From e668e36b49b38d6d422c6e9d4b3eddd374bcbb15 Mon Sep 17 00:00:00 2001 From: nobu Date: Tue, 13 Jan 2009 09:05:29 +0000 Subject: * template/{encdb,transdb}.h.tmpl: moved enc/make_encdb.rb and enc/trans/make_transdb.rb using tool/generic_erb.rb. * common.mk (encdb.h, transdb.h): generates from avobe template. git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@21490 b2dd03c8-39d4-4d8f-98ff-823fe69b080e --- template/encdb.h.tmpl | 79 +++++++++++++++++++++++++++++++++++++++++++++++++ template/transdb.h.tmpl | 59 ++++++++++++++++++++++++++++++++++++ 2 files changed, 138 insertions(+) create mode 100644 template/encdb.h.tmpl create mode 100644 template/transdb.h.tmpl (limited to 'template') diff --git a/template/encdb.h.tmpl b/template/encdb.h.tmpl new file mode 100644 index 0000000000..e65b651050 --- /dev/null +++ b/template/encdb.h.tmpl @@ -0,0 +1,79 @@ +<% +# +# OnigEncodingDefine(foo, Foo) = { +# .. +# "Shift_JIS", /* Canonical Name */ +# .. +# }; +# ENC_ALIAS("SJIS", "Shift_JIS") +# ENC_REPLICATE("Windows-31J", "Shift_JIS") +# ENC_ALIAS("CP932", "Windows-31J") +# + +def check_duplication(defs, name, fn, line) + if defs[name] + raise ArgumentError, "%s:%d: encoding %s is already registered(%s:%d)" % + [fn, line, name, *defs[name]] + else + defs[name.upcase] = [fn,line] + end +end + +count = 0 +lines = [] +encodings = [] +defs = {} +encdirs = ARGV.dup +encdirs << 'enc' if encdirs.empty? +files = {} +encdirs.each do |encdir| + next unless File.directory?(encdir) + Dir.open(encdir) {|d| d.grep(/.+\.[ch]\z/)}.sort_by {|e| + e.scan(/(\d+)|(\D+)/).map {|n,a| a||[n.size,n.to_i]}.flatten + }.each do |fn| + next if files[fn] + files[fn] = true + open(File.join(encdir,fn)) do |f| + orig = nil + name = nil + f.each_line do |line| + if (/^OnigEncodingDefine/ =~ line)..(/"(.*?)"/ =~ line) + if $1 + check_duplication(defs, $1, fn, $.) + encodings << $1 + count += 1 + end + else + case line + when /^\s*rb_enc_register\(\s*"([^"]+)"/ + count += 1 + line = nil + when /^ENC_REPLICATE\(\s*"([^"]+)"\s*,\s*"([^"]+)"/ + raise ArgumentError, + '%s:%d: ENC_REPLICATE: %s is not defined yet. (replica %s)' % + [fn, $., $2, $1] unless defs[$2.upcase] + count += 1 + when /^ENC_ALIAS\(\s*"([^"]+)"\s*,\s*"([^"]+)"/ + raise ArgumentError, + '%s:%d: ENC_ALIAS: %s is not defined yet. (alias %s)' % + [fn, $., $2, $1] unless defs[$2.upcase] + when /^ENC_DUMMY\(\s*"([^"]+)"/ + count += 1 + else + next + end + check_duplication(defs, $1, fn, $.) + lines << line.sub(/;.*/m, "").chomp + ";" if line + end + end + end + end +end +encodings.each do |e| +%>ENC_DEFINE("<%=e%>"); +% end +% lines.each do |line| +<%=line%> +% end + +#define ENCODING_COUNT <%=count%> diff --git a/template/transdb.h.tmpl b/template/transdb.h.tmpl new file mode 100644 index 0000000000..d0cf101344 --- /dev/null +++ b/template/transdb.h.tmpl @@ -0,0 +1,59 @@ +<% +# +# static const rb_transcoder +# rb_from_US_ASCII = { +# "US-ASCII", "UTF-8", &from_US_ASCII, 1, 0, +# NULL, NULL, +# }; +# + +count = 0 +converters = {} +transdirs = ARGV.dup +transdirs << 'enc/trans' if transdirs.empty? + +transdirs = transdirs.sort_by {|td| + -td.length +}.inject([]) {|tds, td| + next tds unless File.directory?(td) + tds << td if tds.all? {|td2| !File.identical?(td2, td) } + tds +} + +files = {} +names_t = [] +converter_list = [] +transdirs.each do |transdir| + names = Dir.entries(transdir) + names_t += names.map {|n| /(?!\A)\.trans\z/ =~ n ? $` : nil }.compact + names_c = names.map {|n| /(?!\A)\.c\z/ =~ n ? $` : nil }.compact + (names_t & names_c).map {|n| + "#{n}.c" + }.sort_by {|e| + e.scan(/(\d+)|(\D+)/).map {|n,a| a||[n.size,n.to_i]}.flatten + }.each do |fn| + next if files[fn] + files[fn] = true + path = File.join(transdir,fn) + open(path) do |f| + f.each_line do |line| + if (/^static const rb_transcoder/ =~ line)..(/"(.*?)"\s*,\s*"(.*?)"/ =~ line) + if $1 && $2 + from_to = "%s to %s" % [$1, $2] + if converters[from_to] + raise ArgumentError, '%s:%d: transcode "%s" is already registered at %s:%d' % + [path, $., from_to, *converters[from_to].values_at(3, 4)] + else + converters[from_to] = [$1, $2, fn[0..-3], path, $.] + converter_list << from_to + end + end + end + end + end + end +end +converter_list.each do |from_to| + from, to, fn = *converters[from_to] +%>rb_declare_transcoder("<%=from%>", "<%=to%>", "<%=fn%>"); +% end -- cgit v1.2.3