Improving my text processing script

[pruebauno]

I am sure there is a better way of writing this, but how?

import re
f=file('tlst')
tlst=f.read().split('\n')
f.close()
f=file('plst')
sep=re.compile('Identifier "(.*?)"')
plst=[]
for elem in f.read().split('Identifier'):
content='Identifier'+elem
match=sep.search(content)
if match:
plst.append((match.group(1),content))
f.close()
flst=[]
for table in tlst:
for prog,content in plst:
if content.find(table)>0:
flst.append('"%s","%s"'%(prog,table))
flst.sort()
for elem in flst:
print elem

What would be the best way of writing this program. BTW find>0 to check
in case table=='' (empty line) so I do not include everything.

tlst is of the form:

tablename1
tablename2

....

plst is of the form:

Identifier "Program1"
Name "Random Stuff"
Value "tablename2"
....other random properties
Name "More Random Stuff"
Identifier "Program 2"
Name "Yet more stuff"
Value "tablename2"
....
I want to know in what programs are the tables in tlst (and only those)
used.

[Paul McGuire]

Even though you are using re's to try to look for specific substrings
(which you sort of fake in by splitting on "Identifier", and then
prepending "Identifier" to every list element, so that the re will
match...), this program has quite a few holes.

What if the word "Identifier" is inside one of the quoted strings?
What if the actual value is "tablename10"? This will match your
"tablename1" string search, but it is certainly not what you want.
Did you know there are trailing blanks on your table names, which could
prevent any program name from matching?

So here is an alternative approach using, as many have probably
predicted by now if they've spent any time on this list, the pyparsing
module. You may ask, "isn't a parser overkill for this problem?" and
the answer will likely be "probably", but in the case of pyparsing, I'd
answer "probably, but it is so easy, and takes care of so much junk
like dealing with quoted strings and intermixed data, so, who cares if
it's overkill?"

So here is the 20-line pyparsing solution, insert it into your program
after you have read in tlst, and read in the input data using something
like data = file('plst).read(). (The first line strips the whitespace
from the ends of your table names.)

tlist = map(str.rstrip, tlist)

from pyparsing import quotedString,LineStart,LineEnd,removeQuotes
quotedString.setParseAction( removeQuotes )

identLine = (LineStart() + "Identifier" + quotedString +
LineEnd()).setResultsName("identifier")
tableLine = (LineStart() + "Value" + quotedString +
LineEnd()).setResultsName("tableref")

interestingLines = ( identLine | tableLine )
thisprog = ""
for toks,start,end in interestingLines.scanString( data ):
toktype = toks.getName()
if toktype == 'identifier':
thisprog = toks[1]
elif toktype == 'tableref':
thistable = toks[1]
if thistable in tlist:
print '"%s","%s"' % (thisprog, thistable)
else:
print "Not", thisprog, "contains wrong table
("+thistable+")"

This program will print out:
"Program1","tablename2"
"Program 2","tablename2"
Download pyparsing at http://pyparsing.sourceforge.net.

-- Paul

[Miki Tebeka]

Hello pruebauno,

import re
f=file('tlst')
tlst=f.read().split('\n')
f.close() tlst = open("tlst").readlines()
f=file('plst')
sep=re.compile('Identifier "(.*?)"')
plst=[]
for elem in f.read().split('Identifier'):
content='Identifier'+elem
match=sep.search(content)
if match:
plst.append((match.group(1),content))
f.close() Look at re.findall, I think it'll be easier.
flst=[]
for table in tlst:
for prog,content in plst:
if content.find(table)>0: if table in content: flst.append('"%s","%s"'%(prog,table)) flst.sort()
for elem in flst:
print elem

print "\n".join(sorted(flst))

HTH.
--
------------------------------------------------------------------------
Miki Tebeka <mi*********@zoran.com>
http://tebeka.bizhat.com
The only difference between children and adults is the price of the toys

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Cygwin)

iD8DBQFDFrzO8jAdENsUuJsRAk42AJ0Q2CEr8e+1/ZLLhadgxtz879oROACggk24
/2SSAFEgEVbS/SmT6cl17xo=
=OF21
-----END PGP SIGNATURE-----

[pruebauno]

Paul McGuire wrote:

match...), this program has quite a few holes.

