start regular expression from 2nd word only

Ask Question Forum:

Model Library:2025-02-08 Updated:A.I. model is online for auto reply question page

Show

ASK RECENT

start regular expression from 2nd word only

Asked by PeterBaileyUk

at 2024-07-12 22:45:48

Point:500 Replies:15 POST_ID:828969USER_ID:11868

Topic:

Regular Expressions;;

How can I instruct regular expression to ignore the first word, then start its search for a pattern.

Expert: kaufmed replied at 2024-07-25 05:01:11

Not that I don't appreciate the gesture, but I didn't really contribute anything toward the actual solution.

Author: PeterBaileyUk replied at 2024-07-24 23:55:25

Ive shared the points just out of fairness hope i didnt miss anyone.

I post a new question about adding banding to current pattern as its off topic on this question.

Author: PeterBaileyUk replied at 2024-07-24 23:52:33

Whats happening certainly in the first model case Audi "A3" was that if the string contained c02 value and bhp value then c02 had to be removed first hence the c02 removal call. the c02 call was very specific 107G whereas bhp wasnt always so neat.

It appears that if after the first word it finds another number lets say as part of valves 16V then bhp that it stops after identifying the first number thus ignoring the bhp and leaving it in. It almost suggests that the other technical details should be removed first and then the bhp pattern should then be called. it might explain why some bhp get left in the string.

here is an example: 207 GTi 1.6 16V 175 3dr the output string is : 207 GTi 1.6 V 175 3dr so it took out the 16 off the valves then stopped and left in the bhp.

same here: Insignia 1.8 16V 140 Elite 6Spd becomes Insignia 1.8 V 140 Elite 6Spd so took the 16 of valves out then stopped and left in the bhp.

If i can band the pattern for bhp within a range that would cure part of this.

I wonder now to close this question and put in a new as it does currently ignore the first word.

Accepted Solution

Expert: Terry Woods replied at 2024-07-24 19:39:15

125 points EXCELLENT

Ok, so the pattern you've shown:

StrPattern = "^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3})"

1:

Select All Code

captures the first word in the (S+(?=s).*?) part of the pattern. It then gets inserted back into the string as a replacement (it's the first of two capturing groups that are put back in, being $1 and $2), so the end result is that the first word shouldn't be removed.

Are you sure it's this pattern that's giving the problem? (or is the data something other than "207 HATCHBACK XR 1.6 VTi [120] 5dr Auto"?)

I also see you're making a call to the removeCO2 function, which could be a cause of the problem if my above suggestions don't help

txt = removeC02(txt)

1:

Select All Code

Author: PeterBaileyUk replied at 2024-07-23 23:57:59

Yes it has the bit you mentioned

I believe (and reason why I started to try simplify it) was that it allows some through and that it would be better to check bands.

Function removeBHP(ByVal txt As String) As StringDim StrPattern As StringIf CountWordsInString(txt) > 1 Then'StrPattern = "^s*(S+s.*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3})"'StrPattern = "^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3})"txt = removeC02(txt)'StrPattern = "^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3}|d{2})"'StrPattern ="^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{2})"StrPattern = "^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3})"   With CreateObject("VBScript.RegExp")       .IgnoreCase = True       .Pattern = StrPattern       .Global = True       removeBHP = .Replace(txt, "$1$2")'       removeBHP = .Replace(removeBHP, "$1$2")       Debug.Print "Input: " & txt & " Output: " & removeBHP'       Debug.Print txt   End With   Else   removeBHP = txt   End IfEnd Function

1:2:3:4:5:6:7:8:9:10:11:12:13:14:15:16:17:18:19:20:21:22:23:24:25:26:27:28:29:30:31:32:33:34:35:36:37:

Select All Code

Expert: Terry Woods replied at 2024-07-23 19:02:34

Are you doing a replace with the replacement string being something like "$1"? We may need to change that as part of the solution. In fact, if you weren't doing a replace with "$1" as the replacement string, then try doing that as it may solve the problem :-)

Author: PeterBaileyUk replied at 2024-07-23 01:42:21

I have a pattern which evolved around one manufacturer and model.

Ignoring the very first word is fundamental in case the actual vehicle was named with digits like puegeot 207 but had the BHP which could be from 2 digits to 3 digits long i.e 90 or 109 and or have the word BHP as a separate word or attached 110bhp.

