Fix String#similarity to divide by the longer length

similarity divided the shared length by (longest length + 1), so 9 of
10 matching characters scored 0.818 instead of the 0.9 its docs showed,
and non-identical strings could never approach 1.0. It now divides by
the longest length.

The +1 did stop one thing: comparing an empty String with a zero-length
non-String (an empty Symbol or Array) gets past the equality check, and
0.0/0 is NaN. That case now returns 0.0 explicitly.

Restores 0.9 in the docs and demo, updates the unit test's expected
value (0.75 -> 0.818), and adds tests for the 9-of-10 case and the empty
non-String arguments.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01E2Njj4j1r9twkVcXsMySLW

Author: claude

HISTORY.md

@@ -62,6 +62,9 @@ Changes:
     quietly changed both.
   * `Enumerable::Argumentable#min` and `#max` now pass their arguments to
     `#each`, as documented, rather than to Ruby's own `min(n)`/`max(n)`.
+  * `String#similarity` now divides the shared length by the longer string's
+    length, not length + 1, so 9 of 10 matching characters score 0.9 as its
+    docs said (it gave 0.818). Scores for non-identical strings rise slightly.
   * `require 'facets'` no longer crashes on Ruby 4.1, which removed
     `ObjectSpace._id2ref`. `ObjectSpace[]` is only defined where `_id2ref`
     exists.

demo/core/string/similarity.md

@@ -5,7 +5,7 @@
 A fuzzy matching mechanism. Returns a score from 0-1, based on the number of
 shared edges. To be effective, the strings must be of length 2 or greater.
 
-    "Alexsander".similarity("Aleksander").round(2).assert == 0.82
+    "Alexsander".similarity("Aleksander").assert == 0.9
 
     "Alexander".similarity("Alexander").assert == 1.0
 

lib/core/facets/string/similarity.rb

@@ -4,7 +4,7 @@ class String
   # on the number of shared edges. To be effective, the strings
   # must be of length 2 or greater.
   #
-  #     "Alexsander".similarity("Aleksander")  #=> 0.8181818181818182
+  #     "Alexsander".similarity("Aleksander")  #=> 0.9
   #
   # The way it works:
   #
@@ -82,8 +82,9 @@ def similarity(str_in)
 
     score = 0.0
     matches.each{ |mm| score += mm.length }
-    self.length > str_in.length ? largest = self.length : largest = str_in.length
-    return score/(largest+1)
+    largest = [self.length, str_in.length].max
+    return 0.0 if largest.zero?   # e.g. an empty Symbol or Array
+    return score/largest
   end
 
 end

test/core/string/test_similarity.rb

@@ -5,7 +5,7 @@
   method :similarity do
 
     test do
-      "Alexsanders".similarity("Aleksander").assert == 0.75
+      "Alexsanders".similarity("Aleksander").round(3).assert == 0.818
     end
 
     test do
@@ -16,6 +16,15 @@
       "Alexander".similarity("").assert == 0.0
     end
 
+    test "9 of 10 characters shared" do
+      "Alexsander".similarity("Aleksander").assert == 0.9
+    end
+
+    test "empty non-String argument" do
+      "".similarity(:"").assert == 0.0
+      "".similarity([]).assert == 0.0
+    end
+
   end
 
 end