Search all files containing text

Question

With Busybox, how do you search for an expression within a bunch of files recursively through a bunch of directories, but only look through text files?

We don't know what the file's suffix is going to be; it could be .sh, it could be nothing, it could be something else. I was considering somehow basing the search on encoding although I am not quite sure what the encoding would be either.

I've tried busybox grep -r but it searches through binary files too, which wastes a lot of time.

Answer 1

grep supports the option --binary-files to treat binary files as binary, text or without-match. The last option should skip binary files. It is equivalent to grep -I. I did a test and it seems Busybox's grep supports -I to skip binary files.

grep is not "magical" in determining if a file is binary or text: it simply checks the first few bytes to see if they seem to be text or not and assumes the same for the rest of the file (according to man grep).

---

Sidenotes: one could emulate this by e.g. comparing if head -1 myfile of a file equals head -1 myfile | strings. If file is available it is even more reliant, but it is not in Busybox, and most likely not on the system if Busybox is primarily used.

Answer 2

What is the difference between a binary file with the struct

struct MyBin
{
    byte a;
    byte b;
    byte c;
}

with the values

myBin.a = 70;
myBin.b = 111;
myBin.c = 111;

And a text file with the text Foo?

All a text file is, is a binary file that you interpret using special look up codes called Character Encodings (ASCII, UTF-8, ect...). So there is no easy way to tell "Binary Files" apart from "Text Files".

There may be a way to exclude files that have the execute bit set, or only search files under a file-size (I doubt your text file is over 1 MB) but I do not have enough knowledge on how to filter the grep results to give a example of how to do it.