figshare_api_v2 0.9.9 → 0.9.10

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 5a59303e131674b6b915f07bf0e0aa317a3d5d594c8e87586fe273d8ed83697d
-  data.tar.gz: b8bdb004509e69e295b110f60f98ac6da83ea2ea3c5d2bf72e9d46ec041d274a
+  metadata.gz: 1b02d428dc8877f75c9bf9ba05d54b1465944745ec2ff1ca0c962bc6803cdc0a
+  data.tar.gz: 6095dffd50dbd105b0ccadeffd3f9243c0434f44cb79bcec32872008add97fab
 SHA512:
-  metadata.gz: ebdcee71baa8b80d0c434c9c041255a8843c897574c00725c60c288670854c0d3bce3e404805f03201951b981148ea8bfc49acf465b686af75cabd2bd47655f9
-  data.tar.gz: 5cdd11942289e5cb75e4e09afd730ebd12209b161fc68221aa5dadc347a90fc04370d9af464c419b69ed7cb794097254d796cd318ffc9957402cd3736728328b
+  metadata.gz: 59456d19b2df1624bd7f276c6672326ce888fa967f51ff8aace4b8317ac67990e579aaa85affcd098f4b390b46073353df479d50cd9a65b1b690e04d2a96fcc1
+  data.tar.gz: c06980357bed4295c410ae9f08a04de00108620647268445990c8d7c5a82221d4d824c0a32e1baa08918f9be9c628057afce07dca3c5da2b0013258f1258e485

data/History.txt

@@ -1,3 +1,23 @@
+robertburrowes	Fri May 13 08:40:05 2022 +1200
+	Bumped minor version
+robertburrowes	Thu May 12 17:25:07 2022 +1200
+	Moved the accounts_wrapper() into institution.accounts(), less the debugging stuff
+robertburrowes	Thu May 12 17:06:48 2022 +1200
+	Comment
+robertburrowes	Thu May 12 16:52:57 2022 +1200
+	Check returned account IDs are in order (they do seem to be)
+robertburrowes	Thu May 12 16:32:58 2022 +1200
+	Incorporate the new id_gte into institute.accounts, as page/page_size pagination has a 9000 record limit.
+robertburrowes	Thu May 12 16:22:48 2022 +1200
+	return earlier, in _paginate, to not set 'page'
+robertburrowes	Thu May 12 15:56:44 2022 +1200
+	Added 'once_only' to the post and get _pagination calls, so we don't automatically move onto the next page.
+robertburrowes	Thu May 12 11:18:45 2022 +1200
+	Refactored _pagination response code to see if it helped fix the duplicate record issue. It didn't.
+robertburrowes	Thu May 12 09:43:45 2022 +1200
+	Added debug output of updated Figshare API.
+robertburrowes	Wed May 11 09:07:42 2022 +1200
+	update comments
 robertburrowes	Tue May 10 10:42:27 2022 +1200
 	Comment
 robertburrowes	Tue May 10 10:40:02 2022 +1200

data/lib/base.rb

@@ -71,41 +71,53 @@ module Figshare
     # @param by_offset [Boolean] use offset/limit rather than page/page_size in API calls
     # @yield [String] if given a block, iterates through the result from figshare
     # @return [Integer] number of results.
-    private def get_paginate(api_query:, args: {}, debug: false, by_offset: false, &block)
+    private def get_paginate(api_query:, args: {}, debug: false, by_offset: false, once_only: false, &block)
       args = {} if args.nil?
-      if ! args.is_a?(Hash)
-        raise 'get_paginate(): Expecting args to be a Hash'
-      end
-
-      # Loop variables, if we are using pages
-      page = args['page'].nil? ? 1 : args['page']
-      page_size = args['page_size'].nil? ? 100 : args['page_size']
-
-      # Loop variables, if we are using offsets
-      offset = args['offset'].nil? ? 0 : args['offset']
-      limit = args['limit'].nil? ? 100 : args['limit']
+      raise 'get_paginate(): Expecting args to be a Hash' unless args.is_a?(Hash)
 
       by_offset = true if args['page'].nil? && ! args['offset'].nil?
+      if by_offset
+        # Loop variables, if we are using offsets
+        offset = args['offset'].nil? ? 0 : args['offset']
+        limit = args['limit'].nil? ? 100 : args['limit']
+        args['offset'] = offset
+        args['limit'] = limit
+      else
+        page = args['page'].nil? ? 1 : args['page']
+        page_size = args['page_size'].nil? ? 100 : args['page_size']
+        args['page'] = page
+        args['page_size'] = page_size
+      end
 
