Merge pull request #72 from mlibrary/remove_punctuation_from_browse_s…

…tring

Clean up the user-submitted author browse string

catalog-browse.rb

@@ -6,6 +6,7 @@
 
 require_relative "lib/catalog_solr_client"
 require_relative "lib/utilities/browse_solr_client"
+require_relative "lib/utilities/string_cleaner"
 require_relative "lib/models/browse_list"
 require_relative "lib/models/browse_list_presenter"
 require_relative "lib/models/callnumber_list"
@@ -21,7 +22,7 @@
 
 if ENV.fetch("AUTHOR_ON") == "true"
   get "/author" do
-    author = params[:query]
+    author = StringCleaner.cleanup_author_browse_string(params[:query])
     reference_id = params[:reference_id] || author
     begin
       list = AuthorList.for(direction: params[:direction], reference_id: reference_id, num_rows_to_display: 20, original_reference: author, banner_reference: params[:banner_reference])

lib/utilities/string_cleaner.rb

@@ -0,0 +1,43 @@
+module StringCleaner
+  def self.prefixes
+    # are there more of these?
+    [
+      "academic_discipline",
+      "author",
+      "call_number_starts_with",
+      "contains",
+      "contributor",
+      "date",
+      "isbn",
+      "isn",
+      "issn",
+      "journal_title",
+      "keyword",
+      "pmid",
+      "publication_date",
+      "publisher",
+      "realuth",
+      "series",
+      "subject",
+      "title",
+      "title_starts_with"
+    ]
+  end
+
+  def self.strip_symbols(str)
+    # str.gsub(/[\p{P}\p{Sm}\p{Sc}\p{So}^`]/, "")
+    str.gsub(/["']/, "")
+  end
+
+  # TODO: add bits to remove field prefix (e.g., 'author:') as defined in 00-catalog.yml
+  # this is where the author browse specific cleaning goes
+  def self.cleanup_author_browse_string(str)
+    prefixes.each do |x|
+      if str.match?(/^#{x}:/)
+        str.sub!(/^#{x}:/, "")
+        break
+      end
+    end
+    strip_symbols(str)
+  end
+end

spec/utilities/string_cleaner_spec.rb

@@ -0,0 +1,15 @@
+describe StringCleaner do
+  context ".strip_symbols" do
+    it "removes certain punctuation" do
+      expect(described_class.strip_symbols('"\'')).to eq("")
+    end
+  end
+  context ".cleanup_author_browse_string" do
+    it "removes 'author:' prefix" do
+      expect(described_class.cleanup_author_browse_string('author:"Author Name"')).to eq("Author Name")
+    end
+    it "removes 'isn(' prefix" do
+      expect(described_class.cleanup_author_browse_string("isn:(123-456)")).to eq("(123-456)")
+    end
+  end
+end