summaryrefslogtreecommitdiff
path: root/data/data_generator.rb
diff options
context:
space:
mode:
authorSteven G. Johnson <stevenj@mit.edu>2018-07-24 13:25:51 -0400
committerSteven G. Johnson <stevenj@mit.edu>2018-07-24 13:25:51 -0400
commite0295be467d15e7abec2af275bcca30dc816bc9e (patch)
tree370dcfc22d38a3ed302c1beeafa2ee5bf1b52db7 /data/data_generator.rb
parent98e5529a0a6cd4dd09a8885029253f26c677c85f (diff)
parentd4a58cfec5345bbb2bb0db1e85172a8cff278da7 (diff)
downloadlibutf8proc-e0295be467d15e7abec2af275bcca30dc816bc9e.tar.gz
libutf8proc-e0295be467d15e7abec2af275bcca30dc816bc9e.tar.bz2
Merge branch 'master' of https://github.com/JuliaLang/utf8proc
Diffstat (limited to 'data/data_generator.rb')
-rw-r--r--data/data_generator.rb13
1 files changed, 13 insertions, 0 deletions
diff --git a/data/data_generator.rb b/data/data_generator.rb
index 9932f74..8bd87e8 100644
--- a/data/data_generator.rb
+++ b/data/data_generator.rb
@@ -87,6 +87,19 @@ $grapheme_boundclass_list.each_line do |entry|
end
end
+$emoji_data_list = File.read("emoji-data.txt")
+$emoji_data_list.each_line do |entry|
+ if entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s*;\s*Extended_Pictographic\W/
+ $1.hex.upto($2.hex) { |e2| $grapheme_boundclass[e2] = "UTF8PROC_BOUNDCLASS_EXTENDED_PICTOGRAPHIC" }
+ elsif entry =~ /^([0-9A-F]+)\s*;\s*Extended_Pictographic\W/
+ $grapheme_boundclass[$1.hex] = "UTF8PROC_BOUNDCLASS_EXTENDED_PICTOGRAPHIC"
+ elsif entry =~ /^([0-9A-F]+)\.\.([0-9A-F]+)\s*;\s*Emoji_Modifier\W/
+ $1.hex.upto($2.hex) { |e2| $grapheme_boundclass[e2] = "UTF8PROC_BOUNDCLASS_EXTEND" }
+ elsif entry =~ /^([0-9A-F]+)\s*;\s*Emoji_Modifier\W/
+ $grapheme_boundclass[$1.hex] = "UTF8PROC_BOUNDCLASS_EXTEND"
+ end
+end
+
$charwidth_list = File.read("CharWidths.txt")
$charwidth = Hash.new(0)
$charwidth_list.each_line do |entry|