Lesson 23 · Parsing a section of records

Parsing a section of records

A response section is simply a count of resource records back to back. Today you parse a run of N records into a list, each one advancing the cursor past its RDATA - the demo that ties the whole chapter together.

The goal

Parse a run of N resource records into a list, tracking the cursor across each.

Start here - the target

TO DO

Scenario: A count of records parses into a list

Givena message where offset 12 holds www.example.com, followed by two A records that both point their name to offset 12 (C0 0C), type A, class IN, TTL 300, RDLENGTH 4, with RDATA 5D B8 D8 22 and 5D B8 D8 23

Whentwo records are parsed starting just after the name

Thenthe list has 2 records, both named "www.example.com", with A addresses "93.184.216.34" and "93.184.216.35"

Andeach record advances the cursor past its own RDATA so the second parses right after the first

Background

A response does not hold one record - it holds sections, and each section is just a count of resource records laid end to end (the counts come from the header: ANCOUNT, NSCOUNT, ARCOUNT). Parsing a section is therefore a loop: parse a record, advance the cursor by however many bytes it consumed, and repeat for the count. The per-record consumed length you have tracked all chapter is exactly what makes the cursor land on the next record.

The two A records here share a name via compression (both C0 0C pointing at www.example.com at offset 12), which is the common case in a real answer section listing several addresses for one host. Parsing them yields a two-element list with addresses 93.184.216.34 and 93.184.216.35. This is the chapter’s payoff: you can now read an entire block of answers. Next chapter uses it to parse a full response and start resolving names for real.

Make it work

func parseRecords(msg []byte, off, count int) ([]RR, int, error) {
  var rrs []RR
  for k := 0; k < count; k++ {
    rr, n, err := parseRR(msg, off) // record, byte length, err
    if err != nil { return nil, off, err }
    rrs = append(rrs, rr)
    off += n
  }
  return rrs, off, nil
}

CheckpointDONE

You can parse a whole section of records. Commit and stop here.