Mercurial > minori

comparison dep/utf8proc/data/charwidths.jl @ 343:1faa72660932

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
*: transfer back to cmake from autotools autotools just made lots of things more complicated than they should have and many things broke (i.e. translations)
author Paper <paper@paper.us.eu.org>
date Thu, 20 Jun 2024 05:56:06 -0400
parents
children
comparison
equal deleted inserted replaced
342:adb79bdde329 343:1faa72660932
1 # Following work by @jiahao, we compute character widths using a combination of
2 # * character category
3 # * UAX 11: East Asian Width
4 # * a few exceptions as needed
5 # Adapted from http://nbviewer.ipython.org/gist/jiahao/07e8b08bf6d8671e9734
6 #
7 # We used to also use data from GNU Unifont, but that has proven unreliable
8 # and unlikely to match widths assumed by terminals.
9 #
10 # Requires Julia (obviously) and FontForge.
11
12 #############################################################################
13 CharWidths = Dict{Int,Int}()
14
15 #############################################################################
16 # Use ../libutf8proc for category codes, rather than the one in Julia,
17 # to minimize bootstrapping complexity when a new version of Unicode comes out.
18 catcode(c) = ccall((:utf8proc_category,"../libutf8proc"), Cint, (Int32,), c)
19
20 # utf8proc category constants (must match h)
21 const UTF8PROC_CATEGORY_CN = 0
22 const UTF8PROC_CATEGORY_LU = 1
23 const UTF8PROC_CATEGORY_LL = 2
24 const UTF8PROC_CATEGORY_LT = 3
25 const UTF8PROC_CATEGORY_LM = 4
26 const UTF8PROC_CATEGORY_LO = 5
27 const UTF8PROC_CATEGORY_MN = 6
28 const UTF8PROC_CATEGORY_MC = 7
29 const UTF8PROC_CATEGORY_ME = 8
30 const UTF8PROC_CATEGORY_ND = 9
31 const UTF8PROC_CATEGORY_NL = 10
32 const UTF8PROC_CATEGORY_NO = 11
33 const UTF8PROC_CATEGORY_PC = 12
34 const UTF8PROC_CATEGORY_PD = 13
35 const UTF8PROC_CATEGORY_PS = 14
36 const UTF8PROC_CATEGORY_PE = 15
37 const UTF8PROC_CATEGORY_PI = 16
38 const UTF8PROC_CATEGORY_PF = 17
39 const UTF8PROC_CATEGORY_PO = 18
40 const UTF8PROC_CATEGORY_SM = 19
41 const UTF8PROC_CATEGORY_SC = 20
42 const UTF8PROC_CATEGORY_SK = 21
43 const UTF8PROC_CATEGORY_SO = 22
44 const UTF8PROC_CATEGORY_ZS = 23
45 const UTF8PROC_CATEGORY_ZL = 24
46 const UTF8PROC_CATEGORY_ZP = 25
47 const UTF8PROC_CATEGORY_CC = 26
48 const UTF8PROC_CATEGORY_CF = 27
49 const UTF8PROC_CATEGORY_CS = 28
50 const UTF8PROC_CATEGORY_CO = 29
51
52 #############################################################################
53 # Use a default width of 1 for all character categories that are
54 # letter/symbol/number-like, as well as for unassigned/private-use chars.
55 # This can be overridden by UAX 11
56 # below, but provides a useful nonzero fallback for new codepoints when
57 # a new Unicode version has been released but Unifont hasn't been updated yet.
