summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authornobu <nobu@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2016-07-14 16:32:13 +0000
committernobu <nobu@b2dd03c8-39d4-4d8f-98ff-823fe69b080e>2016-07-14 16:32:13 +0000
commit0fd7666d57f9919962ae28ae7a6758451e10cd11 (patch)
tree5cf12995db62290e280ba8c19b979330c4cb41fb
parent230a91c133faa009f95ea3a47d075921c7d9c467 (diff)
enc-unicode.rb: check Unicode version
* tool/enc-unicode.rb (data_foreach): check Unicode version in data files, and yield each lines. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@55685 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
-rwxr-xr-xtool/enc-unicode.rb32
1 files changed, 25 insertions, 7 deletions
diff --git a/tool/enc-unicode.rb b/tool/enc-unicode.rb
index a8075a34a64..4eea2108ec2 100755
--- a/tool/enc-unicode.rb
+++ b/tool/enc-unicode.rb
@@ -15,10 +15,11 @@ if ARGV[0] == "--header"
ARGV.shift
end
unless ARGV.size == 1
- $stderr.puts "Usage: #{$0} data_directory"
- exit(1)
+ abort "Usage: #{$0} data_directory"
end
+$unicode_version = File.basename(ARGV[0])[/\A[.\d]+\z/]
+
POSIX_NAMES = %w[NEWLINE Alpha Blank Cntrl Digit Graph Lower Print Punct Space Upper XDigit Word Alnum ASCII]
def pair_codepoints(codepoints)
@@ -137,7 +138,7 @@ def parse_scripts(data, categories)
cps = []
names = {}
files.each do |file|
- IO.foreach(get_file(file[:fn])) do |line|
+ data_foreach(file[:fn]) do |line|
if /^# Total code points: / =~ line
data[current] = cps
categories[current] = file[:title]
@@ -158,12 +159,12 @@ end
def parse_aliases(data)
kv = {}
- IO.foreach(get_file('PropertyAliases.txt')) do |line|
+ data_foreach('PropertyAliases.txt') do |line|
next unless /^(\w+)\s*; (\w+)/ =~ line
data[$1] = data[$2]
kv[normalize_propname($1)] = normalize_propname($2)
end
- IO.foreach(get_file('PropertyValueAliases.txt')) do |line|
+ data_foreach('PropertyValueAliases.txt') do |line|
next unless /^(sc|gc)\s*; (\w+)\s*; (\w+)(?:\s*; (\w+))?/ =~ line
if $1 == 'gc'
data[$3] = data[$2]
@@ -188,7 +189,7 @@ def parse_age(data)
last_constname = nil
cps = []
ages = []
- IO.foreach(get_file('DerivedAge.txt')) do |line|
+ data_foreach('DerivedAge.txt') do |line|
if /^# Total code points: / =~ line
constname = constantize_agename(current)
# each version matches all previous versions
@@ -211,7 +212,7 @@ def parse_block(data)
last_constname = nil
cps = []
blocks = []
- IO.foreach(get_file('Blocks.txt')) do |line|
+ data_foreach('Blocks.txt') do |line|
if /^([0-9a-fA-F]+)\.\.([0-9a-fA-F]+);\s*(.*)/ =~ line
cps = ($1.to_i(16)..$2.to_i(16)).to_a
constname = constantize_blockname($3)
@@ -276,6 +277,23 @@ def get_file(name)
File.join(ARGV[0], name)
end
+def data_foreach(name, &block)
+ fn = get_file(name)
+ warn "Reading #{name}"
+ pat = /^# #{name.sub(/\./, '-([\\d.]+)\\.')}/
+ File.open(fn) do |f|
+ line = f.gets
+ unless pat =~ line
+ raise ArgumentError, "#{name}: no Unicode version"
+ end
+ if !$unicode_version
+ $unicode_version = $1
+ elsif $unicode_version != $1
+ raise ArgumentError, "#{name}: Unicode version mismatch: #$1"
+ end
+ f.each(&block)
+ end
+end
# Write Data
class Unifdef