What if the word "Identifier" is inside one of the quoted strings?
What if the actual value is "tablename10"? This will match your
"tablename1" string search, but it is certainly not what you want.
Did you know there are trailing blanks on your table names, which could
prevent any program name from matching?
Good point. I did not think about that. I got lucky because none of the
table names had trailing blanks (google groups seems to add those) the
word identifier is not used inside of quoted strings anywhere and I do
not have tablename10, but I do have "dba.tablename1" and that one has
to match with tablename1 (and magically did).

So here is an alternative approach using, as many have probably
predicted by now if they've spent any time on this list, the pyparsing
module. You may ask, "isn't a parser overkill for this problem?" and

You had to plug pyparsing! :-). Thanks for the info I did not know
something like pyparsing existed. Thanks for the code too, because
looking at the module it was not totally obvious to me how to use it. I
tried run it though and it is not working for me. The following code
runs but prints nothing at all:

import pyparsing as prs

f=file('tlst'); tlst=[ln.strip() for ln in f if ln]; f.close()
f=file('plst'); plst=f.read() ; f.close()

prs.quotedString.setParseAction(prs.removeQuotes)

identLine=(prs.LineStart()
+ 'Identifier'
+ prs.quotedString
+ prs.LineEnd()
).setResultsName('prog')

tableLine=(prs.LineStart()
+ 'Value'
+ prs.quotedString
+ prs.LineEnd()
).setResultsName('table')

interestingLines=(identLine | tableLine)

for toks,start,end in interestingLines.scanString(plst):
print toks,start,end

[pruebauno]

Miki Tebeka wrote:

Look at re.findall, I think it'll be easier.

Minor changes aside the interesting thing, as you pointed out, would be
using re.findall. I could not figure out how to.

[pruebauno]

pr*******@latinmail.com wrote:

Paul McGuire wrote:

match...), this program has quite a few holes.

tried run it though and it is not working for me. The following code
runs but prints nothing at all:

import pyparsing as prs

And this is the point where I have to post the real stuff because your
code works with the example i posted and not with the real thing. The
identifier I am interested in is (if I understood the the requirements
correctly) the one after the "title with the stars"

So here is the "real" data for tlst some info replaced with z to
protect privacy:

************************************************** ***************************
Identifier "zzz0main"
************************************************** ***************************
Identifier "zz501"
Value "zzz_CLCL_zzzz,zzzzzz_ID"
Name "zzzzz"
Name "zzzzzz"
************************************************** ***************************
Identifier "zzzz3main"
************************************************** ***************************
Identifier "zzz505"
Value "dba.zzz_CKPY_zzzz_SUM"
Name "xxx_xxx_xxx_DT"
----------------------------------
Value "zzz_zzzz_zzz_zzz"
Name "zzz_zz_zzz"
----------------------------------
Value "zzz_zzz_zzz_HIST"
Name "zzz_zzz"
----------------------------------

[Paul McGuire]

Yes indeed, the real data often has surprising differences from the
simulations! :)

It turns out that pyparsing LineStart()'s are pretty fussy. Usually,
pyparsing is very forgiving about whitespace between expressions, but
it turns out that LineStart *must* be followed by the next expression,
with no leading whitespace.

Fortunately, your syntax is really quite forgiving, in that your
key-value pairs appear to always be an unquoted word (for the key) and
a quoted string (for the value). So you should be able to get this
working just by dropping the LineStart()'s from your expressions, that
is:

identLine=('Identifier'
+ prs.quotedString
+ prs.LineEnd()
).setResultsName('prog')
tableLine=('Value'
+ prs.quotedString
+ prs.LineEnd()
).setResultsName('table')

See if that works any better for you.

-- Paul