some of the comments
;
; The initial table is our default table (nominally GUID 00000001-57EE-1E5C-00B4-D0000BB1E11E)
; Weights are 4 bytes: SM, AW, DW & CW. Together the SM + AW are "Unicode Weight"
; SM - Script Member
; AW - Alphabetic Weight
; DW - Diacritic Weight
; CW - Case Weight
;
; For the most part, strings are sorted with all of the "Unicode Weights" being primary, then
; the diacritic and case weights as being secondary and tertiary weights. (a few more extra
; weights follow that).
;
; The sort keys generated are the "Unicode Weight" of the entire string, delimited by 01, then
; the Diacritic (or secondary) weight, then another 01 delimiter, then the Case (or tertiary)
; weight. The code also adds two more sections, again delimited by 01. Weights are 0 terminated.
;
; So for Äpple the weights would be 0e 02 0e 7e 0e 7e 0e 48 0e 21 01 13 01 12 01 01 00
; 0e 02 0e 7e 0e 7e 0e 48 0e 21 "apple" Unicode Weight
; 01 delimeter
; 13 "ä" diacritic weight
; 01
; 12 "A" case weight
; 01 01 00 (two sections with no weight in this string and terminator)
;
; Special cases:
; Our sort keys are delimited by 1 and terminated by 0, so no weights by use 0 or 1.
; In these tables, 0 0 0 0 is used as a flag for "no weight". Those characters are effectively
; ignored by NLS.
;
; East Asian locales may have "3 byte weights". In that case, DW is promoted to be included
; in UW, so UW then is SM, AW, and DW. This only happens if 192 <= SW <= 239.
; PUA are always 3 byte weights
;
; SM values:
; 0 - reserved for no weight code points
; 1 - flag for non-spacing marks, those weights are added to the previous base character's weights
; (Note: DW are a minimum of 2, so all characters have 2 added to their DW, however
; non-spacing characters presume the base already had that 2, so these weights are only
; the additional diacritic difference, without the base 2
; example: A has a DW of 2, ̈ has a DW of 17, so Ä has a DW of 19 (2 + 17))
; 2 - Unused ??? reserved because our minimums are 2 (Does SortGetSortKey generate any 2s?)
; 3 -
; 14 - Latin script
;
; AW values:
; 0-1 - reserved for delimiter/terminator
; 2 - "first letter in script". Eg: the AW for A.
;
; DW Values:
; 0-1 - reserved for delimiter/terminator
;
; CW Values:
; 0-1 - reserved for delimiter/terminator
; 0x01 bit - Full Width (if set). (1 == Full Width, 0 == Half/Normal Width)
; 0x02 bit - Set by default, can be cleared in some case (if another higher bit is set)
; 0x04 bit - Super/Subscript?
; 0x08 bit -
; 0x10 bit - Upper Case. (16 == upper case, 0 == lower case)
; 0x20 bit -
; 0x40/0x80 bits - Reserved for nlstrans, which uses these as flags for characters that may compress
;
; Flags to NLSTrans
; After the GUID
; HAS_3_BYTE_WEIGHTS - This ID has 3 byte weights. Must be tagged on the EXCEPTION AND on the COMPRESSION
; LINGUISTIC_CASING - Tagged (only) on the EXCEPTION table that applies to the linguistic case.
; Should also have another untagged EXCEPTION table for non-linguistic casing.
