bind-to-tinydns 0.4.3 Daniel Erat <dan-tinydns@erat.org>
================================================================================
What it does / Why it does it
================================================================================
This program attempts to convert a BIND 9 zone file into the tinydns-data
format natively used by the tinydns component of Dan Bernstein's djbdns
package (http://cr.yp.to/djbdns.html). This can be useful if you need to
periodically convert zones that are stored in BIND format to tinydns format
and you don't have the willingness or the oodles of memory needed to run a
copy of BIND from which you can slave the zones with axfr-get (think >100
MB DNSbl zones, for instance). For one-time conversions, you should
instead use the AXFR procedure described by Dan
(http://cr.yp.to/djbdns/run-server-bind.html).
Getting / Compiling
================================================================================
The newest version of this program is available from my homepage
(http://erat.org/). After downloading the source, extract it and run
"make". Copy bind-to-tinydns wherever you want after this (I would
recommend /usr/local/bin). Edit the Makefile and source or email me if it
doesn't work.
Usage
================================================================================
bind-to-tinydns is invoked in the following manner:
bind-to-tinydns <origin> <output file> <temp file>
The BIND zone is read from STDIN. To convert a BIND zone file named
"input" containing the zone "example.com" to a tinydns-data file named
"output", you would run:
bind-to-tinydns example.com output output.tmp <input
If the conversion is successful, the program exits with a return value of
0. Otherwise, it exits with a return value of 1 (after deleting the
temporary file). It will abort if the temporary file already exists.
Portability
================================================================================
I've only tested this program on Linux. I hope that it will work on most
UNIX systems. Please let me know if it doesn't.
Details
================================================================================
The zone file format described in RFC 1035, and by extension, the one used
by BIND, is not parser-friendly. Without making this program a bigger mess
than it already is, I've tried to make it as close to BIND 9 as possible in
what it accepts and how it interprets data. I am aware of the following
differences:
- Missing SOA records in zone files do not cause failure.
- Missing NS records in zone files do not cause failure.
- RRs that occur before $TTL directives and before any SOA records do
not cause failure or warnings (a default TTL of 86400, rather than
that of the MINIMUM value of the SOA's rdata, is used).
- The parser does not reject or modify RRs with the same owner that
have different TTLs.
- IN is the only supported record class. Records in other classes are
ignored.
- The only supported record types are SOA, NS, MX, A, CNAME, PTR, TXT,
and SRV. Records of other types are ignored.
- The $INCLUDE directive is not supported, and causes failure.
If you find additional differences (or worse yet, input that makes the
program crash or go into an infinite loop), or if any of these differences
bug you, please let me know so I can fix the problem.
Version History / Changelog
================================================================================
0.3 20030521 Initial release.
0.4 20030601 Added a string structure and modified functions to use it,
yielding simpler, safer code and a 33% performance increase
when processing the TXT-record heavy relays.osirusoft.com
zone. Spun off tokenizing and parsing/emitting sections of
main() into new functions. Tried to mitigate
aforementioned safety by adding a bunch of nasty code to
handle $GENERATE directives (including ${offset,width,base}
parameters). The program now accepts lines with
unterminated double- or backslash-quoted strings, but
prints a warning. Fixed a silly bug that would allow
specific out-of-zone data through.
0.4.1 20030608 Fixed a stupid typo that prevented too-large TTLs in RRs
from being detected. Changed str_to_uint() to behave
closer to the way that my other string-handling functions
behave. Added size checks for time values in SOA RDATA.
0.4.2 20040326 Mariano Absatz <baby@baby.com.ar> wrote that he came
across some zone files generated by Microsoft's DNS software
that had invalid (too large) TTL values. Made his
recommended changes, so the program spits out a warning and
uses the default TTL instead of exiting with a fatal error.
I also made the program a bit more lenient in other places
(RRs of unknown type are skipped with a warning message
instead of causing failure, for example). Thanks Mariano!
0.4.3 20050817 Ick, I can't believe that I wrote this ugly code. :(
Fixed a bug that Rein <o7rein@uninet.ee> ran into, where an
origin of "." resulted in all records being declared
out-of-zone.
