fix: safety sheets are not imported and significantly slow down the i…

…mport (#1779)

* define timeout for safety data sheet requests

* modify header for safety data sheet merck request and improve product number extraction

* fix regex for extracting merck product number

* add sleep time between fetch sds merck api requests

* fix: handle empty safety phrases objects in chemicalTab component

* refactor: remove check_if_safety_sheet_already_saved function in chemicalsService class

app/packs/src/components/ChemicalTab.js

@@ -219,19 +219,23 @@ export default class ChemicalTab extends React.Component {
   // eslint-disable-next-line class-methods-use-this
   stylePhrases = (str) => {
     const HazardPhrases = [];
-    // eslint-disable-next-line no-restricted-syntax
-    for (const [key, value] of Object.entries(str.h_statements)) {
-      // eslint-disable-next-line react/jsx-one-expression-per-line
-      const st = <p key={key}> {key}:{value} </p>;
-      HazardPhrases.push(st);
+    if (str && str.h_statements && str.h_statements.length !== 0) {
+      // eslint-disable-next-line no-restricted-syntax
+      for (const [key, value] of Object.entries(str.h_statements)) {
+        // eslint-disable-next-line react/jsx-one-expression-per-line
+        const st = <p key={key}> {key}:{value} </p>;
+        HazardPhrases.push(st);
+      }
     }
 
     const precautionaryPhrases = [];
-    // eslint-disable-next-line no-restricted-syntax
-    for (const [key, value] of Object.entries(str?.p_statements || {})) {
-      // eslint-disable-next-line react/jsx-one-expression-per-line
-      const st = <p key={key}>{key}:{value}</p>;
-      precautionaryPhrases.push(st);
+    if (str && str.p_statements && str?.p_statements?.length !== 0) {
+      // eslint-disable-next-line no-restricted-syntax
+      for (const [key, value] of Object.entries(str?.p_statements || {})) {
+        // eslint-disable-next-line react/jsx-one-expression-per-line
+        const st = <p key={key}>{key}:{value}</p>;
+        precautionaryPhrases.push(st);
+      }
     }
 
     const pictogramsArray = str.pictograms?.map((i) => (
@@ -582,7 +586,7 @@ export default class ChemicalTab extends React.Component {
       vendor_product: vendorProduct
     };
     ChemicalFetcher.saveSafetySheets(params).then((result) => {
-      if (result || result === 'file is already saved') {
+      if (result) {
         const value = `/safety_sheets/${productInfo.productNumber}_${productInfo.vendor}.pdf`;
         const chemicalData = chemical._chemical_data;
         const pathArr = [];

lib/chemotion/chemicals_service.rb

@@ -4,10 +4,14 @@ module Chemotion
   class ChemicalsService
     def self.request_options
       { headers: {
-        'Access-Control-Request-Method' => 'GET',
-        'Accept' => '*/*',
-        'User-Agent': 'Google Chrome',
-      } }
+          'Access-Control-Request-Method' => 'GET',
+          'Accept' => '*/*',
+          'User-Agent' => 'Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120.0.0.0',
+          'Accept-Encoding': 'gzip, deflate, br',
+          Connection: 'keep-alive',
+        },
+        timeout: 15,
+        follow_redirects: true }
     end
 
     def self.merck_request(name)
@@ -50,37 +54,26 @@ def self.alfa(name, language)
       'Could not find safety data sheet from Thermofisher'
     end
 
