Position pointer to specific line in a file

[Anyname Really]

TL;DR Summary

Is it possible, and if so, in what language, to read a specific line of a file without reading the lines before it?

I would like to read a specific line of a file. Everything I have seen does it by reading the file from the top, ignoring what is read until the line of interest is reached. Is it possible to move the pointer of the reader to the line directly, say by counting the number of line end sequences. Is the problem easier to solve if every line of data has the same format?

 

[jedishrfu]

If the lines are all of a fixed length then it is possible to seek into a file to a specific line start otherwise no.

In most text files, lines are of varying length and it is not possible to seek into the file at the start of a given line without some sort of prebuilt index of seek values. One can create an index and as long as the file doesn't change ie lines added, updated or deleted then it can be used to locate a specific line.

Python, Java, C and Go come to mind as languages with a seek function.

Here's a discussion using python:

https://pynative.com/python-file-seek/

and java

https://www.tutorialspoint.com/java/io/randomaccessfile_seek.htm

and in c:

https://www.scaler.com/topics/c/random-access-file-in-c/

and in golang:

https://golang.hotexamples.com/examples/os/File/Seek/golang-file-seek-method-examples.html

other languages that read files should have similar strategies.

Lastly, python programmers usually read in the whole file into an array that can be easily accessed which of course takes up a larger amount of memory than the seek+index table would.

 

[pbuk]

Is it possible to move the pointer of the reader to the line directly, say by counting the number of line end sequences.

How are you going to count line endings without reading the file?

Is the problem easier to solve if every line of data has the same format?

Yes, most programming languages/operating systems can read directly from a specified position in a file.

 

[Anyname Really]

Interesting. I just got two responses which seem to contradict one another. The first makes more sense. Maybe I am asking for the impossible. A partial solution would be to know how a computer knows where it is in a file. I suppose, the complier, in a language such as C++, has some internal variable which indicates where it is in a file. If so, is it possible to access this variable directly?

 

[pbuk]

Interesting. I just got two responses which seem to contradict one another.

I don't think they do, they both state the obvious which is that if you know what position you want to go to then you can go to it but if you don't you can't.

I suppose, the complier, in a language such as C++, has some internal variable which indicates where it is in a file. If so, is it possible to access this variable directly?

Why don't you follow the links in @jedishrfu' s post, these will tell you all you need to know. I would not recommend C++ for someone starting out in programming.

 

[Ibix]

I don't see any contradiction either. A text file is a string of characters, each taking up one byte. Thus you can easily skip to (e.g.) the 3047th character in a file without reading the previous 3046 in most languages just by asking the file system to give you the character 3046 characters after the start position of the file.

What you can't do without reading the 3046 characters is know how many "end of line" characters you skipped, so you can't generally skip to the 453rd line. You can do it if you know that all your lines are guaranteed to be (e.g.) 80 characters long, by skipping to character 80×453. Alternatively, if you have an index file (perhaps with fixed length entries) that tells you what character number line 453 starts at you can then skip to that character. But both of these strategies rely on having extra information about the file.

 

[pbuk]

A text file is a string of characters, each taking up one byte.

Many text files use UTF-8 encoding so some characters may take up more than one byte. Because of this you need to be careful when seeking into a text file.

 

[hmmm27]

Interesting. I just got two responses which seem to contradict one another.

They don't contradict : you just don't have enough basic knowledge of the subject matter.

You can look up "relative record data set" for IBM's implementation. It's a system level component of VSAM (probably ISAM as well). I'd be mildly surprised (not shocked though) if you couldn't find PC software that does it. No clue what keywords to use for that search.

But, if the file is small enough that it can be stuffed into an array, it's almost trivial to write a file handler as a program subroutine, in pretty much any language. Could easily do variable-length records, as well, using another array as an index.

As was previously mentioned in the "contradictory" posts.

 

[jedishrfu]

Many text files use UTF-8 encoding so some characters may take up more than one byte. Because of this you need to be careful when seeking into a text file.

One further point is that you might start in the middle of a multi-byte character and thus get buggy results.