summaryrefslogtreecommitdiff
path: root/data
diff options
context:
space:
mode:
Diffstat (limited to 'data')
-rw-r--r--data/Makefile62
-rw-r--r--data/charwidths.jl161
-rw-r--r--data/data_generator.rb18
3 files changed, 237 insertions, 4 deletions
diff --git a/data/Makefile b/data/Makefile
new file mode 100644
index 0000000..409e03a
--- /dev/null
+++ b/data/Makefile
@@ -0,0 +1,62 @@
+# Unicode data generation rules. Except for the test data files, most
+# users will not use these Makefile rules, which are primarily to re-generate
+# unicode_data.c when we get a new Unicode version or charwidth data; they
+# require ruby, fontforge, and julia to be installed.
+
+# programs
+CURL=curl
+RUBY=ruby
+PERL=perl
+MAKE=make
+JULIA=julia
+CURLFLAGS = --retry 5 --location
+
+# use JuliaLang caching (https://github.com/staticfloat/cache.julialang.org)
+# so that Travis builds do not depend on anyone's flaky servers but our own
+URLCACHE=https://cache.e.ip.saba.us/
+
+.PHONY: clean
+
+.DELETE_ON_ERROR:
+
+utf8proc_data.c.new: data_generator.rb UnicodeData.txt GraphemeBreakProperty.txt DerivedCoreProperties.txt CompositionExclusions.txt CaseFolding.txt CharWidths.txt
+ $(RUBY) data_generator.rb < UnicodeData.txt > $@
+
+# GNU Unifont version for font-metric calculations:
+UNIFONT_VERSION=7.0.06
+
+unifont-$(UNIFONT_VERSION).ttf:
+ $(CURL) $(CURLFLAGS) -o $@ $(URLCACHE)http://unifoundry.com/pub/unifont-$(UNIFONT_VERSION)/font-builds/unifont-$(UNIFONT_VERSION).ttf
+
+unifont_upper-$(UNIFONT_VERSION).ttf:
+ $(CURL) $(CURLFLAGS) -o $@ $(URLCACHE)http://unifoundry.com/pub/unifont-$(UNIFONT_VERSION)/font-builds/unifont_upper-$(UNIFONT_VERSION).ttf
+
+CharWidths.txt: charwidths.jl unifont-$(UNIFONT_VERSION).ttf unifont_upper-$(UNIFONT_VERSION).ttf EastAsianWidth.txt
+ UNIFONT_VERSION=$(UNIFONT_VERSION) $(JULIA) charwidths.jl > $@
+
+UnicodeData.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/UnicodeData.txt
+
+EastAsianWidth.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt
+
+GraphemeBreakProperty.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakProperty.txt
+
+DerivedCoreProperties.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/DerivedCoreProperties.txt
+
+CompositionExclusions.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/CompositionExclusions.txt
+
+CaseFolding.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
+
+NormalizationTest.txt:
+ $(CURL) $(CURLFLAGS) -o $@ -O $(URLCACHE)http://www.unicode.org/Public/UNIDATA/NormalizationTest.txt
+
+GraphemeBreakTest.txt:
+ $(CURL) $(CURLFLAGS) $(URLCACHE)http://www.unicode.org/Public/UCD/latest/ucd/auxiliary/GraphemeBreakTest.txt | $(PERL) -pe 's,÷,/,g;s,×,+,g' > $@
+
+clean:
+ rm -f UnicodeData.txt EastAsianWidth.txt DerivedCoreProperties.txt CompositionExclusions.txt CaseFolding.txt NormalizationTest.txt GraphemeBreakTest.txt CharWidths.txt unifont*.ttf unifont*.sfd
diff --git a/data/charwidths.jl b/data/charwidths.jl
new file mode 100644
index 0000000..72d695f
--- /dev/null
+++ b/data/charwidths.jl
@@ -0,0 +1,161 @@
+# Following work by @jiahao, we compute character widths using a combination of
+# * advance widths from GNU Unifont (advance width 512 = 1 en)
+# * UAX 11: East Asian Width
+# * a few exceptions as needed
+# Adapted from http://nbviewer.ipython.org/gist/jiahao/07e8b08bf6d8671e9734
+#
+# Requires Julia (obviously) and FontForge.
