Newbie code review of parsing program Please

L

len

I have created the following program to read a text file which happens
to be a cobol filed definition. The program then outputs to a file
what is essentially a file which is a list definition which I can
later
copy and past into a python program. I will eventually expand the
program
to also output an SQL script to create a SQL file in MySQL

The program still need a little work, it does not handle the following
items
yet;

1. It does not handle OCCURS yet.
2. It does not handle REDEFINE yet.
3. GROUP structures will need work.
4. Does not create SQL script yet.

It is my anticipation that any files created out of this program may
need
manual tweeking but I have a large number of cobol file definitions
which
I may need to work with and this seemed like a better solution than
hand
typing each list definition and SQL create file script by hand.

What I would like is if some kind soul could review my code and give
me
some suggestions on how I might improve it. I think the use of
regular
expression might cut the code down or at least simplify the parsing
but
I'm just starting to read those chapters in the book;)

*** SAMPLE INPUT FILE ***

000100 FD SALESMEN-FILE
000200 LABEL RECORDS ARE STANDARD
000300 VALUE OF FILENAME IS "SALESMEN".
000400
000500 01 SALESMEN-RECORD.
000600 05 SALESMEN-NO PIC 9(3).
000700 05 SALESMEN-NAME PIC X(30).
000800 05 SALESMEN-TERRITORY PIC X(30).
000900 05 SALESMEN-QUOTA PIC S9(7) COMP.
001000 05 SALESMEN-1ST-BONUS PIC S9(5)V99 COMP.
001100 05 SALESMEN-2ND-BONUS PIC S9(5)V99 COMP.
001200 05 SALESMEN-3RD-BONUS PIC S9(5)V99 COMP.
001300 05 SALESMEN-4TH-BONUS PIC S9(5)V99 COMP.

*** PROGRAM CODE ***

#!/usr/bin/python

import sys

f_path = '/home/lenyel/Bruske/MCBA/Internet/'
f_name = sys.argv[1]

fd = open(f_path + f_name, 'r')

def fmtline(fieldline):
size = ''
type = ''
dec = ''
codeline = []
if fieldline.count('COMP.') > 0:
left = fieldline[3].find('(') + 1
right = fieldline[3].find(')')
num = fieldline[3][left:right].lstrip()
if fieldline[3].count('V'):
left = fieldline[3].find('V') + 1
dec = int(len(fieldline[3][left:]))
size = ((int(num) + int(dec)) / 2) + 1
else:
size = (int(num) / 2) + 1
dec = 0
type = 'Pdec'
elif fieldline[3][0] in ('X', '9'):
dec = 0
left = fieldline[3].find('(') + 1
right = fieldline[3].find(')')
size = int(fieldline[3][left:right].lstrip('0'))
if fieldline[3][0] == 'X':
type = 'Xstr'
else:
type = 'Xint'
else:
dec = 0
left = fieldline[3].find('(') + 1
right = fieldline[3].find(')')
size = int(fieldline[3][left:right].lstrip('0'))
if fieldline[3][0] == 'X':
type = 'Xint'
codeline.append(fieldline[1].replace('-', '_').replace('.',
'').lower())
codeline.append(size)
codeline.append(type)
codeline.append(dec)
return codeline

wrkfd = []
rec_len = 0

for line in fd:
if line[6] == '*': # drop comment lines
continue
newline = line.split()
if len(newline) == 1: # drop blank line
continue
newline = newline[1:]
if 'FILENAME' in newline:
filename = newline[-1].replace('"','').lower()
filename = filename.replace('.','')
output = open('/home/lenyel/Bruske/MCBA/Internet/'+filename
+'.fd', 'w')
code = filename + ' = [\n'
output.write(code)
elif newline[0].isdigit() and 'PIC' in newline:
wrkfd.append(fmtline(newline))
rec_len += wrkfd[-1][1]

fd.close()

fmtfd = []

for wrkline in wrkfd[:-1]:
fmtline = str(tuple(wrkline)) + ',\n'
output.write(fmtline)

fmtline = tuple(wrkfd[-1])
fmtline = str(fmtline) + '\n'
output.write(fmtline)

lastline = ']\n'
output.write(lastline)

lenrec = filename + '_len = ' + str(rec_len)
output.write(lenrec)

output.close()

