Regex help needed

[rh0dium]

Hi all,

I am using python to drive another tool using pexpect. The values
which I get back I would like to automatically put into a list if there
is more than one return value. They provide me a way to see that the
data is in set by parenthesising it.

This is all generated as I said using pexpect - Here is how I use it..
child = pexpect.spawn( _buildCadenceExe(), timeout=timeout)
child.sendline("somefunction()")
child.expect("> ")
data=child.before

Given this data can take on several shapes:

Single return value -- THIS IS THE ONE I CAN'T GET TO WORK..
data = 'somefunction()\r\n"@(#)$CDS: icfb.exe version 5.1.0 05/22/2005
23:36 (cicln01) $"\r\n'

Multiple return value
data = 'somefunction()\r\n("." "~"
"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile")\r\n'

It may take up several lines...
data = 'somefunction()\r\n("." "~"
\r\n"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"\r\n"foo")\r\n'

So if you're still reading this I want to parse out data. Here are the
rules...
- Line 1 ALWAYS is the calling function whatever is there (except
"\r\n") should be kept as "original"
- Anything may occur inside the quotations - I don't care what's in
there per se but it must be maintained.
- Parenthesed items I want to be pushed into a list. I haven't run
into a case where you have nested paren's but that not to say it won't
happen...

So here is my code.. Pardon my hack job..

import os,re

def main(data=None):

# Get rid of the annoying \r's
dat=data.split("\r")
data="".join(dat)

# Remove the first line - that is the original call
dat = data.split("\n")
original=dat[0]
del dat[0]

print "Original", original
# Now join all of the remaining lines
retl="".join(dat)

# self.logger.debug("Original = \'%s\'" % original)

try:
# Get rid of the parenthesis
parmatcher = re.compile( r'\(([^()]*)\)' )
parmatch = parmatcher.search(retl)

# Get rid of the first and last quotes
qrmatcher = re.compile( r'\"([^()]*)\"' )
qrmatch = qrmatcher.search(parmatch.group(1))

# Split the items
qmatch=re.compile(r'\"\s+\"')
results = qmatch.split(qrmatch.group(1))
except:
qrmatcher = re.compile( r'\"([^()]*)\"' )
qrmatch = qrmatcher.search(retl)

# Split the items
qmatch=re.compile(r'\"\s+\"')
results = qmatch.split(qrmatch.group(1))

print "Orig", original, "Results", results
return original,results


# General run..
if __name__ == '__main__':


# data = 'someFunction\r\n "test" "foo"\r\n'
# data = 'someFunction\r\n "test foo"\r\n'
data = 'getVersion()\r\n"@(#)$CDS: icfb.exe version 5.1.0
05/22/2005 23:36 (cicln01) $"\r\n'
# data = 'someFunction\r\n ("test" "test1" "foo aasdfasdf"\r\n
"newline" "test2")\r\n'

main(data)

CAN SOMEONE PLEASE CLEAN THIS UP?

 

[Paul McGuire]

Hi all,

I am using python to drive another tool using pexpect. The values
which I get back I would like to automatically put into a list if there
is more than one return value. They provide me a way to see that the
data is in set by parenthesising it.

<snip>

Well, you asked for regex help, but a pyparsing rendition may be easier to
read and maintain.

-- Paul
(Download pyparsing at http://pyparsing.sourceforge.net.)


# test data strings
test1 = """somefunction()
"@(#)$CDS: icfb.exe version 5.1.0 05/22/2005 23:36 (cicln01) $"
"""

test2 = """somefunction()
("." "~"
"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"
"foo")
"""

test3 = """somefunctionWithNestedlist()
("." "~"
"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"
("Hey!"
"this is a nested"
"list")
"foo")
"""

"""
So if you're still reading this I want to parse out data. Here are the
rules...
- Line 1 ALWAYS is the calling function whatever is there (except
"\r\n") should be kept as "original"
- Anything may occur inside the quotations - I don't care what's in
there per se but it must be maintained.
- Parenthesed items I want to be pushed into a list. I haven't run
into a case where you have nested paren's but that not to say it won't
happen...
"""

from pyparsing import Literal, Word, alphas, alphanums, \
dblQuotedString, OneOrMore, Group, Forward

LPAR = Literal("(")
RPAR = Literal(")")

# assume function identifiers must start with alphas, followed by zero or
more
# alphas, numbers, or '_' - expand this defn as needed
ident = Word(alphas,alphanums+"_")

# define a list as one or more quoted strings, inside ()'s - we'll tackle
nesting
# in a minute
quoteList = Group( LPAR.suppress() +
OneOrMore(dblQuotedString) +
RPAR.suppress() )

