Match First Sequence in Regular Expression?

[Roger L. Cauvin]

Say I have some string that begins with an arbitrary sequence of characters
and then alternates repeating the letters 'a' and 'b' any number of times,
e.g.

"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the first, and only the
first, sequence of the letter 'a', and only if the length of the sequence is
exactly 3.

Does such a regular expression exist? If so, any ideas as to what it could
be?

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christoph Conrad]

Hello Roger,

I'm looking for a regular expression that matches the first, and only
the first, sequence of the letter 'a', and only if the length of the
sequence is exactly 3.

import sys, re, os

if __name__=='__main__':

m = re.search('a{3}', 'xyz123aaabbaaabbbbababbbbaabb')
print m.group(0)
print "Preceded by: \"" + m.string[0:m.start(0)] + "\""

Best wishes,
Christoph

[Sybren Stuvel]

Roger L. Cauvin enlightened us with:

I'm looking for a regular expression that matches the first, and
only the first, sequence of the letter 'a', and only if the length
of the sequence is exactly 3.

Your request is ambiguous:

1) You're looking for the first, and only the first, sequence of the
letter 'a'. If the length of this first, and only the first,
sequence of the letter 'a' is not 3, no match is made at all.

2) You're looking for the first, and only the first, sequence of
length 3 of the letter 'a'.

What is it?

Sybren
--
The problem with the world is stupidity. Not saying there should be a
capital punishment for stupidity, but why don't we just take the
safety labels off of everything and let the problem solve itself?
Frank Zappa

[Tim Chase]

> Say I have some string that begins with an arbitrary

sequence of characters and then alternates repeating the
letters 'a' and 'b' any number of times, e.g.
"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the
first, and only the first, sequence of the letter 'a', and
only if the length of the sequence is exactly 3.

Does such a regular expression exist? If so, any ideas as
to what it could be?

I'm not quite sure what your intent here is, as the
resulting find would obviously be "aaa", of length 3.

If you mean that you want to test against a number of
things, and only find items where "aaa" is the first "a" on
the line, you might try something like

import re
listOfStringsToTest = [
'helloworld',
'xyz123aaabbaabababbab',
'cantalopeaaabababa',
'baabbbaaabbbbb',
'xyzaa123aaabbabbabababaa']
r = re.compile("[^a]*(a{3})b+(a+b+)*")
matches = [s for s in listOfStringsToTest if r.match(s)]
print repr(matches)

If you just want the *first* triad of "aaa", you can change
the regexp to

r = re.compile(".*?(a{3})b+(a+b+)*")

With a little more detail as to the gist of the problem,
perhaps a better solution can be found. In particular, are
there items in the listOfStringsToTest that should be found
but aren't with either of the regexps?

-tkc

[Alex Martelli]

Tim Chase <py*********@tim.thechases.com> wrote:
...

I'm not quite sure what your intent here is, as the
resulting find would obviously be "aaa", of length 3.

But that would also match 'aaaa'; I think he wants negative loobehind
and lookahead assertions around the 'aaa' part. But then there's the
spec about matching only if the sequence is the first occurrence of
'a's, so maybe he wants '$[^a]*' instead of the lookbehind (and maybe
parentheses around the 'aaa' to somehow 'match' is specially?).

It's definitely not very clear what exactly the intent is, no...
Alex

[Roger L. Cauvin]

"Sybren Stuvel" <sy*******@YOURthirdtower.com.imagination> wrote in message
news:sl**********************@schuimige.unrealtowe r.org...

Roger L. Cauvin enlightened us with:

I'm looking for a regular expression that matches the first, and
only the first, sequence of the letter 'a', and only if the length
of the sequence is exactly 3.

Your request is ambiguous:

1) You're looking for the first, and only the first, sequence of the
letter 'a'. If the length of this first, and only the first,
sequence of the letter 'a' is not 3, no match is made at all.

2) You're looking for the first, and only the first, sequence of
length 3 of the letter 'a'.

What is it?

