IANA language subtag registry: make registry download & parsing reusable

Part of the implementation of #2919.

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
 

rake.d/iana_language_subtag_registry.rb

@@ -1,11 +1,10 @@
-def create_iana_language_subtag_registry_list_file
-  txt_file      = "language-subtag-registry.txt"
+module Mtx::IANALanguageSubtagRegistry
+  def self.with_registry &func
     url               = "https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry"
-  cpp_file_name = "src/common/iana_language_subtag_registry_list.cpp"
+    registry          = "language-subtag-registry"
+    download_registry = !FileTest.exists?(registry)
 
-  File.unlink(txt_file) if FileTest.exists?(txt_file)
-
-  runq "wget", url, "wget --quiet -O #{txt_file} #{url}"
+    runq "wget", url, "wget --quiet -O #{registry} #{url}" if download_registry
 
     entry   = {}
     entries = {}
@@ -20,7 +19,7 @@ def create_iana_language_subtag_registry_list_file
       entry = {}
     end
 
-  IO.readlines(txt_file).
+    IO.readlines(registry).
       map(&:chomp).
       each do |line|
 
@@ -38,7 +37,15 @@ def create_iana_language_subtag_registry_list_file
 
     process.call
 
-  formatter = lambda do |entry|
+    func.call(entries)
+
+  ensure
+    File.unlink(registry) if download_registry && FileTest.exists?(registry)
+  end
+
+  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
@@ -51,17 +58,18 @@ def create_iana_language_subtag_registry_list_file
       ]
     end
 
-  rows = entries["extlang"].map(&formatter)
+    formatter = lambda do |type, name|
+      rows = entries[type].map(&entry_formatter)
 
-  extlangs = "std::vector<entry_t> const g_extlangs{\n" +
+      "std::vector<entry_t> const g_#{name}{\n" +
         format_table(rows.sort, :column_suffix => ',', :row_prefix => "  { ", :row_suffix => "  },").join("\n") +
         "\n};\n"
+    end
 
-  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"
+    formatted = [
+      formatter.call("extlang", "extlangs"),
+      formatter.call("variant", "variants"),
+    ]
 
     header = <<EOT
 /*
@@ -95,13 +103,13 @@ EOT
 EOT
 
     content = header +
-    extlangs +
-    "\n" +
-    variants +
+      formatted.join("\n") +
       footer
 
     runq("write", cpp_file_name) { IO.write("#{$source_dir}/#{cpp_file_name}", content); 0 }
-
-ensure
-  File.unlink(txt_file) if FileTest.exists?(txt_file)
+  end
+
+  def self.create_cpp
+    self.with_registry { |entries| do_create_cpp(entries) }
+  end
 end