*** RESULTING OUTPUT ***

salesmen = [
('salesmen_no', 3, 'Xint', 0),
('salesmen_name', 30, 'Xstr', 0),
('salesmen_territory', 30, 'Xstr', 0),
('salesmen_quota', 4, 'Pdec', 0),
('salesmen_1st_bonus', 4, 'Pdec', 2),
('salesmen_2nd_bonus', 4, 'Pdec', 2),
('salesmen_3rd_bonus', 4, 'Pdec', 2),
('salesmen_4th_bonus', 4, 'Pdec', 2)
]
salesmen_len = 83

If you find this code useful please feel free to use any or all of it
at your own risk.

Thanks
Len S
 
L

len

I have created the following program to read a text file which happens
to be a cobol filed definition.  The program then outputs to a file
what is essentially a file which is a list definition which I can
later
copy and past into a python program.  I will eventually expand the
program
to also output an SQL script to create a SQL file in MySQL
The program still need a little work, it does not handle the following
items
yet;
1.  It does not handle OCCURS yet.
2.  It does not handle REDEFINE yet.
3.  GROUP structures will need work.
4.  Does not create SQL script yet.
It is my anticipation that any files created out of this program may
need
manual tweeking but I have a large number of cobol file definitions
which
I may need to work with and this seemed like a better solution than
hand
typing each list definition and SQL create file script by hand.
What I would like is if some kind soul could review my code and give
me
some suggestions on how I might improve it.  I think the use of
regular
expression might cut the code down or at least simplify the parsing
but
I'm just starting to read those chapters in the book;)
*** SAMPLE INPUT FILE ***
000100 FD  SALESMEN-FILE
000200     LABEL RECORDS ARE STANDARD
000300     VALUE OF FILENAME IS "SALESMEN".
000400
000500 01  SALESMEN-RECORD.
000600     05  SALESMEN-NO                PIC 9(3).
000700     05  SALESMEN-NAME              PIC X(30)..
000800     05  SALESMEN-TERRITORY         PIC X(30).
000900     05  SALESMEN-QUOTA             PIC S9(7) COMP.
001000     05  SALESMEN-1ST-BONUS         PIC S9(5)V99 COMP.
001100     05  SALESMEN-2ND-BONUS         PIC S9(5)V99 COMP.
001200     05  SALESMEN-3RD-BONUS         PIC S9(5)V99 COMP.
001300     05  SALESMEN-4TH-BONUS         PIC S9(5)V99 COMP.
*** PROGRAM CODE ***

import sys
f_path = '/home/lenyel/Bruske/MCBA/Internet/'
f_name = sys.argv[1]
fd = open(f_path + f_name, 'r')
def fmtline(fieldline):
   size = ''
   type = ''
   dec = ''
   codeline = []
   if fieldline.count('COMP.') > 0:
       left = fieldline[3].find('(') + 1
       right = fieldline[3].find(')')
       num = fieldline[3][left:right].lstrip()
       if fieldline[3].count('V'):
           left = fieldline[3].find('V') + 1
           dec = int(len(fieldline[3][left:]))
           size = ((int(num) + int(dec)) / 2) + 1
       else:
           size = (int(num) / 2) + 1
           dec = 0
       type = 'Pdec'
   elif fieldline[3][0] in ('X', '9'):
       dec = 0
       left = fieldline[3].find('(') + 1
       right = fieldline[3].find(')')
       size = int(fieldline[3][left:right].lstrip('0'))
       if fieldline[3][0] == 'X':
           type = 'Xstr'
       else:
           type = 'Xint'
   else:
       dec = 0
       left = fieldline[3].find('(') + 1
       right = fieldline[3].find(')')
       size = int(fieldline[3][left:right].lstrip('0'))
       if fieldline[3][0] == 'X':
           type = 'Xint'
   codeline.append(fieldline[1].replace('-', '_').replace('.',
'').lower())
   codeline.append(size)
   codeline.append(type)
   codeline.append(dec)
   return codeline
wrkfd = []
rec_len = 0
for line in fd:
   if line[6] == '*':      # drop comment lines
       continue
   newline = line.split()
   if len(newline) == 1:   # drop blank line
       continue
   newline = newline[1:]
   if 'FILENAME' in newline:
       filename = newline[-1].replace('"','').lower()
       filename = filename.replace('.','')
       output = open('/home/lenyel/Bruske/MCBA/Internet/'+filename
+'.fd', 'w')
       code = filename + ' = [\n'
       output.write(code)
   elif newline[0].isdigit() and 'PIC' in newline:
       wrkfd.append(fmtline(newline))
       rec_len += wrkfd[-1][1]
fd.close()

