telegram-bot/bin/fetch-telegram-methods

#!/usr/bin/env ruby
# frozen_string_literal: true

#
# Fetch list of methods from Telegram docs.
# Use it to update client.rb.

require 'net/http'
require 'nokogiri'

DOCS_URL = 'https://core.telegram.org/bots/api'

page_html = Net::HTTP.get(URI(DOCS_URL))
doc = Nokogiri::HTML(page_html)

# Select `h4`s, but use `h3`s to group them.
headers = doc.css('h3, h4').
  chunk_while { |_, x| x.name == 'h4' }.
  map { |g| g.select { |x| x.name == 'h4' } }.
  map { |g| g.map(&:text) }

# Method starts with lowercase and does not have spaces.
NOT_METHOD_REGEXP = /(\A[^a-z])|\s/.freeze

# Filter method names.
method_list = headers.
  map { |g| g.grep_v(NOT_METHOD_REGEXP) }.
  reject(&:empty?)

api_version = doc.text.match(/^(?:Introducing )?(Bot API ([\d\.]+))\.?$/)

result = ['# Generated with bin/fetch-telegram-methods']
result << "# #{api_version[1]}" if api_version
result << ''
result << method_list.map { |g| g.join("\n") }.join("\n\n")
result << ''
result_txt = result.join("\n")

puts result_txt

API_METHODS_FILE = File.expand_path('../lib/telegram/bot/client/api_methods.txt', __dir__).freeze
File.write(API_METHODS_FILE, result_txt)
puts '', "Updated #{API_METHODS_FILE}"

puts '', 'Payload types:'
update_tbody = doc.css('tbody').
  find { |tbody| tbody.css('td').any? { |td| td.text == 'update_id' } }
payload_types = update_tbody.css('td:first-child').map(&:text).reject { |x| x == 'update_id' }

result = ['# Generated with bin/fetch-telegram-methods']
result << "# #{api_version[1]}" if api_version
result << payload_types.join("\n")
result << ''
result_txt = result.join("\n")

puts result_txt

PAYLOAD_TYPES_FILE = File.expand_path(
  '../lib/telegram/bot/updates_controller/payload_types.txt',
  __dir__,
).freeze
File.write(PAYLOAD_TYPES_FILE, result_txt)
puts '', "Updated #{PAYLOAD_TYPES_FILE}"