Reimplement fraction string parsing without slowish regular expressions. · scoder/quicktions@cc034e0 (original) (raw)

`@@ -91,21 +91,6 @@ except AttributeError: # pre Py3.2

91

` _PyHASH_INF = hash(float('+inf'))

92

93

94

cdef object _parse_rational = re.compile(r"""

95

\A\s* # optional whitespace at the start, then

96

(?P[-+]?) # an optional sign, then

97

(?=\d|.\d) # lookahead for digit or .digit

98

(?P\d*) # numerator (possibly empty)

99

(?: # followed by

100

(?:/(?P\d+))? # an optional denominator

101

| # or

102

(?:.(?P\d*))? # an optional fractional part

103

(?:E(?P[-+]?\d+))? # and optional exponent

104

)

105

\s*\Z # and optional whitespace to finish

106

""", re.VERBOSE | re.IGNORECASE).match

107

-

108

-

109

94

`cdef class Fraction:

110

95

`"""A Rational number.

111

96

`@@ -161,34 +146,12 @@ cdef class Fraction:

161

146

`self._denominator = (numerator)._denominator

162

147

`return

163

148

164

elif isinstance(numerator, basestring):

165

Handle construction from strings.

166

m = _parse_rational(numerator)

167

if m is None:

168

raise ValueError('Invalid literal for Fraction: %r' %

169

numerator)

170

group = m.group

171

numerator = int(group('num') or 0)

172

denom = group('denom')

173

if denom:

174

denominator = int(denom)

175

else:

176

decimal = group('decimal')

177

if decimal:

178

scale = 10 ** len(decimal)

179

numerator = numerator * scale + int(decimal)

180

denominator = scale

181

else:

182

denominator = 1

183

exp = group('exp')

184

if exp:

185

exp = int(exp)

186

if exp >= 0:

187

numerator *= 10**exp

188

else:

189

denominator = 10*-exp

190

if group('sign') == '-':

191

numerator = -numerator

149

elif isinstance(numerator, unicode):

150

numerator, denominator = _parse_fraction(numerator)

151

fall through to normalisation below

152

+

153

elif PY_MAJOR_VERSION < 3 and isinstance(numerator, bytes):

154

numerator, denominator = _parse_fraction(numerator)

192

155

`# fall through to normalisation below

193

156

194

157

`elif isinstance(numerator, (Fraction, Rational)):

`@@ -234,7 +197,7 @@ cdef class Fraction:

234

197

`"Rational instances")

235

198

236

199

`if denominator == 0:

237

raise ZeroDivisionError('Fraction(%s, 0)' % numerator)

200

raise ZeroDivisionError(f'Fraction({numerator}, 0)')

238

201

`if _normalize:

239

202

` g = _gcd(numerator, denominator)

240

203

`# NOTE: 'is' tests on integers are generally a bad idea, but

`@@ -268,11 +231,10 @@ cdef class Fraction:

268

231

`if isinstance(f, numbers.Integral):

269

232

`return cls(f)

270

233

`elif not isinstance(f, float):

271

raise TypeError("%s.from_float() only takes floats, not %r (%s)" %

272

(cls.name, f, type(f).name))

234

raise TypeError(f"{cls.name}.from_float() only takes floats, not {f!r} ({type(f).name})")

273

235

`if math.isinf(f):

274

raise OverflowError("Cannot convert %r to %s." % (f, cls.name))

275

raise ValueError("Cannot convert %r to %s." % (f, cls.name))

236

raise OverflowError(f"Cannot convert {f!r} to {cls.name}.")

237

raise ValueError(f"Cannot convert {f!r} to {cls.name}.")

276

238

277

239

`@classmethod

278

240

`def from_decimal(cls, dec):

`@@ -281,13 +243,11 @@ cdef class Fraction:

281

243

` dec = Decimal(int(dec))

282

244

`elif not isinstance(dec, Decimal):

283

245

`raise TypeError(

284

"%s.from_decimal() only takes Decimals, not %r (%s)" %

285

(cls.name, dec, type(dec).name))

246

f"{cls.name}.from_decimal() only takes Decimals, not {dec!r} ({type(dec).name})")

286

247

`if dec.is_infinite():

287

raise OverflowError(

288

"Cannot convert %s to %s." % (dec, cls.name))

248

raise OverflowError(f"Cannot convert {dec} to {cls.name}.")