The first option describes what I want, with the additional restriction that
the "first sequence of the letter 'a'" is defined as 1 or more consecutive
occurrences of the letter 'a', followed directly by the letter 'b'.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Christoph Conrad" <no****@spamgourmet.com> wrote in message
news:up***********@ID-24456.user.uni-berlin.de...

Hello Roger,

I'm looking for a regular expression that matches the first, and only
the first, sequence of the letter 'a', and only if the length of the
sequence is exactly 3.

import sys, re, os

if __name__=='__main__':

m = re.search('a{3}', 'xyz123aaabbaaabbbbababbbbaabb')
print m.group(0)
print "Preceded by: \"" + m.string[0:m.start(0)] + "\""

The correct pattern should reject the string:

'xyz123aabbaaab'

since the length of the first sequence of the letter 'a' is 2. Yours
accepts it, right?

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Alex Martelli" <al***@mail.comcast.net> wrote in message
news:1h9reyq.z7u4ziv8itblN%al***@mail.comcast.net. ..

Tim Chase <py*********@tim.thechases.com> wrote:
...

I'm not quite sure what your intent here is, as the
resulting find would obviously be "aaa", of length 3.

But that would also match 'aaaa'; I think he wants negative loobehind
and lookahead assertions around the 'aaa' part. But then there's the
spec about matching only if the sequence is the first occurrence of
'a's, so maybe he wants '$[^a]*' instead of the lookbehind (and maybe
parentheses around the 'aaa' to somehow 'match' is specially?).

It's definitely not very clear what exactly the intent is, no...

Sorry for the confusion. The correct pattern should reject all strings
except those in which the first sequence of the letter 'a' that is followed
by the letter 'b' has a length of exactly three.

Hope that's clearer . . . .

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christos Georgiou]

On Thu, 26 Jan 2006 14:09:54 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Say I have some string that begins with an arbitrary sequence of characters
and then alternates repeating the letters 'a' and 'b' any number of times,
e.g.

"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the first, and only the
first, sequence of the letter 'a', and only if the length of the sequence is
exactly 3.

Does such a regular expression exist? If so, any ideas as to what it could
be?

Is this what you mean?

^[^a]*(a{3})(?:[^a].*)?$

This fits your description.
--
TZOTZIOY, I speak England very best.
"Dear Paul,
please stop spamming us."
The Corinthians

[Christoph Conrad]

Hello Roger,

since the length of the first sequence of the letter 'a' is 2. Yours
accepts it, right?

Yes, i misunderstood your requirements. So it must be modified
essentially to that what Tim Chase wrote:

m = re.search('^[^a]*a{3}b', 'xyz123aabbaaab')

Best wishes from germany,
Christoph

[Tim Chase]

> Sorry for the confusion. The correct pattern should reject

all strings except those in which the first sequence of the
letter 'a' that is followed by the letter 'b' has a length of
exactly three.

Ah...a little more clear.

r = re.compile("[^a]*a{3}b+(a+b*)*")
matches = [s for s in listOfStringsToTest if r.match(s)]

or (as you've only got 3 of 'em)

r = re.compile("[^a]*aaab+(a+b*)*")
matches = [s for s in listOfStringsToTest if r.match(s)]

should do the trick. To exposit:

[^a]* a bunch of stuff that's not "a"

a{3} or aaa three letter "a"s

b+ one or more "b"s

(a+b*) any number of "a"s followed optionally by "b"s

Hope this helps,

-tkc

[Alex Martelli]

Tim Chase <py*********@tim.thechases.com> wrote:

Sorry for the confusion. The correct pattern should reject
all strings except those in which the first sequence of the
letter 'a' that is followed by the letter 'b' has a length of
exactly three.

Ah...a little more clear.

r = re.compile("[^a]*a{3}b+(a+b*)*")
matches = [s for s in listOfStringsToTest if r.match(s)]

Unfortunately, the OP's spec is even more complex than this, if we are
to take to the letter what you just quoted; e.g.
aazaaab
SHOULD match, because the sequence 'aaz' (being 'a' NOT followed by the
letter 'b') should not invalidate the match that follows. I don't think
he means the strings contain only a's and b's.

