app/lib/extractor.rb

   1 # frozen_string_literal: true
   2
   3 module Extractor
   4   extend Twitter::Extractor
   5
   6   module_function
   7
   8   # :yields: username, list_slug, start, end
   9   def extract_mentions_or_lists_with_indices(text)
  10     return [] unless text =~ Twitter::Regex[:at_signs]
  11
  12     possible_entries = []
  13
  14     text.to_s.scan(Account::MENTION_RE) do |screen_name, _|
  15       match_data = $LAST_MATCH_INFO
  16       after = $'
  17       unless after =~ Twitter::Regex[:end_mention_match]
  18         start_position = match_data.char_begin(1) - 1
  19         end_position = match_data.char_end(1)
  20         possible_entries << {
  21           screen_name: screen_name,
  22           indices: [start_position, end_position],
  23         }
  24       end
  25     end
  26
  27     if block_given?
  28       possible_entries.each do |mention|
  29         yield mention[:screen_name], mention[:indices].first, mention[:indices].last
  30       end
  31     end
  32     possible_entries
  33   end
  34
  35   def extract_hashtags_with_indices(text, **)
  36     return [] unless text =~ /#/
  37
  38     tags = []
  39     text.scan(Tag::HASHTAG_RE) do |hash_text, _|
  40       match_data = $LAST_MATCH_INFO
  41       start_position = match_data.char_begin(1) - 1
  42       end_position = match_data.char_end(1)
  43       after = $'
  44       if after =~ %r{\A://}
  45         hash_text.match(/(.+)(https?\Z)/) do |matched|
  46           hash_text = matched[1]
  47           end_position -= matched[2].char_length
  48         end
  49       end
  50
  51       tags << {
  52         hashtag: hash_text,
  53         indices: [start_position, end_position],
  54       }
  55     end
  56
  57     tags.each { |tag| yield tag[:hashtag], tag[:indices].first, tag[:indices].last } if block_given?
  58     tags
  59   end
  60
  61   def extract_cashtags_with_indices(_text)
  62     [] # always returns empty array
  63   end
  64 end