mirror of
https://github.com/ilyakooo0/urbit.git
synced 2024-11-29 04:04:02 +03:00
WIP simplify parsing constants to ++soft, comment placement
This commit is contained in:
Anton Dyudin
parent
ccfb11bda9
commit
40050a490e
@ -15,24 +15,28 @@
|
||||
^- line:unicode-data
|
||||
(rash b line)
|
||||
::
|
||||
:> parses a single line of the unicode data file.
|
||||
:> parses a single character information line of the unicode data file.
|
||||
++ line
|
||||
;~ (glue sem)
|
||||
hex
|
||||
name-string
|
||||
general-category
|
||||
(bass 10 (plus sid:ab))
|
||||
bidi-category
|
||||
decomposition-mapping
|
||||
string-number
|
||||
string-number
|
||||
string-number
|
||||
yes-or-no
|
||||
name-string
|
||||
name-string
|
||||
optional-hex
|
||||
optional-hex
|
||||
optional-hex
|
||||
hex :: code/@c codepoint in hex format
|
||||
name-string :: name/tape character name
|
||||
general-category :: gen/general type of character
|
||||
(bass 10 (plus dit)) :: can/@ud canonical combining class
|
||||
bidi-category :: bi/bidi bidirectional category
|
||||
decomposition-mapping :: de/decomp decomposition mapping
|
||||
::
|
||||
:: todo: decimal/digit/numeric need to be parsed.
|
||||
::
|
||||
string-number :: decimal/tape decimal digit value (or ~)
|
||||
string-number :: digit/tape digit value, even if non-decimal
|
||||
string-number :: numeric/tape numeric value, including fractions
|
||||
::
|
||||
(flag 'Y' 'N') :: mirrored/? is char mirrored in bidi text?
|
||||
name-string :: old-name/tape unicode 1.0 compatibility name
|
||||
name-string :: iso/tape iso 10646 comment field
|
||||
(punt hex) :: up/(unit @c) uppercase mapping codepoint
|
||||
(punt hex) :: low/(unit @c) lowercase mapping codepoint
|
||||
(punt hex) :: title/(unit @c) titlecase mapping codepoint
|
||||
==
|
||||
::
|
||||
:> parses a single name or comment string.
|
||||
@ -40,122 +44,37 @@
|
||||
%+ cook
|
||||
|=(a/tape a)
|
||||
(star ;~(less sem prn))
|
||||
:: (star ;~(pose alp (mask " <>()")))
|
||||
::
|
||||
:> parses a unicode general category abbreviation to symbol
|
||||
++ general-category
|
||||
;~ pose
|
||||
(cold %lu (jest 'Lu'))
|
||||
(cold %ll (jest 'Ll'))
|
||||
(cold %lt (jest 'Lt'))
|
||||
(cold %mn (jest 'Mn'))
|
||||
(cold %mc (jest 'Mc'))
|
||||
(cold %me (jest 'Me'))
|
||||
(cold %nd (jest 'Nd'))
|
||||
(cold %nl (jest 'Nl'))
|
||||
(cold %no (jest 'No'))
|
||||
(cold %zs (jest 'Zs'))
|
||||
(cold %zl (jest 'Zl'))
|
||||
(cold %zp (jest 'Zp'))
|
||||
(cold %cc (jest 'Cc'))
|
||||
(cold %cf (jest 'Cf'))
|
||||
(cold %cs (jest 'Cs'))
|
||||
(cold %co (jest 'Co'))
|
||||
(cold %cn (jest 'Cn'))
|
||||
(cold %lm (jest 'Lm'))
|
||||
(cold %lo (jest 'Lo'))
|
||||
(cold %pc (jest 'Pc'))
|
||||
(cold %pd (jest 'Pd'))
|
||||
(cold %ps (jest 'Ps'))
|
||||
(cold %pe (jest 'Pe'))
|
||||
(cold %pi (jest 'Pi'))
|
||||
(cold %pf (jest 'Pf'))
|
||||
(cold %po (jest 'Po'))
|
||||
(cold %sm (jest 'Sm'))
|
||||
(cold %sc (jest 'Sc'))
|
||||
(cold %sk (jest 'Sk'))
|
||||
(cold %so (jest 'So'))
|
||||
==
|
||||
%+ sear (soft general:unicode-data)
|
||||
:(cook crip cass ;~(plug hig low (easy ~)))
|
||||
::
|
||||
:> parses a bidirectional category abbreviation to symbol.
|
||||
++ bidi-category
|
||||
;~ pose
|
||||
(cold %fsi (jest 'FSI'))
|
||||
(cold %lre (jest 'LRE'))
|
||||
(cold %lri (jest 'LRI'))
|
||||
(cold %lro (jest 'LRO'))
|
||||
(cold %nsm (jest 'NSM'))
|
||||
(cold %pdf (jest 'PDF'))
|
||||
(cold %pdi (jest 'PDI'))
|
||||
(cold %rle (jest 'RLE'))
|
||||
(cold %rli (jest 'RLI'))
|
||||
(cold %rlo (jest 'RLO'))
|
||||
(cold %al (jest 'AL'))
|
||||
(cold %an (jest 'AN'))
|
||||
(cold %bn (jest 'BN'))
|
||||
(cold %cs (jest 'CS'))
|
||||
(cold %en (jest 'EN'))
|
||||
(cold %es (jest 'ES'))
|
||||
(cold %et (jest 'ET'))
|
||||
(cold %on (jest 'ON'))
|
||||
(cold %ws (jest 'WS'))
|
||||
(cold %b (jest 'B'))
|
||||
(cold %l (jest 'L'))
|
||||
(cold %r (jest 'R'))
|
||||
(cold %s (jest 'S'))
|
||||
==
|
||||
%+ sear (soft bidi:unicode-data)
|
||||
:(cook crip cass (star hig))
|
||||
::
|
||||
:: TODO: This seems to be where the nest-fail is. There's an extra @ here?