Locating 'the first sequence of a followed by b' is easy, and reasonably
easy to check the sequence is exactly of length 3 (e.g. with a negative
lookbehind) -- but I don't know how to tell a RE to *stop* searching for
more if the check fails.

If a little more than just REs and matching was allowed, it would be
reasonably easy, but I don't know how to fashion a RE r such that
r.match(s) will succeed if and only if s meets those very precise and
complicated specs. That doesn't mean it just can't be done, just that I
can't do it so far. Perhaps the OP can tell us what constrains him to
use r.match ONLY, rather than a little bit of logic around it, so we can
see if we're trying to work in an artificially overconstrained domain?
Alex

[Alex Martelli]

Christoph Conrad <no****@spamgourmet.com> wrote:

Hello Roger,

since the length of the first sequence of the letter 'a' is 2. Yours
accepts it, right?

Yes, i misunderstood your requirements. So it must be modified
essentially to that what Tim Chase wrote:

m = re.search('^[^a]*a{3}b', 'xyz123aabbaaab')

....but that rejects 'aazaaab' which should apparently be accepted.
Alex

[Christoph Conrad]

Hallo Alex,

r = re.compile("[^a]*a{3}b+(a+b*)*") matches = [s for s in
listOfStringsToTest if r.match(s)]

Unfortunately, the OP's spec is even more complex than this, if we are
to take to the letter what you just quoted; e.g. aazaaab SHOULD match,

Then it's again "a{3}b", isn't it?

Freundliche Grüße,
Christoph

[Roger L. Cauvin]

"Tim Chase" <py*********@tim.thechases.com> wrote in message
news:ma***************************************@pyt hon.org...

Sorry for the confusion. The correct pattern should reject
all strings except those in which the first sequence of the
letter 'a' that is followed by the letter 'b' has a length of
exactly three.

Ah...a little more clear.

r = re.compile("[^a]*a{3}b+(a+b*)*")
matches = [s for s in listOfStringsToTest if r.match(s)]

Wow, I like it, but it allows some strings it shouldn't. For example:

"xyz123aabbaaab"

(It skips over the two-letter sequence of 'a' and matches 'bbaaab'.)

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:bo********************************@4ax.com...

On Thu, 26 Jan 2006 14:09:54 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Say I have some string that begins with an arbitrary sequence of
characters
and then alternates repeating the letters 'a' and 'b' any number of times,
e.g.

"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the first, and only the
first, sequence of the letter 'a', and only if the length of the sequence
is
exactly 3.

Does such a regular expression exist? If so, any ideas as to what it
could
be?

Is this what you mean?

^[^a]*(a{3})(?:[^a].*)?$

Close, but the pattern should allow "arbitrary sequence of characters" that
precede the alternating a's and b's to contain the letter 'a'. In other
words, the pattern should accept:

"xayz123aaabbab"

since the 'a' between the 'x' and 'y' is not directly followed by a 'b'.

Your proposed pattern rejects this string.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Tim Chase]

>>r = re.compile("[^a]*a{3}b+(a+b*)*")

matches = [s for s in listOfStringsToTest if r.match(s)]

Wow, I like it, but it allows some strings it shouldn't. For example:

"xyz123aabbaaab"

(It skips over the two-letter sequence of 'a' and matches 'bbaaab'.)

Anchoring it to the beginning/end might solve that:

r = re.compile("^[^a]*a{3}b+(a+b*)*$")

this ensures that no "a"s come before the first 3x"a" and nothing
but "b" and "a" follows it.

-tkc
(who's translating from vim regexps which are just diff. enough
to throw a wrench in works...)

[Peter Hansen]

Roger L. Cauvin wrote:

Sorry for the confusion. The correct pattern should reject all strings
except those in which the first sequence of the letter 'a' that is followed
by the letter 'b' has a length of exactly three.

Hope that's clearer . . . .

Examples are a *really* good way to clarify ambiguous or complex
requirements. In fact, when made executable they're called "test cases"
:-), and supplying a few of those (showing input values and expected
output values) would help, not only to clarify your goals for the
humans, but also to let the proposed solutions easily be tested.

(After all, are you going to just trust that whatever you are handed
here is correctly implemented, and based on a perfect understanding of
your apparently unclear requirements?)