289

249

`if dec.is_nan():

290

raise ValueError("Cannot convert %s to %s." % (dec, cls.name))

250

raise ValueError(f"Cannot convert {dec} to {cls.name}.")

291

251

` sign, digits, exp = dec.as_tuple()

292

252

` digits = int(''.join(map(str, digits)))

293

253

`if sign:

`@@ -683,6 +643,10 @@ cdef class Fraction:

683

643

`return type(self)(self._numerator, self._denominator)

684

644

685

645

646

Register with Python's numerical tower.

647

Rational.register(Fraction)

648

+

649

+

686

650

`cdef _pow(an, ad, bn, bd):

687

651

`if bd == 1:

688

652

`if bn >= 0:

`@@ -842,4 +806,101 @@ cdef reverse(a, b, math_func monomorphic_operator, str pyoperator_name):

842

806

`return NotImplemented

843

807

844

808

845

Rational.register(Fraction)

809

ctypedef fused AnyString:

810

bytes

811

unicode

812

+

813

+

814

cdef _raise_invalid_input(s):

815

raise ValueError(f'Invalid literal for Fraction: {s!r}') from None

816

+

817

+

818

cdef tuple _parse_fraction(AnyString s):

819

cdef size_t i, dec_pos = 0, exp_pos = 0, dec_len = 0

820

cdef Py_UCS4 c

821

cdef AnyString numerator, denominator

822

+

823

allow_sign = True

824

require_separator = False

825

number_seen = False

826

has_underscores = False

827

for i, c in enumerate(s):

828

if c == u'/':

829

if not number_seen or dec_pos > 0 or exp_pos > 0:

830

_raise_invalid_input(s)

831

try:

832

return int(s[:i].replace('', '') if has_underscores else s[:i]), int(s[i+1:].replace('', ''))

833

except ValueError:

834

_raise_invalid_input(s)

835

elif c == u'.':

836

if dec_pos > 0 or exp_pos > 0:

837

_raise_invalid_input(s)

838

require_separator = False

839

number_seen = False

840

allow_sign = False

841

dec_pos = i

842

elif c in u'eE':

843

if exp_pos > 0 or i == 0:

844

_raise_invalid_input(s)

845

require_separator = False

846

number_seen = False

847

allow_sign = True

848

exp_pos = i

849

elif c in u'0123456789':

850

if require_separator:

851

_raise_invalid_input(s)

852

number_seen = True

853

require_separator = False

854

allow_sign = False

855

if dec_pos > 0 and exp_pos == 0:

856

dec_len += 1

857

elif c in u'+-':

858

if require_separator:

859

_raise_invalid_input(s)

860

require_separator = False

861

if not allow_sign:

862

_raise_invalid_input(s)

863

allow_sign = False

864

elif c == u'_':

865

if not number_seen or require_separator:

866

_raise_invalid_input(s)

867

has_underscores = True

868

else:

869

if c.isspace():

870

if number_seen or not allow_sign:

871

require_separator = True

872

else:

873

_raise_invalid_input(s)

874

+

875

if exp_pos > 0:

876

shift = int(s[exp_pos+1:])

877

numerator = s[:exp_pos]

878

else:

879

shift = 0

880

numerator = s

881

+

882

neg = False

883

if dec_pos > 0:

884

numerator, decimal = numerator[:dec_pos], numerator[dec_pos+1:]

885

numerator = numerator.strip()

886

decimal = decimal.strip()

887

shift -= dec_len

888

numerator += decimal

889

+

890

if has_underscores:

891

numerator = numerator.replace('_', '')

892

+

893

try:

894

num = int(numerator)

895

except ValueError:

896

_raise_invalid_input(s)

897

+

898

denom = 1

899

if neg:

900

num = -num

901

if shift > 0:

902

num *= 10 ** shift

903

elif shift < 0:

904

denom = 10 ** -shift

905

+

906

return num, denom