Make all regexes constants lazy loaded from pregenerated files

data/generate_constants.rb

@@ -0,0 +1,188 @@
+require_relative '../lib/unicode/emoji/constants'
+require_relative '../lib/unicode/emoji/index'
+require_relative '../lib/unicode/emoji/lazy_constants'
+
+include Unicode::Emoji
+
+pack = ->(ord){ Regexp.escape(Array(ord).pack("U*")) }
+join = -> (*strings){ "(?:" + strings.join("|") + ")" }
+pack_and_join = ->(ords){  join[*ords.map{ |ord| pack[ord] }] }
+
+emoji_character     = pack_and_join[EMOJI_CHAR]
+emoji_modifier      = pack_and_join[EMOJI_MODIFIERS]
+emoji_modifier_base = pack_and_join[EMOJI_MODIFIER_BASES]
+emoji_component     = pack_and_join[EMOJI_COMPONENT]
+emoji_presentation  = pack_and_join[EMOJI_PRESENTATION]
+picto               = pack_and_join[EXTENDED_PICTOGRAPHIC]
+picto_no_emoji      = pack_and_join[EXTENDED_PICTOGRAPHIC_NO_EMOJI]
+
+emoji_presentation_sequence = \
+  join[
+    pack_and_join[TEXT_PRESENTATION] + pack[EMOJI_VARIATION_SELECTOR],
+    emoji_presentation + "(?!" + pack[TEXT_VARIATION_SELECTOR] + ")" + pack[EMOJI_VARIATION_SELECTOR] + "?",
+  ]
+
+non_component_emoji_presentation_sequence = \
+  "(?!" + emoji_component + ")" + emoji_presentation_sequence
+
+text_presentation_sequence = \
+  join[
+    pack_and_join[TEXT_PRESENTATION]+ "(?!" + join[emoji_modifier, pack[EMOJI_VARIATION_SELECTOR]] + ")" + pack[TEXT_VARIATION_SELECTOR] + "?",
+    emoji_presentation + pack[TEXT_VARIATION_SELECTOR]
+  ]
+
+emoji_modifier_sequence = \
+  emoji_modifier_base + emoji_modifier
+
+emoji_keycap_sequence = \
+  pack_and_join[EMOJI_KEYCAPS] + pack[[EMOJI_VARIATION_SELECTOR, EMOJI_KEYCAP_SUFFIX]]
+
+emoji_valid_flag_sequence = \
+  pack_and_join[VALID_REGION_FLAGS]
+
+emoji_well_formed_flag_sequence = \
+  "(?:" +
+    pack_and_join[REGIONAL_INDICATORS] +
+    pack_and_join[REGIONAL_INDICATORS] +
+  ")"
+
+emoji_valid_core_sequence = \
+  join[
+    # emoji_character,
+    emoji_keycap_sequence,
+    emoji_modifier_sequence,
+    non_component_emoji_presentation_sequence,
+    emoji_valid_flag_sequence,
+  ]
+
+emoji_well_formed_core_sequence = \
+  join[
+    # emoji_character,
+    emoji_keycap_sequence,
+    emoji_modifier_sequence,
+    non_component_emoji_presentation_sequence,
+    emoji_well_formed_flag_sequence,
+  ]
+
+emoji_rgi_tag_sequence = \
+  pack_and_join[RECOMMENDED_SUBDIVISION_FLAGS]
+
+emoji_valid_tag_sequence = \
+  "(?:" +
+    pack[EMOJI_TAG_BASE_FLAG] +
+    "(?:" + VALID_SUBDIVISIONS.map{ |sd| Regexp.escape(sd.tr("\u{20}-\u{7E}", "\u{E0020}-\u{E007E}"))}.join("|") + ")" +
+    pack[CANCEL_TAG] +
+  ")"
+
+emoji_well_formed_tag_sequence = \
+  "(?:" +
+    join[
+      non_component_emoji_presentation_sequence,
+      emoji_modifier_sequence,
+    ] +
+    pack_and_join[TAGS] + "+" +
+    pack[CANCEL_TAG] +
+  ")"
+
+emoji_rgi_zwj_sequence = \
+  pack_and_join[RECOMMENDED_ZWJ_SEQUENCES]
+
+emoji_valid_zwj_element = \
+  join[
