Module: PinterestDL::Extractors::Video

Defined in:
lib/pinterest_dl/extractors/video.rb

Overview

Extracts the direct video URL for a pin from raw HTML.

Class Method Summary collapse

Class Method Details

.extract(html) ⇒ Object 



10
11
12
13
14
15
 
# File 'lib/pinterest_dl/extractors/video.rb', line 10

def extract(html)
  url = from_json_ld(html) || from_open_graph(html) || from_content_url(html)
  return nil unless url

  unescape(url)
end

.from_content_url(html) ⇒ Object 



26
27
28
29
 
# File 'lib/pinterest_dl/extractors/video.rb', line 26

def from_content_url(html)
  match = html.match(/"contentUrl"\s*:\s*"([^"]+\.mp4[^"]*)"/)
  match && match[1]
end

.from_json_ld(html) ⇒ Object 



17
18
19
 
# File 'lib/pinterest_dl/extractors/video.rb', line 17

def from_json_ld(html)
  PinterestDL::Extractors::JsonLd.content_url(html, 'VideoObject')
end

.from_open_graph(html) ⇒ Object 



21
22
23
24
 
# File 'lib/pinterest_dl/extractors/video.rb', line 21

def from_open_graph(html)
  match = html.match(/<meta[^>]*property=["']og:video["'][^>]*content=["']([^"']+)["']/)
  match && match[1]
end

.unescape(url) ⇒ Object 



31
32
33
 
# File 'lib/pinterest_dl/extractors/video.rb', line 31

def unescape(url)
  url.gsub('\\u002F', '/').gsub('\\', '')
end