fmtfd = []
for wrkline in wrkfd[:-1]:
   fmtline = str(tuple(wrkline)) + ',\n'
   output.write(fmtline)
fmtline = tuple(wrkfd[-1])
fmtline = str(fmtline) + '\n'
output.write(fmtline)
lastline = ']\n'
output.write(lastline)
lenrec = filename + '_len = ' + str(rec_len)
output.write(lenrec)

*** RESULTING OUTPUT ***
salesmen = [
('salesmen_no', 3, 'Xint', 0),
('salesmen_name', 30, 'Xstr', 0),
('salesmen_territory', 30, 'Xstr', 0),
('salesmen_quota', 4, 'Pdec', 0),
('salesmen_1st_bonus', 4, 'Pdec', 2),
('salesmen_2nd_bonus', 4, 'Pdec', 2),
('salesmen_3rd_bonus', 4, 'Pdec', 2),
('salesmen_4th_bonus', 4, 'Pdec', 2)
]
salesmen_len = 83
If you find this code useful please feel free to use any or all of it
at your own risk.
Thanks
Len S

You might want to check out the pyparsing library.

-Mark

Thanks Mark I will check in out right now.

Len
 
S

Steve Holden

Mark said:
news:fc3ef718-edc4-4892-8418-3eeff0975edc@u18g2000pro.googlegroups.com... [...]

You might want to check out the pyparsing library.
And you might want to trim your messages to avoid quoting irrelevant
stuff. This is not directed personally at Mark, but at all readers.

Loads of us do it, and I wish we'd stop it. It's poor netiquette because
it forces people to skip past stuff that isn't relevant to the point
being made. It's also a global wste of bandwidth and storage space,
though that's less important than it used to be.

regards
Steve
 
L

Lawrence D'Oliveiro

len said:
if fieldline.count('COMP.') > 0:

I take it you're only handling a particular subset of COBOL constructs: thus, "COMP" is never "COMPUTATIONAL" or "USAGE IS COMPUTATIONAL", and it always occurs just before the full-stop (can't remember enough COBOL syntax to be sure if anything else can go afterwards).
elif newline[0].isdigit() and 'PIC' in newline:

Similarly, "PIC" is never "PICTURE" or "PICTURE IS".

Aargh, I think I have to stop. I'm remembering more than I ever wanted to about COBOL. Must ... rip ... brain ... out ...
 
L

Lawrence D'Oliveiro

Mark said:
Point taken...or I could top post ;^)

A: A Rolls seats six.
Q: What's the saddest thing about seeing a Rolls with five top-posters in it going over a cliff?
 
J

John Machin

A: A Rolls seats six.
Q: What's the saddest thing about seeing a Rolls with five top-posters in it going over a cliff?

+1 but you forgot the boot & the roof rack AND if it was a really old
one there'd be space for a few on the running boards (attached like
the Norwegian Blue parrot)
 
P

Paul McGuire

Thanks Mark I will check in out right now.

Len

Len -

Here is a rough pyparsing starter for your problem:

from pyparsing import *

COMP = Optional("USAGE IS") + oneOf("COMP COMPUTATIONAL")
PIC = oneOf("PIC PICTURE") + Optional("IS")
PERIOD,LPAREN,RPAREN = map(Suppress,".()")

ident = Word(alphanums.upper()+"_-")
integer = Word(nums).setParseAction(lambda t:int(t[0]))
lineNum = Suppress(Optional(LineEnd()) + LineStart() + Word(nums))

rep = LPAREN + integer + RPAREN
repchars = "X" + rep
repchars.setParseAction(lambda tokens: ['X']*tokens[1])
strdecl = Combine(OneOrMore(repchars | "X"))

