Extracting certain lines from a text file containing both characters and numbers

I am trying to load the following data into MATLAB. I then want to extract the line starting with "length area (&)" until "16 0.9 " into a new text file. I have a large number of files exactly like this but where the only difference is the values. Is there a piece of code I could use to extract these lines that can be used for different files? I am pretty new to MATLAB so suggestions welcome.

volume(ml)					122.278 							6.064 	--	9.460 	6.064 


Regional Enhancement (AHA segmentation)
	rel.
Data starting
length	area (&)
1	0.000 
2	0.3 
3	0.3 
4	158 
5	57 
6	17 
7	0.00 
8	0.02 
9	13 
10	1.15 
11	10.20 
12	3.68 
13	18 
14	16 
15	0.6 
16	0.9 

Area of lifetime isotope (all slices)
			rel.
	slice		ending
slice	place.	segment	volume (%)
10	-3.37 	*	--
20	-4.37 	*	--
30	-5337 	*	--
40	-6.37 	1	0.0 

[dpb]

I took a guess at formatting the code/text...don't know where actual line breaks are for sure, however, so don't know how many header lines are actually in the file.

Always helps to actually attach a sample file instead of trying to paste in if is possibility of line wrap.

[Guillaume]

Rather than pasting the content of the text file, attach the file to the question. Text files can have different text encodings and line ending characters which are all lost when you paste but may make a difference to the parsing.

Plus it makes your comment wayyyyy too long (please delete the paste after attaching your file).

Answer 1

[dpb]

Presuming records (including blank lines) are as above,

fid=fopen('HB.txt');
data=cell2mat(textscan(fid,'%f%f','headerlines',5));
fid=fclose(fid);

returned

data =
    1.0000         0
    2.0000    0.3000
    3.0000    0.3000
    4.0000  158.0000
    5.0000   57.0000
    6.0000   17.0000
    7.0000         0
    8.0000    0.0200
    9.0000   13.0000
   10.0000    1.1500
   11.0000   10.2000
   12.0000    3.6800
   13.0000   18.0000
   14.0000   16.0000
   15.0000    0.6000
   16.0000    0.9000

May need to adjust the number of header lines; when I copied from the text above, the apparent blank records disappeared so don't know if they're real or not in the original.

Using the specific format causes textscan to fail on the first record it finds that doesn't match which is the next after the last data you happen to be interested in--serendipity since on failure it returns the data that it did successfully convert.

[HB]

Hi,

Thank you for your reply.

I may have miscommunicated my question. I did not paste the entire text file as I didn't think it would be necessary, clearly it is so apologies!! I have attached the entire file below (apologies for the length of text). Any suggestions? Serendipity didn't come my way after all.

EDIT: all 1636 lines of text data deleted from this comment, and uploaded as a text file (with absolutely no guarantee that it has the same EOL characters, file format, etc. as the original data file).

[dpb]

That is NOT attaching the file, that's just pasting in more. Use the paperclip icon instead.

[dpb]

The above answer still works, you just have to adjust the 'headerlines' argument to account for how much stuff there is before the section of interest.

If the file format is always identical, you can just use a constant empirically determined by loading a file in the editor and looking.

If there can be a different number of lines in preceding data sections, then you'll have to read through the file until you find that header line preceding the section of interest, then read from that point using the correct 'headerlines' argument count for just the extra lines at that point.

I posted a solution to another question within last couple weeks at outside that illustrates identically this--unfortunately, I can't recall the actual Q? title to point at it immediately, but it is essentially just

fid=fopen('HB.txt');
while ~feof(fid)
  l=fgetl(fid);
  if strfind(l,'yoursearchpattern'),break,end
end
data=cell2mat(textscan(fid,'%f%f','headerlines',N));
fid=fclose(fid);

where you need to pick the unique search pattern and N lines to skip after the section is found.

[Stephen23]

"Any suggestions?"

Upload data files, rather than pasting them as text in your questions/comments.

"apologies for the length of text"

Files can be uploaded by clicking the paperclip button.