Don't treat * dynamic width/precision as a fixed specifier

`%*d` / `%.*f` consume an extra `int` argument (the width/precision) that the
one-slot-per-specifier signature model can't represent, so the regex matched
them but undercounted — making `%d` and `%*d` compare as compatible. Concretely,
changing a base string from `%.1f km` to `%.*f km` shifts the call site from one
argument to two, yet the check would wave it through, so every unchanged
translation would render the precision where the value belongs.

Stop matching `*` in the width/precision, leaving such a specifier unrecognized:
a change to or from a `*` form now surfaces as a difference and gets flagged
rather than silently passing. Fixed width/precision/length specifiers are
unaffected.

Author: jkmassel

lib/fastlane/plugin/wpmreleasetoolkit/helper/string_placeholders_helper.rb

@@ -38,7 +38,12 @@ module StringPlaceholdersHelper
       # flags, width, precision, an optional length modifier, then the conversion
       # character. `%%` (a literal percent) is matched too so it can be explicitly
       # skipped rather than mistaken for a placeholder.
-      SPECIFIER = /%(?<position>\d+\$)?[-+ 0#]*(?:\d+|\*)?(?:\.(?:\d+|\*))?(?:hh|h|ll|l|q|L|z|t|j)?(?<conversion>[@diouxXeEfgGaAcCsSpn%])/
+      #
+      # `*` dynamic width/precision (`%*d`, `%.*f`) is deliberately NOT matched: it consumes an extra
+      # `int` argument that this one-slot-per-specifier model can't represent. Leaving it unrecognized
+      # is the safe choice — rather than silently treating `%*d` as compatible with `%d`, a change to
+      # or from a `*` specifier surfaces as a difference and gets flagged.
+      SPECIFIER = /%(?<position>\d+\$)?[-+ 0#]*\d*(?:\.\d+)?(?:hh|h|ll|l|q|L|z|t|j)?(?<conversion>[@diouxXeEfgGaAcCsSpn%])/
 
       # A canonical signature of the placeholders in a string value.
       #

spec/string_placeholders_helper_spec.rb

@@ -107,6 +107,16 @@
         expect(described_class.placeholder_signature('%@ and %@')).to eq('1:object,2:object')
       end
     end
+
+    context 'with `*` dynamic width or precision' do
+      # `%*d`/`%.*f` consume an extra `int` (the width/precision) that the one-slot-per-specifier
+      # model can't represent, so they are intentionally left unrecognized rather than silently
+      # treated as compatible with a fixed-width/precision specifier.
+      it 'does not recognize a `*` specifier (so it cannot masquerade as its fixed counterpart)' do
+        expect(described_class.placeholder_signature('%.*f km')).to eq('')
+        expect(described_class.placeholder_signature('%*d items')).to eq('')
+      end
+    end
   end
 
   describe '.placeholders_compatible?' do
@@ -141,6 +151,13 @@
     it 'treats adding a genuinely new positional argument as incompatible' do
       expect(described_class.placeholders_compatible?('%1$@', '%1$@ in %2$@')).to be(false)
     end
+
+    it 'flags a change between a fixed and a `*` dynamic width/precision specifier' do
+      # `%.1f km` consumes one float; `%.*f km` consumes an int (precision) + a float. Treating them
+      # as compatible would let an existing translation render the precision where the value belongs.
+      expect(described_class.placeholders_compatible?('%.1f km', '%.*f km')).to be(false)
+      expect(described_class.placeholders_compatible?('%5d items', '%*d items')).to be(false)
+    end
   end
 
   describe '.incompatible_placeholder_changes' do