[SOLVED] Regexp Expert: How to exctract number

[michaelslamet]

Hi RegExp expert

I need to extract numbers from some sentences that seems to have no exact pattern. I think using regexp will solve my problem.

Example:

This sentence:

Cotton +- 400, 600, 1000, 3000, Sugar +- 800, 1800, 2700,Move Car +- 5000, SIX, Top: 0813.7771.2114 (TS)

I need to extract this number: 081377712114 (remove the dot between them)

This sentence: Ls.Nevergone M.Silky Expensive car less nop, AXD,is.begin.72m @ 1.6km,APP: 7211122, 2000051

I need to extract this number: 7211122 and 2000051

This sentence: Si Cheaper Good No.1(more Gd Papaya)Call ads less bulky Lt 479mUUTnomorethan0890-111219

I need to extract this number: 0890-111219

This sentence: ASX Top.Bulk Exclsv Ultimo dbl hik paper TL BD, B:599m2 (10x20x3). OOP: 0815141151(P)

I need to extract this number: 0815141151

This sentence: ATP Tirte Golf 100-300m; SX price market less; 46530181 Jen0811001918 Koles33119221

I need to extract this number: 46530181 and 0811001918 and 33119221

The rules are:

1. Take numbers that begin with 08 and at least has 9 digits

2. If the number not begin with 08 then take it only if it has at least 7 digits

How to do it in AutoIT? Try few days and found no solution

Thanks in advance

Edited February 17, 2011 by michaelslamet

[jchd]

Your rules are not enough to work in all the cases above.

Removing of dots will have to be done separately.

The closest we can get is: (?m)(08[\d\.]{9,}|[-\d]{7,})

Applied to the text

Cotton +- 400, 600, 1000, 3000, Sugar +- 800, 1800, 2700,Move Car +- 5000, SIX, Top: 0813.7771.2114 (TS)
Ls.Nevergone M.Silky Expensive car less nop, AXD,is.begin.72m @ 1.6km,APP: 7211122, 2000051
Si Cheaper Good No.1(more Gd Papaya)Call ads less bulky Lt 479mUUTnomorethan0890-111219
ASX Top.Bulk Exclsv Ultimo dbl hik paper TL BD, B:599m2 (10x20x3). OOP: 0815141151(P)
ATP Tirte Golf 100-300m; SX price market less; 46530181 Jen0811001918 Koles33119221

gives

0813.7771.2114
7211122
2000051
0890-111219
0815141151
100-300
46530181
0811001918
33119221

The problem is with the unwanted capture of 100-300 in the last input line. Perhaps could you be a little more specific about the allowed charset and/ornumber of chars between dashes.

Another way to process input could be to first remove dots and dashes, so the number of digits will be fully significant.

[michaelslamet]

Wow, thanks a lot jchd!! You're hero!!

This is all I need, I will work to enhanced it. Your Regexp is working most of the time, I only need to modify it a little to suit my need.

Many thanks again, jchd

Your rules are not enough to work in all the cases above.

Removing of dots will have to be done separately.

The closest we can get is: (?m)(08[\d\.]{9,}|[-\d]{7,})

Applied to the text

Cotton +- 400, 600, 1000, 3000, Sugar +- 800, 1800, 2700,Move Car +- 5000, SIX, Top: 0813.7771.2114 (TS)
Ls.Nevergone M.Silky Expensive car less nop, AXD,is.begin.72m @ 1.6km,APP: 7211122, 2000051
Si Cheaper Good No.1(more Gd Papaya)Call ads less bulky Lt 479mUUTnomorethan0890-111219
ASX Top.Bulk Exclsv Ultimo dbl hik paper TL BD, B:599m2 (10x20x3). OOP: 0815141151(P)
ATP Tirte Golf 100-300m; SX price market less; 46530181 Jen0811001918 Koles33119221

gives

0813.7771.2114
7211122
2000051
0890-111219
0815141151
100-300
46530181
0811001918
33119221

The problem is with the unwanted capture of 100-300 in the last input line. Perhaps could you be a little more specific about the allowed charset and/ornumber of chars between dashes.

