How can I convert all files to txt?

I'm working on a project which consists on developing a matlab version of the lz77 algorithm and it's extension lzss. The algortihms are already implemented and fully working for text file imported as char array. In the project I should also zip and unzip all the other types of data (like images or sounds) but I haven't found anything like that on the internet, in this forum, etc.. Is there a way to convert all files to text and than back to their original extension in MatLab? I've attemped to open files from a text editor but the algortihm seems not to work for the huge number of symbols. Thank you for all the help.

[Rik]

What do you mean with converting to txt? Changing the extensions is simple enough, but I suspect you don't mean that. Usually whether or not you can convert something to plain text is highly dependent on the source data file format. Generally there will not be an obvious way to do this, which is why there are functions like imread to read images and there is a similar function to read .wav files ( wavread ).

[Simone De Vecchi]

I need to have an array of char to use my algorithm so I need to represent audio and images as that. So, is there a tool which converts this kinds of data in text or do I need to develop my own kind of software which maps the RGB values to char? And for the audio file how can I do that? When I import .wav files I have complex numbers, how can I map them into chars?

[Guillaume]

To me the question shows a lack of understanding over what a data type is or how data is encoded on a computer.

Fundamentaly, 6here is no difference between text and binary data such as an image or audio file. They are both encoded as numbers. The only difference comes from how the reader interprets these numbers.

The sequence of number 65:72 in a file can be interpreted as char as the sequence 'ABCDEFGH'. It could of course be interpreted as the sequence of 8-bit integers 65 to 72. It could also be interpreted as the sequence of 16-bit integers [16961, 17475, 17989, 18503], or it could be interpreted as the double 1.5839e40. All of these are encoded exactly the same way on a (little-endian) computer.

Therefore, saying that an algorithm only work with text is meaningless. text is just a sequence of number like any other data type. The only difference may be that the sequence of numbers used for text is limited to a certain range of values. For example, text using US-ASCII encoding is limited to the range of 8-bit values 0 to 127. However, since that wasn't specified in the question, it must be assumed that text means the char type of matlab which has no restriction on the range.

Note that in matlab, text is encoded as a sequence of 16-bit integers.

So, really, to convert a file to text, you just have to read the sequence of numbers and tell matlab they are characters, as per dpb's answer.

[Simone De Vecchi]

Thank you Guillaume for your entusiasm. I know how data is stored and all the things you wrote. Nevertheless I developed an algorithm that works with chars because the requirement of the project was to compress text files. As an extension of that algorithm I need to make the algorithm work with also images and sounds. My question was: is there in matlab a routine which automatically converts images and sounds in chars? And then back into their original data type?

To be more precise and to follow your comment: is there a Matlab routine which manipulates the bits of any data type and then group them 8 by 8 so I can interpret them as chars and then al the way back to reconsider them as their original data type?

Hope that everything is clear, I'm italian, I know my english is not perfect, sorry for that.

Answer 1

[dpb]

To consider any file as simply a stream of bytes (char), write

dat=fread(fid,inf,'*char');

What this ignores with special file formats such as audio, image, etc., etc., is that there is header information in the file as well as just the data.

So it depends upon whether you're wanting to compress the file itself or the content of the file.

[Simone De Vecchi]

Thank you dpb for your answer, but I'm looking for something different, if you want check my last comment. Thank you anyway

Answer 2

[Walter Roberson]

"is there a Matlab routine which manipulates the bits of any data type and then group them 8 by 8 so I can interpret them as chars and then all the way back to reconsider them as their original data type?"

No. All of the MATLAB data types except char, logical, and the numeric ones, use the internal equivalent of pointers into memory. You might be able to get access to the pointer value, but when you reconstruct it afterwards, it is not likely to be pointing to a useful area of memory.

Because of that, some people have written serialization and deserialization routines, such as https://www.mathworks.com/matlabcentral/fileexchange/34564-fast-serialize-deserialize . Those are able to encode data as streams of bytes, and when they are deserialized, the objects are reconstructed as best feasible. But aspects such as internal serialization numbers might not get reconstructed, and the memory pointers are not going to be the same, so "==" might not work between the original version and the reconstructed version.

"As an extension of that algorithm I need to make the algorithm work with also images and sounds"

If those images or sounds are stored in files, the same way that your text is stored in files, then you do not need to do any of the serialization that I discuss above. Instead, just read the files as a sequence of chars like Duane showed, dat=fread(fid,inf,'*char'); or using dat = fileread('NameOfFile');

If you have data that has already been read into memory and is stored in numeric arrays (e.g., you are not trying to encode a drawing of an image, just the data of the image), then you can use

Class_of_variable = class(YourNumericVariable);
Variable_as_char = char( typecast(YourNumericVariable, 'uint8') )

and to reverse that,

Restored_numeric_variable = typecast( uint8(0 + Variable_as_char), Class_of_variable );