       # How many results so far.
       result_count = 0
 
       loop do
         content_type = response = nil
-        form_args = by_offset ? { 'limit' => limit, 'offset' => offset } : { 'page_size' => page_size, 'page' => page }
         WIKK::WebBrowser.https_session( host: @hostname, verify_cert: false ) do |ws|
           response = ws.get_page( query: "#{@api_url}#{api_query}",
                                   authorization: "token #{@auth_token}",
-                                  form_values: form_args.merge(args)
+                                  form_values: args
                                 )
           content_type = ws.header_value(key: 'Content-Type')
         end
+        # File.write("/tmp/debug_#{args['id_gte']}_#{page}.json", response)
         page_count = iterate_json_response(response: response, content_type: content_type, debug: debug, &block)
         result_count += page_count
-        break if page_count < page_size # Got less results than we asked for, so it was the last page
 
-        page += 1 # Ready to fetch next page
-        offset += limit # if we use offset, then mor
+        return result_count if once_only
+
+        if by_offset
+          break if page_count < limit # Got less results than we asked for, so it was the last page
+
+          offset += limit # if we use offset
+          args['offset'] = offset
+        else
+          break if page_count < page_size # Got less results than we asked for, so it was the last page
+
+          page += 1 # Ready to fetch next page
+          args['page'] = page
+        end
       end
 
       return result_count
@@ -156,42 +168,54 @@ module Figshare
     # @param by_offset [Boolean] use offset/limit rather than page/page_size in API calls
     # @yield [String] if given a block, iterates through the result from figshare
     # @return [Integer] number of results.
-    private def post_paginate(api_query:, args: {}, debug: false, by_offset: false, &block)
+    private def post_paginate(api_query:, args: {}, debug: false, by_offset: false, once_only: false, &block)
       # Loop variables, if we are using pages
-      page = args['page'].nil? ? 1 : args['page']
-      page_size = args['page_size'].nil? ? 100 : args['page_size']
+      args = {} if args.nil?
+      raise 'post_paginate(): Expecting args to be a Hash' unless args.is_a?(Hash)
 
-      # Loop variables, if we are using offsets
-      offset = args['offset'].nil? ? 0 : args['offset']
-      limit = args['limit'].nil? ? 100 : args['limit']
+      by_offset = true if args['page'].nil? && ! args['offset'].nil?
+      if by_offset
+        # Loop variables, if we are using offsets
+        offset = args['offset'].nil? ? 0 : args['offset']
+        limit = args['limit'].nil? ? 100 : args['limit']
+        args['offset'] = offset
+        args['limit'] = limit
+      else
+        page = args['page'].nil? ? 1 : args['page']
+        page_size = args['page_size'].nil? ? 100 : args['page_size']
+        args['page'] = page
+        args['page_size'] = page_size
+      end
 
       # How many results so far.
       result_count = 0
 
-      by_offset = true if args['page'].nil? && ! args['offset'].nil?
-
-      args = {} if args.nil?
-      if ! args.is_a?(Hash)
-        raise 'post_paginate(): Expecting args to be a Hash'
-      end
-
       loop do
         content_type = response = nil
-        form_args = by_offset ? { 'limit' => limit, 'offset' => offset } : { 'page_size' => page_size, 'page' => page }
         WIKK::WebBrowser.https_session( host: @hostname, verify_cert: false ) do |ws|
           response = ws.post_page(  query: "#{@api_url}#{api_query}",
                                     content_type: 'application/json; charset=UTF-8',
                                     authorization: "token #{@auth_token}",
-                                    data: args.merge(form_args).to_j
+                                    data: args.to_j
                                  )
           content_type = ws.header_value(key: 'Content-Type')
         end
         page_count = iterate_json_response(response: response, content_type: content_type, debug: debug, &block)
         result_count += page_count
-        break if page_count < page_size # Got less results than we asked for, so it was the last page
 