# define format of a line of data - don't bother with \n's or \r's,
# pyparsing just skips 'em
dataFormat = ident + LPAR + RPAR + ( dblQuotedString | quoteList )

def test(t):
print dataFormat.parseString(t)

print "Parse flat lists"
test(test1)
test(test2)

# modifications for nested lists
quoteList = Forward()
quoteList << Group( LPAR.suppress() +
OneOrMore(dblQuotedString | quoteList) +
RPAR.suppress() )
dataFormat = ident + LPAR + RPAR + ( dblQuotedString | quoteList )

print
print "Parse using nested lists"
test(test1)
test(test2)
test(test3)

Parsing results:
Parse flat lists
['somefunction', '(', ')', '"@(#)$CDS: icfb.exe version 5.1.0 05/22/2005
23:36 (cicln01) $"']
['somefunction', '(', ')', ['"."', '"~"',
'"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"', '"foo"']]

Parse using nested lists
['somefunction', '(', ')', '"@(#)$CDS: icfb.exe version 5.1.0 05/22/2005
23:36 (cicln01) $"']
['somefunction', '(', ')', ['"."', '"~"',
'"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"', '"foo"']]
['somefunctionWithNestedlist', '(', ')', ['"."', '"~"',
'"/eda/ic_5.10.41.500.1.18/tools.lnx86/dfII/samples/techfile"', ['"Hey!"',
'"this is a nested"', '"list"'], '"foo"']]

 

[rh0dium]

-- Paul
(Download pyparsing at http://pyparsing.sourceforge.net.)

Done.


Hey this is pretty cool! I have one small problem that I don't know
how to resolve. I want the entire contents (whatever it is) of line 1
to be the ident. Now digging into the code showed a method line,
lineno and LineStart LineEnd. I tried to use all three but it didn't
work for a few reasons ( line = type issues, lineno - I needed the data
and could't get it to work, LineStart/End - I think it matches every
line and I need the scope to line 1 )

So here is my rendition of the code - But this is REALLY slick..

I think the problem is the parens on line one....

def main(data=None):

LPAR = Literal("(")
RPAR = Literal(")")

# assume function identifiers must start with alphas, followed by
zero or more
# alphas, numbers, or '_' - expand this defn as needed
ident = LineStart + LineEnd

# define a list as one or more quoted strings, inside ()'s - we'll
tackle nesting
# in a minute
quoteList = Group( LPAR.suppress() + OneOrMore(dblQuotedString) +
RPAR.suppress())

# define format of a line of data - don't bother with \n's or \r's,

# pyparsing just skips 'em
dataFormat = ident + ( dblQuotedString | quoteList )

return dataFormat.parseString(data)


# General run..
if __name__ == '__main__':


# data = 'someFunction\r\n "test" "foo"\r\n'
# data = 'someFunction\r\n "test foo"\r\n'
data = 'getVersion()\r\n"@(#)$CDS: icfb.exe version 5.1.0
05/22/2005 23:36 (cicln01) $"\r\n'
# data = 'someFunction\r\n ("test" "test1" "foo aasdfasdf"\r\n
"newline" "test2")\r\n'

foo = main(data)

print foo

 

[Paul McGuire]

Done.


Hey this is pretty cool! I have one small problem that I don't know
how to resolve. I want the entire contents (whatever it is) of line 1
to be the ident. Now digging into the code showed a method line,
lineno and LineStart LineEnd. I tried to use all three but it didn't
work for a few reasons ( line = type issues, lineno - I needed the data
and could't get it to work, LineStart/End - I think it matches every
line and I need the scope to line 1 )

So here is my rendition of the code - But this is REALLY slick..

I think the problem is the parens on line one....

def main(data=None):

LPAR = Literal("(")
RPAR = Literal(")")

# assume function identifiers must start with alphas, followed by
zero or more
# alphas, numbers, or '_' - expand this defn as needed
ident = LineStart + LineEnd

# define a list as one or more quoted strings, inside ()'s - we'll
tackle nesting
# in a minute
quoteList = Group( LPAR.suppress() + OneOrMore(dblQuotedString) +
RPAR.suppress())

# define format of a line of data - don't bother with \n's or \r's,

# pyparsing just skips 'em
dataFormat = ident + ( dblQuotedString | quoteList )

return dataFormat.parseString(data)


# General run..
if __name__ == '__main__':


# data = 'someFunction\r\n "test" "foo"\r\n'
# data = 'someFunction\r\n "test foo"\r\n'
data = 'getVersion()\r\n"@(#)$CDS: icfb.exe version 5.1.0
05/22/2005 23:36 (cicln01) $"\r\n'
# data = 'someFunction\r\n ("test" "test1" "foo aasdfasdf"\r\n
"newline" "test2")\r\n'

foo = main(data)

print foo

LineStart() + LineEnd() will only match an empty line.


