number recognition

[memoryoverflow]

Having stumbled over the side effects of implicit type casting way too often, I'm going to check expected numbers that might come as strings explicitely before using them.

Here's where I don't see the obvious:

Local $i = 0, $asResult[3] = ['hex', 'NaN', 'dec']
Local $asTest = StringSplit('-123.45e+12|-.1|23.E-9|0.|+.0|  12|001 || |0xfedcba98|-|2l0|10O0|12,345.67|9e|e8|7-6|+ 5|4.3.21|9 8|7+e6|5e4.3|2f1|1.2e-', '|')
Local $sRxNaDec = '(?i)[^-+.e\d\s]|\..*\.|e.*[^-+\d]|\A[^-+.\d\s]|[^.\d\s]\z|[^\A\s]\s[^\z\s]|[^e\A\s][-+]'
Local $sRxIsHex = '(?i)0x[\da-f]{1,8}\z'
For $i = 1 To $asTest[0]
    ConsoleWrite(StringFormat('%15s  %s', $asTest[$i], $asResult[2 - StringRegExp($asTest[$i], $sRxIsHex) - StringRegExp($asTest[$i], $sRxNaDec)]) & '   ''Number(…)'' = ' & Number($asTest[$i]) & @crlf)
Next

Well, it works, but...

... isn't there a much simpler way, I just can't see atm?

edit: $sRxHex amended

Edited October 18, 2009 by memoryoverflow

[martin]

Having stumbled over the side effects of implicit type casting way too often, I'm going to check expected numbers that might come as strings explicitely before using them.

Here's where I don't see the obvious:

Local $i = 0, $asResult[3] = ['hex', 'NaN', 'dec']
Local $asTest = StringSplit('-123.45e+12|-.1|23.E-9|0.|+.0| 12|001 || |0xfedcba98|-|2l0|10O0|12,345.67|9e|e8|7-6|+ 5|4.3.21|9 8|7+e6|5e4.3|2f1|1.2e-', '|')
Local $sRxNaDec = '(?i)[^-+.e\d\s]|\..*\.|e.*[^-+\d]|\A[^-+.\d\s]|[^.\d\s]\z|[^\A\s]\s[^\z\s]|[^e\A\s][-+]'
Local $sRxIsHex = '(?i)0x[\da-f]{1,8}'
For $i = 1 To $asTest[0]
    ConsoleWrite(StringFormat('%15s %s', $asTest[$i], $asResult[2 - StringRegExp($asTest[$i], $sRxIsHex) - StringRegExp($asTest[$i], $sRxNaDec)]) & ' ''Number(…)'' = ' & Number($asTest[$i]) & @crlf)
Next

Well, it works, but...

... isn't there a much simpler way, I just can't see atm?

Yes, it can be difficult. This gets near

Local $i = 0, $asResult[3] = ['hex', 'NaN', 'dec']
Local $asTest = StringSplit('-123.45e+12|-.1|23.E-9|0.|+.0| 12|001 || |0xfedcba98|-|2l0|10O0|12,345.67|9e|e8|7-6|+ 5|4.3.21|9 8|7+e6|5e4.3|2f1|1.2e-', '|')
Local $sRxNaDec = '(?i)[^-+.e\d\s]|\..*\.|e.*[^-+\d]|\A[^-+.\d\s]|[^.\d\s]\z|[^\A\s]\s[^\z\s]|[^e\A\s][-+]'
Local $sRxIsHex = '(?i)0x[\da-f]{1,8}'
For $i = 1 To $asTest[0]

    ConsoleWrite(StringFormat('%15s %s', $asTest[$i], $asResult[2 - StringRegExp($asTest[$i], $sRxIsHex) - StringRegExp($asTest[$i], $sRxNaDec)]) & ' ''Number(…)'' = ' & _Number($asTest[$i]) & @CRLF)

Next


Func _Number($sN)
    $sN = StringReplace($sN, ',', '')
    $res1 = Execute($sN)
    If $res1 = '' and not isNumber($sN) Then
    Return 'NAN'
    Else
    Return Number($res1)
    EndIf

EndFunc ;==>_Number

[memoryoverflow]

Thanks for the quick answer, Martin

While it doesn't exactly fit my choices of permissiveness (which are to be revisted, anyway - except for the evaluation of expressions, which I don't want to allow at this place), it's a good inspiration to look into the execute function, anyway.

And yes, those commas are another story. I thought about allowing them, but decided not to unless I process the locale settings, too (which I have postponed).

p.s. & ot: Just seen that the topic still displays 0 views

But at least Martin and I viewed it - how that?

Edited October 18, 2009 by memoryoverflow

[GEOSoft]

p.s. & ot: Just seen that the topic still displays 0 views

But at least Martin and I viewed it - how that?

when Jon makes changs to the forum, the counter refresh time gets set back to the default which I believe is every 45 minutes and he keeps forgetting to change them again. Very likely because he's too busy to worry about the minor details.

[memoryoverflow]

Thanks for bothering, George.

John has answered himself in the 'chat' sub forum. Think to remember something like default = 3 hours, but well... memory overflow... and it wouldn't have mattered that much for sure. Overlooking such minor details isn't surprising. I myself appreciate it the more, if someone who sees such things gives me a hint. (E.g. the 2nd pattern in my opening here has a flaw, forgot to check that '0x...' must be the start: '\A0x...')