pulse-downloader 0.1.23 → 0.1.28

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: e354feaf3ad68d91b2b58b1410ad297582ace9b6f3493b17cf8ae11d11c18d8d
-  data.tar.gz: 3e036681abb31307b958bba6038d1fda3a9634195d71c61af3e8896bd695b22a
+  metadata.gz: 2a9e6c9185000eb45fe03fcf5ec6739aa0b4df48cd9733654cc2f312ca156b54
+  data.tar.gz: f03dfbbbecbd75ffe795b194c0ac19689021e1a74a4668ffb6573b6193eb0627
 SHA512:
-  metadata.gz: cf401b59b944a0bdca308998c19f4c8b996fadf60cc8118c06a909b5eaa1ca98d9f0f3d234af7050ce71611475d8f9a9c4a79503003649870a5d7f18dc62caa0
-  data.tar.gz: 98bd01ea727425dad6a932fd431c241b023d954134827ac781faab9b90559a3d05c75916eb4e0bf99d4f8f4c4244b2ce7b019df8ced9c474ad8efe18fa4e7e7d
+  metadata.gz: 706da41a31ad0fda7b4a17b748f8271f758973510039d2e3eac79115761ef71e970eee042d3ef30ed2a080ae7e64f49584d8110c249022a4bca3029c48cc1024
+  data.tar.gz: 97c3c92845700c69343eba19fb93a9b9de71fe44ceca0ff986e25df8cc69ffb71a48c778eaa145df4ae802295b148feaee768814c52124e4ca74079e8ca9e807

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    pulse-downloader (0.1.11)
+    pulse-downloader (0.1.26)
       active_attr (~> 0.15)
       httparty (~> 0.18)
       nokogiri (~> 1.11)
@@ -69,7 +69,7 @@ GEM
       ruby-progressbar
     mocha (1.11.2)
     multi_xml (0.6.0)
-    nokogiri (1.11.3-x86_64-linux)
+    nokogiri (1.11.5-x86_64-linux)
       racc (~> 1.4)
     options (2.3.2)
     progress_bar (1.3.3)

data/README.md

@@ -33,7 +33,8 @@ client = Pulse::Downloader::Client.new(
   drop_exitsing_files_in_path: false,
   save_and_dont_return: true,
   report_time: false,
-  progress_bar: true
+  progress_bar: true,
+  scrape_images: true
 )
 
 client.call!

data/lib/pulse/downloader/client.rb

@@ -13,6 +13,7 @@ module Pulse
         :save_path,
         :read_from_save_path,
         :verify_ssl,
+        :headers,
         :drop_exitsing_files_in_path,
         :save_and_dont_return,
         :report_time,
@@ -36,6 +37,7 @@ module Pulse
         save_path: '',
         read_from_save_path: false,
         verify_ssl: true,
+        headers: nil,
         drop_exitsing_files_in_path: false,
         save_and_dont_return: true,
         report_time: false,
@@ -48,6 +50,7 @@ module Pulse
         @save_path = save_path
         @read_from_save_path = read_from_save_path
         @verify_ssl = verify_ssl
+        @headers = headers
         @drop_exitsing_files_in_path = drop_exitsing_files_in_path
         @save_and_dont_return = save_and_dont_return
         @report_time = report_time

data/lib/pulse/downloader/file_downloader.rb

@@ -8,7 +8,10 @@ module Pulse
 
         @start_time = get_micro_second_time
 
-        file_data = HTTParty.get(escape(compute_file_link(file_path)), verify: verify_ssl)
+        file_data = HTTParty.get(
+          escape(compute_file_link(file_path)),
+          verify: verify_ssl
+        )
 
         @end_time = get_micro_second_time
 

data/lib/pulse/downloader/version.rb

@@ -1,5 +1,5 @@
 module Pulse
   module Downloader
-    VERSION = "0.1.23"
+    VERSION = "0.1.28"
   end
 end

data/lib/pulse/downloader/web_page_parser.rb

@@ -4,7 +4,7 @@ module Pulse
       def fetch_file_paths(custom_path_root=nil)
         @start_time = get_micro_second_time
 
-        response = HTTParty.get(url, verify: verify_ssl)
+        response = HTTParty.get(url, verify: verify_ssl, headers: headers)
 
         @end_time = get_micro_second_time
 
@@ -12,30 +12,48 @@ module Pulse
           print_time
         end
 
-        extract_file_urls(response, custom_path_root)
+        if file_type.is_a?(Array)
+          file_type.flat_map do |type|
+            extract_file_urls(response, custom_path_root, type)
+          end
+        else
+          extract_file_urls(response, custom_path_root, file_type)
+        end
       end
 
       private
 
-      def extract_file_urls(response, custom_path_root)
+      def extract_file_urls(response, custom_path_root, type)
         return [] if response.body.nil? || response.body.empty?
-        (
-          extract_download_links(response, custom_path_root) +
-            extract_embedded_images(response, custom_path_root)
+
+        remove_base64(
+          extract_all_urls(response, custom_path_root, type) +
+            extract_download_links(response, type) +
+            extract_embedded_images(response, type)
         ).uniq
       end
 
-      def extract_download_links(response, custom_path_root)
+      def extract_all_urls(response, custom_path_root, type)
+        parse_html(response.body)
+          .to_s
+          .split(/\s+/)
+          .find_all { |u| u =~ /^https?:/ }
+          .compact
+          .select { |link| (link.include? type || link.include?(custom_path_root)) }
+          .map { |link| add_base_url(link) }
+      end
+
+      def extract_download_links(response, type)
         parse_html(response.body)
           .css('a')
           .to_a
           .map { |link| link['href'] }
           .compact
-          .select { |link| (link.include? file_type || link.include?(custom_path_root)) }
+          .select { |link| (link.include? type) }
           .map { |link| add_base_url(link) }
       end
 
-      def extract_embedded_images(response, custom_path_root)
+      def extract_embedded_images(response, type)
         return [] unless scrape_images
 
         parse_html(response.body)
@@ -43,10 +61,16 @@ module Pulse
           .to_a
           .map { |e| e["src"] }
           .compact
-          .select { |link| (link.include? file_type || link.include?(custom_path_root)) }
+          .select { |link| (link.include? type) }
           .map { |link| add_base_url(link) }
       end
 
+      def remove_base64(urls)
+        urls.reject do |url|
+          url.include?(':image/') || url.include?('base64')
+        end
+      end
+
       def parse_html(raw_html)
         Nokogiri::HTML(raw_html)
       end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: pulse-downloader
 version: !ruby/object:Gem::Version
-  version: 0.1.23
+  version: 0.1.28
 platform: ruby
 authors:
 - trex22
 autorequire: 
 bindir: exe
 cert_chain: []
-date: 2021-05-10 00:00:00.000000000 Z
+date: 2021-05-24 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: httparty