Splitting words in different combinations in Excel - excel

Question

I would like to get only bold part of the text in new column i.e
Czechowice - Dziedzice AMBRA
Białystok DEF
Komorniki
Bielsko Biała EC
Farmacja Luboń
Gorzów Wlkp.
Grudziądz EC
Kędzierzyn-Koźle EC
Ostrowiec Świętokrzyski EC
Puck T
Przeworsk+Sklep
Białystok + sklep
Kielce (Masłów)
Barlinek + Myślibórz
Lublin TR
Biała Podlaska TR
Puławy II TR
Toruń DLS TR
Kraków SJ TR
I tried to use IF(ISNUMBER(SEARCH("AMBRA";B2));LEFT(B2;LEN(B2)-6) for all options but it's very inefficient. Any help is appreciated.

Answer 1

This needs VBA to create a custom function. You can use the Characters property of the Range to return information about individual characters within a cell:
Public Function getBoldText(cellReference As Range) As String
Dim i As Long
'Loop through each character in the cell
For i = 0 To cellReference.Characters.Count
'If the character is bold then...
If cellReference.Characters(i, 1).Font.Bold Then
'...add it to the output
getBoldText = getBoldText & cellReference.Characters(i, 1).Text
End If
Next i
End Function
Create a module, paste in this code, and then you can use e.g. =getBoldText(A1) within your worksheet to return only bold text from a cell. This function only works with single cells, and returns #VALUE! if the cell contains anything other than text.
Note I would have used For Each...Next for the loop, but despite appearances Characters isn't actually a collection so you can't iterate over it.

Answer 2

#simoco has made the very pertinent point:
How do you imagine Excel should detect which part of string should be left, if there is not any pattern in your values?
However OP has mentioned I have about 10000 rows so though Excel may not be able to provide a full solution it may still be of some help. OP seems, understandably, to have lost interest (perhaps is still working through the 10,000 manually?) but the problem is not unusual and demonstrating an approach, even if one that is only partially successful, may be of some use for others.
So I put the OP's list in A1:A19. From observation, most of what is no emboldened starts after the last space, so in B1:
=LEN(A1)-LEN(SUBSTITUTE(A1," ",""))
By comparing the original length of the strings with their length after removing spaces we obtain the number of spaces in the original string.
In C1:
=IF(RIGHT(A1)=".",A1,SUBSTITUTE(A1," ",REPT(" ",LEN(A1)),B1))
we start at the last space and replace it with as many spaces as the length of the original string. (There is reason for doing so!). Also noticing that Gorzów Wlkp. ends in a full stop (to be retained) whereas nothing to be removed does, we make a specific exception for strigs ending ..
In D1:
=IF(ISERROR(C1),A1,LEFT(C1,LEN(A1)))
Having inserted a large number of spaces between the text to be kept and not to be kept we now select from the left the number of characters that we started with - so mostly what we want to keep plus a lot of blanks. Where there were no spaces to start with the formula in C1 returns an error, so in those cases we take the whole of the original string instead.
In E1:
=IFERROR(LEFT(D1,FIND("+",D1)),D1)
This attempts to cope with at least some of the data containing plus signs +. Which are, with one exception, Barlinek + Myślibórz, in the data sample, to be removed along with any following characters. The above removes those following characters.
In F1:
=TRIM(SUBSTITUTE(E1,"+",""))
This is mainly tidying because there is not much more that can be 'automated'. The +s are removed and the surplus from the spaces that were inserted earlier.
So although the original data was 'unnormalised', of the sample of 19 only the exception mentioned above and Toruń DLS TR and Kraków SJ TR are not as required. For these last two DLS and SJ are retained where they should not be - whether worth a further processing step may depend on what is in the rest of the 10,000 entries - but it may be better anyway to fall short by omission rather than commission.
Stripping Myślibórz from Barlinek + Myślibórz may be considered more of an issue but it may be possible to review all entries containing a + and substitute say & for + and append a space in cases such as Barlinek + Myślibórz, which would then result in Barlinek & Myślibórz (unless the choice is made to continue with a step that reverses the replacement).
So for the example data formulae can handle all but about two of the 19 cases. Extrapolated to the 10,000 rows 8,947 as required might be considered at least a good start and further 'rules' might be added in other columns to deal with any other observable patterns in data in ColumnF that at least is close to what is required.