Adding support for thesis department

config/thesisdeptmap.csv

@@ -0,0 +1,80 @@
+campus_unitid,dept_unitid,thesis_department
+ucsc_etd,ucsc_anthro_etd,Anthropology
+ucsc_etd,ucsc_anthro_etd,Anthropology (Feminist Studies)
+ucsc_etd,ucsc_anthro_etd,Anthropology (Latin American and Latino Studies)
+ucsc_etd,ucsc_appms_etd,Applied Mathematics and Statistics
+ucsc_etd,ucsc_appms_etd,Applied Mathematics and Statistics (Robotics and Control)
+ucsc_etd,ucsc_astro_etd ,Astronomy & Astrophysics
+ucsc_etd,ucsc_astro_etd ,Astronomy and Astrophysics (Statistics)
+ucsc_etd,ucsc_chem_etd ,Biochemistry
+ucsc_etd,ucsc_chem_etd ,Bioinformatics
+ucsc_etd,ucsc_biomeng_etd,Biomolecular Engineering and Bioinformatics
+ucsc_etd,ucsc_chem_etd ,Chemistry
+ucsc_etd,ucsc_compmed_etd,Computational Media
+ucsc_etd,ucsc_cseng_etd,Computer Engineering
+ucsc_etd,ucsc_cseng_etd,Computer Engineering (Robotics and Control)
+ucsc_etd,ucsc_cseng_etd,Computer Science
+ucsc_etd,ucsc_digartmedia_etd,Digital Arts and New Media
+ucsc_etd,ucsc_earth_etd,Earth Science
+ucsc_etd,ucsc_eeb_etd,Ecology and Evolutionary Biology
+ucsc_etd,ucsc_eeb_etd,Ecology and Evolutionary Biology (Coastal Science and Policy)
+ucsc_etd,ucsc_eeb_etd,Ecology and Evolutionary Biology (Education)
+ucsc_etd,ucsc_eeb_etd,Ecology and Evolutionary Biology (Environmental Studies)
+ucsc_etd,ucscecon_etd,Economics
+ucsc_etd,ucsced_etd,Education
+ucsc_etd,ucsced_etd,Education (Sociology)
+ucsc_etd,ucsc_eceng_etd,Electrical Engineering
+ucsc_etd,ucsc_eceng_etd,Electrical Engineering (Robotics and Control)
+ucsc_etd,ucsc_env_etd,Environmental Studies
+ucsc_etd,ucsc_env_etd,Environmental Studies (Ecology & Evolutionary Biology)
+ucsc_etd,ucsc_env_etd,Environmental Studies (Sociology)
+ucsc_etd,ucsc_env_etd,Environmental Toxicology
+ucsc_etd,ucscfeministstudies_etd,Feminist Studies
+ucsc_etd,ucsc_film_etd,Film & Digital Media (Visual Studies)
+ucsc_etd,ucsc_film_etd,Film and Digital Media
+ucsc_etd,ucsc_hist_etd,History
+ucsc_etd,ucsc_hist_etd,History (Feminist Studies)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (American Studies)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (Anthropology)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (Feminist Studies)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (Literature)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (Philosophy)
+ucsc_etd,ucsc_histcons_etd,History of Consciousness (Visual Studies)
+ucsc_etd,ucscecon_etd,International Economics
+ucsc_etd,ucsc_latin_etd,Latin American Latino Studies
+ucsc_etd,ucsc_ling_etd,Linguistics
+ucsc_etd,ucsc_lit_etd,Literature
+ucsc_etd,ucsc_lit_etd,Literature (American Studies)
+ucsc_etd,ucsc_lit_etd,Literature (Feminist Studies)
+ucsc_etd,ucscmath_etd,Mathematics
+ucsc_etd,metx_etd,Microbiology and Environmental Toxicology
+ucsc_etd,ucsc_cellbio_etd,Molecular Cell and Developmental Biology
+ucsc_etd,ucsc_music_etd ,Music
+ucsc_etd,ucsc_music_etd ,Music Composition
+ucsc_etd,ucsc_ocean_etd,Ocean Sciences
+ucsc_etd,ucsc_philo_etd,Philosophy
+ucsc_etd,ucsc_phys_etd,Physics
+ucsc_etd,ucsc_pol_etd,Politics
+ucsc_etd,ucsc_pol_etd,Politics (American Studies)
+ucsc_etd,ucsc_pol_etd,Politics (Environmental Studies)
+ucsc_etd,ucsc_pol_etd,Politics (Feminist Studies and Latin American & Latino Studies)
+ucsc_etd,ucsc_pol_etd,Politics (Feminist Studies)
+ucsc_etd,ucsc_pol_etd,Politics (History of Consciousness and Latin American & Latino Studies)
+ucsc_etd,ucsc_pol_etd,Politics (History of Consciousness)
+ucsc_etd,ucsc_pol_etd,Politics (Latin American and Latino Studies)
+ucsc_etd,ucsc_psych_etd,Psychology
+ucsc_etd,ucsc_psych_etd,Psychology (Feminist Studies)
+ucsc_etd,ucsc_psych_etd,Psychology (Latin American and Latino Studies)
+ucsc_etd,ucsc_cseng_etd,Scientific Computing and Applied Mathematics
+ucsc_etd,ucsc_socio_etd,Sociology
+ucsc_etd,ucsc_socio_etd,Sociology (Anthropology)
+ucsc_etd,ucsc_socio_etd,Sociology (Feminist Studies and Latin American & Latino Studies)
+ucsc_etd,ucsc_socio_etd,Sociology (Feminist Studies)
+ucsc_etd,ucsc_socio_etd,Sociology (Latin American and Latino Studies)
+ucsc_etd,ucsc_socio_etd,Sociology (Philosophy)
+ucsc_etd,ucsc_stats_etd,Statistical Science
+ucsc_etd,ucsc_stats_etd,Statistics and Applied Mathematics
+ucsc_etd,ucsc_cseng_etd,Technology and Information Management
+ucsc_etd,ucsc_ppd_etd,Theater Arts
+ucsc_etd,ucsc_arthist_etd ,Visual Studies

