Mercurial > minori

annotate dep/utf8proc/data/charwidths.jl @ 345:d085c05247bd

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
build: fix windows build
author Paper <paper@paper.us.eu.org>
date Thu, 20 Jun 2024 07:39:01 -0400
parents 1faa72660932
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
343
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
1 # Following work by @jiahao, we compute character widths using a combination of
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
2 # * character category
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
3 # * UAX 11: East Asian Width
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
4 # * a few exceptions as needed
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
5 # Adapted from http://nbviewer.ipython.org/gist/jiahao/07e8b08bf6d8671e9734
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
6 #
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
7 # We used to also use data from GNU Unifont, but that has proven unreliable
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
8 # and unlikely to match widths assumed by terminals.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
9 #
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
10 # Requires Julia (obviously) and FontForge.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
11
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
12 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
13 CharWidths = Dict{Int,Int}()
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
14
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
15 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
16 # Use ../libutf8proc for category codes, rather than the one in Julia,
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
17 # to minimize bootstrapping complexity when a new version of Unicode comes out.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
18 catcode(c) = ccall((:utf8proc_category,"../libutf8proc"), Cint, (Int32,), c)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
19
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
20 # utf8proc category constants (must match h)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
21 const UTF8PROC_CATEGORY_CN = 0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
22 const UTF8PROC_CATEGORY_LU = 1
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
23 const UTF8PROC_CATEGORY_LL = 2
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
24 const UTF8PROC_CATEGORY_LT = 3
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
25 const UTF8PROC_CATEGORY_LM = 4
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
26 const UTF8PROC_CATEGORY_LO = 5
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
27 const UTF8PROC_CATEGORY_MN = 6
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
28 const UTF8PROC_CATEGORY_MC = 7
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
29 const UTF8PROC_CATEGORY_ME = 8
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
30 const UTF8PROC_CATEGORY_ND = 9
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
31 const UTF8PROC_CATEGORY_NL = 10
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
32 const UTF8PROC_CATEGORY_NO = 11
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
33 const UTF8PROC_CATEGORY_PC = 12
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
34 const UTF8PROC_CATEGORY_PD = 13
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
35 const UTF8PROC_CATEGORY_PS = 14
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
36 const UTF8PROC_CATEGORY_PE = 15
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
37 const UTF8PROC_CATEGORY_PI = 16
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
38 const UTF8PROC_CATEGORY_PF = 17
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
39 const UTF8PROC_CATEGORY_PO = 18
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
40 const UTF8PROC_CATEGORY_SM = 19
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
41 const UTF8PROC_CATEGORY_SC = 20
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
42 const UTF8PROC_CATEGORY_SK = 21
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
43 const UTF8PROC_CATEGORY_SO = 22
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
44 const UTF8PROC_CATEGORY_ZS = 23
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
45 const UTF8PROC_CATEGORY_ZL = 24
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
46 const UTF8PROC_CATEGORY_ZP = 25
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
47 const UTF8PROC_CATEGORY_CC = 26
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
48 const UTF8PROC_CATEGORY_CF = 27
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
49 const UTF8PROC_CATEGORY_CS = 28
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
50 const UTF8PROC_CATEGORY_CO = 29
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
51
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
52 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
53 # Use a default width of 1 for all character categories that are
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
54 # letter/symbol/number-like, as well as for unassigned/private-use chars.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
55 # This can be overridden by UAX 11
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
56 # below, but provides a useful nonzero fallback for new codepoints when
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
57 # a new Unicode version has been released but Unifont hasn't been updated yet.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
58
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
59 zerowidth = Set{Int}() # categories that may contain zero-width chars
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
60 push!(zerowidth, UTF8PROC_CATEGORY_MN)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
61 push!(zerowidth, UTF8PROC_CATEGORY_MC)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
62 push!(zerowidth, UTF8PROC_CATEGORY_ME)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
63 # push!(zerowidth, UTF8PROC_CATEGORY_SK) # see issue #167
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
64 push!(zerowidth, UTF8PROC_CATEGORY_ZL)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
65 push!(zerowidth, UTF8PROC_CATEGORY_ZP)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
66 push!(zerowidth, UTF8PROC_CATEGORY_CC)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
67 push!(zerowidth, UTF8PROC_CATEGORY_CF)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
68 push!(zerowidth, UTF8PROC_CATEGORY_CS)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
69 for c in 0x0000:0x110000
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
70 if catcode(c) ∉ zerowidth
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
71 CharWidths[c] = 1
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
72 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
73 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
74
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
75 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
76 # Widths from UAX #11: East Asian Width
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
77 # .. these take precedence for all codepoints
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
78 # listed explicitly as wide/full/narrow/half-width
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
79
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
80 for line in readlines(open("EastAsianWidth.txt"))
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
81 #Strip comments
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
82 (isempty(line) || line[1] == '#') && continue
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
83 precomment = split(line, '#')[1]
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
84 #Parse code point range and width code
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
85 tokens = split(precomment, ';')
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
86 length(tokens) >= 2 || continue
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
87 charrange = tokens[1]
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
88 width = strip(tokens[2])
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
89 #Parse code point range into Julia UnitRange
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
90 rangetokens = split(charrange, "..")