-Peter

[Roger L. Cauvin]

"Tim Chase" <py*********@tim.thechases.com> wrote in message
news:ma***************************************@pyt hon.org...

r = re.compile("[^a]*a{3}b+(a+b*)*")
matches = [s for s in listOfStringsToTest if r.match(s)]

Wow, I like it, but it allows some strings it shouldn't. For example:

"xyz123aabbaaab"

(It skips over the two-letter sequence of 'a' and matches 'bbaaab'.)

Anchoring it to the beginning/end might solve that:

r = re.compile("^[^a]*a{3}b+(a+b*)*$")

this ensures that no "a"s come before the first 3x"a" and nothing but "b"
and "a" follows it.

Anchoring may be the key here, but this pattern rejects

"xayz123aaabab"

which it should accept, since the 'a' between the 'x' and the 'y' is not
directly followed by the letter 'b'.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Peter Hansen" <pe***@engcorp.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

Sorry for the confusion. The correct pattern should reject all strings
except those in which the first sequence of the letter 'a' that is
followed by the letter 'b' has a length of exactly three.

Hope that's clearer . . . .

Examples are a *really* good way to clarify ambiguous or complex
requirements. In fact, when made executable they're called "test cases"
:-), and supplying a few of those (showing input values and expected
output values) would help, not only to clarify your goals for the humans,
but also to let the proposed solutions easily be tested.

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christos Georgiou]

On Thu, 26 Jan 2006 16:26:57 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:bo********************************@4ax.com.. .

On Thu, 26 Jan 2006 14:09:54 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Say I have some string that begins with an arbitrary sequence of
characters
and then alternates repeating the letters 'a' and 'b' any number of times,
e.g.

"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the first, and only the
first, sequence of the letter 'a', and only if the length of the sequence
is
exactly 3.

Does such a regular expression exist? If so, any ideas as to what it
could
be?

Is this what you mean?

^[^a]*(a{3})(?:[^a].*)?$

Close, but the pattern should allow "arbitrary sequence of characters" that
precede the alternating a's and b's to contain the letter 'a'. In other
words, the pattern should accept:

"xayz123aaabbab"

since the 'a' between the 'x' and 'y' is not directly followed by a 'b'.

Your proposed pattern rejects this string.

1.

(a{3})(?:b[ab]*)?$

This finds the first (leftmost) "aaa" either at the end of the string or
followed by 'b' and then arbitrary sequences of 'a' and 'b'.

This will also match "aaaa" (from second position on).

2.

If you insist in only three 'a's and you can add the constraint that:

* let s be the "arbitrary sequence of characters" at the start of your
searched text
* len(s) >= 1 and not s.endswith('a')

then you'll have this reg.ex.

(?<=[^a])(a{3})(?:b[ab]*)?$

3.

If you want to allow for a possible empty "arbitrary sequence of characters"
at the start and you don't mind search speed

^(?:.?*[^a])?(a{3})(?:b[ab]*)?$

This should cover you:

s="xayzbaaa123aaabbab"
r=re.compile(r"^(?:.*?[^a])?(a{3})(?:b[ab]*)?$")
m= r.match(s)
m.group(1) 'aaa' m.start(1) 11 s[11:]

'aaabbab'
--
TZOTZIOY, I speak England very best.
"Dear Paul,
please stop spamming us."
The Corinthians

[Roger L. Cauvin]

"Alex Martelli" <al***@mail.comcast.net> wrote in message
news:1h9rhxf.vlrqwp1l99n66N%al***@mail.comcast.net ...

Tim Chase <py*********@tim.thechases.com> wrote:

> Sorry for the confusion. The correct pattern should reject
> all strings except those in which the first sequence of the
> letter 'a' that is followed by the letter 'b' has a length of
> exactly three.

....
.... If a little more than just REs and matching was allowed, it would be
reasonably easy, but I don't know how to fashion a RE r such that
r.match(s) will succeed if and only if s meets those very precise and
complicated specs. That doesn't mean it just can't be done, just that I
can't do it so far. Perhaps the OP can tell us what constrains him to
use r.match ONLY, rather than a little bit of logic around it, so we can
see if we're trying to work in an artificially overconstrained domain?