Another way to process input could be to first remove dots and dashes, so the number of digits will be fully significant.

[jchd]

Try removing dots and hyphen first:

$SimpleText = StringRegExpReplace($text, '\.-', '')

then apply a simpler regexp to grab your pretty numbers

$nums = StringRegExp($SimpleText, '(?m)(08\d{9,}|\d{7,})', 3)

Should work better.

[michaelslamet]

I format the result as your advice and it look great

But I found there is situation where this RegExp give a wrong result.

This text: Computer Easy Clean Closed Always love car 50he &amx; Less 700hi 0808 1191 7722

I should get 0808 1191 7722 as a result, but the RegExp give nothing as result.

Can you please modify the RegExp to capture this situation?

Thanks a lot

Try removing dots and hyphen first:

$SimpleText = StringRegExpReplace($text, '\.-', '')

then apply a simpler regexp to grab your pretty numbers

$nums = StringRegExp($SimpleText, '(?m)(08\d{9,}|\d{7,})', 3)

Should work better.

[Xenobiologist]

Hi,

try this:

#include<Array.au3>
Global $str = 'Cotton +- 400, 600, 1000, 3000, Sugar +- 800, 1800, 2700,Move Car +- 5000, SIX, Top: 0813.7771.2114 (TS)' & @CRLF & _
        'This sentence: Ls.Nevergone M.Silky Expensive car less nop, AXD,is.begin.72m @ 1.6km,APP: 7211122, 2000051' & @CRLF & _
        'This sentence: Si Cheaper Good No.1(more Gd Papaya)Call ads less bulky Lt 479mUUTnomorethan0890-111219' & @CRLF & _
        'This sentence: ASX Top.Bulk Exclsv Ultimo dbl hik paper TL BD, B:599m2 (10x20x3). OOP: 0815141151(P)' & @CRLF & _
        'This sentence: ATP Tirte Golf 100-300m; SX price market less; 46530181 Jen0811001918 Koles33119221' & @CRLF & _
        'This text: Computer Easy Clean Closed Always love car 50he &amx; Less 700hi 0808 1191 7722'

;~ 'I need to extract this number: 081377712114 (remove the dot between them)' & @CRLF
;~ 'I need to extract this number: 7211122 and 2000051' & @CRLF
;~ 'I need to extract this number: 0890-111219' & @CRLF
;~ 'I need to extract this number: 0815141151' & @CRLF
;~ 'I need to extract this number: 46530181 and 0811001918 and 33119221'
;~ I should get 0808 1191 7722 as a result, but the RegExp give nothing as result

;~ The rules are:
;~ 1. Take numbers that begin with 08 and at least has 9 digits
;~ 2. If the number not begin with 08 then take it only if it has at least 7 digits
$nstr = StringRegExpReplace($str, '[\.-\h]+', '')
ConsoleWrite($nstr & @CRLF)
$re = StringRegExp($nstr, '08\d{7,}|\d{7,}', 3)
_ArrayDisplay($re)

Mega

[jchd]

Hi michaelslamet

Xenobiologist made it for you, this case was easy to deal with.

You see, to build a working RegExp, you have first to determine exactly what your rules are. You need to do so very precisely, since forgetting one case and later changing what seems a marginal part of a condition can demand a complete rewrite of the patern.

One very powerful way to test RegExp is using the RegExp toolkit offered by GeoSoft: locate a post of him and look in his signature. Have fun!

[michaelslamet]

Mega and jchd,

Thanks a lot for your help. The code running well and the problem solved!

I combine RegExp from jchd and Mega and it's work great.

Thanks a lot for you both!!

Hi michaelslamet

Xenobiologist made it for you, this case was easy to deal with.

You see, to build a working RegExp, you have first to determine exactly what your rules are. You need to do so very precisely, since forgetting one case and later changing what seems a marginal part of a condition can demand a complete rewrite of the patern.

One very powerful way to test RegExp is using the RegExp toolkit offered by GeoSoft: locate a post of him and look in his signature. Have fun!