Changing the header in a file which include binary content
显示 更早的评论
My ply file contains an ascii header and the rest of the mesh data content is written as binary. I would like to change lines 10, 11, 12 in the header with my specific content. I do not want to touch to the binary data which contains the mesh data. When I use the script below, my mesh data becomes like a flat surface instead of a 3D surface curvature. Why am I changing the rest of the content? How can I change the three header lines, without damaging the rest of the file?
A = regexp( fileread(ply), '\n', 'split');
A{10} = sprintf('%s',"header line 1");
A{11} = sprintf('%s',"header line 2");
A{12} = sprintf('%s',"header line 3");
fid = fopen(ply, 'wb');
fprintf(fid, A{:});
fclose(fid);
采纳的回答
Guillaume
2018-4-11
For a start, you never put back all the \n characters that you've removed.
fprintf(fid, strjoin(A, '\n'));
may work. However, you're playing with fire here as you're interpreting binary data as text. There's no guarantee that the binary data will survive the round trip unaltered (e.g. the binary data forms an invalid character that could be ignored by regex).
And of course, it's possible that he binary data includes offset indicating where to find other binary data into the file, if you change the length of the header the offsets will point to the wrong location.
It's very unusual to have a file that's text and binary. It may be that some of the binary content is some textual data, but it should still be interpreted as binary. Usually text in a binary file is either fixed length, prefixed by binary data indicating how long the text is, or \0 terminated.
6 个评论
Beril Sirmacek
2018-4-12
Guillaume
2018-4-12
At this point, the best would be for you to provide an example file and if possible, a link to a document specifying the format of the file.
As both Walter and I said, treating the binary part as text is dangerous anyway.
Beril Sirmacek
2018-4-12
Guillaume
2018-4-12
Does the attached file show the problem as well?
It was created with this code:
fid = fopen('result2.0.ply', 'r');
text = {];
while true
tline = fgetl(fid);
assert(ischar(tline), 'end of file reached prematurely');
text = [text; {tline}];
if strcmp(tline, 'end_header')
break;
end;
end;
binarydata = fread(fid);
fclose(fid);
text(11:13) = {'line 1'; 'line 2'; 'line3'}
fid = fopen('newresult.ply', 'w');
fprintf(fid, '%s\n', strjoin(text, '\n'));
fwrite(fid, binarydata);
fclose(fid);
Beril Sirmacek
2018-4-12
Guillaume
2018-4-13
Well, it was meant to be:
text = {};
but either work.
更多回答（1 个）
Walter Roberson
2018-4-11
When you split, the newlines are removed, and you are not putting them back.
fprintf(fid, '%s\n', A{1:end-1});
fprintf(fid, '%s', A{end}); %do not add extra \n at the end
but splitting the binary content is kinda dicey.
I would recommend that you instead switch to reading a limited number of lines using fgetl(), and write out the revised versions as required, and then fread() to end of file and fwrite() it to the new file.
类别
在 帮助中心 和 File Exchange 中查找有关 String Parsing 的更多信息
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!
Translated by 
