From b3a4ad089217617fd15ecba40b2d5e392b14aec3 Mon Sep 17 00:00:00 2001
From: Adam Coffman <acc@fastmail.com>
Date: Fri, 25 Oct 2024 10:22:23 -0500
Subject: [PATCH] update clinical trial scraper to use new api, auto fetch
 titles and descriptions

---
 .../jobs/populate_clinical_trial_record.rb    |  5 ++++
 server/app/lib/scrapers/clinical_trial.rb     | 27 +++++++++++++++++++
 .../lib/scrapers/clinical_trial_response.rb   | 24 +++++++++++++++++
 server/app/models/clinical_trial.rb           | 12 ++++++++-
 4 files changed, 67 insertions(+), 1 deletion(-)
 create mode 100644 server/app/jobs/populate_clinical_trial_record.rb
 create mode 100644 server/app/lib/scrapers/clinical_trial.rb
 create mode 100644 server/app/lib/scrapers/clinical_trial_response.rb

diff --git a/server/app/jobs/populate_clinical_trial_record.rb b/server/app/jobs/populate_clinical_trial_record.rb
new file mode 100644
index 000000000..755335dbe
--- /dev/null
+++ b/server/app/jobs/populate_clinical_trial_record.rb
@@ -0,0 +1,5 @@
+class PopulateClinicalTrialRecord < ApplicationJob
+  def perform(clinical_trial)
+    Scrapers::ClinicalTrial.populate_fields(clinical_trial)
+  end
+end
diff --git a/server/app/lib/scrapers/clinical_trial.rb b/server/app/lib/scrapers/clinical_trial.rb
new file mode 100644
index 000000000..6a40ec8ac
--- /dev/null
+++ b/server/app/lib/scrapers/clinical_trial.rb
@@ -0,0 +1,27 @@
+module Scrapers
+  module ClinicalTrial
+    def self.populate_fields(clinical_trial)
+      resp = call_clinical_trials_api(clinical_trial.nct_id)
+      clinical_trial.description = resp.description
+      clinical_trial.name = resp.name
+      clinical_trial.save
+    end
+
+    def self.call_clinical_trials_api(nct_id)
+      http_resp = Util.make_get_request(url_for_nct_id(nct_id))
+      ClinicalTrialResponse.new(http_resp)
+    end
+
+    def self.url_for_nct_id(nct_id)
+      "https://clinicaltrials.gov/api/v2/studies/#{nct_id}?format=json&fields=#{fields}"
+    end
+
+    def self.fields
+      [
+        "protocolSection.identificationModule.nctId",
+        "protocolSection.identificationModule.briefTitle",
+        "protocolSection.descriptionModule.briefSummary"
+      ].join(",")
+    end
+  end
+end
diff --git a/server/app/lib/scrapers/clinical_trial_response.rb b/server/app/lib/scrapers/clinical_trial_response.rb
new file mode 100644
index 000000000..0e6729458
--- /dev/null
+++ b/server/app/lib/scrapers/clinical_trial_response.rb
@@ -0,0 +1,24 @@
+module Scrapers
+  class ClinicalTrialResponse
+    attr_reader :json
+    def initialize(response_body)
+      @json = JSON.parse(response_body)
+    end
+
+    def nct_id
+      json.dig("protocolSection", "identificationModule", "nctId")
+    end
+
+    def name
+      json.dig("protocolSection", "identificationModule", "briefTitle")
+    end
+
+    def description
+      if desc = json.dig("protocolSection", "descriptionModule", "briefSummary")
+        desc.squish
+      else
+        nil
+      end
+    end
+  end
+end
diff --git a/server/app/models/clinical_trial.rb b/server/app/models/clinical_trial.rb
index d46abab0c..8eb19074e 100644
--- a/server/app/models/clinical_trial.rb
+++ b/server/app/models/clinical_trial.rb
@@ -1,15 +1,25 @@
 class ClinicalTrial < ActiveRecord::Base
   has_and_belongs_to_many :sources
 
+  after_create :populate_additional_fields_if_needed
+
   def self.url_for(nct_id:)
     if nct_id.blank?
       nil
     else
-      "https://clinicaltrials.gov/ct2/show/#{nct_id}"
+      "https://clinicaltrials.gov/study/#{nct_id}"
     end
   end
 
   def link
     Rails.application.routes.url_helpers.url_for("/clinical-trials/#{self.id}")
   end
+
+ 
+  private
+  def populate_additional_fields_if_needed
+    if self.name.blank? || self.description.blank?
+      PopulateClinicalTrialRecord.perform_later(self)
+    end
+  end
 end