add leading zeros and underscores support for Decimal parsing

Author: 9il

source/mir/bignum/decimal.d

@@ -119,7 +119,7 @@ struct Decimal(size_t maxSize64)
  Returns: false in case of overflow or incorrect string.
  Precondition: non-empty coefficients.
  +/
- bool fromStringImpl(C, bool allowSpecialValues = true, bool allowDotOnBounds = true, bool allowDExponent = true, bool allowStartingPlus = true, bool checkEmpty = true)(scope const(C)[] str, out DecimalExponentKey key)
+ bool fromStringImpl(C, bool allowSpecialValues = true, bool allowDotOnBounds = true, bool allowDExponent = true, bool allowStartingPlus = true, bool checkEmpty = true, bool allowUnderscores = true, bool allowLeadingZeros = true)(scope const(C)[] str, out DecimalExponentKey key)
  @safe pure @nogc nothrow
  if (isSomeChar!C)
  {
@@ -163,14 +163,20 @@ struct Decimal(size_t maxSize64)
  ulong v;
  uint afterDot;
  bool dot;
-
- if (d == 0)
+ static if (allowUnderscores)
  {
- if (str.length == 0)
- goto R;
- if (str[0] >= '0' && str[0] <= '9')
- return false;
- goto S;
+ bool recentUnderscore;
+ }
+ static if (!allowLeadingZeros)
+ {
+ if (d == 0)
+ {
+ if (str.length == 0)
+ goto R;
+ if (str[0] >= '0' && str[0] <= '9')
+ return false;
+ goto S;
+ }
  }
 
  if (d < 10)
@@ -206,6 +212,10 @@ struct Decimal(size_t maxSize64)
 
  if (_expect(d <= 10, true))
  {
+ static if (allowUnderscores)
+ {
+ recentUnderscore = false;
+ }
  {
  import core.checkedint: addu, mulu;
  bool overflow;
@@ -223,18 +233,35 @@ struct Decimal(size_t maxSize64)
  R:
  coefficient.data[0] = v;
  coefficient.length = v != 0;
- return true;
+ static if (allowUnderscores)
+ {
+ return !recentUnderscore;
+ }
+ else
+ {
+ return true;
+ }
+ }
+ key = cast(DecimalExponentKey)d;
+ static if (allowUnderscores)
+ {
+ if (recentUnderscore)
+ return false;
  }
  switch (d)
  {
- D:
  case DecimalExponentKey.dot:
- if (dot)
+ if (_expect(dot, false))
  break;
- key = cast(DecimalExponentKey)d;
  dot = true;
  if (str.length)
+ {
+ static if (allowUnderscores)
+ {
+ recentUnderscore = true;
+ }
  continue;
+ }
  static if (allowDotOnBounds)
  {
  goto R;
@@ -251,11 +278,18 @@ struct Decimal(size_t maxSize64)
  }
  case DecimalExponentKey.e:
  case DecimalExponentKey.E:
- key = cast(DecimalExponentKey)d;
  import mir.parse: parse;
  if (parse(str, exponent) && str.length == 0)
  goto E;
  break;
+ static if (allowUnderscores)
+ {
+ case '_' - '0':
+ recentUnderscore = true;
+ if (str.length)
+ continue;
+ break;
+ }
  default:
  }
  break;
@@ -287,7 +321,7 @@ struct Decimal(size_t maxSize64)
  {
  import mir.bignum.low_level_view: DecimalView, BigUIntView, MaxWordPow10;
  auto work = DecimalView!size_t(false, 0, BigUIntView!size_t(coefficient.data));
- auto ret = work.fromStringImpl!(C, allowSpecialValues, allowDExponent, allowStartingPlus)(str, key);
+ auto ret = work.fromStringImpl!(C, allowSpecialValues, allowDExponent, allowStartingPlus, checkEmpty, allowUnderscores, allowLeadingZeros)(str, key);
  coefficient.length = cast(uint) work.coefficient.coefficients.length;
  coefficient.sign = work.sign;
  exponent = work.exponent;
@@ -359,13 +393,13 @@ struct Decimal(size_t maxSize64)
  assert(decimal.fromStringImpl("4.", key));
  assert(!decimal.fromStringImpl(".", key));
  assert(decimal.fromStringImpl("0.", key));
- assert(!decimal.fromStringImpl("00", key));
+ assert(decimal.fromStringImpl("00", key));
  assert(!decimal.fromStringImpl("0d", key));
  }
 
  static if (maxSize64 == 3)
  version(mir_bignum_test)
- // @safe pure nothrow @nogc
+ @safe pure nothrow @nogc
  unittest
  {
  import mir.conv: to;
@@ -424,7 +458,7 @@ struct Decimal(size_t maxSize64)
  assert(decimal.fromStringImpl("4.", key));
  assert(!decimal.fromStringImpl(".", key));
  assert(decimal.fromStringImpl("0.", key));
- assert(!decimal.fromStringImpl("00", key));
+ assert(decimal.fromStringImpl("00", key));
  assert(!decimal.fromStringImpl("0d", key));
  }
 