Alex, you seem to grasp exactly what the requirements are in this case. I
of course don't *have* to use regular expressions only, but I'm working with
an infrastructure that uses regexps in configuration files so that the code
doesn't have to change to add or change patterns. Before throwing up my
hands and re-architecting, I wanted to see if regexps would handle the job
(they have in every case but one).

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christos Georgiou]

On Thu, 26 Jan 2006 16:41:08 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Applying my last regex to your test cases:

r.match("xyz123aaabbab") <_sre.SRE_Match object at 0x00B47F60> r.match("xyz123aabbaab")
r.match("xayz123aaabab") <_sre.SRE_Match object at 0x00B50020> r.match("xaaayz123abab")
r.match("xaaayz123aaabab") <_sre.SRE_Match object at 0x00B47F60> print r.pattern

^(?:.*?[^a])?(a{3})(?:b[ab]*)?$

You should also remember to check the (match_object).start(1) to verify that
it matches the "aaa" you want.
--
TZOTZIOY, I speak England very best.
"Dear Paul,
please stop spamming us."
The Corinthians

[Dave Hansen]

On Thu, 26 Jan 2006 16:26:57 GMT in comp.lang.python, "Roger L.
Cauvin" <ro***@deadspam.com> wrote:

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:bo********************************@4ax.com.. .

[...]

Is this what you mean?

^[^a]*(a{3})(?:[^a].*)?$

Close, but the pattern should allow "arbitrary sequence of characters" that
precede the alternating a's and b's to contain the letter 'a'. In other
words, the pattern should accept:

"xayz123aaabbab"

since the 'a' between the 'x' and 'y' is not directly followed by a 'b'.

I don't know an RE is the best solution to this problem. If I
understand the problem correctly, building a state machine to solve
this is trivial. The following took about 5 minutes of coding:

---begin included file
# Define our states.
# state[0] is next state if character is 'a'
# state[1] is next state if character is 'b'
# state[2] is next state for any other character

# Accept state means we've found a match
Accept = []
for i in range(3):
Accept.append(Accept)

# Reject state means the string cannot match
Reject = []
for i in range(3):
Reject.append(Reject)

# Remaining states: Start, 'a' found, 'aa', 'aaa', and 'aaaa'
Start = [0,1,2]
a1 = [0,1,2]
a2 = [0,1,2]
a3 = [0,1,2]
a4 = [0,1,2]

# Start: looking for first 'a'
Start[0] = a1
Start[1] = Start
Start[2] = Start

# a1: 1 'a' found so far
a1[0] = a2
a1[1] = Reject
a1[2] = Start

# a2: 'aa' found
a2[0] = a3
a2[1] = Reject
a2[2] = Start

# a3: 'aaa' found
a3[0] = a4
a3[1] = Accept
a3[2] = Start

# a4: four or more 'a' in a row
a4[0] = a4
a4[1] = Reject
a4[2] = Start

def detect(s):
"""
Return 1 if first substring aa*b has exactly 3 a's
Return 0 otherwise
"""
state = Start
for c in s:
if c == 'a':
state = state[0]
elif c == 'b':
state = state[1]
else:
state = state[2]
if state is Accept:
return 1
return 0

print detect("xyza123abc")
print detect("xyzaaa123aabc")
print detect("xyzaa123aaabc")
print detect("xyza123aaaabc")

--- end included file ---

And I'm pretty sure it does what you need, though it's pretty naive.
Note that if '3' isn't a magic number, states a1, a2, a3, and a4 could
be re-implemented as a single state with a counter, but the logic
inside detect gets a little hairier.

I haven't timed it, but it's not doing anything other than simple
comparisons and assignments. It's a little (OK, a lot) more code than
a simple RE, but I know it works.

HTH,
-=Dave

--
Change is inevitable, progress is not.

