Prev: Fix for Isosurface problems
Next: How to change a gui's figure icon
From: Sydney Hai on 16 Jun 2010 15:56
Dear all,

I'm new to Matlab, so here is a problem I encountered when trying to modify a text file.

1. My file has four lines in the beginning that do not contain the data I want so I want to skip all the them when reading my file, it seems like skip header function can only skip one? and the four lines are all of different lengths, contain both strings and numeric values, is there a way to specify my function so when I do textscan or dlmread I can skip all four of them?

2. After all the calculations, I generated new data that I want to append to my text file,
I used fseek to the end of the file and then fprintf to append, but the end of my file has an "END" that I want to keep, so is there a way to fseek to the end of the file, but before the "END". How do I append right after my last row of data but before the "END"?
example:
my data looks like this:
AAAA 1 2 3 4
BBBBB 1 2 3 4
END

3. when using randi(64) to generate a random integer, is there a way to make sure that every time I run the function, the number I generated is unique, different from previous session? I don't know if matlab even has a way to remember what number it generated the last time and avoid that. T.T

4. I'm trying to do calculation using two rows of data in one file, if i randomly choose the first one A(X,Y,Z), when choosing the second one, I'm using textscan as well, but is there a way to skip point A and only choose the second one B from the rest of the file?

Sorry for so many questions! Thank you all in advance for all the help!

Sincerely,
Syd
From: dpb on 16 Jun 2010 16:12
Sydney Hai wrote:
....

> I'm new to Matlab, so here is a problem I encountered when trying to
> modify a text file.

Ewww....I foresee "issues", but let's go on for now...

> 1. My file has four lines in the beginning that do not contain the data
> I want so I want to skip all the them when reading my file, it seems
> like skip header function can only skip one? ...

I don't know about dlmread() much, but certainly the 'headerlines'
argument in textscan() accepts a following numeric value for the number
of lines to skip. It doesn't care what's in the line(s), only that they
are properly terminated for the specific platform.

> 2. After all the calculations, I generated new data that I want to
> append to my text file, I used fseek to the end of the file and then
> fprintf to append, but the end of my file has an "END" that I want to
> keep, so is there a way to fseek to the end of the file, but before the
> "END". How do I append right after my last row of data but before the
> "END"? example:
> my data looks like this:
> AAAA 1 2 3 4 BBBBB 1 2 3 4 END

And, indeed, here it is w/ "modifying" a text file...sequential text
files are, by their nature, er...., sequential.

There is no "backspace" operator in a Matlab file, you could only seek
"not so far" as the end and then read until you found the \n for the
next-to-last line. Then, you would need to read the last line,
backspace (fseek(fid,-len,0) the length of that line bytes (yet again),
then rewrite the line w/o the "END" string, then continue on.

The upshot is, as is usual w/ sequential text files, it's probably
simply easier to read the file line by line until find "END" (meanwhile
writing what found to a new file) and the when find the last line modify
and write, then keep on in the new file w/ the new data. Once done,
close both and (optionally) rename the new to the old name if that is
the desired result.

> 3. when using randi(64) to generate a random integer, is there a way to
> make sure that every time I run the function, the number I generated is
> unique, different from previous session? I don't know if matlab even has
> a way to remember what number it generated the last time and avoid that.

doc randi % discusses all those things about using the rng's in ML

> 4. I'm trying to do calculation using two rows of data in one file, if i
> randomly choose the first one A(X,Y,Z), when choosing the second one,
> I'm using textscan as well, but is there a way to skip point A and only
> choose the second one B from the rest of the file?
....

Yes, but...I'd suggest probably easier/faster to read the file and
simply select the desired row from memory rather than trying to do it w/
i/o operations. (Or at least read the amount of the file from which the
selection will be made if the entire file is much larger)

--
 | 
Pages: 1
Prev: Fix for Isosurface problems
Next: How to change a gui's figure icon