+
+#############################################################################
+# Julia 0.3/0.4 compatibility (taken from Compat package)
+if VERSION < v"0.4.0-dev+1419"
+ const UInt16 = Uint16
+end
+
+CharWidths = Dict{Int,Int}()
+
+#############################################################################
+# Widths from GNU Unifont
+
+universion=get(ENV, "UNIFONT_VERSION", "7.0.06")
+for fontfile in ["unifont-$universion", "unifont_upper-$universion"]
+ isfile("$fontfile.ttf") || download("http://unifoundry.com/pub/unifont-$universion/font-builds/$fontfile.ttf", "$fontfile.ttf")
+ isfile("$fontfile.sfd") || run(`fontforge -lang=ff -c "Open(\"$fontfile.ttf\");Save(\"$fontfile.sfd\");Quit(0);"`)
+end
+
+#Read sfdfile for character widths
+function parsesfd(filename::String, CharWidths::Dict{Int,Int}=Dict{Int,Int}())
+ state=:seekchar
+ lineno = 0
+ for line in readlines(open(filename))
+ lineno += 1
+ if state==:seekchar #StartChar: nonmarkingreturn
+ if contains(line, "StartChar: ")
+ codepoint = nothing
+ width = nothing
+ state = :readdata
+ end
+ elseif state==:readdata #Encoding: 65538 -1 2, Width: 1024
+ contains(line, "Encoding:") && (codepoint = int(split(line)[3]))
+ contains(line, "Width:") && (width = int(split(line)[2]))
+ if codepoint!=nothing && width!=nothing && codepoint >= 0
+ CharWidths[codepoint]=div(width, 512) # 512 units to the en
+ state = :seekchar
+ end
+ end
+ end
+ CharWidths
+end
+CharWidths=parsesfd("unifont-$universion.sfd", CharWidths)
+CharWidths=parsesfd("unifont_upper-$universion.sfd", CharWidths)
+
+#############################################################################
+# Widths from UAX #11: East Asian Width
+# .. these take precedence over the Unifont width for all codepoints
+# listed explicitly as wide/full/narrow/half-width
+
+isfile("EastAsianWidth.txt") || download("http://www.unicode.org/Public/UNIDATA/EastAsianWidth.txt", "EastAsianWidth.txt")
+for line in readlines(open("EastAsianWidth.txt"))
+ #Strip comments
+ line[1] == '#' && continue
+ precomment = split(line, '#')[1]
+ #Parse code point range and width code
+ tokens = split(precomment, ';')
+ length(tokens) >= 2 || continue
+ charrange = tokens[1]
+ width = strip(tokens[2])
+ #Parse code point range into Julia UnitRange
+ rangetokens = split(charrange, "..")
+ charstart = uint32("0x"*rangetokens[1])
+ charend = uint32("0x"*rangetokens[length(rangetokens)>1 ? 2 : 1])
+
+ #Assign widths
+ for c in charstart:charend
+ if width=="W" || width=="F" # wide or full
+ CharWidths[c]=2
+ elseif width=="Na"|| width=="H" # narrow or half
+ CharWidths[c]=1
+ end
+ end
+end
+
+#############################################################################
+# A few exceptions to the above cases, found by manual comparison
+# to other wcwidth functions and similar checks.
+
+# Use ../libutf8proc for category codes, rather than the one in Julia,
+# to minimize bootstrapping complexity when a new version of Unicode comes out.
+function catcode(c)
+ uint(c) > 0x10FFFF && return 0x0000 # see utf8proc_get_property docs
+ return unsafe_load(ccall((:utf8proc_get_property,"../libutf8proc"), Ptr{UInt16}, (Int32,), c))
+end
+
+# use Base.UTF8proc module to get category codes constants, since
+# we aren't goint to change these in utf8proc.
+import Base.UTF8proc
+
+for c in keys(CharWidths)
+ cat = catcode(c)
+
+ # make sure format control character (category Cf) have width 0,
+ # except for the Arabic characters 0x06xx (see unicode std 6.2, sec. 8.2)
+ if cat==UTF8proc.UTF8PROC_CATEGORY_CF && c ∉ [0x0601,0x0602,0x0603,0x06dd]
+ CharWidths[c]=0
+ end
+
+ # Unifont has nonzero width for a number of non-spacing combining
+ # characters, e.g. (in 7.0.06): f84,17b4,17b5,180b,180d,2d7f, and
+ # the variation selectors
+ if cat==UTF8proc.UTF8PROC_CATEGORY_MN
+ CharWidths[c]=0
+ end
+
+ # We also assign width of zero to unassigned and private-use
+ # codepoints (Unifont includes ConScript Unicode Registry PUA fonts,
+ # but since these are nonstandard it seems questionable to recognize them).