+    emoji_modifier_sequence,
+    emoji_presentation_sequence,
+    emoji_character,
+  ]
+
+emoji_valid_zwj_sequence = \
+  "(?:" +
+    "(?:" + emoji_valid_zwj_element + pack[ZWJ] + ")+" + emoji_valid_zwj_element +
+  ")"
+
+emoji_rgi_sequence = \
+  join[
+    emoji_rgi_zwj_sequence,
+    emoji_rgi_tag_sequence,
+    emoji_valid_core_sequence,
+  ]
+
+emoji_valid_sequence = \
+  join[
+    emoji_valid_zwj_sequence,
+    emoji_valid_tag_sequence,
+    emoji_valid_core_sequence,
+  ]
+
+emoji_well_formed_sequence = \
+  join[
+    emoji_valid_zwj_sequence,
+    emoji_well_formed_tag_sequence,
+    emoji_well_formed_core_sequence,
+  ]
+
+def write_regex(const_name, regex)
+  filename = const_name.to_s.downcase
+  filepath = File.expand_path("../lib/unicode/emoji/generated/#{filename}.rb", __dir__)
+
+  File.write(filepath, <<~CONTENT)
+    # This file was generated. Please, do not edit this file by hand.
+    module Unicode
+      module Emoji
+        #{const_name} = #{regex.inspect}
+      end
+    end
+  CONTENT
+  puts "#{const_name} written to #{filepath}"
+end
+
+# Matches basic singleton emoji and all kind of sequences, but restrict zwj and tag sequences to known sequences (rgi)
+REGEX = Regexp.compile(emoji_rgi_sequence)
+write_regex(:REGEX, REGEX)
+
+# Matches basic singleton emoji and all kind of valid sequences
+REGEX_VALID = Regexp.compile(emoji_valid_sequence)
+write_regex(:REGEX_VALID, REGEX_VALID)
+
+# Matches basic singleton emoji and all kind of sequences
+REGEX_WELL_FORMED = Regexp.compile(emoji_well_formed_sequence)
+write_regex(:REGEX_WELL_FORMED, REGEX_WELL_FORMED)
+
+# Matches only basic single, non-textual emoji
+# Ignores "components" like modifiers or simple digits
+REGEX_BASIC = Regexp.compile(
+  "(?!" + emoji_component + ")" + emoji_presentation_sequence
+)
+write_regex(:REGEX_BASIC, REGEX_BASIC)
+
+# Matches only basic single, textual emoji
+# Ignores "components" like modifiers or simple digits
+REGEX_TEXT = Regexp.compile(
+  "(?!" + emoji_component + ")" + text_presentation_sequence
+)
+write_regex(:REGEX_TEXT, REGEX_TEXT)
+
+# Matches any emoji-related codepoint - Use with caution (returns partil matches)
+REGEX_ANY = Regexp.compile(
+  emoji_character
+)
+write_regex(:REGEX_ANY, REGEX_ANY)
+
+# Combined REGEXes which also match for TEXTUAL emoji
+REGEX_INCLUDE_TEXT = Regexp.union(REGEX, REGEX_TEXT)
+write_regex(:REGEX_INCLUDE_TEXT, REGEX_INCLUDE_TEXT)
+
+REGEX_VALID_INCLUDE_TEXT = Regexp.union(REGEX_VALID, REGEX_TEXT)
+write_regex(:REGEX_VALID_INCLUDE_TEXT, REGEX_VALID_INCLUDE_TEXT)
+
+REGEX_WELL_FORMED_INCLUDE_TEXT = Regexp.union(REGEX_WELL_FORMED, REGEX_TEXT)
+write_regex(:REGEX_WELL_FORMED_INCLUDE_TEXT, REGEX_WELL_FORMED_INCLUDE_TEXT)
+
+REGEX_PICTO = Regexp.compile(
+  picto
+)
+write_regex(:REGEX_PICTO, REGEX_PICTO)
+
+REGEX_PICTO_NO_EMOJI = Regexp.compile(
+  picto_no_emoji
+)
+write_regex(:REGEX_PICTO_NO_EMOJI, REGEX_PICTO_NO_EMOJI)