I wanted to try to create a less complex pattern that was not vehicle specific.

StrPattern = "^s*(S+(?=s).*?)(?:([^d.])d+$|(?d* ?bhp?d*)?|(d+)|[d{2,3}]|d{3})"

I have discovered a boundary so it must find values between 39-302 inclusive but not pick up for example 3020 or 390
skipping the first word is paramount. the pattern must find the values if they are at the end of the string also.

Expert: Terry Woods replied at 2024-07-22 18:47:14

Peter, are you able to please post the code you've currently got? Any existing capturing groups (and their replacement) may need to be adjusted once we provide a change to your existing pattern.

I have noticed that the first word is never preceded by a space so is that how some of these work?

If the pattern you're talking about, with regards to that comment, is the one starting with

^s*w+

1:

Select All Code

then the s* part of the pattern can match zero occurrences of a space character. To match one or more occurrences, rather than zero or more, you would use s+ instead of s*

The pattern

^s*w+s*(d+)

1:

Select All Code

is matching the first word, but capturing the second one (well, at least it is if the second word is only made up of numbers, as it's d+ rather than w+). As I mentioned above, we need to know what's happening with the capturing groups to determine what happens from there on. I suspect this pattern isn't one that will work for you, as it only matches when the second "word" is made up of numbers.

Author: PeterBaileyUk replied at 2024-07-21 23:42:10

I am in vb access using VBScript.RegExp object

Assisted Solution

Expert: kaufmed replied at 2024-07-14 12:01:55

125 points EXCELLENT

What programming language or text editor are you using to execute this? If we don't know what engine you use, then we could be offering bad advice. For example, what duncanb7 suggested is a PERL-based expression. It would not work (the way you think it would) in a .NET regex.

Author: PeterBaileyUk replied at 2024-07-14 00:23:48

I have noticed that the first word is never preceded by a space so is that how some of these work?

Author: PeterBaileyUk replied at 2024-07-14 00:18:44

i put a simple pattern in with respect to idID: 40192739
~ s/^S+s*//d+ i've added the d+ just a simple pattern

Test data 207 HATCHBACK XR 1.6 VTi [120] 5dr Auto I had hoped the 120 would be highlighted after ignoring the first word the 207.

i put a simple pattern in with respect to idID: 40192765 ^s*w+s*(d+)
207 HATCHBACK XR 1.6 VTi [120] 5dr Auto that highlights the first word, it should have ignored that and find the 120

I wanted to ignore the first word as the manufacturer have names their models with integers ie 207

would it be simpler maybe to reverse the search start from end in?

Author: PeterBaileyUk replied at 2024-07-13 23:28:50

ok i will take a look. Ive been involved in a project recently involving 750000 word groups and asked a lot of the experts. I am widening the data sets for testing and trying now to create shorter patterns that cover most of what i need as its not possible to find a pattern that does all that i need.

its become evident that the first word has more weight in the description.

Assisted Solution

Expert: Dan Craciun replied at 2024-07-13 00:49:34

125 points EXCELLENT

Depending on your Regexp engine, this usually works:

^.*?(your search pattern here)

1:

Select All Code

.*? will match the first word (sequence of letters, numbers or _)

or

^s*w+s*(your search pattern here)

1:

Select All Code

s*w+s* will match some (optional) space characters, 1 or more word characters, then some more optional space characters.

HTH,
Dan

Assisted Solution

Expert: duncanb7 replied at 2024-07-12 23:44:43

125 points EXCELLENT

you could try this, remove first word of each line first and then search pattern

~ s/^S+s*//

s/.../.../ # Substitute command.
^ # (Zero-width) Begin of line.
S+ # Non-space characters.
s* # Blank-space characters.
// # Substitute with nothing, so remove them.

Hope understand your question completely.If not, please point it out

Duncan

新聞財經娛樂♬資訊 News Economy

About Us|Service|Pricing

Stock Index全球即時指數期貨:

&nbps;

News即時新聞:↓

News即時新聞:↑

On Balance Volume (OBV):

Youtube Video♬ Sharing by members:↓

Youtube Video♬ Sharing by members:↑

Ask Question Forum:

start regular expression from 2nd word only

Your Information

Members/Profolio