CODEWITHSUNDEEP
cipcpipe
ajay_t
ajay_t

Reputation: 2385

Reading data of variable size in buffer using read system call in C

I have an application which writes data (control data, access information etc) to one end of the pipe at parent process. At the child process, I want to read that data as it is.

Parent process performs many write() operation at many location. For reading the data into the buffer, we need to specify the length of the data

read(int fd, buffer, len).

My problem is, parent process writes variable size data every time. So how would the child process came to know the length of data.

I have tried to read single character and add it to buff as, 

char ch;
int n = 0;
while(n >= 0)
{
    n = read(int fd, ch, 1);
    *buff = ch; buff++;
}

But it doesn't seems to way to do it

Please tell me how read variable size data in child process?

Upvotes: 2

Views: 677

Answers (3)

Nick
Nick

Reputation: 5198

You'll need to put formatted data in the pipe, that can be decoded on the read end. That is, you'll have to specify a format/protocol to be able to understand the data that's coming out. You could specify length, or use lines ending with a \n or \0 character, whatever suites your data.

Upvotes: 4

DigitalRoss
DigitalRoss

Reputation: 146093

Heh, you've asked one of the oldest questions there is. Many answers have been developed over the years...

  • write formatted data organized as lines then read one line at a time and parse as much data as is there
  • write binary data, but write a record type first
  • as above, but in two levels, use binary data, wrapped in a record with a type field, and all that preceded by a record length word. This allows you to structure the reader in layers, a lower I/O layer can easily read the records and return their length to a higher layer that is responsible for application-specific logic.
  • write formatted data in lines, but identify the "record type" with a leading identifier
  • create a language and write a parser for it; the parser probably reads the input as text byte-by-byte
  • organize the data as XML
  • organize the data as YAML
  • organize the data as JSON

All of these techniques have one thing in common: the reader must already know the encoding type and must be prepared, using program logic, to read the data conditionally according to its structure. There are, of course, libraries already written that can read XML, YAML, and JSON.

Upvotes: 4

Some programmer dude
Some programmer dude

Reputation: 409206

Two most common and simplest methods are to either write the length first in a fixed-size, or to have a special record-terminator that tells that the record has ended.

Upvotes: 4

Related Questions