tools/convert.rb

@@ -35,6 +35,8 @@
 require 'sequel'
 require 'time'
 require 'unindent'
+require 'csv'
+require 'cgi'
 require_relative '../util/nailgun.rb'
 require_relative '../util/normalize.rb'
 require_relative '../util/sanitize.rb'
@@ -111,6 +113,16 @@ def getEnv(name)
 $issueCoverCache = {}
 $issueNumberingCache = {}
 
+# Initialize the nested map for thesis department
+$thesisdeptmap = Hash.new { |h, k| h[k] = {} }
+mappath = File.join(__dir__, '..', 'config', 'thesisdeptmap.csv')
+CSV.foreach(mappath, headers: true) do |row|
+  campus = row['campus_unitid'].strip
+  department = CGI.unescapeHTML(row['thesis_department']).strip
+  dept_unitid = row['dept_unitid'].strip
+  $thesisdeptmap[campus][department] = dept_unitid
+end
+#puts $thesisdeptmap
 # Make puts thread-safe, and prepend each line with the thread it's coming from. While we're at it,
 # let's auto-flush the output.
 $stdoutMutex = Mutex.new
@@ -1001,6 +1013,7 @@ def tryMainAndSequester(path)
 
 ###################################################################################################
 def parseUCIngest(itemID, inMeta, fileType, isPending)
+  puts "IN Parse"
   attrs = {}
   attrs[:addl_info] = inMeta.html_at("./comments") and sanitizeHTML(inMeta.html_at("./comments"))
   attrs[:author_hide] = !!inMeta.at("./authors[@hideAuthor]")   # Only journal items can have this attribute
@@ -1023,7 +1036,9 @@ def parseUCIngest(itemID, inMeta, fileType, isPending)
   attrs[:pub_submit] = parseDate(inMeta.text_at("./context/dateSubmitted"))
   attrs[:pub_accept] = parseDate(inMeta.text_at("./context/dateAccepted"))
   attrs[:pub_publish] = parseDate(inMeta.text_at("./context/datePublished"))
-
+  attrs[:thesis_dept] = inMeta.text_at("./context/department")
+  puts "ATTRS THESIS Dept #{attrs[:thesis_dept]}"
+  #
   # Record submitter (especially useful for forensics)
   attrs[:submitter] = inMeta.xpath("./history/stateChange").map { |sc|
     sc[:state] =~ /^(new|uploaded|pending|published)/ && sc[:who] ? sc[:who] : nil
@@ -1261,7 +1276,7 @@ def parseUCIngest(itemID, inMeta, fileType, isPending)
                            dbItem[:content_type].nil? &&
                            attrs[:supp_files]) ? "multimedia" :
                           fileType == "ETD" ? "dissertation" :
-                          inMeta[:type] ? inMeta[:type].sub("paper", "article") :
+                          inMeta[:type] ? inMeta[:type].sub("paper", "article").sub("etd","dissertation"):
                           "article"
   dbItem[:submitted]    = submissionDate
   dbItem[:added]        = addDate
@@ -1442,6 +1457,18 @@ def indexItem(itemID, batch, nailgun)
       parseUCIngest(itemID, rawMeta, "UCIngest", isPending)
   end
 
+  # Add department unit id if possible
+  if attrs[:thesis_dept]
+    puts "found thesis dept"
+    etd_units = units.select { |unit| unit.include?('_etd') }
+    etd_units.each do |campus|
+       dept_unitid = $thesisdeptmap.dig(campus, attrs[:thesis_dept])
+       puts dept_unitid
+       units << dept_unitid  unless dept_unitid.nil?
+       end
+    puts "All units"
+    puts units
+    end
   text = $noCloudSearchMode ? "" : grabText(itemID, dbItem.content_type)
 
   # Create JSON for the full text index

tools/updIndex.rb

@@ -100,6 +100,11 @@
                                                               search_enabled:    true,
                                                               facet_enabled:     true,
                                                               sort_enabled:      false } },
+  { index_field_name: "thesis_dept", index_field_type:      "literal",
+                                     literal_options:         { facet_enabled:    true,
+                                                              search_enabled: true,
+                                                              return_enabled: true,
+                                                              sort_enabled:      true } },
 ]
 
 ###################################################################################################