parse audience based on presence in description

lib/ingestors/event_ingestion.rb

@@ -22,6 +22,37 @@ def convert_location(input)
       input
     end
 
+    def parse_audience(description)
+      audience_mapping = {
+        'post-docs': 'researchers',
+        "PhD's candidate": 'researchers',
+        'PhD student': 'researchers',
+        'principal investigator': 'researchers',
+        'professor': 'researchers',
+        'scientist': 'researchers',
+        'library staff': 'research support staff',
+        'research librarian': 'research support staff',
+        'information specialist': 'research support staff',
+        'archivist': 'research support staff',
+        'repository manager': 'research support staff',
+        'data steward': 'research support staff',
+        'data manager': 'research support staff',
+        'data professional': 'research support staff',
+        'data engineer': 'research support staff',
+        'software engineer': 'research support staff',
+        'data librarian': 'research support staff',
+        'bachelor': 'students',
+        'master': 'students',
+        'teacher': 'trainers',
+        'coaches': 'trainers',
+        'educator': 'trainers',
+      }
+      audience_mapping
+        .select{ |key, val| description.downcase.include?(key.to_s.downcase) }
+        .values
+        .uniq
+    end
+
     def parse_dates(input, timezone = nil)
       Time.use_zone(timezone) do
         # try to split on obvious interval markers

lib/ingestors/taxila/dans_ingestor.rb

@@ -64,6 +64,7 @@ def process_dans(url)
 
             event.source = 'DANS'
             event.timezone = 'Amsterdam'
+            event.target_audience = parse_audience(event.description)
 
             add_event(event)
           rescue Exception => e

lib/ingestors/taxila/dtls_ingestor.rb

@@ -72,6 +72,7 @@ def process_dtls(url)
             event.set_default_times
             event.source = 'DTL'
             event.timezone = 'Amsterdam'
+            event.target_audience = parse_audience(event.description)
             add_event(event)
           rescue Exception => e
             @messages << "Extract event fields failed with: #{e.message}"

lib/ingestors/taxila/han_ingestor.rb

@@ -50,6 +50,7 @@ def process_han(_url)
           event.venue = "#{venue_super_css.text} #{venue_sub_css.text}"
           event.source = "HAN"
           event.timezone = 'Amsterdam'
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e

lib/ingestors/taxila/lcrdm_ingestor.rb

@@ -58,6 +58,7 @@ def process_lcrdm(url)
           event.source = 'LCRDM'
           event.timezone = 'Amsterdam'
           event.set_default_times
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e

lib/ingestors/taxila/leiden_ingestor.rb

@@ -83,6 +83,7 @@ def process_leiden(url)
             # does TeSS support that?
 
             event.source = 'Universiteit Leiden'
+            event.target_audience = parse_audience(event.description)
 
             add_event(event)
           rescue Exception => e

lib/ingestors/taxila/maastricht_ingestor.rb

@@ -58,6 +58,7 @@ def process_maastricht(url)
             event.timezone = 'Europe/Amsterdam' # how to get this from Icalendar Event object?
 
             event.source = 'Maastricht University'
+            event.target_audience = parse_audience(event.description)
 
             add_event(event)
           rescue Exception => e

lib/ingestors/taxila/nwo_ingestor.rb

@@ -47,6 +47,7 @@ def process_nwo(url)
             event.url = "https://www.nwo.nl#{event_data.css('h3.card__title > a').attribute('href').value}"
 
             event.source = 'NWO'
+            event.target_audience = parse_audience(event.description)
 
             add_event(event)
           rescue Exception => e

lib/ingestors/taxila/odissei_ingestor.rb

@@ -59,6 +59,7 @@ def process_odissei(_url)
           event.source = 'ODISSEI'
           event.timezone = 'Amsterdam'
           event.set_default_times
+          event.target_audience = parse_audience(event.description)
           add_event(event)
         rescue Exception => e
           @messages << "Extract event fields failed with: #{e.message}"

lib/ingestors/taxila/oscd_ingestor.rb

@@ -57,6 +57,7 @@ def process_oscd(url)
             end
             if div&.next_sibling&.next_sibling.nil? || (div&.next_sibling&.next_sibling&.name == 'h1')
               event.set_default_times
+              event.target_audience = parse_audience(event.description)
               add_event(event)
             end
           end

lib/ingestors/taxila/oscm_ingestor.rb

@@ -57,6 +57,7 @@ def process_oscm(url)
             # it's not really needed since dtstart and dtend contain timezone information
             event.source = 'OSCM'
             event.online = true
+            event.target_audience = parse_audience(event.description)
 
             # add event to events array
             add_event(event)

lib/ingestors/taxila/rdnl_ingestor.rb

@@ -56,6 +56,7 @@ def process_rdnl(url)
           event.source = 'RDNL'
           event.timezone = 'Amsterdam'
           event.set_default_times
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e

lib/ingestors/taxila/rug_ingestor.rb

@@ -50,6 +50,7 @@ def process_rug(url)
           event.source = 'RUG'
           event.timezone = 'Amsterdam'
           event.set_default_times
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e

lib/ingestors/taxila/surf_ingestor.rb

@@ -56,6 +56,7 @@ def process_surf(url)
             event.source = 'SURF'
             event.online = true
             event.timezone = 'Amsterdam'
+            event.target_audience = parse_audience(event.description)
 
             # add event to events array
             add_event(event)

lib/ingestors/taxila/tdcc_ingestor.rb

@@ -59,6 +59,7 @@ def process_tdcc(url)
             event.start += 1.year
             event.end += 1.year
           end
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e

lib/ingestors/taxila/utwente_ingestor.rb

@@ -52,6 +52,7 @@ def process_utwente(url)
           event.timezone = 'Amsterdam'
           event.organizer = 'University of Twente'
           event.source = 'University of Twente'
+          event.target_audience = parse_audience(event.description)
           add_event(event)
         rescue Exception => e
           @messages << "Extract event fields failed with: #{e.message}"

lib/ingestors/taxila/uu_ingestor.rb

@@ -106,6 +106,7 @@ def process_uu(url)
             event.timezone = 'Amsterdam'
             # UU wants opt-in instead of opt-out for this scraper
             event.visible = false
+            event.target_audience = parse_audience(event.description)
 
             # the below code allows fetching the long description, at the cost of a
             # page load per event.

lib/ingestors/taxila/uva_ingestor.rb

@@ -57,6 +57,7 @@ def process_uva(url)
           event.keywords = attr.fetch('taxonomy', []).map(&:values).flatten
 
           event.event_types = attr.fetch('eventType', []).map { |t| convert_event_types(t) }
+          event.target_audience = parse_audience(event.description)
 
           # add event to events array
           add_event(event)

lib/ingestors/taxila/wur_ingestor.rb

@@ -82,6 +82,7 @@ def process_wur(url)
           event.set_default_times
           event.source = 'WUR'
           event.timezone = 'Amsterdam'
+          event.target_audience = parse_audience(event.description)
 
           add_event(event)
         rescue Exception => e