Python module for Dbase files?

[William Dandreta]

Hi Anders,

>Why don't you set up a Linux server and a TCP/IP net with DOS clients.
>Then you could have the database on the server and you would have much
>more to choose from!

That would be a good idea but we are not using plain DOS. It is Concurrent
DOS, amultitasking/mulituser DOS system and each user has a terminal. It is
a legacy system that is no longer being upgraded. It does not support
networking.

>>
>It's probably nothing wrong with xDB, but I think it's rather a big
>task to port it to DOS and create a Python Wrapper. So maybe there are
>other solutions that are better?


I don't think it will be as difficult as you imagine. I have documentation
on the dbase file structure. I wrote a simple Python program today that
takes the data in a Dbase file an converts it to a comma delimited ASCII
file. (A copy is below, I haven't finished all the data types yet.)

The only part I haven't a clue about is the indexing. Our needs are limited.
If I can add to the database, update indexes and access data using indexes,
I would be satisfied. So, I would only need to port a small part of it.
-------------------------------------------
# converts .dbf file to comma delimited ascii.

from struct import unpack
from sys import exit

DBFfile = 'tier_5.dbf'
DBF = open(DBFfile,'rb')
hdra = DBF.read(32)
numRecs,hdrSize,recSize = unpack('lhh',hdra[4:12])
hdrbSize = hdrSize-32
hdrb = DBF.read(hdrbSize)
numFields = (hdrbSize-1)/32
if '\x0D' <> hdrb[-1]:
  print 'Cannot read DBF file, header is corrupted.'
  exit(1)
line = ''
varFields = {}
for i in range(11,32*numFields+11,32):
  if (hdrb[i] == 'C') or (hdrb[i] == 'N'):
    line = line + '%s, '
    varFields[i] = ord(hdrb[i+5])
  elif hdrb[i] == 'D':
    pass #date stuff here
  elif hdrb[i] == 'L':
    pass #logic stuff here
  else:
    line = line + 'NOT A TEXT FIELD ,'
line = line[:-2] + '\n'
txtFile = DBFfile[:-3]+'txt'
txt = open(txtFile,'w')
for i in range(numRecs):
  record = DBF.read(recSize)
  vf = ()
  j = 0
  k = 1  #first byte is a tag
  while j < 32*numFields:
    fieldSize = varFields[11+j]
    vf = vf + (record[k:k+fieldSize],)
    k = k + fieldSize
    j = j + 32
  txt.write(line % vf)
--------------------------------------------
Bill