Merge pull request #106 from sneakers-the-rat/handle-10_5555-prefix

Special case `10.5555` DOIs (and fix ambiguous DOI checker results)

Author: xuanxu

app/lib/doi_checker.rb

@@ -9,32 +9,52 @@ def initialize(entries=[])
   end
 
   def check_dois
-    doi_summary = {ok: [], missing: [], invalid: []}
+    doi_summary = {ok: [], skip: [], missing: [], invalid: []}
 
     if @entries.any?
       @entries.each do |entry|
-        if entry.has_field?('doi') && !entry.doi.empty?
+        # handle special cases first
+        special_case = self.handle_special_case(entry)
+        if special_case
+          doi_validity = special_case
+        elsif entry.has_field?('doi') && !entry.doi.empty?
+          # Validate entries with DOIs
           doi_validity = validate_doi(entry.doi.value)
-          doi_summary[doi_validity[:validity]].push(doi_validity[:msg])
-        # If there's no DOI present, check Crossref to see if we can find a candidate DOI for this entry.
         elsif entry.has_field?('title')
-            candidate_doi = crossref_lookup(entry.title.value)
-            truncated_title = entry.title.to_s[0,50]
-            truncated_title += "..." if truncated_title.length < entry.title.to_s.length
-            if candidate_doi == "CROSSREF-ERROR"
-              doi_summary[:missing].push("Errored finding suggestions for \"#{truncated_title}\", please try later")
-            elsif candidate_doi
-              doi_summary[:missing].push("#{candidate_doi} may be a valid DOI for title: #{truncated_title}")
-            else
-              doi_summary[:missing].push("No DOI given, and none found for title: #{truncated_title}")
-            end
+          # Try and find candidate entries if doi absent, but title present
+          doi_validity = handle_missing_doi(entry)
         else
-          doi_summary[:missing].push("Entry without DOI or title found")
+          doi_validity = {validity: :missing, msg: "Entry without DOI or title found"}
         end
+
+        doi_summary[doi_validity[:validity]].push(doi_validity[:msg])
       end
     end
 
     doi_summary
+    end
+
+  # any special case should return false if not applicable, and an object like
+  # {:validity => :ok, :msg => "whatever"} otherwise.
+  # Add additional special cases as private methods and chain in a tidy sequence plz <3
+  def handle_special_case(entry)
+    validity = acm_105555_prefix(entry) and return validity
+    false
+  end
+
+
+  # If there's no DOI present, check Crossref to see if we can find a candidate DOI for this entry.
+  def handle_missing_doi(entry)
+    candidate_doi = crossref_lookup(entry.title.value)
+    truncated_title = entry.title.to_s[0,50]
+    truncated_title += "..." if truncated_title.length < entry.title.to_s.length
+    if candidate_doi == "CROSSREF-ERROR"
+      { validity: :missing, msg: "Errored finding suggestions for \"#{truncated_title}\", please try later" }
+    elsif candidate_doi
+      { validity: :missing, msg: "#{candidate_doi} may be a valid DOI for title: #{truncated_title}" }
+    else
+      { validity: :skip, msg: "No DOI given, and none found for title: #{truncated_title}" }
+    end
   end
 
   def validate_doi(doi_string)
@@ -112,4 +132,16 @@ def levenshtein_distance(s, t)
   def similar?(string_1, string_2)
     levenshtein_distance(string_1, string_2) < 3
   end
+
+  private
+
+  def acm_105555_prefix(entry)
+    if entry.has_field?('doi') && entry.doi.include?("10.5555/")
+      { validity: :invalid, msg: "#{entry.doi} is INVALID - 10.5555 is a known broken prefix, replace with https://dl.acm.org/doi/{doi} in the {url} field" }
+    elsif entry.has_field?('url') && entry.url.include?("https://dl.acm.org/doi/10.5555")
+      { validity: :skip, msg: "#{entry.url} - correctly put 10.5555 prefixed doi in the url field, editor should ensure this resolves" }
+    else
+      false
+    end
+  end
 end

app/responses/doi_checks.erb

@@ -1,9 +1,13 @@
 ```
 Reference check summary (note 'MISSING' DOIs are suggestions that need verification):
 <% doi_summary.each do |type, messages| -%>
-
-<%= type.to_s.upcase %> DOIs
-
+<% if type.to_s === "ok" %>
+✅ - <%= type.to_s.upcase %> DOIs
+<% elsif type.to_s === "skip" %>
+❔ - <%= type.to_s.upcase %> DOIs
+<% else %>
+❌ - <%= type.to_s.upcase %> DOIs
+<% end %>
 <%  if messages.empty? -%>
 - None
 <%  else -%>

spec/doi_checker_spec.rb

@@ -93,7 +93,7 @@
 
       expect(doi_summary[:ok]).to be_empty
       expect(doi_summary[:invalid]).to be_empty
-      expect(doi_summary[:missing][0]).to eq("No DOI given, and none found for title: #{title}")
+      expect(doi_summary[:skip][0]).to eq("No DOI given, and none found for title: #{title}")
     end
 
     it "should report entries with no DOI or title as missing both" do
@@ -107,6 +107,33 @@
     end
   end
 
+  describe "#handle_special_case" do
+    it "should treat DOIs with a 10.5555 prefix as invalid" do
+      entry = BibTeX::Entry.new(doi: "10.5555/xxxxxxx.yyyyyyyyy")
+      validity = subject.handle_special_case(entry)
+      expect(validity[:validity]).to eq(:invalid)
+      expect(validity[:msg]).to include("replace with https://dl.acm.org/doi")
+    end
+
+    it "should treat URLs with a 10.5555 prefix as a skip" do
+      entry = BibTeX::Entry.new(url: "https://dl.acm.org/doi/10.5555/2827719.2827740")
+      validity = subject.handle_special_case(entry)
+      expect(validity[:validity]).to eq(:skip)
+      expect(validity[:msg]).to eq("https://dl.acm.org/doi/10.5555/2827719.2827740 - correctly put 10.5555 prefixed doi in the url field, editor should ensure this resolves")
+    end
+
+    it "should handle special cases separately from normal DOI checking" do
+      entry = BibTeX::Entry.new(doi: "10.5555/xxxxxxx.yyyyyyyyy")
+      doi_checker = DOIChecker.new([entry])
+
+      doi_summary = doi_checker.check_dois
+      expect(doi_summary[:ok]).to be_empty
+      expect(doi_summary[:missing]).to be_empty
+      expect(doi_summary[:skip]).to be_empty
+      expect(doi_summary[:invalid][0]).to include("is INVALID - 10.5555 is a known broken prefix, replace with https://dl.acm.org/doi/")
+    end
+  end
+
   describe "#validate_doi" do
 
     it "should invalidate empty doi strings" do