;
SORTKEY
DEFAULT ; Characters in this table are sorted by: SM, AW, DW, CW, Codepoint #
; Weightless code points (these will be ignored when comparing strings)
0x0000 0 0 0 0 ;<control>
0x00ad 0 0 0 0 ;Soft Hyphen
0x034f 0 0 0 0 ;Combining Grapheme Joiner
0x0640 0 0 0 0 ;Arabic Tatweel
0x0ecc 0 0 0 0 ;Lao Cancellation Mark
0x1806 0 0 0 0 ;Mongolian Todo Soft Hyphen
0x180b 0 0 0 0 ;Mongolian Free Variation Selector One
0x180c 0 0 0 0 ;Mongolian Free Variation Selector Two
0x180d 0 0 0 0 ;Mongolian Free Variation Selector Three
0x200c 0 0 0 0 ;Zero Width Non-joiner
0x200d 0 0 0 0 ;Zero Width Joiner
0x200e 0 0 0 0 ;Left-to-right Mark
0x200f 0 0 0 0 ;Right-to-left Mark
0x202a 0 0 0 0 ;Left-to-right Embedding
0x202b 0 0 0 0 ;Right-to-left Embedding
0x202c 0 0 0 0 ;Pop Directional Formatting
0x202d 0 0 0 0 ;Left-to-right Override
0x202e 0 0 0 0 ;Right-to-left Override
0x2060 0 0 0 0 ;Word Joiner
0x2061 0 0 0 0 ;Function Application
0x2062 0 0 0 0 ;Invisible Times
0x2063 0 0 0 0 ;Invisible Separator
0x2064 0 0 0 0 ;INVISIBLE PLUS
0x206a 0 0 0 0 ;Inhibit Symmetric Swapping
0x206b 0 0 0 0 ;Activate Symmetric Swapping
0x206c 0 0 0 0 ;Inhibit Arabic Form Shaping
0x206d 0 0 0 0 ;Activate Arabic Form Shaping
0x206e 0 0 0 0 ;National Digit Shapes
0x206f 0 0 0 0 ;Nominal Digit Shapes
0x3190 0 0 0 0 ;Ideographic Annotation Linking Mark
0x3191 0 0 0 0 ;Ideographic Annotation Reverse Mark
0xdb40 0 0 0 0 ;Variation Selector High Surrogate in range 0xE0100-0xE01EF
0xfe00 0 0 0 0 ;Variation Selector-1
0xfe01 0 0 0 0 ;Variation Selector-2
0xfe02 0 0 0 0 ;Variation Selector-3
0xfe03 0 0 0 0 ;Variation Selector-4
0xfe04 0 0 0 0 ;Variation Selector-5
0xfe05 0 0 0 0 ;Variation Selector-6
0xfe06 0 0 0 0 ;Variation Selector-7
0xfe07 0 0 0 0 ;Variation Selector-8
0xfe08 0 0 0 0 ;Variation Selector-9
0xfe09 0 0 0 0 ;Variation Selector-10
0xfe0a 0 0 0 0 ;Variation Selector-11
0xfe0b 0 0 0 0 ;Variation Selector-12
0xfe0c 0 0 0 0 ;Variation Selector-13
0xfe0d 0 0 0 0 ;Variation Selector-14
0xfe0e 0 0 0 0 ;Variation Selector-15
0xfe0f 0 0 0 0 ;Variation Selector-16
0xfeff 0 0 0 0 ;ZWNBSP - Byte Order Mark
0xfff9 0 0 0 0 ;Interlinear Annotation Anchor
0xfffa 0 0 0 0 ;Interlinear Annotation Separator
0xfffb 0 0 0 0 ;Interlinear Annotation Terminator
0xfffc 0 0 0 0 ;Object Replacement Character
0xfffd 0 0 0 0 ;Replacement Character
; 1 - flag for non-spacing marks, these weights are added to the previous base character's weights
; (Note: DW are a minimum of 2, so all characters have 2 added to their DW, however
; non-spacing characters presume the base already had that 2, so these weights are only
; the additional diacritic difference, without the base 2
; example: A has a DW of 2, ̈ has a DW of 17, so Ä has a DW of 19 (2 + 17))
; We ignore leading/trailing DWs of 2 when we construct the sort key in order to save memory.
;CP SCRIPT ALPHA DIACRITIC CASING COMMENT
;0x0b70 had a DW of 0 in sort version 00060304, but we ignore leading/trailing DWs <= 2. Hence we changed the DW to 3 in version 00060305.
0x0b70 1 0 3 0 ;Odia Isshar
;The following six characters had a DW of 1 in sort version 00060304, but we ignore leading/trailing DWs <= 2. Hence we changed the DW to 3 in version 00060305.
