Fix strformat neg zero (#7954)

* fix strformat handling of neg zero with sign

* better tests for neg zero with sign

* use inplace insertion of the sign as suggested by Varriount

Author: skilchen

lib/pure/strformat.nim

@@ -527,8 +527,13 @@ proc format*(value: SomeFloat; specifier: string; res: var string) =
   var sign = false
   if value >= 0.0:
     if spec.sign != '-':
-      f = spec.sign & f
       sign = true
+      if  value == 0.0:
+        if 1.0 / value == Inf:
+          # only insert the sign if value != negZero
+          f.insert($spec.sign, 0)
+      else:
+        f.insert($spec.sign, 0)
   else:
     sign = true
 

tests/stdlib/tstrformat.nim

@@ -46,3 +46,11 @@ doAssert fmt"{-1.5:0>8}" == "0000-1.5" # even that does not work for negative fl
 doAssert fmt"{-1.5:08}" == "-00001.5" # works
 doAssert fmt"{1.5:+08}" == "+00001.5" # works
 doAssert fmt"{1.5: 08}" == " 00001.5" # works
+
+# only add explicitly requested sign if value != -0.0 (neg zero)
+doAssert fmt"{-0.0:g}" == "-0"
+doassert fmt"{-0.0:+g}" == "-0"
+doassert fmt"{-0.0: g}" == "-0"
+doAssert fmt"{0.0:g}" == "0"
+doAssert fmt"{0.0:+g}" == "+0"
+doAssert fmt"{0.0: g}" == " 0"