-    def self.check_if_safety_sheet_already_saved(file_name, safety_sheet_files_names)
-      saved = false
-      unless  safety_sheet_files_names.empty?
-        safety_sheet_files_names.each do |file|
-          if file == file_name
-            saved = true
-            break
-          end
-        end
-      end
-      saved
-    end
-
     def self.write_file(file_path, link)
-      req_safety_sheet = HTTParty.get(link, request_options)
+      options = request_options.dup
+      options[:headers]['Origin'] = 'https://www.sigmaaldrich.com'
+      req_safety_sheet = HTTParty.get(link, options)
       file_name = "public/safety_sheets/#{file_path}"
       if req_safety_sheet.headers['Content-Type'] == 'application/pdf'
         File.binwrite(file_name, req_safety_sheet)
         true
       else
         'there is no file to save'
       end
+    rescue HTTParty::RedirectionTooDeep => e
+      "Redirection limit exceeded: #{e}"
+    rescue HTTParty::TimeoutError => e
+      "Request timed out: #{e}"
     end
 
     def self.create_sds_file(file_path, link)
-      safety_sheet_files_names = Dir.children('public/safety_sheets')
-      if check_if_safety_sheet_already_saved(file_path, safety_sheet_files_names) == false
-        write_file(file_path, link)
-      else
-        'file is already saved'
-      end
+      write_file(file_path, link)
+      sleep 1
     rescue StandardError
       'could not save safety data sheet'
     end

lib/import/import_chemicals.rb

@@ -106,7 +106,9 @@ def self.extract_product_number(url)
         match[1]
       else
         path = url.split('/')
-        path.last
+        filter = path.last
+        # extract digits from the string
+        filter.match(/.*?(?=[$%&?]|$)/)&.[](0)
       end
     end
 
@@ -115,7 +117,7 @@ def self.create_safety_sheet_path(vendor, value, product_number, chemical)
       chemical['chemical_data'][0]['safetySheetPath'] ||= []
       sheet_path = { "#{vendor}_link" => "#{SAFETY_SHEET_PATH}#{file_path}" }
       is_created = Chemotion::ChemicalsService.create_sds_file(file_path, value)
-      result = [true, 'file is already saved'].include?(is_created)
+      result = [true].include?(is_created)
       chemical['chemical_data'][0]['safetySheetPath'] << sheet_path if result
       chemical
     end

spec/lib/chemotion/chemicals_service_spec.rb

@@ -4,26 +4,6 @@
 
 describe Chemotion::ChemicalsService do
   describe Chemotion::ChemicalsService do
-    context 'when check_if_safety_sheet_already_saved is called' do
-      it 'returns true when file is already saved' do
-        file_name = 'safety_file.pdf'
-        safety_sheet_files_names = ['safety_file.pdf', 'other_file.pdf']
-        expect(described_class.check_if_safety_sheet_already_saved(file_name, safety_sheet_files_names)).to be_truthy
-      end
-
-      it 'returns false when file is not already saved' do
-        file_name = 'safety_file.pdf'
-        safety_sheet_files_names = ['other_file.pdf', 'another_file.pdf']
-        expect(described_class.check_if_safety_sheet_already_saved(file_name, safety_sheet_files_names)).to be_falsey
-      end
-
-      it 'returns false when safety_sheet_files_names is empty' do
-        file_name = 'safety_file.pdf'
-        safety_sheet_files_names = []
-        expect(described_class.check_if_safety_sheet_already_saved(file_name, safety_sheet_files_names)).to be_falsey
-      end
-    end
-
     context 'when write_file is called' do
       let(:link) { 'https://www.sigmaaldrich.com/DE/en/sds/sigald/383112' }
       let(:file_path) { '252549_Merck.pdf' }
@@ -55,15 +35,6 @@
         result = described_class.create_sds_file(file_path, link)
         expect(result).to be_truthy
       end
-
-      it 'returns file is already saved if already saved' do
-        file_path = '252549_Merck.pdf'
-        link = 'https://www.sigmaaldrich.com/US/en/sds/sial/252549'
-        allow(described_class).to receive(:check_if_safety_sheet_already_saved)
-          .with(file_path, anything).and_return(true)
-        result = described_class.create_sds_file(file_path, link)
-        expect(result).to eq('file is already saved')
-      end
     end
 
     context 'when health_section is called' do