From 6718ddef228e265e1229ab24336b433213b81db2 Mon Sep 17 00:00:00 2001
From: Karl Berry <karl@freefriends.org>
Date: Thu, 26 Mar 2020 21:59:35 +0000
Subject: hyph-utf8, with tlpsrc from arthur

git-svn-id: svn://tug.org/texlive/trunk@54568 c570f23f-e606-0410-a88d-b1316a301751
---
 .../generic/hyph-utf8/lib/tex/hyphen/converter.rb  | 39 ++++++++++++++++++++++
 .../hyph-utf8/lib/tex/hyphen/texlive/loader.rb     |  2 +-
 .../hyph-utf8/lib/tex/hyphen/texlive/package.rb    |  6 +++-
 .../hyph-utf8/lib/tex/hyphen/texlive/source.rb     |  6 ++--
 4 files changed, 48 insertions(+), 5 deletions(-)
 create mode 100644 Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/converter.rb

(limited to 'Master/texmf-dist/source/generic/hyph-utf8/lib/tex')

diff --git a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/converter.rb b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/converter.rb
new file mode 100644
index 00000000000..3b966d45a40
--- /dev/null
+++ b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/converter.rb
@@ -0,0 +1,39 @@
+# encoding: UTF-8
+
+require 'scanf'
+require 'byebug'
+
+module TeX
+  module Hyphen
+    class Converter
+      def read(conversion)
+        @mapping = { }
+        File.read(conversion).each_line do |line|
+          next if line =~ /^#/
+
+          eightbit, usv = line.scanf "0x%02X\tU+%04X"
+          @mapping[eightbit] = usv.chr(Encoding::UTF_8)
+        end
+      end
+
+      def convert(filename)
+        raise "Please define the encoding mapping first with #read" unless @mapping
+        doconvert = false
+        File.open(filename, external_encoding: Encoding::ASCII_8BIT).each_line do |line|
+          if line =~ /\\patterns{/
+            doconvert = true
+            next
+          end
+          doconvert = false if doconvert && line =~ /}/
+
+          if doconvert
+            puts (line.strip.each_byte.map do |byte|
+              byebug unless @mapping[byte]
+              @mapping[byte]
+            end || '').join
+          end
+        end
+      end
+    end
+  end
+end
diff --git a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/loader.rb b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/loader.rb
index 2fcb3945d30..878bf0ef0d3 100644
--- a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/loader.rb
+++ b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/loader.rb
@@ -4,7 +4,7 @@ module TeX
       module Loader
 
         def unicode_only? # TODO Spec out
-          ['cu', 'sa','as','bn','gu','hi','hy','kn','lo','mul-ethi','ml','mr','or','pa','ta','te', 'pi'].include? @bcp47
+          ['cu', 'sa','as','bn','gu','hi','hy','kn','lo','mul-ethi','ml','mr','or','pa','ta','te', 'pi', 'mk'].include? @bcp47
         end
 
         def string_enc
diff --git a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/package.rb b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/package.rb
index 69481a41774..1d40dc512e8 100644
--- a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/package.rb
+++ b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/package.rb
@@ -104,7 +104,11 @@ module TeX
           unless has_dependency?
             languages.each do |language|
               if language.use_old_patterns_comment and language.legacy_patterns != "zerohyph.tex" and language.bcp47 != 'cop'
-                files << sprintf("tex/generic/hyphen/%s", language.legacy_patterns)
+                if language.bcp47 == 'la-x-classic'
+                  files << sprintf("tex/generic/hyph-utf8/patterns/tex-8bit/%s", language.legacy_patterns)
+                else
+                  files << sprintf("tex/generic/hyphen/%s", language.legacy_patterns)
+                end
               end
             end
           end
diff --git a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/source.rb b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/source.rb
index 61292dcd3bb..dce5be00b41 100644
--- a/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/source.rb
+++ b/Master/texmf-dist/source/generic/hyph-utf8/lib/tex/hyphen/texlive/source.rb
@@ -82,7 +82,7 @@ module TeX
           end
 
           files << File.join(PATH::HYPHU8, 'patterns', 'tex', sprintf('hyph-%s.tex', bcp47))
-          if encoding && encoding != "ascii" then
+          if encoding && encoding != "ascii" && bcp47 != 'la-x-classic' then
             files << File.join(PATH::HYPHU8, 'patterns', 'ptex', sprintf('hyph-%s.%s.tex', bcp47, encoding))
           elsif bcp47 == "cop"
             files << File.join(PATH::HYPHU8, 'patterns', 'tex-8bit', legacy_patterns)
@@ -91,14 +91,14 @@ module TeX
           # we skip the mongolian language for luatex files
           return files if bcp47 == "mn-cyrl-x-lmc"
 
-          ['chr', 'pat', 'hyp', 'lic'].each do |t|
+          ['pat', 'hyp'].each do |t|
             files << File.join(PATH::HYPHU8, 'patterns', 'txt', sprintf('hyph-%s.%s.txt', bcp47, t))
           end
 
           if bcp47 =~ /^sh-/
             # duplicate entries (will be removed later)
             files << File.join(PATH::HYPHU8, 'patterns', 'tex', 'hyph-sr-cyrl.tex')
-            ['chr', 'pat', 'hyp', 'lic'].each do |t|
+            ['pat', 'hyp'].each do |t|
               # duplicate entries (will be removed later)
               files << File.join(PATH::HYPHU8, 'patterns', 'txt', sprintf('hyph-sr-cyrl.%s.txt', t))
             end
-- 
cgit v1.2.3