0x303c 1 0 3 0 ;Masu Mark
0x303d 1 0 3 0 ;Part Alternation Mark
0x303e 1 0 3 0 ;Ideographic Variation Indicator
0x3099 1 0 3 0 ;Non-Spacing Kana Daku-On
0x309b 1 0 3 0 ;Kana Daku-On
0xff9e 1 0 3 0 ;Halfwidth Kana Daku-On
;0x05a2 had a DW of 2 in sort version 00060304, but we ignore leading/trailing DWs <= 2. Hence we changed the DW to 3 in version 00060305.
0x05a2 1 0 3 0 ;Hebrew Accent Atnah Hafukh
;The following characters had a DW of 2 in sort version 00060304, but we ignore leading/trailing DWs <= 2.
;Nonspacing marks in the same script are already taking DW 3, hence we changed the DW to 4 in version 00060305.
0x309a 1 0 4 0 ;Non-Spacing Kana HanDaku-On
0x309c 1 0 4 0 ;Kana HanDaku-On
0xff9f 1 0 4 0 ;Halfwidth Kana HanDaku-On
;The following five characters had a DW of 3 prior to updating the above DWs.
0x030d 1 0 3 0 ;Non-Spacing Vertical Line Above
0x0e48 1 0 3 0 ;Thai Tone Mai Ek
0x1a74 1 0 3 0 ;TAI THAM SIGN MAI KANG
0x1c37 1 0 3 0 ;LEPCHA SIGN NUKTA
0xa92b 1 0 3 0 ;KAYAH LI TONE PLOPHU
0x0591 1 0 4 0 ;Hebrew Accent Etnahta
0x09bc 1 0 4 0 ;Bengali Sign Nukta
0x0a3c 1 0 4 0 ;Gurmukhi Sign Nukta
0x0b3c 1 0 4 0 ;Odia Sign Nukta
0x0cbc 1 0 4 0 ;Kannada Sign Nukta
0x0cd5 1 0 4 0 ;Kannada Length Mark
0x0e49 1 0 4 0 ;Thai Tone Mai Tho
0x1a75 1 0 4 0 ;TAI THAM SIGN TONE-1
0xa92c 1 0 4 0 ;KAYAH LI TONE CALYA
0x0592 1 0 5 0 ;Hebrew Accent Tipeha
0x0cd6 1 0 5 0 ;Kannada Ai Length Mark
0x0e4a 1 0 5 0 ;Thai Tone Mai Tri
0x1a76 1 0 5 0 ;TAI THAM SIGN TONE-2
0xa92d 1 0 5 0 ;KAYAH LI TONE CALYA PLOPHU
0x0593 1 0 6 0 ;Hebrew Accent Dehi
0x093c 1 0 6 0 ;Devanagari Sign Nukta
0x0a71 1 0 6 0 ;Gurmukhi Addak
0x0e4b 1 0 6 0 ;Thai Tone Mai Chattawa
0x1a77 1 0 6 0 ;TAI THAM SIGN KHUEN TONE-3
0x0594 1 0 7 0 ;Hebrew Accent Mahapakh
0x0971 1 0 7 0 ;DEVANAGARI SIGN HIGH SPACING DOT
0x0e4d 1 0 7 0 ;Thai Nikkhahit
0x1a78 1 0 7 0 ;TAI THAM SIGN KHUEN TONE-4
0x302a 1 0 7 0 ;Ideographic Level Tone Mark
0x302e 1 0 7 0 ;Hangul Single Dot Tone Mark
0x0595 1 0 8 0 ;Hebrew Accent Yetiv
0x0e47 1 0 8 0 ;Thai Vowel Sign Mai Tai Khu
0x1a79 1 0 8 0 ;TAI THAM SIGN KHUEN TONE-5
0x302b 1 0 8 0 ;Ideographic Rising Tone Mark
0x302f 1 0 8 0 ;Hangul Double Dot Tone Mark
0x0596 1 0 9 0 ;Hebrew Accent Tevir
0x0e4c 1 0 9 0 ;Thai Thanthakhat
0x1a7a 1 0 9 0 ;TAI THAM SIGN RA HAAM
0x302c 1 0 9 0 ;Ideographic Departing Tone Mark