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
91 charstart = parse(UInt32, "0x"*rangetokens[1])
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
92 charend = parse(UInt32, "0x"*rangetokens[length(rangetokens)>1 ? 2 : 1])
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
93
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
94 #Assign widths
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
95 for c in charstart:charend
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
96 if width=="W" || width=="F" # wide or full
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
97 CharWidths[c]=2
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
98 elseif width=="Na"|| width=="H"
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
99 CharWidths[c]=1
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
100 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
101 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
102 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
103
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
104 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
105 # A few exceptions to the above cases, found by manual comparison
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
106 # to other wcwidth functions and similar checks.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
107
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
108 for c in keys(CharWidths)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
109 cat = catcode(c)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
110
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
111 # make sure format control character (category Cf) have width 0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
112 # (some of these, like U+0601, can have a width in some cases
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
113 # but normally act like prepended combining marks. U+fff9 etc
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
114 # are also odd, but have zero width in typical terminal contexts)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
115 if cat==UTF8PROC_CATEGORY_CF
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
116 CharWidths[c]=0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
117 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
118
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
119 # Unifont has nonzero width for a number of non-spacing combining
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
120 # characters, e.g. (in 7.0.06): f84,17b4,17b5,180b,180d,2d7f, and
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
121 # the variation selectors
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
122 if cat==UTF8PROC_CATEGORY_MN
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
123 CharWidths[c]=0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
124 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
125
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
126 # We also assign width of one to unassigned and private-use
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
127 # codepoints (Unifont includes ConScript Unicode Registry PUA fonts,
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
128 # but since these are nonstandard it seems questionable to use Unifont metrics;
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
129 # if they are printed as the replacement character U+FFFD they will have width 1).
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
130 if cat==UTF8PROC_CATEGORY_CO || cat==UTF8PROC_CATEGORY_CN
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
131 CharWidths[c]=1
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
132 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
133
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
134 # for some reason, Unifont has width-2 glyphs for ASCII control chars
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
135 if cat==UTF8PROC_CATEGORY_CC
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
136 CharWidths[c]=0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
137 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
138 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
139
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
140 #Soft hyphen is typically printed as a hyphen (-) in terminals.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
141 CharWidths[0x00ad]=1
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
142
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
143 #By definition, should have zero width (on the same line)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
144 #0x002028 '
' category: Zl name: LINE SEPARATOR/
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
145 #0x002029 '
' category: Zp name: PARAGRAPH SEPARATOR/
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
146 CharWidths[0x2028]=0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
147 CharWidths[0x2029]=0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
148
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
149 #############################################################################
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
150 # Output (to a file or pipe) for processing by data_generator.rb,
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
151 # encoded as a sequence of intervals.
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
152
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
153 firstc = 0x000000
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
154 lastv = 0
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
155 uhex(c) = uppercase(string(c,base=16,pad=4))
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
156 for c in 0x0000:0x110000
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
157 global firstc, lastv
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
158 v = get(CharWidths, c, 0)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
159 if v != lastv || c == 0x110000
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
160 v < 4 || error("invalid charwidth $v for $c")
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
161 if firstc+1 < c
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
162 println(uhex(firstc), "..", uhex(c-1), "; ", lastv)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
163 else
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
164 println(uhex(firstc), "; ", lastv)
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
165 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
166 firstc = c
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
167 lastv = v
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
168 end
1faa72660932 *: transfer back to cmake from autotools
Paper <paper@paper.us.eu.org>
parents:
diff changeset
169 end