source/mir/bignum/low_level_view.d

@@ -2048,7 +2048,7 @@ struct DecimalView(W, WordEndian endian = TargetEndian, Exp = sizediff_t)
  Precondition: non-empty coefficients
  Note: doesn't support signs.
  +/
- bool fromStringImpl(C, bool allowSpecialValues = true, bool allowDotOnBounds = true, bool allowDExponent = true, bool allowStartingPlus = true, bool checkEmpty = true)(scope const(C)[] str, out DecimalExponentKey key)
+ bool fromStringImpl(C, bool allowSpecialValues = true, bool allowDotOnBounds = true, bool allowDExponent = true, bool allowStartingPlus = true, bool checkEmpty = true, bool allowUnderscores = true, bool allowLeadingZeros = true)(scope const(C)[] str, out DecimalExponentKey key)
  @safe pure @nogc nothrow
  if (isSomeChar!C)
  {
@@ -2097,16 +2097,24 @@ struct DecimalView(W, WordEndian endian = TargetEndian, Exp = sizediff_t)
  uint afterDot;
  bool dot;
 
- if (d == 0)
+ static if (allowUnderscores)
  {
- if (str.length == 0)
+ bool recentUnderscore;
+ }
+
+ static if (!allowLeadingZeros)
+ {
+ if (d == 0)
  {
- coefficient = coefficient.init;
- return true;
+ if (str.length == 0)
+ {
+ coefficient = coefficient.init;
+ return true;
+ }
+ if (str[0] >= '0' && str[0] <= '9')
+ return false;
+ goto S;
  }
- if (str[0] >= '0' && str[0] <= '9')
- return false;
- goto S;
  }
 
  if (d < 10)
@@ -2143,6 +2151,10 @@ struct DecimalView(W, WordEndian endian = TargetEndian, Exp = sizediff_t)
 
  if (_expect(d <= 10, true))
  {
+ static if (allowUnderscores)
+ {
+ recentUnderscore = false;
+ }
  v *= 10;
  S:
  t *= 10;
@@ -2172,22 +2184,39 @@ struct DecimalView(W, WordEndian endian = TargetEndian, Exp = sizediff_t)
  M:
  exponent -= afterDot;
  coefficient = work.mostSignificant == 0 ? coefficient.init : work;
- return true;
+ static if (allowUnderscores)
+ {
+ return !recentUnderscore;
+ }
+ else
+ {
+ return true;
+ }
  }
  }
 
  continue;
  }
  key = cast(DecimalExponentKey)d;
+ static if (allowUnderscores)
+ {
+ if (recentUnderscore)
+ return false;
+ }
  switch (d)
  {
- D:
  case DecimalExponentKey.dot:
  if (_expect(dot, false))
  break;
  dot = true;
  if (str.length)
+ {
+ static if (allowUnderscores)
+ {
+ recentUnderscore = true;
+ }
  continue;
+ }
  static if (allowDotOnBounds)
  {
  goto L;
@@ -2212,6 +2241,14 @@ struct DecimalView(W, WordEndian endian = TargetEndian, Exp = sizediff_t)
  goto M;
  }
  break;
+ static if (allowUnderscores)
+ {
+ case '_' - '0':
+ recentUnderscore = true;
+ if (str.length)
+ continue;
+ break;
+ }
  default:
  }
  break;

source/mir/parse.d

@@ -96,6 +96,16 @@ version(mir_bignum_test)
  assert("-12.3e-30".fromString!double == -12.3e-30);
  assert("2.9802322387695312E-8".fromString!double == 2.9802322387695312E-8);
 
+ // default support of underscores
+ assert("123_456.789_012".fromString!double == 123_456.789_012);
+ assert("12_34_56_78_90_12e-6".fromString!double == 123_456.789_012);
+
+ // default support of leading zeros
+ assert("010".fromString!double == 10.0);
+ assert("000010".fromString!double == 10.0);
+ assert("0000.10".fromString!double == 0.1);
+ assert("0000e10".fromString!double == 0);
+
  /// Test CTFE support 
  static assert("-12.3e-30".fromString!double == -0x1.f2f280b2414d5p-97);
  static assert("+12.3e+30".fromString!double == 0x1.367ee3119d2bap+103);
@@ -120,7 +130,16 @@ version(mir_bignum_test)
 @safe pure unittest
 {
  import std.exception: assertThrown;
- assertThrown("010".fromString!float);
+ assertThrown("1_".fromString!float);
+ assertThrown("1__2".fromString!float);
+ assertThrown("_1".fromString!float);
+ assertThrown("123_.456".fromString!float);
+ assertThrown("123_e0".fromString!float);
+ assertThrown("123._456".fromString!float);
+ assertThrown("12__34.56".fromString!float);
+ assertThrown("123.456_".fromString!float);
+ assertThrown("-_123.456".fromString!float);
+ assertThrown("_123.456".fromString!float);
 }
 
 /++