-        page += 1 # Ready to fetch next page
-        offset += limit # if we use offset
+        return result_count if once_only
+
+        if by_offset
+          break if page_count < limit # Got less results than we asked for, so it was the last page
+
+          offset += limit # if we use offset
+          args['offset'] = offset
+        else
+          break if page_count < page_size # Got less results than we asked for, so it was the last page
+
+          page += 1 # Ready to fetch next page
+          args['page'] = page
+        end
       end
       return result_count
     end

data/lib/figshare_api_v2.rb

@@ -1,7 +1,7 @@
 module Figshare # :nodoc:
   # Figshare module initialization
   #
-  VERSION = '0.9.9'
+  VERSION = '0.9.10'
 
   require 'wikk_webbrowser'
   require 'wikk_json'

data/lib/institutions.rb

@@ -14,9 +14,9 @@ module Figshare
       end
     end
 
-    # Get the institional account details
+    # Get the institional account's details (not a person's account details)
     #
-    # @yield [Hash]
+    # @yield [Hash] {id:, name: "institute"}
     def account(&block)
       get(api_query: 'account/institution', &block)
     end
@@ -137,12 +137,13 @@ module Figshare
     end
 
     # Get the accounts for which the account has administrative privileges (assigned and inherited).
+    # Accounts are returned in account id order, and there is a 9000 user upper limit. See id_gte.
     #
     # @param is_active [Boolean] user account is active
     # @param institution_user_id [String] As set in the HR upload
     # @param email [String] as set in the HR upload
-    # @param id_lte [Integer] ID is <=
-    # @param id_gte [Integer] ID is >=
+    # @param id_lte [Integer] ID is <= Introduced to get around the buffer limit of 9000 users
+    # @param id_gte [Integer] ID is >= Introduced to get around the buffer limit of 9000 users
     # @param page [Numeric] Pages start at 1. Page and Page size go together
     # @param page_size [Numeric]
     # @param offset [Numeric] offset is 0 based.  Offset and Limit go together
@@ -157,6 +158,7 @@ module Figshare
                   page_size: nil,
                   offset: nil,
                   limit: nil,
+                  cursor_pagination: true,
                   &block
                 )
       args = {}
@@ -169,7 +171,26 @@ module Figshare
       args['page_size'] = page_size unless page_size.nil?
       args['offset'] = offset unless offset.nil?
       args['limit'] = limit unless limit.nil?
-      get_paginate(api_query: 'account/institution/accounts', args: args, &block)
+      if cursor_pagination
+        highest_id_gte = 0
+        args['id_lte'] = nil
+        args['page_size'] = 100
+        loop do
+          count = 0
+          args['page'] = 1
+          args['id_gte'] = highest_id_gte + 1
+          get_paginate(api_query: 'account/institution/accounts', args: args, once_only: true) do |account|
+            next if account.nil? || account['id'].nil?
+
+            count += 1
+            highest_id_gte = account['id'].to_i if highest_id_gte < account['id'].to_i
+            yield account
+          end
+          break if count < 100 # Didn't reach the page_size limit.
+        end
+      else # Do it the old broken way (Pagination only works for the first 9000 entries)
+        get_paginate(api_query: 'account/institution/accounts', args: args, &block)
+      end
     end
 
     # Create new Institution Account

data/lib/private_projects.rb

@@ -363,7 +363,7 @@ module Figshare
     # @param project_id [Integer] Figshare id of the project
     # @param article_id [Integer] Figshare id of the article
     # @param impersonate [Integer] Figshare account_id of the user we are making this call on behalf of
-    # @yield [Array] See docs.figshare.com for article hash
+    # @yield [Array] See docs.figshare.com
     def artilce_files(project_id:, article_id:, impersonate: nil, &block)
       args = {}
       args['impersonate'] = impersonate unless impersonate.nil?

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: figshare_api_v2
 version: !ruby/object:Gem::Version
-  version: 0.9.9
+  version: 0.9.10
 platform: ruby
 authors:
 - Rob Burrowes
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2022-05-10 00:00:00.000000000 Z
+date: 2022-05-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: wikk_json