+ if cat==UTF8proc.UTF8PROC_CATEGORY_CO || cat==UTF8proc.UTF8PROC_CATEGORY_CN
+ CharWidths[c]=0
+ end
+
+ # for some reason, Unifont has width-2 glyphs for ASCII control chars
+ if cat==UTF8proc.UTF8PROC_CATEGORY_CC
+ CharWidths[c]=0
+ end
+end
+
+#By definition, should have zero width (on the same line)
+#0x002028 '
' category: Zl name: LINE SEPARATOR/
+#0x002029 '
' category: Zp name: PARAGRAPH SEPARATOR/
+CharWidths[0x2028]=0
+CharWidths[0x2029]=0
+
+#By definition, should be narrow = width of 1 en space
+#0x00202f ' ' category: Zs name: NARROW NO-BREAK SPACE/
+CharWidths[0x202f]=1
+
+#By definition, should be wide = width of 1 em space
+#0x002001 ' ' category: Zs name: EM QUAD/
+#0x002003 ' ' category: Zs name: EM SPACE/
+CharWidths[0x2001]=2
+CharWidths[0x2003]=2
+
+#############################################################################
+# Output (to a file or pipe) for processing by data_generator.rb
+# ... don't bother to output zero widths since that will be the default.
+
+firstc = 0x000000
+lastv = 0
+uhex(c) = uppercase(hex(c,4))
+for c in 0x0000:0x110000
+ v = get(CharWidths, c, 0)
+ if v != lastv || c == 0x110000
+ v < 4 || error("invalid charwidth $v for $c")
+ if firstc+1 < c
+ println(uhex(firstc), "..", uhex(c-1), "; ", lastv)
+ else
+ println(uhex(firstc), "; ", lastv)
+ end
+ firstc = c
+ lastv = v
+ end
+end
diff --git a/data/data_generator.rb b/data/data_generator.rb
index 1439038..b24caad 100644
--- a/data/data_generator.rb
+++ b/data/data_generator.rb
@@ -85,14 +85,23 @@ $grapheme_boundclass_list.each_line do |entry|
end
end
+$charwidth_list = File.read("CharWidths.txt")
+$charwidth = Hash.new(0)
+$charwidth_list.each_line do |entry|
+ if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s*;\s*([0-9]+)/
+ $1.hex.upto($2.hex) { |e2| $charwidth[e2] = $3.to_i }
+ elsif entry =~ /^([0-9A-F]+)\s*;\s*([0-9]+)/
+ $charwidth[$1.hex] = $2.to_i
+ end
+end
+
$exclusions = File.read("CompositionExclusions.txt")[/# \(1\) Script Specifics.*?# Total code points:/m]
$exclusions = $exclusions.chomp.split("\n").collect { |e| e.hex }
$excl_version = File.read("CompositionExclusions.txt")[/# \(2\) Post Composition Version precomposed characters.*?# Total code points:/m]
$excl_version = $excl_version.chomp.split("\n").collect { |e| e.hex }
-$case_folding_string = File.open("CaseFolding.txt").read
-
+$case_folding_string = File.open("CaseFolding.txt", :encoding => 'utf-8').read
$case_folding = {}
$case_folding_string.chomp.split("\n").each do |line|
next unless line =~ /([0-9A-F]+); [CFS]; ([0-9A-F ]+);/i
@@ -172,7 +181,8 @@ class UnicodeChar
"#{$exclusions.include?(code) or $excl_version.include?(code)}, " <<
"#{$ignorable.include?(code)}, " <<
"#{%W[Zl Zp Cc Cf].include?(category) and not [0x200C, 0x200D].include?(category)}, " <<
- "#{$grapheme_boundclass[code]}},\n"
+ "#{$grapheme_boundclass[code]}, " <<
+ "#{$charwidth[code]}},\n"
end
end
@@ -295,7 +305,7 @@ end
$stdout << "};\n\n"
$stdout << "const utf8proc_property_t utf8proc_properties[] = {\n"
-$stdout << " {0, 0, 0, 0, NULL, NULL, -1, -1, -1, -1, -1, false,false,false,false, UTF8PROC_BOUNDCLASS_OTHER},\n"
+$stdout << " {0, 0, 0, 0, NULL, NULL, -1, -1, -1, -1, -1, false,false,false,false, UTF8PROC_BOUNDCLASS_OTHER, 0},\n"
properties.each { |line|
$stdout << line
}