Class: Archaeo::UrlRewriter

Inherits:
Object
  • Object
show all
Defined in:
lib/archaeo/url_rewriter.rb

Overview

Rewrites Wayback Machine archive URLs to local file paths.

Used for saving archived pages and their assets for offline browsing. Converts absolute archive URLs into relative paths rooted at a configurable local directory.

Constant Summary collapse

URL_ATTRS =
%w[src href data-src poster].freeze

Instance Method Summary collapse

Constructor Details

#initialize(archive_prefix, local_prefix) ⇒ UrlRewriter

Returns a new instance of UrlRewriter.



14
15
16
17
# File 'lib/archaeo/url_rewriter.rb', line 14

def initialize(archive_prefix, local_prefix)
  @archive_prefix = archive_prefix.to_s
  @local_prefix = local_prefix.to_s
end

Instance Method Details

#rewrite(url) ⇒ Object



19
20
21
22
23
24
# File 'lib/archaeo/url_rewriter.rb', line 19

def rewrite(url)
  return url unless url.start_with?(@archive_prefix)

  relative = url.sub(@archive_prefix, "")
  File.join(@local_prefix, relative)
end

#rewrite_batch(urls) ⇒ Object



26
27
28
# File 'lib/archaeo/url_rewriter.rb', line 26

def rewrite_batch(urls)
  urls.map { |url| rewrite(url) }
end

#rewrite_html(html_content) ⇒ Object



30
31
32
33
34
35
# File 'lib/archaeo/url_rewriter.rb', line 30

def rewrite_html(html_content)
  doc = Nokogiri::HTML(html_content)
  rewrite_url_attrs(doc)
  rewrite_srcset_attrs(doc)
  doc.to_html
end

#rewrite_srcset_attrs(doc) ⇒ Object



48
49
50
51
52
# File 'lib/archaeo/url_rewriter.rb', line 48

def rewrite_srcset_attrs(doc)
  doc.css("[srcset]").each do |el|
    el["srcset"] = rewrite_srcset(el["srcset"])
  end
end

#rewrite_url_attrs(doc) ⇒ Object



37
38
39
40
41
42
43
44
45
46
# File 'lib/archaeo/url_rewriter.rb', line 37

def rewrite_url_attrs(doc)
  URL_ATTRS.each do |attr|
    doc.css("[#{attr}]").each do |el|
      original = el[attr]
      next unless original&.start_with?(@archive_prefix)

      el[attr] = rewrite(original)
    end
  end
end