From 19e61a2d2eba97a62e40c50406fbefea7f1b4db4 Mon Sep 17 00:00:00 2001
From: Moritz Bunkus <moritz@bunkus.org>
Date: Sun, 6 Sep 2020 19:51:58 +0200
Subject: [PATCH] IANA language subtag registry: make registry download &
 parsing reusable

Part of the implementation of #2919.
---
 Rakefile                                |   2 +-
 rake.d/iana_language_subtag_registry.rb | 132 +++++++++++++-----------
 2 files changed, 71 insertions(+), 63 deletions(-)

diff --git a/Rakefile b/Rakefile
index 39d036f7f..9c58da949 100644
--- a/Rakefile
+++ b/Rakefile
@@ -870,7 +870,7 @@ namespace :dev do
 
   desc "Create iana_language_subtag_registry_list.cpp from official list"
   task :iana_language_subtag_registry_list do
-    create_iana_language_subtag_registry_list_file
+    Mtx::IANALanguageSubtagRegistry.create_cpp
   end
 end
 
diff --git a/rake.d/iana_language_subtag_registry.rb b/rake.d/iana_language_subtag_registry.rb
index 9214eeb55..4edd0c795 100644
--- a/rake.d/iana_language_subtag_registry.rb
+++ b/rake.d/iana_language_subtag_registry.rb
@@ -1,69 +1,77 @@
-def create_iana_language_subtag_registry_list_file
-  txt_file      = "language-subtag-registry.txt"
-  url           = "https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry"
-  cpp_file_name = "src/common/iana_language_subtag_registry_list.cpp"
+module Mtx::IANALanguageSubtagRegistry
+  def self.with_registry &func
+    url               = "https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry"
+    registry          = "language-subtag-registry"
+    download_registry = !FileTest.exists?(registry)
 
-  File.unlink(txt_file) if FileTest.exists?(txt_file)
+    runq "wget", url, "wget --quiet -O #{registry} #{url}" if download_registry
 
-  runq "wget", url, "wget --quiet -O #{txt_file} #{url}"
+    entry   = {}
+    entries = {}
+    process = lambda do
+      type = entry.delete(:type)
 
-  entry   = {}
-  entries = {}
-  process = lambda do
-    type = entry.delete(:type)
+      if type
+        entries[type] ||= []
+        entries[type]  << entry
+      end
 
-    if type
-      entries[type] ||= []
-      entries[type]  << entry
+      entry = {}
     end
 
-    entry = {}
+    IO.readlines(registry).
+      map(&:chomp).
+      each do |line|
+
+      if line == '%%'
+        process.call
+
+      elsif %r{^(Type|Subtag|Description): *(.+)}i.match(line)
+        entry[$1.downcase.to_sym] = $2
+
+      elsif %r{^Prefix: *(.+)}i.match(line)
+        entry[:prefix] ||= []
+        entry[:prefix]  << $1
+      end
+    end
+
+    process.call
+
+    func.call(entries)
+
+  ensure
+    File.unlink(registry) if download_registry && FileTest.exists?(registry)
   end
 
-  IO.readlines(txt_file).
-    map(&:chomp).
-    each do |line|
+  def self.do_create_cpp entries
+    cpp_file_name   = "src/common/iana_language_subtag_registry_list.cpp"
+    entry_formatter = lambda do |entry|
+      if entry[:prefix]
+        prefix = '{ ' + entry[:prefix].sort.map(&:to_cpp_string).join(', ') + ' }'
+      else
+        prefix = '{}'
+      end
 
-    if line == '%%'
-      process.call
-
-    elsif %r{^(Type|Subtag|Description): *(.+)}i.match(line)
-      entry[$1.downcase.to_sym] = $2
-
-    elsif %r{^Prefix: *(.+)}i.match(line)
-      entry[:prefix] ||= []
-      entry[:prefix]  << $1
-    end
-  end
-
-  process.call
-
-  formatter = lambda do |entry|
-    if entry[:prefix]
-      prefix = '{ ' + entry[:prefix].sort.map(&:to_cpp_string).join(', ') + ' }'
-    else
-      prefix = '{}'
+      [ entry[:subtag].downcase.to_cpp_string,
+        entry[:description].to_u8_cpp_string,
+        prefix,
+      ]
     end
 
-    [ entry[:subtag].downcase.to_cpp_string,
-      entry[:description].to_u8_cpp_string,
-      prefix,
+    formatter = lambda do |type, name|
+      rows = entries[type].map(&entry_formatter)
+
+      "std::vector<entry_t> const g_#{name}{\n" +
+        format_table(rows.sort, :column_suffix => ',', :row_prefix => "  { ", :row_suffix => "  },").join("\n") +
+        "\n};\n"
+    end
+
+    formatted = [
+      formatter.call("extlang", "extlangs"),
+      formatter.call("variant", "variants"),
     ]
-  end
 
-  rows = entries["extlang"].map(&formatter)
-
-  extlangs = "std::vector<entry_t> const g_extlangs{\n" +
-    format_table(rows.sort, :column_suffix => ',', :row_prefix => "  { ", :row_suffix => "  },").join("\n") +
-    "\n};\n"
-
-  rows = entries["variant"].map(&formatter)
-
-  variants = "std::vector<entry_t> const g_variants{\n" +
-    format_table(rows.sort, :column_suffix => ',', :row_prefix => "  { ", :row_suffix => "  },").join("\n") +
-    "\n};\n"
-
-  header = <<EOT
+    header = <<EOT
 /*
    mkvmerge -- utility for splicing together matroska files
    from component media subtypes
@@ -89,19 +97,19 @@ namespace mtx::iana::language_subtag_registry {
 
 EOT
 
-  footer = <<EOT
+    footer = <<EOT
 
 } // namespace mtx::iana::language_subtag_registry
 EOT
 
-  content = header +
-    extlangs +
-    "\n" +
-    variants +
-    footer
+    content = header +
+      formatted.join("\n") +
+      footer
 
-  runq("write", cpp_file_name) { IO.write("#{$source_dir}/#{cpp_file_name}", content); 0 }
+    runq("write", cpp_file_name) { IO.write("#{$source_dir}/#{cpp_file_name}", content); 0 }
+  end
 
-ensure
-  File.unlink(txt_file) if FileTest.exists?(txt_file)
+  def self.create_cpp
+    self.with_registry { |entries| do_create_cpp(entries) }
+  end
 end