Module: MMonitor::Spider

Defined in:
lib/mmonitor/spider.rb

Overview

蜘蛛，负责http请求处理

Class Method Summary collapse

Class Method Details

.get_html(url, params = {}) ⇒ Object

抓取HTML



12
13
14
15
16
 
# File 'lib/mmonitor/spider.rb', line 12

def get_html(url, params={})
  body = get(url, params)
  $body = body
  ::Nokogiri::HTML(body)
end

.get_json(url, params = {}) ⇒ Object

抓取JSON



18
19
20
21
 
# File 'lib/mmonitor/spider.rb', line 18

def get_json(url, params={})
  body = get(url, params)
  ::Oj.load(body) rescue {}
end

.get_ocr(photo_url) ⇒ Object

抓取图片上的文字



23
24
25
26
27
28
29
30
31
32
33
34
35
 
# File 'lib/mmonitor/spider.rb', line 23

def get_ocr(photo_url)
  image = MiniMagick::Image.open(photo_url)
  image.combine_options do |c|
    c.background '#FFFFFF'
    c.colorspace 'GRAY'
    c.alpha 'remove'
  end
  image.format 'jpg'
  ocr = RTesseract.new(image.path, processor: 'mini_magick')
  str = ocr.to_s
  image.destroy!
  return str
end

.number_page(total, limit) ⇒ Object

分页



37
38
39
40
41
 
# File 'lib/mmonitor/spider.rb', line 37

def number_page(total, limit)
  count = total / limit
  count += 1 if total % limit > 0
  count
end