0x0597 1 0 10 0 ;Hebrew Accent Munah
0x07EB 1 0 10 0 ;NKO COMBINING SHORT HIGH TONE
0x0951 1 0 10 0 ;Devanagari Stress Sign Udatta
0x0a01 1 0 10 0 ;Gurmukhi Sign Adak Bindi
0x0a02 1 0 10 0 ;Gurmukhi Sign Bindi
0x1a7b 1 0 10 0 ;TAI THAM SIGN MAI SAM
0x0598 1 0 11 0 ;Hebrew Accent Merkha
0x07EC 1 0 11 0 ;NKO COMBINING SHORT LOW TONE
0x0abc 1 0 11 0 ;Gujarati Sign Nukta
0x0f84 1 0 11 0 ;Tibetan Mark Halanta
0x1a7c 1 0 11 0 ;TAI THAM SIGN KHUEN-LUE KARAN
0x2d7f 1 0 11 0 ;TIFINAGH CONSONANT JOINER
0x302d 1 0 11 0 ;Ideographic Entering Tone Mark
0x02b9 1 0 12 0 ;Modifier Prime
0x0301 1 0 12 0 ;Non-Spacing Acute Accent
0x0341 1 0 12 0 ;Non-Spacing Acute Tone Mark
0x0599 1 0 12 0 ;Hebrew Accent Merkha Kefula
0x07ED 1 0 12 0 ;NKO COMBINING SHORT RISING TONE
0x0952 1 0 12 0 ;Devanagari Stress Sign Anudatta
0x1a7f 1 0 12 0 ;TAI THAM COMBINING CRYPTOGRAMMIC DOT
0x0300 1 0 13 0 ;Non-Spacing Grave Accent
0x0340 1 0 13 0 ;Non-Spacing Grave Tone Mark
0x059a 1 0 13 0 ;Hebrew Accent Darga
0x07EE 1 0 13 0 ;NKO COMBINING LONG DESCENDING TONE
0x0c55 1 0 13 0 ;Telugu Length Mark
0x0f71 1 0 13 0 ;Tibetan Vowel Sign Aa
0x0307 1 0 14 0 ;Non-Spacing Dot Above
0x059b 1 0 14 0 ;Hebrew Accent Yerah Ben Yomo
0x07EF 1 0 14 0 ;NKO COMBINING LONG HIGH TONE
0x0953 1 0 14 0 ;Devanagari Grave Accent
0x0a70 1 0 14 0 ;Gurmukhi Tippi
0x0c56 1 0 14 0 ;Telugu Ai Length Mark
0x059c 1 0 15 0 ;Hebrew Accent Segol
0x07F0 1 0 15 0 ;NKO COMBINING LONG LOW TONE
0x0f39 1 0 15 0 ;Tibetan Mark Tsa -phru
0x0302 1 0 16 0 ;Non-Spacing Circumflex
0x059d 1 0 16 0 ;Hebrew Accent Shalshelet
0x07F1 1 0 16 0 ;NKO COMBINING LONG RISING TONE
0x0954 1 0 16 0 ;Devanagari Acute Accent
0x0308 1 0 17 0 ;Non-Spacing Diaeresis
0x059e 1 0 17 0 ;Hebrew Accent Zaqef Qatan
0x07F2 1 0 17 0 ;NKO COMBINING NASALIZATION MARK
0x0f7f 1 0 17 0 ;Tibetan Sign Rnam Bcad
0x030c 1 0 18 0 ;Non-Spacing Caron (Hacek)
0x059f 1 0 18 0 ;Hebrew Accent Zaqef Gadol
0x0306 1 0 19 0 ;Non-Spacing Breve
0x05a0 1 0 19 0 ;Hebrew Accent Revia
0x07F3 1 0 19 0 ;NKO COMBINING DOUBLE DOT ABOVE
0x0f85 1 0 19 0 ;Tibetan Mark Paluta
0x035d 1 0 20 0 ;Combining Double Breve
0x0304 1 0 21 0 ;Non-Spacing Macron
0x035e 1 0 21 0 ;Combining Double Macron
0x035f 1 0 21 0 ;Combining Double Macron Below
0x05a1 1 0 21 0 ;Hebrew Accent Zarqa
0x0f88 1 0 21 0 ;Tibetan Sign Lce Tsa Can
string.Compare("A","a")为什么等于1