SIGN = Optional("S")
repdigits = "9" + rep
repdigits.setParseAction(lambda tokens: ['9']*tokens[1])
intdecl = SIGN("sign") + Combine(OneOrMore(repdigits | "9"))
("intpart")
realdecl = SIGN("sign") + Combine(OneOrMore(repdigits | "9"))
("intpart") + "V" + \
Combine(OneOrMore("9" + rep | "9"))("realpart")

type = Group((strdecl | realdecl | intdecl) +
Optional(COMP("COMP")))

fieldDecl = lineNum + "05" + ident("name") + \
PIC + type("type") + PERIOD
structDecl = lineNum + "01" + ident("name") + PERIOD + \
OneOrMore(Group(fieldDecl))("fields")

It prints out:

SALESMEN-RECORD
SALESMEN-NO ['999']
SALESMEN-NAME ['XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX']
SALESMEN-TERRITORY ['XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX']
SALESMEN-QUOTA ['S', '9999999', 'COMP']
SALESMEN-1ST-BONUS ['S', '99999', 'V', '99', 'COMP']
SALESMEN-2ND-BONUS ['S', '99999', 'V', '99', 'COMP']
SALESMEN-3RD-BONUS ['S', '99999', 'V', '99', 'COMP']
SALESMEN-4TH-BONUS ['S', '99999', 'V', '99', 'COMP']

I too have some dim, dark, memories of COBOL. I seem to recall having
to infer from the number of digits in an integer or real what size the
number would be. I don't have that logic implemented, but here is an
extension to the above program, which shows you where you could put
this kind of type inference logic (insert this code before the call to
searchString):

class TypeDefn(object):
@staticmethod
def intType(tokens):
self = TypeDefn()
self.str = "int(%d)" % (len(tokens.intpart),)
self.isSigned = bool(tokens.sign)
return self
@staticmethod
def realType(tokens):
self = TypeDefn()
self.str = "real(%d.%d)" % (len(tokens.intpart),len
(tokens.realpart))
self.isSigned = bool(tokens.sign)
return self
@staticmethod
def charType(tokens):
self = TypeDefn()
self.str = "char(%d)" % len(tokens)
self.isSigned = False
self.isComp = False
return self
def __repr__(self):
return ("+-" if self.isSigned else "") + self.str
intdecl.setParseAction(TypeDefn.intType)
realdecl.setParseAction(TypeDefn.realType)
strdecl.setParseAction(TypeDefn.charType)

This prints:

SALESMEN-RECORD
SALESMEN-NO [int(3)]
SALESMEN-NAME [char(1)]
SALESMEN-TERRITORY [char(1)]
SALESMEN-QUOTA [+-int(7), 'COMP']
SALESMEN-1ST-BONUS [+-real(5.2), 'COMP']
SALESMEN-2ND-BONUS [+-real(5.2), 'COMP']
SALESMEN-3RD-BONUS [+-real(5.2), 'COMP']
SALESMEN-4TH-BONUS [+-real(5.2), 'COMP']

You can post more questions about pyparsing on the Discussion tab of
the pyparsing wiki home page.

Best of luck!
-- Paul
 
L

len

len said:
    if fieldline.count('COMP.') > 0:

I take it you're only handling a particular subset of COBOL constructs: thus, "COMP" is never "COMPUTATIONAL" or "USAGE IS COMPUTATIONAL", and it always occurs just before the full-stop (can't remember enough COBOL syntax to  be sure if anything else can go afterwards).
    elif newline[0].isdigit() and 'PIC' in newline:

Similarly, "PIC" is never "PICTURE" or "PICTURE IS".

Aargh, I think I have to stop. I'm remembering more than I ever wanted to about COBOL. Must ... rip ... brain ... out ...

Most of the cobol code originally comes from packages and is
relatively consistant.

Thanks
Len
 
L

len

Thanks Paul

I will be going over your code today. I started looking at Pyparsing
last night
and it just got to late and my brain started to fog over. I would
really like
to thank you for taking the time to provide me with the code sample
I'm sure it
will really help. Again thank you very much.

Len

Thanks Mark I will check in out right now.

