[strings] Add search category consistence tool

- Review fixes

tools/ruby/category_consistency/check_consistency.rb

@@ -0,0 +1,65 @@
+#!/usr/bin/env ruby
+
+require_relative './omim_parsers'
+
+ROOT = File.expand_path(File.dirname(__FILE__))
+OMIM_ROOT = File.join(ROOT, '..', '..', '..')
+CPP_CATEGORIES_FILENAME = File.join(OMIM_ROOT, 'search', 'displayed_categories.cpp')
+CATEGORIES_FILENAME = File.join(OMIM_ROOT, 'data', 'categories.txt')
+STRINGS_FILENAME = File.join(OMIM_ROOT, 'strings.txt')
+CATEGORIES_MATCHER = /m_keys = \{(.*)\};/m
+
+def load_categories_from_cpp(filename)
+  raw_categories = File.read(CPP_CATEGORIES_FILENAME)
+  match = CATEGORIES_MATCHER.match(raw_categories)
+  if match
+    cpp_categories = match[1].split(/,\s+/)
+    # Delete quotes
+    cpp_categories.map { |cat| cat.gsub!(/^"|"$/, '') }
+    cpp_categories
+  end
+end
+
+def compare_categories(string_cats, search_cats)
+  inconsistent_strings = {}
+
+  string_cats.each do |category_name, category|
+    if !search_cats.include? category_name
+      puts "Category '#{category_name}' not found in categories.txt"
+      next
+    end
+    category.each do |lang, translation|
+      if search_cats[category_name].include? lang
+        if !search_cats[category_name][lang].include? translation
+          not_found_cats_list = search_cats[category_name][lang]
+          (inconsistent_strings[category_name] ||= {})[lang] = [translation, not_found_cats_list]
+        end
+      end
+    end
+  end
+
+  inconsistent_strings.each do |name, languages|
+    puts "\nInconsistent category \"#{name}\""
+    languages.each do |lang, values|
+      string_value, category_value = values
+      puts "\t#{lang} : \"#{string_value}\" is not matched by #{category_value}"
+    end
+  end
+  inconsistent_strings.empty?
+end
+
+def check_search_categories_consistent
+  cpp_categories = load_categories_from_cpp(CPP_CATEGORIES_FILENAME)
+  categories_txt_parser = OmimParsers::CategoriesParser.new cpp_categories
+  strings_txt_parser = OmimParsers::StringsParser.new cpp_categories
+
+  search_categories = categories_txt_parser.parse_file(CATEGORIES_FILENAME)
+  string_categories = strings_txt_parser.parse_file(STRINGS_FILENAME)
+
+  compare_categories(string_categories, search_categories) ? 0 : 1
+end
+
+
+if __FILE__ == $0
+  exit check_search_categories_consistent()
+end

tools/ruby/category_consistency/omim_parsers.rb

@@ -0,0 +1,99 @@
+module OmimParsers
+  LANGUAGES = %w(en ru ar cs da nl fi fr de hu id it ja ko nb pl
+                 pt ro es sv th tr uk vi zh-Hans zh-Hant he sk)
+
+  class AbstractParser
+    def initialize(keys)
+      @keys = keys
+    end
+
+    def parse_line(line)
+      raise NotImplementedError.new("You must implement parse_file.")
+    end
+
+    def match_category(line, result)
+      category_match = category.match(line)
+      if !category_match.nil?
+        category = category_match[1]
+        if @keys.include? category
+          result[category] ||= {}
+        end
+      end
+    end
+
+    def parse_file(filename)
+      current_string = nil
+      result = {}
+      File.open(filename, 'r:UTF-8').each do |line|
+        line.strip!
+        next if should_exclude_line? line
+
+        # If line is empty -> next category block started
+        if line.empty?
+          current_string = nil
+          next
+        end
+
+        current_string ||= match_category(line, result)
+
+        parsed = parse_line(line)
+        if !parsed.nil? and !current_string.nil?
+          lang, translation = parsed
+          current_string[lang] = translation
+        end
+      end
+      result
+    end
+
+    def category
+      raise NotImplementedError.new("You must implement category.")
+    end
+
+    def should_exclude_line?(line)
+      false
+    end
+  end
+
+  class CategoriesParser < AbstractParser
+    def parse_line(line)
+      line_match = /^([^:]+):(\S+)$/u.match(line)
+      return if !line_match
+
+      lang = $1.strip
+      return if !LANGUAGES.include? lang
+
+      translation = $2.strip
+      synonyms = []
+      translation.split('|').each do |token|
+        token_match  = /\d?\^?(.*)$/.match(token)
+        synonyms.push(token_match[1]) if token_match
+      end
+      [lang, synonyms]
+    end
+
+    def should_exclude_line?(line)
+      line.start_with? '#'
+    end
+
+    def category
+      # We match only global categories ('food', 'bank'...)
+      /^@([A-Za-z0-9]+)$/
+    end
+  end
+
+  class StringsParser < AbstractParser
+    def parse_line(line)
+      line_match = /^([^=]+)=(.*)$/.match(line)
+      if line_match
+        lang = $1.strip
+        if LANGUAGES.include? lang
+          [lang, $2.strip]
+        end
+      end
+    end
+
+    def category
+      /^\[(.+)\]/
+    end
+  end
+end