deleted the old datacite graphql code and replace with the rest api

src/lambdas/harvesters/datacite/.aws-sam/build.toml

@@ -1,6 +1,6 @@
 # This file is auto generated by SAM CLI build command
 
-[function_build_definitions.66e880ba-8a1c-4a71-9af7-3212a8fef2f9]
+[function_build_definitions.d040dc14-b551-4b81-b69b-29ae4079ae63]
 codeuri = "/Users/briley/Documents/workspace/dmsp_aws_prototype/src/lambdas/harvesters/datacite"
 runtime = "ruby3.2"
 architecture = "x86_64"

src/lambdas/harvesters/datacite/.aws-sam/build/Function/template.yaml

@@ -63,7 +63,7 @@ Resources:
           DYNAMO_INDEX_TABLE: !Ref DynamoIndexTableName
           SNS_FATAL_ERROR_TOPIC: !Ref SnsTopicEmailArn
       Events:
-        DataCiteHarvesting:
+        DataCiteHarvester:
           Type: 'EventBridgeRule'
           Properties:
             DeadLetterConfig:

src/lambdas/harvesters/datacite/.aws-sam/build/template.yaml

@@ -55,7 +55,7 @@ Resources:
           SNS_FATAL_ERROR_TOPIC:
             Ref: SnsTopicEmailArn
       Events:
-        DataCiteHarvesting:
+        DataCiteHarvester:
           Type: EventBridgeRule
           Properties:
             DeadLetterConfig:

src/lambdas/harvesters/datacite/template.yaml

@@ -63,7 +63,7 @@ Resources:
           DYNAMO_INDEX_TABLE: !Ref DynamoIndexTableName
           SNS_FATAL_ERROR_TOPIC: !Ref SnsTopicEmailArn
       Events:
-        DataCiteHarvesting:
+        DataCiteHarvester:
           Type: 'EventBridgeRule'
           Properties:
             DeadLetterConfig:

src/lambdas/harvesters/harvestable_dmps/app.rb

@@ -21,6 +21,39 @@ module Functions
   class HarvestableDmps
     SOURCE = 'Harvestable DMPs'
 
+    PILOT_DMPS = [
+      # Northwestern University
+      'doi.org/10.48321/D10B3E54E4',
+      'doi.org/10.48321/D1143FD15F',
+      'doi.org/10.48321/D139D84658',
+      'doi.org/10.48321/D1944C8215',
+      'doi.org/10.48321/D1A04A9B1D',
+
+      # University of California, Berkeley
+      'doi.org/10.48321/D114471AC3',
+      'doi.org/10.48321/D18F9B93B8',
+      'doi.org/10.48321/D1BA48FBC9',
+      'doi.org/10.48321/D1CE350633',
+      'doi.org/10.48321/D1DF9DDDAF',
+
+      # University of California, Riverside
+      'doi.org/10.48321/D13BEA529C',
+      'doi.org/10.48321/D14406894e',
+      'doi.org/10.48321/D145457051',
+      'doi.org/10.48321/D1FCB77AF0',
+      'doi.org/10.48321/D1FFBFF8FE',
+
+      # University of California, Santa Barbara
+      'doi.org/10.48321/D154FA23E9',
+      'doi.org/10.48321/D1A90CCC2B',
+      'doi.org/10.48321/D1BAD5B94D',
+      'doi.org/10.48321/D1FFE5D7FD',
+
+      # University of Colorado Boulder
+      'doi.org/10.48321/D14F38aa13',
+      'doi.org/10.48321/D1B581751F'
+    ]
+
     class << self
       def process(event:, context:)
         # No need to validate the source and detail-type because that is done by the EventRule
@@ -41,16 +74,18 @@ def process(event:, context:)
         docs = _fetch_relevant_dmps_from_dynamo(client: dynamo_client, table:, logger:)
         logger.debug(message: 'Relevant DMP search results: ', details: docs) if logger.respond_to?(:debug)
 
-        rors = docs.map { |doc| doc.fetch('affiliation_ids', []) }.flatten.compact.uniq
+        # rors = docs.map { |doc| doc.fetch('affiliation_ids', []) }.flatten.compact.uniq
+        affils = docs.map { |doc| doc.fetch('affiliations', []) }.flatten.compact.uniq
 
         # Kick off harvesters for each unique ROR id
         publisher = Uc3DmpEventBridge::Publisher.new
-        rors.each do |ror|
-          dmps = docs.select { |doc| doc.fetch('affiliation_ids', []).include?(ror) }
+        affils.each do |affil|
+          # dmps = docs.select { |doc| doc.fetch('affiliation_ids', []).include?(ror) }
+          dmps = docs.select { |doc| doc.fetch('affiliations', []).include?(affil) }
 
           # limit the number of DMPs we send at one time because SNS has a size limit
           dmps.each_slice(50) do |dmp_entries|
-            _kick_off_harvester(ror:, dmps: dmp_entries, publisher:, logger:)
+            _kick_off_harvester(affil:, dmps: dmp_entries, publisher:, logger:)
           end
 
           # Pause for a second. Publishing these messages kicks off multiple Lambda harvesters and we
@@ -91,10 +126,10 @@ def _open_search_connect(logger:)
         puts e.backtrace
       end
 