[Fredrik Lundh]

Roger L. Cauvin wrote:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

$ more test.py

import re

print "got expected"
print "------ --------"

testsuite = (
("xyz123aaabbab", "accept"),
("xyz123aabbaab", "reject"),
("xayz123aaabab", "accept"),
("xaaayz123abab", "reject"),
("xaaayz123aaabab", "accept"),
)

for string, result in testsuite:
m = re.search("aaab", string)
if m:
print "accept",
else:
print "reject",
print result
$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept

</F>

[Roger L. Cauvin]

"Fredrik Lundh" <fr*****@pythonware.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

$ more test.py

import re

print "got expected"
print "------ --------"

testsuite = (
("xyz123aaabbab", "accept"),
("xyz123aabbaab", "reject"),
("xayz123aaabab", "accept"),
("xaaayz123abab", "reject"),
("xaaayz123aaabab", "accept"),
)

for string, result in testsuite:
m = re.search("aaab", string)
if m:
print "accept",
else:
print "reject",
print result
$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept

Thanks, but the second test case I listed contained a typo. It should have
contained a sequence of three of the letter 'a'. The test cases should be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Your pattern fails the second test.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christos Georgiou]

On Thu, 26 Jan 2006 18:01:07 +0100, rumours say that "Fredrik Lundh"
<fr*****@pythonware.com> might have written:

Roger L. Cauvin wrote:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept
$ more test.py

[snip of code] m = re.search("aaab", string) [snip of more code]
$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept

You're right, Fredrik, but we (graciously as a group :) take also notice of
the other requirements that the OP has provided elsewhere and that are not
covered by the simple test that he specified.

The code above works for "aaaab" too, which the OP has already ruled out,
and it doesn't work for "aaa".
--
TZOTZIOY, I speak England very best.
"Dear Paul,
please stop spamming us."
The Corinthians

[Roger L. Cauvin]

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:ie********************************@4ax.com...

On Thu, 26 Jan 2006 16:41:08 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Applying my last regex to your test cases:

r.match("xyz123aaabbab") <_sre.SRE_Match object at 0x00B47F60> r.match("xyz123aabbaab")
r.match("xayz123aaabab") <_sre.SRE_Match object at 0x00B50020> r.match("xaaayz123abab")
r.match("xaaayz123aaabab") <_sre.SRE_Match object at 0x00B47F60> print r.pattern

^(?:.*?[^a])?(a{3})(?:b[ab]*)?$

You should also remember to check the (match_object).start(1) to verify
that
it matches the "aaa" you want.

Thanks, but the second test case I listed contained a typo. It should have
contained a sequence of three of the letter 'a'. The test cases should be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Your pattern fails the second test.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:t9********************************@4ax.com...

On Thu, 26 Jan 2006 16:26:57 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:bo********************************@4ax.com. ..

On Thu, 26 Jan 2006 14:09:54 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:Say I have some string that begins with an arbitrary sequence of
characters
and then alternates repeating the letters 'a' and 'b' any number of
times,
e.g.

"xyz123aaabbaabbbbababbbbaaabb"

I'm looking for a regular expression that matches the first, and only
the
first, sequence of the letter 'a', and only if the length of the
sequence
is
exactly 3.

Does such a regular expression exist? If so, any ideas as to what it
could
be?

Is this what you mean?

^[^a]*(a{3})(?:[^a].*)?$

Close, but the pattern should allow "arbitrary sequence of characters"
that
precede the alternating a's and b's to contain the letter 'a'. In other
words, the pattern should accept:

"xayz123aaabbab"

since the 'a' between the 'x' and 'y' is not directly followed by a 'b'.

Your proposed pattern rejects this string.

1.

(a{3})(?:b[ab]*)?$

This finds the first (leftmost) "aaa" either at the end of the string or
followed by 'b' and then arbitrary sequences of 'a' and 'b'.

This will also match "aaaa" (from second position on).

2.

If you insist in only three 'a's and you can add the constraint that:

* let s be the "arbitrary sequence of characters" at the start of your
searched text
* len(s) >= 1 and not s.endswith('a')

then you'll have this reg.ex.

(?<=[^a])(a{3})(?:b[ab]*)?$

3.