If you describe in words what you want ident to be, it may be more natural
to translate to pyparsing.

"A word starting with an alpha, followed by zero or more alphas, numbers, or
'_'s, with a trailing pair of parens"

ident = Word(alpha,alphanums+"_") + LPAR + RPAR


If you want the ident all combined into a single token, use:

ident = Combine( Word(alpha,alphanums+"_") + LPAR + RPAR )


LineStart and LineEnd are geared more for line-oriented or
whitespace-sensitive grammars. Your example doesn't really need them, I
don't think.

If you *really* want everything on the first line to be the ident, try this:

ident = Word(alpha,alphanums+"_") + restOfLine
or
ident = Combine( Word(alpha,alphanums+"_") + restOfLine )


Now the next step is to assign field names to the results:

dataFormat = ident.setResultsName("ident") + ( dblQuotedString |
quoteList ).setResultsName("contents")

test = "blah blah test string"

results = dataFormat.parseString(test)
print results.ident, results.contents

I'm glad pyparsing is working out for you! There should be a number of
examples that ship with pyparsing that may give you some more ideas on how
to proceed from here.

-- Paul

 

[Michael Spencer]

Hi all,

I am using python to drive another tool using pexpect. The values
which I get back I would like to automatically put into a list if there
is more than one return value. They provide me a way to see that the
data is in set by parenthesising it.
....


CAN SOMEONE PLEASE CLEAN THIS UP?

How about using the Python tokenizer rather than re:
... allowed_tokens = (tokenize.STRING, tokenize.OP)
... src = cStringIO.StringIO(source).readline
... src = tokenize.generate_tokens(src)
... return (token[1] for token in src if token[0] in allowed_tokens)
... ... output = []
... for token in tokens:
... if token == "(":
... output.append(rest_eval(tokens))
... elif token == ")":
... return output
... else:
... output.append(token[1:-1])
... return output
... ... source = source.splitlines()
... original, rest = source[0], "\n".join(source[1:])
... return original, rest_eval(get_tokens(rest))
...

>>> sources = [

... 'someFunction\r\n "test" "foo"\r\n',
... 'someFunction\r\n "test foo"\r\n',
... 'getVersion()\r\n"@(#)$CDS: icfb.exe version 5.1.0 05/22/2005 23:36
(cicln01) $"\r\n',
... 'someFunction\r\n ("test" "test1" "foo aasdfasdf"\r\n "newline"
"test2")\r\n'] ...
('someFunction', ['test', 'foo'])
('someFunction', ['test foo'])
('getVersion()', ['@(#)$CDS: icfb.exe version 5.1.0 05/22/2005 23:36 (cicln01)
$'])
('someFunction', [['test', 'test1', 'foo aasdfasdf', 'newline', 'test2']])
Cheers

Michael

 

[rh0dium]

ident = Combine( Word(alpha,alphanums+"_") + LPAR + RPAR )

This will only work for a word with a parentheses ( ie. somefunction()
)

If you *really* want everything on the first line to be the ident, try this:

ident = Word(alpha,alphanums+"_") + restOfLine
or
ident = Combine( Word(alpha,alphanums+"_") + restOfLine )

This nicely grabs the "\r".. How can I get around it?

Now the next step is to assign field names to the results:

dataFormat = ident.setResultsName("ident") + ( dblQuotedString |
quoteList ).setResultsName("contents")

This is super cool!!

So let's take this for example

test= 'fprintf( outFile "leSetInstSelectable( t )\n" )\r\n ("test"
"test1" "foo aasdfasdf"\r\n "newline" "test2")\r\n'

Now I want the ident to pull out 'fprintf( outFile
"leSetInstSelectable( t )\n" )' so I tried to do this?

ident = Forward()
ident << Group( Word(alphas,alphanums) + LPAR + ZeroOrMore(
dblQuotedString | ident | Word(alphas,alphanums) ) + RPAR)

Borrowing from the example listed previously. But it bombs out cause
it wants a ")" but it has one.. Forward() ROCKS!!

Also how does it know to do this for just the first line? It would
seem that this will work for every line - No?

 

[rh0dium]

... source = source.splitlines()
... original, rest = source[0], "\n".join(source[1:])
... return original, rest_eval(get_tokens(rest))

This is a very clean and elegant way to separate them - Very nice!! I
like this alot - I will definately use this in the future!!