-      def _kick_off_harvester(ror:, dmps:, publisher: nil, logger: nil)
+      def _kick_off_harvester(affil:, dmps:, publisher: nil, logger: nil)
         # Publish the change to the EventBridge
         publisher = Uc3DmpEventBridge::Publisher.new if publisher.nil?
-        publisher.publish(source: 'HarvestableDmps', event_type: 'Harvest', dmp: {}, detail: { ror:, dmps: }, logger:)
+        publisher.publish(source: 'HarvestableDmps', event_type: 'Harvest', dmp: {}, detail: { ror: affil, dmps: }, logger:)
       end
 
       # Instead of OpenSearch (for now) grab the relevant DMPs from our Dynamo INdex table
@@ -116,7 +151,7 @@ def _dynamo_scan(client:, table:, items: [], last_key: '', logger: nil)
         args = {
           table_name: table,
           consistent_read: false,
-          projection_expression: 'PK, affiliation_ids',
+          projection_expression: 'PK, affiliations',
           # expression_attribute_values: {
           #   ':sk': 'METADATA',
           #   ':not_empty': '',
@@ -127,42 +162,7 @@ def _dynamo_scan(client:, table:, items: [], last_key: '', logger: nil)
           # filter_expression: expr.join(' AND ')
 
           expression_attribute_values: {
-            ':dmp_pks': [
-=begin
-              # Northwestern University
-              'doi.org/10.48321/D10B3E54E4',
-              'doi.org/10.48321/D1944C8215',
-              'doi.org/10.48321/D139D84658',
-              'doi.org/10.48321/D1A04A9B1D',
-
-              # University of Colorado Boulder
-              'doi.org/10.48321/D14F38aa13',
-              'doi.org/10.48321/D1B581751F',
-
-              # University of California, Santa Barbara
-              'doi.org/10.48321/D1BAD5B94D',
-              'doi.org/10.48321/D1FFE5D7FD',
-              'doi.org/10.48321/D1A90CCC2B',
-              'doi.org/10.48321/D154FA23E9',
-=end
-              # University of California, Berkeley
-              'doi.org/10.48321/D114471AC3',
-              'doi.org/10.48321/D1DF9DDDAF',
-              'doi.org/10.48321/D18F9B93B8',
-              'doi.org/10.48321/D1BA48FBC9',
-              'doi.org/10.48321/D1CE350633',
-=begin
-              # University of California, Riverside
-              'doi.org/10.48321/D14406894e',
-              'doi.org/10.48321/D145457051',
-              'doi.org/10.48321/D1FFBFF8FE',
-              'doi.org/10.48321/D1FCB77AF0',
-              'doi.org/10.48321/D13BEA529C',
-
-              # Boston University
-              'doi.org/10.48321/D1A04A9B1D'
-=end
-            ]
+            ':dmp_pks': PILOT_DMPS
           },
           filter_expression: 'contains(:dmp_pks, dmp_id)'
         }

src/lambdas/layers/baseline/Gemfile.lock

@@ -5,16 +5,16 @@ GEM
       public_suffix (>= 2.0.2, < 6.0)
     ast (2.4.2)
     aws-eventstream (1.3.0)
-    aws-partitions (1.923.0)
+    aws-partitions (1.925.0)
     aws-sdk-cognitoidentityprovider (1.90.0)
       aws-sdk-core (~> 3, >= 3.193.0)
       aws-sigv4 (~> 1.1)
-    aws-sdk-core (3.194.0)
+    aws-sdk-core (3.194.1)
       aws-eventstream (~> 1, >= 1.3.0)
       aws-partitions (~> 1, >= 1.651.0)
       aws-sigv4 (~> 1.8)
       jmespath (~> 1, >= 1.6.1)
-    aws-sdk-dynamodb (1.107.0)
+    aws-sdk-dynamodb (1.108.0)
       aws-sdk-core (~> 3, >= 3.193.0)
       aws-sigv4 (~> 1.1)
     aws-sdk-eventbridge (1.58.0)
@@ -120,7 +120,7 @@ GEM
       rubocop (~> 1.41)
     rubocop-factory_bot (2.25.1)
       rubocop (~> 1.41)
-    rubocop-rspec (2.29.1)
+    rubocop-rspec (2.29.2)
       rubocop (~> 1.40)
       rubocop-capybara (~> 2.17)
       rubocop-factory_bot (~> 2.22)
@@ -153,7 +153,7 @@ GEM
       aws-sdk-ssm (~> 1.150)
       httparty (~> 0.21.0)
       json (~> 2.6)
-    uc3-dmp-id (0.1.37)
+    uc3-dmp-id (0.1.39)
       json (~> 2.6)
       json-schema (~> 3.0)
       text (~> 1.3)

src/sam/gems/uc3-dmp-id/lib/uc3-dmp-id/comparator.rb

@@ -78,7 +78,7 @@ def compare(hash:)
         response = _text_match?(type: 'title', text: hash['title'], dmp:, response:) if response[:score].positive?
         response = _text_match?(type: 'abstract', text: hash['description'], dmp:, response:) if response[:score].positive?
         # If the score is less than 3 then we have no confidence that it is a match
-        next if response[:score] <= 2
+        # next if response[:score] <= 2
 
         # Set the confidence level based on the score
         response[:dmp_id] = "DMP##{dmp['dmp_id']}"

src/sam/gems/uc3-dmp-id/lib/uc3-dmp-id/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Uc3DmpId
-  VERSION = '0.1.38'
+  VERSION = '0.1.39'
 end