|
||||
++ decomposition-mapping
|
||||
;~ pose
|
||||
:: a tag and a list of characters to decompose to
|
||||
%+ stag ~
|
||||
;~ plug
|
||||
;~ pose
|
||||
(stag ~ (ifix [gal ;~(plug gar ace)] decomp-tag))
|
||||
(easy ~)
|
||||
==
|
||||
(cook |=(a/(list @c) a) (most ace hex))
|
||||
==
|
||||
:: no decomposition information
|
||||
(easy ~)
|
||||
%- punt :: optional
|
||||
:: a tag and a list of characters to decompose to
|
||||
;~ plug
|
||||
(punt (ifix [gal ;~(plug gar ace)] decomp-tag))
|
||||
(cook |=(a/(list @c) a) (most ace hex))
|
||||
==
|
||||
::
|
||||
++ decomp-tag
|
||||
;~ pose
|
||||
(cold %font (jest 'font'))
|
||||
(cold %no-break (jest 'noBreak'))
|
||||
(cold %initial (jest 'initial'))
|
||||
(cold %medial (jest 'medial'))
|
||||
(cold %final (jest 'final'))
|
||||
(cold %isolated (jest 'isolated'))
|
||||
(cold %circle (jest 'circle'))
|
||||
(cold %super (jest 'super'))
|
||||
(cold %sub (jest 'sub'))
|
||||
(cold %vertical (jest 'vertical'))
|
||||
(cold %wide (jest 'wide'))
|
||||
(cold %narrow (jest 'narrow'))
|
||||
(cold %small (jest 'small'))
|
||||
(cold %square (jest 'square'))
|
||||
(cold %fraction (jest 'fraction'))
|
||||
(cold %compat (jest 'compat'))
|
||||
==
|
||||
%+ sear (soft decomp-tag:unicode-data)
|
||||
%+ cook |=(a/term ?+(a a $nobreak %no-break)) ::REVIEW or just change the type
|
||||
:(cook crip cass (star alf))
|
||||
::
|
||||
++ string-number
|
||||
%+ cook
|
||||
|=(a/tape a)
|
||||
(star ;~(pose nud fas hep))
|
||||
::
|
||||
++ yes-or-no
|
||||
;~ pose
|
||||
(cold %.y (jest 'Y'))
|
||||
(cold %.n (jest 'N'))
|
||||
==
|
||||
::
|
||||
++ optional-hex
|
||||
;~ pose
|
||||
(stag ~ hex)
|
||||
(easy ~)
|
||||
==
|
||||
--
|
||||
--
|
||||
++ grad %txt
|
||||
|
||||
@ -5,38 +5,23 @@
|
||||
++ line
|
||||
:> an individual codepoint definition
|
||||
:>
|
||||
:> code: the codepoint in hexadecimal format
|
||||
:> name: the character name
|
||||
:> gen: the type of character this is
|
||||
:> can: the canonical combining class for ordering algorithms
|
||||
:> bidi: the bidirectional category of this character
|
||||
:> de: the character decomposition mapping
|
||||
:> decimal: the decimal digit value (or ~)
|
||||
:> digit: the digit value, covering non decimal radix forms
|
||||
:> numeric: the numeric value, including fractions
|
||||
:> mirrored: whether char is mirrored in bidirectional text
|
||||
:> old-name: unicode 1.0 compatibility name
|
||||
:> iso: iso 10646 comment field
|
||||
:> up: uppercase mapping codepoint
|
||||
:> low: lowercase mapping codepoint
|
||||
:> title: titlecase mapping codepoint
|
||||
:>
|
||||
$: code/@c
|
||||
name/tape
|
||||
gen/general
|
||||
can/@ud
|
||||
bi/bidi
|
||||
de/decomp
|
||||
:: todo: decimal/digit/numeric need to be parsed.
|
||||
decimal/tape
|
||||
digit/tape
|
||||
numeric/tape
|
||||
mirrored/?
|
||||
old-name/tape
|
||||
iso/tape
|
||||
up/(unit @c)
|
||||
low/(unit @c)
|
||||
title/(unit @c)
|
||||
$: code/@c :< codepoint in hexadecimal format
|
||||
name/tape :< character name
|
||||
gen/general :< type of character this is
|
||||
:> canonical combining class for ordering algorithms
|
||||
can/@ud
|
||||
bi/bidi :< bidirectional category of this character
|
||||
de/decomp :< character decomposition mapping
|
||||
:: todo: decimal/digit/numeric need to be parsed.
|
||||
decimal/tape :< decimal digit value (or ~)
|
||||
digit/tape :< digit value, covering non decimal radix forms
|
||||
numeric/tape :< numeric value, including fractions
|
||||
mirrored/? :< whether char is mirrored in bidirectional text
|
||||
old-name/tape :< unicode 1.0 compatibility name
|
||||
iso/tape :< iso 10646 comment field
|
||||
up/(unit @c) :< uppercase mapping codepoint
|
||||
low/(unit @c) :< lowercase mapping codepoint
|
||||
title/(unit @c) :< titlecase mapping codepoint
|
||||
==
|
||||
::
|
||||
++ general
|
||||
|
||||
Loading…
Reference in New Issue
Block a user