fix #7717 roundtrip float to string; fix parseFloat for js (#18248)

* refs #7717 roundtrip float to string
* make parseFloat more correct
* improve float tests
* improve float tests
* cleanup

tests/float/tfloat6.nim

@@ -1,28 +0,0 @@
-discard """
-  output: '''
-0.000001 : 0.000001
-0.000001 : 0.000001
-0.001 : 0.001
-0.000001 : 0.000001
-0.000001 : 0.000001
-10.000001 : 10.000001
-100.000001 : 100.000001
-'''
-  disabled: "windows"
-"""
-
-import strutils
-
-echo "0.00_0001".parseFloat(), " : ", 1E-6
-echo "0.00__00_01".parseFloat(), " : ", 1E-6
-echo "0.0_01".parseFloat(), " : ", 0.001
-echo "0.00_000_1".parseFloat(), " : ", 1E-6
-echo "0.00000_1".parseFloat(), " : ", 1E-6
-
-echo "1_0.00_0001".parseFloat(), " : ", 10.000001
-echo "1__00.00_0001".parseFloat(), " : ", 1_00.000001
-
-# bug #18148
-
-var a = 1.1'f32
-doAssert $a == "1.1", $a # fails

tests/float/tfloats.nim

@@ -0,0 +1,84 @@
+discard """
+  targets: "c cpp js"
+"""
+# disabled: "windows"
+
+#[
+xxx merge all or most float tests into this file
+]#
+
+import std/[fenv, math, strutils]
+
+proc equalsOrNaNs(a, b: float): bool =
+  if isNaN(a): isNaN(b)
+  elif a == 0:
+    b == 0 and signbit(a) == signbit(b)
+  else:
+    a == b
+
+template reject(a) =
+  doAssertRaises(ValueError): discard parseFloat(a)
+
+template main =
+  block:
+    proc test(a: string, b: float) =
+      let a2 = a.parseFloat
+      doAssert equalsOrNaNs(a2, b), $(a, a2, b)
+    test "0.00_0001", 1E-6
+    test "0.00__00_01", 1E-6
+    test "0.0_01", 0.001
+    test "0.00_000_1", 1E-6
+    test "0.00000_1", 1E-6
+    test "1_0.00_0001", 10.000001
+    test "1__00.00_0001", 1_00.000001
+    test "inf", Inf
+    test "-inf", -Inf
+    test "-Inf", -Inf
+    test "-INF", -Inf
+    test "NaN", NaN
+    test "-nan", NaN
+    test ".1", 0.1
+    test "-.1", -0.1
+    test "-0", -0.0
+    when false: # pending bug #18246
+      test "-0", -0.0
+    test ".1e-1", 0.1e-1
+    test "0_1_2_3.0_1_2_3E+0_1_2", 123.0123e12
+    test "0_1_2.e-0", 12e0
+    test "0_1_2e-0", 12e0
+    test "-0e0", -0.0
+    test "-0e-0", -0.0
+
+  reject "a"
+  reject ""
+  reject "e1"
+  reject "infa"
+  reject "infe1"
+  reject "_"
+  reject "1e"
+
+  when false: # gray area; these numbers should probably be invalid
+    reject "1_"
+    reject "1_.0"
+    reject "1.0_"
+
+  block: # bug #18148
+    var a = 1.1'f32
+    doAssert $a == "1.1", $a # was failing
+
+  block: # bug #7717
+    proc test(f: float) =
+      let f2 = $f
+      let f3 = parseFloat(f2)
+      doAssert equalsOrNaNs(f, f3), $(f, f2, f3)
+
+    test 1.0 + epsilon(float64)
+    test 1000000.0000000123
+    test log2(100000.0)
+    test maximumPositiveValue(float32)
+    test maximumPositiveValue(float64)
+    test minimumPositiveValue(float32)
+    test minimumPositiveValue(float64)
+
+static: main()
+main()