zorki 0.1.21 → 0.1.23

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: '0228694f5aafe38a9856efe3a4f9f042b94aeecbd5cb7ec919326f48c0b8000e'
-  data.tar.gz: c54da1aa5b98d3166a047a24022b74356538e7dcb974e032bc5236065c5edd4c
+  metadata.gz: 59b7b2ee84517a52b4c4bfaa300fe18d5981789c92c5983cc07d2cbe21323a79
+  data.tar.gz: 6d24c1955520c08164da77783dbbc17d8db09edfa8de30103a3ddfcdaeaf8064
 SHA512:
-  metadata.gz: 30b2953e778ce0a6e36f221350c6760974de7279c764111cfe2327f3bcc5f11a93d19a4747d20c35260038e07c06bad2c3d2fc704b4878c23a41ba309ae689d9
-  data.tar.gz: 99ed979bcc6f4f0758bca49d34580333ba42c35d74ce3ccfe52871acb9665d3841e0d54ed68bb3af2cf1ef366ebb1f3f9ea9136784958778751d7142f09e5b38
+  metadata.gz: 3bb4c7df50e497920f6e21216268a2640128f0b0fee6d443827ffe372f720a63b03fe6ad0c62d2a78862698989e7b62929bd9d0e9c54e5dcf12c8a5ccca85719
+  data.tar.gz: 1cee6eafb78aaf2bcadc855f44676501996ce83c9fc5a3ecafc5303fc7a62d2ee7aa0730b78477c8578fd1edc3e897317ee6014cfd35b38d5743816ebf8307e4

data/lib/zorki/monkeypatch.rb

@@ -41,7 +41,7 @@ module SeleniumMonkeypatch
       data[:sessionId] = @session_id if @session_id
       message = @ws.send_cmd(**data)
       if message.nil? == false && message["error"] && (method != "Fetch.continueRequest")
-        raise Error::WebDriverError, error_message(message["error"])
+        raise Selenium::WebDriver::Error::WebDriverError, error_message(message["error"])
       end
 
       message

data/lib/zorki/scrapers/post_scraper.rb

@@ -14,7 +14,6 @@ module Zorki
           result = attempt_parse(id)
           break
         rescue ImageRequestZeroSize
-          debugger
           # If the image is zero size, we retry
           puts "Zero sized image found, retrying #{count}"
           count += 1
@@ -24,6 +23,9 @@ module Zorki
       raise ImageRequestZeroSize if count == 5
 
       result
+    ensure
+      page.quit
+      # Make sure it's quit? I'm not sure we really want to do this outside of testing.
     end
 
     def attempt_parse(id)
@@ -45,6 +47,7 @@ module Zorki
         "data,xdt_api__v1__media__shortcode__web_info,items"
       )
 
+
       graphql_object = graphql_object.first if graphql_object.kind_of?(Array)
 
       # For pages that have been marked misinfo the structure is very different than not
@@ -62,6 +65,7 @@ module Zorki
         text = graphql_object["articleBody"]
         username = graphql_object["author"]["identifier"]["value"]
         # 2021-04-01T17:07:10-07:00
+
         date = DateTime.strptime(graphql_object["dateCreated"], "%Y-%m-%dT%H:%M:%S%z")
         interactions = graphql_object["interactionStatistic"]
         number_of_likes = interactions.select do |x|

data/lib/zorki/scrapers/scraper.rb

@@ -72,8 +72,9 @@ module Zorki
 
         continue.call(request) do |response|
           # Check if not a CORS prefetch and finish up if not
-          if !response.body.empty? && response.body
+          if !response.body&.empty? && response.body
             check_passed = true
+
             unless additional_search_parameters.nil?
               body_to_check = Oj.load(response.body)
 
@@ -119,7 +120,16 @@ module Zorki
           begin
             element_json = JSON.parse(element.text)
 
-            parsed_element_json = element_json["require"].first.last.first["__bbox"]["require"].first.last.last["__bbox"]["result"]["data"]["xdt_api__v1__media__shortcode__web_info"]
+            if element.text.include?("Dwayne")
+              # if element_json["require"].first.last.first["__bbox"].key?("require")
+
+              #   element_json["require"].first.last.first["__bbox"]["require"].each do |x|
+              #     debugger if x.to_s.include?("Dwayne Johnson")
+              #   end
+              # end
+            end
+
+            parsed_element_json = element_json["require"].last.last.first["__bbox"]["require"].first.last.last["__bbox"]["result"]["data"]["xdt_api__v1__media__shortcode__web_info"]
           rescue StandardError
             next
           end
@@ -137,9 +147,10 @@ module Zorki
       raise ContentUnavailableError.new("Response body nil") if response_body.nil?
       Oj.load(response_body)
     ensure
-      page.quit
+      # page.quit
       # TRY THIS TO MAKE SURE CHROME GETS CLOSED?
       # We may also want to not do this and make sure the same browser is reused instead for cookie purposes
+      # NOW wer'e trying this 2024-05-28
     end
 
   private

data/lib/zorki/scrapers/user_scraper.rb

@@ -18,7 +18,7 @@ module Zorki
       # - *Profile image
       login
 
-      graphql_script = get_content_of_subpage_from_url("https://instagram.com/#{username}/", "?username=")
+      graphql_script = get_content_of_subpage_from_url("https://instagram.com/#{username}/", "graphql", "data,user,full_name")
       graphql_script = graphql_script.first if graphql_script.class == Array
 
       if graphql_script.nil?
@@ -62,13 +62,13 @@ module Zorki
         scraped_username = user["username"]
         raise Zorki::Error unless username == scraped_username
 
-        profile_image_url = user["profile_pic_url_hd"]
+        profile_image_url = user["hd_profile_pic_url_info"]["url"]
         {
           name: user["full_name"],
           username: username,
-          number_of_posts: user["edge_owner_to_timeline_media"]["count"],
-          number_of_followers: user["edge_followed_by"]["count"],
-          number_of_following: user["edge_follow"]["count"],
+          number_of_posts: user["media_count"],
+          number_of_followers: user["follower_count"],
+          number_of_following: user["following_count"],
           verified: user["is_verified"],
           profile: user["biography"],
           profile_link: user["external_url"],

data/lib/zorki/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Zorki
-  VERSION = "0.1.21"
+  VERSION = "0.1.23"
 end

data/lib/zorki.rb

@@ -64,8 +64,6 @@ module Zorki
       return temp_file_name if File.size(temp_file_name) > 100
 
       # Delete the file since we want to retry
-      debugger
-
       File.delete(temp_file_name)
       count += 1
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: zorki
 version: !ruby/object:Gem::Version
-  version: 0.1.21
+  version: 0.1.23
 platform: ruby
 authors:
 - Christopher Guess
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2024-03-29 00:00:00.000000000 Z
+date: 2024-05-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: capybara