 

[Paul McGuire]

This will only work for a word with a parentheses ( ie. somefunction()
)


This nicely grabs the "\r".. How can I get around it?


This is super cool!!

So let's take this for example

test= 'fprintf( outFile "leSetInstSelectable( t )\n" )\r\n ("test"
"test1" "foo aasdfasdf"\r\n "newline" "test2")\r\n'

Now I want the ident to pull out 'fprintf( outFile
"leSetInstSelectable( t )\n" )' so I tried to do this?

ident = Forward()
ident << Group( Word(alphas,alphanums) + LPAR + ZeroOrMore(
dblQuotedString | ident | Word(alphas,alphanums) ) + RPAR)

Borrowing from the example listed previously. But it bombs out cause
it wants a ")" but it has one.. Forward() ROCKS!!

Also how does it know to do this for just the first line? It would
seem that this will work for every line - No?

This works for me:

test4 = r"""fprintf( outFile "leSetInstSelectable( t )\n" )
("test"
"test1" "foo aasdfasdf"
"newline" "test2")
"""

ident = Forward()
ident << Group( Word(alphas,alphanums) + LPAR + ZeroOrMore(
dblQuotedString | ident | Word(alphas,alphanums) ) + RPAR)
dataFormat = ident + ( dblQuotedString | quoteList )

print dataFormat.parseString(test4)

Prints:
[['fprintf', '(', 'outFile', '"leSetInstSelectable( t )\\n"', ')'],
['"test"', '"test1"', '"foo aasdfasdf"', '"newline"', '"test2"']]


1. Is there supposed to be a real line break in the string
"leSetInstSelectable( t )\n", or just a slash-n at the end? pyparsing
quoted strings do not accept multiline quotes, but they do accept escaped
characters such as "\t" "\n", etc. That is, to pyparsing:

"\n this is a valid \t \n string"

"this is not
a valid string"

Part of the confusion is that your examples include explicit \r\n
characters. I'm assuming this is to reflect what you see when listing out
the Python variable containing the string. (Are you opening a text file
with "rb" to read in binary? Try opening with just "r", and this may
resolve your \r\n problems.)

2. If restOfLine is still giving you \r's at the end, you can redefine
restOfLine to not include them, or to include and suppress them. Or (this
is easier) define a parse action for restOfLine that strips trailing \r's:

def stripTrailingCRs(st,loc,toks):
try:
if toks[0][-1] == '\r':
return toks[0][:-1]
except:
pass

restOfLine.setParseAction( stripTrailingCRs )


3. How does it know to only do it for the first line? Presumably you told
it to do so. pyparsing's parseString method starts at the beginning of the
input string, and matches expressions until it finds a mismatch, or runs out
of expressions to match - even if there is more input string to process,
pyparsing does not continue. To search through the whole file looking for
idents, try using scanString which returns a generator; for each match, the
generator gives a tuple containing:
- tokens - the matched tokens
- start - the start location of the match
- end - the end location of the match

If your input file consists *only* of these constructs, you can also just
expand dataFormat.parseString to OneOrMore(dataFormat).parseString.


-- Paul

 

[Michael Spencer]

def parse(source):

... source = source.splitlines()
... original, rest = source[0], "\n".join(source[1:])
... return original, rest_eval(get_tokens(rest))

This is a very clean and elegant way to separate them - Very nice!! I
like this alot - I will definately use this in the future!!

Cheers

Michael

... src = cStringIO.StringIO(source).readline
... src = tokenize.generate_tokens(src)
... return [token[1][1:-1] for token in src if token[0] == tokenize.STRING]
... ... source = source.splitlines()
... original, rest = source[0], "\n".join(source[1:])
... return original, get_tokens2(rest)
...
This matches your main function for the three tests where main works...

>>> for source in sources[:3]: #matches your main function where it works

... assert parse2(source) == main(source)
...
Original someFunction
Orig someFunction Results ['test', 'foo']
Original someFunction
Orig someFunction Results ['test foo']
Original someFunction
Orig someFunction Results ['test', 'test1', 'foo aasdfasdf', 'newline', 'test2']

....and handles the case where main fails (I think correctly, although I'm not
entirely sure what your desired output is in this case:

>>> parse2(sources[3])

('getVersion()', ['@(#)$CDS: icfb.exe version 5.1.0 05/22/2005 23:36 (cicln01)
$'])
If you really do need nested parens, then you'd need the slightly longer version
I posted earlier

Cheers

Michael