58
59 zerowidth = Set{Int}() # categories that may contain zero-width chars
60 push!(zerowidth, UTF8PROC_CATEGORY_MN)
61 push!(zerowidth, UTF8PROC_CATEGORY_MC)
62 push!(zerowidth, UTF8PROC_CATEGORY_ME)
63 # push!(zerowidth, UTF8PROC_CATEGORY_SK) # see issue #167
64 push!(zerowidth, UTF8PROC_CATEGORY_ZL)
65 push!(zerowidth, UTF8PROC_CATEGORY_ZP)
66 push!(zerowidth, UTF8PROC_CATEGORY_CC)
67 push!(zerowidth, UTF8PROC_CATEGORY_CF)
68 push!(zerowidth, UTF8PROC_CATEGORY_CS)
69 for c in 0x0000:0x110000
70 if catcode(c) ∉ zerowidth
71 CharWidths[c] = 1
72 end
73 end
74
75 #############################################################################
76 # Widths from UAX #11: East Asian Width
77 # .. these take precedence for all codepoints
78 # listed explicitly as wide/full/narrow/half-width
79
80 for line in readlines(open("EastAsianWidth.txt"))
81 #Strip comments
82 (isempty(line) || line[1] == '#') && continue
83 precomment = split(line, '#')[1]
84 #Parse code point range and width code
85 tokens = split(precomment, ';')
86 length(tokens) >= 2 || continue
87 charrange = tokens[1]
88 width = strip(tokens[2])
89 #Parse code point range into Julia UnitRange
90 rangetokens = split(charrange, "..")
91 charstart = parse(UInt32, "0x"*rangetokens[1])
92 charend = parse(UInt32, "0x"*rangetokens[length(rangetokens)>1 ? 2 : 1])
93
94 #Assign widths
95 for c in charstart:charend
96 if width=="W" || width=="F" # wide or full
97 CharWidths[c]=2
98 elseif width=="Na"|| width=="H"
99 CharWidths[c]=1
100 end
101 end
102 end
103
104 #############################################################################
105 # A few exceptions to the above cases, found by manual comparison
106 # to other wcwidth functions and similar checks.
107
108 for c in keys(CharWidths)
109 cat = catcode(c)
110
111 # make sure format control character (category Cf) have width 0
112 # (some of these, like U+0601, can have a width in some cases
113 # but normally act like prepended combining marks. U+fff9 etc
114 # are also odd, but have zero width in typical terminal contexts)
115 if cat==UTF8PROC_CATEGORY_CF
116 CharWidths[c]=0
117 end
118
119 # Unifont has nonzero width for a number of non-spacing combining
120 # characters, e.g. (in 7.0.06): f84,17b4,17b5,180b,180d,2d7f, and
121 # the variation selectors
122 if cat==UTF8PROC_CATEGORY_MN
123 CharWidths[c]=0
124 end
125
126 # We also assign width of one to unassigned and private-use
127 # codepoints (Unifont includes ConScript Unicode Registry PUA fonts,
128 # but since these are nonstandard it seems questionable to use Unifont metrics;
129 # if they are printed as the replacement character U+FFFD they will have width 1).
130 if cat==UTF8PROC_CATEGORY_CO || cat==UTF8PROC_CATEGORY_CN
131 CharWidths[c]=1
132 end
133
134 # for some reason, Unifont has width-2 glyphs for ASCII control chars
135 if cat==UTF8PROC_CATEGORY_CC
136 CharWidths[c]=0
137 end
138 end
139
140 #Soft hyphen is typically printed as a hyphen (-) in terminals.
141 CharWidths[0x00ad]=1
142
143 #By definition, should have zero width (on the same line)
144 #0x002028 '
' category: Zl name: LINE SEPARATOR/
145 #0x002029 '
' category: Zp name: PARAGRAPH SEPARATOR/
146 CharWidths[0x2028]=0
147 CharWidths[0x2029]=0
148
149 #############################################################################
150 # Output (to a file or pipe) for processing by data_generator.rb,
151 # encoded as a sequence of intervals.
152
153 firstc = 0x000000
154 lastv = 0
155 uhex(c) = uppercase(string(c,base=16,pad=4))
156 for c in 0x0000:0x110000
157 global firstc, lastv
158 v = get(CharWidths, c, 0)
159 if v != lastv || c == 0x110000
160 v < 4 || error("invalid charwidth $v for $c")
161 if firstc+1 < c
162 println(uhex(firstc), "..", uhex(c-1), "; ", lastv)
163 else
164 println(uhex(firstc), "; ", lastv)
165 end
166 firstc = c
167 lastv = v
168 end
169 end