Len -

Here is a rough pyparsing starter for your problem:

from pyparsing import *

COMP = Optional("USAGE IS") + oneOf("COMP COMPUTATIONAL")
PIC = oneOf("PIC PICTURE") + Optional("IS")
PERIOD,LPAREN,RPAREN = map(Suppress,".()")

ident = Word(alphanums.upper()+"_-")
integer = Word(nums).setParseAction(lambda t:int(t[0]))
lineNum = Suppress(Optional(LineEnd()) + LineStart() + Word(nums))

rep = LPAREN + integer + RPAREN
repchars = "X" + rep
repchars.setParseAction(lambda tokens: ['X']*tokens[1])
strdecl = Combine(OneOrMore(repchars | "X"))

SIGN = Optional("S")
repdigits = "9" + rep
repdigits.setParseAction(lambda tokens: ['9']*tokens[1])
intdecl = SIGN("sign") + Combine(OneOrMore(repdigits | "9"))
("intpart")
realdecl = SIGN("sign") + Combine(OneOrMore(repdigits | "9"))
("intpart") + "V" + \
                Combine(OneOrMore("9" + rep | "9"))("realpart")

type = Group((strdecl | realdecl | intdecl) +
                Optional(COMP("COMP")))

fieldDecl = lineNum + "05" + ident("name") + \
                PIC + type("type") + PERIOD
structDecl = lineNum + "01" + ident("name") + PERIOD + \
                OneOrMore(Group(fieldDecl))("fields")

It prints out:

SALESMEN-RECORD
   SALESMEN-NO ['999']
   SALESMEN-NAME ['XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX']
   SALESMEN-TERRITORY ['XXXXXXXXXXXXXXXXXXXXXXXXXXXXXX']
   SALESMEN-QUOTA ['S', '9999999', 'COMP']
   SALESMEN-1ST-BONUS ['S', '99999', 'V', '99', 'COMP']
   SALESMEN-2ND-BONUS ['S', '99999', 'V', '99', 'COMP']
   SALESMEN-3RD-BONUS ['S', '99999', 'V', '99', 'COMP']
   SALESMEN-4TH-BONUS ['S', '99999', 'V', '99', 'COMP']

I too have some dim, dark, memories of COBOL.  I seem to recall having
to infer from the number of digits in an integer or real what size the
number would be.  I don't have that logic implemented, but here is an
extension to the above program, which shows you where you could put
this kind of type inference logic (insert this code before the call to
searchString):

class TypeDefn(object):
    @staticmethod
    def intType(tokens):
        self = TypeDefn()
        self.str = "int(%d)" % (len(tokens.intpart),)
        self.isSigned = bool(tokens.sign)
        return self
    @staticmethod
    def realType(tokens):
        self = TypeDefn()
        self.str = "real(%d.%d)" % (len(tokens.intpart),len
(tokens.realpart))
        self.isSigned = bool(tokens.sign)
        return self
    @staticmethod
    def charType(tokens):
        self = TypeDefn()
        self.str = "char(%d)" % len(tokens)
        self.isSigned = False
        self.isComp = False
        return self
    def __repr__(self):
        return ("+-" if self.isSigned else "") + self.str
intdecl.setParseAction(TypeDefn.intType)
realdecl.setParseAction(TypeDefn.realType)
strdecl.setParseAction(TypeDefn.charType)

This prints:

SALESMEN-RECORD
   SALESMEN-NO [int(3)]
   SALESMEN-NAME [char(1)]
   SALESMEN-TERRITORY [char(1)]
   SALESMEN-QUOTA [+-int(7), 'COMP']
   SALESMEN-1ST-BONUS [+-real(5.2), 'COMP']
   SALESMEN-2ND-BONUS [+-real(5.2), 'COMP']
   SALESMEN-3RD-BONUS [+-real(5.2), 'COMP']
   SALESMEN-4TH-BONUS [+-real(5.2), 'COMP']

You can post more questions about pyparsing on the Discussion tab of
the pyparsing wiki home page.

Best of luck!
-- Paul
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Members online

Forum statistics

Threads
473,821
Messages
2,569,725
Members
45,511
Latest member
Osiris-Team

Latest Threads

Top