If you want to allow for a possible empty "arbitrary sequence of
characters"
at the start and you don't mind search speed

^(?:.?*[^a])?(a{3})(?:b[ab]*)?$

This should cover you:

s="xayzbaaa123aaabbab"
r=re.compile(r"^(?:.*?[^a])?(a{3})(?:b[ab]*)?$")
m= r.match(s)
m.group(1) 'aaa' m.start(1) 11 s[11:]

'aaabbab'

Thanks for continuing to follow up, Christos. Please see my reply to your
other post (in which you applied the test cases).

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Christos Georgiou]

On Thu, 26 Jan 2006 17:09:18 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Thanks, but the second test case I listed contained a typo. It should have
contained a sequence of three of the letter 'a'. The test cases should be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject

Here I object to either you or your need for a regular expression. You see,
before the "aaa" in your second test case, you have an "arbitrary sequence
of characters", so your requirements are met.
--
TZOTZIOY, I speak England very best.
"Dear Paul,
please stop spamming us."
The Corinthians

[Roger L. Cauvin]

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:cf********************************@4ax.com...

On Thu, 26 Jan 2006 18:01:07 +0100, rumours say that "Fredrik Lundh"
<fr*****@pythonware.com> might have written:

Roger L. Cauvin wrote:

Good suggestion. Here are some "test cases":

"xyz123aaabbab" accept
"xyz123aabbaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

$ more test.py

[snip of code]

m = re.search("aaab", string)

[snip of more code]

$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept

You're right, Fredrik, but we (graciously as a group :) take also notice
of
the other requirements that the OP has provided elsewhere and that are not
covered by the simple test that he specified.

My fault, guys. The second test case should be

"xyz123aabbaaab" reject

instead of

"xyz123aabbaab" reject

Fredrik's pattern fails this test case.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Fredrik Lundh]

Roger L. Cauvin wrote:

$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept

Thanks, but the second test case I listed contained a typo. It should have
contained a sequence of three of the letter 'a'. The test cases should be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Your pattern fails the second test.

$ more test.py

import re

print "got expected"
print "------ --------"

testsuite = (
("xyz123aaabbab", "accept"),
("xyz123aabbaaab", "reject"),
("xayz123aaabab", "accept"),
("xaaayz123abab", "reject"),
("xaaayz123aaabab", "accept"),
)

for string, result in testsuite:
m = re.search("a+b", string)
if m and len(m.group()) == 4:
print "accept",
else:
print "reject",
print result

$ python test.py

got expected
------ --------
accept accept
reject reject
accept accept
reject reject
accept accept

</F>

[Roger L. Cauvin]

"Christos Georgiou" <tz**@sil-tec.gr> wrote in message
news:9m********************************@4ax.com...

On Thu, 26 Jan 2006 17:09:18 GMT, rumours say that "Roger L. Cauvin"
<ro***@deadspam.com> might have written:

Thanks, but the second test case I listed contained a typo. It should
have
contained a sequence of three of the letter 'a'. The test cases should
be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject

Here I object to either you or your need for a regular expression. You
see,
before the "aaa" in your second test case, you have an "arbitrary sequence
of characters", so your requirements are met.

Well, thank you for your efforts so far, Christos.

My purpose is to determine whether it's possible to do this using regular
expressions, since my application is already architected around
configuration files that use regular expressions. It may not be the best
architecture, but I still don't know the answer to my question. Is it
*possible* to fulfill my requirements with regular expressions, even if it's
not the best way to do it?

The requirements are not met by your regular expression, since by definition
the "arbitrary sequence of characters" stops once the sequences of a's and
b's starts.

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Fredrik Lundh" <fr*****@pythonware.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

> $ python test.py
> got expected
> ---------------
> accept accept
> reject reject
> accept accept
> reject reject
> accept accept

Thanks, but the second test case I listed contained a typo. It should
have
contained a sequence of three of the letter 'a'. The test cases should
be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept

Your pattern fails the second test.

$ more test.py

import re

print "got expected"
print "------ --------"

testsuite = (
("xyz123aaabbab", "accept"),
("xyz123aabbaaab", "reject"),
("xayz123aaabab", "accept"),
("xaaayz123abab", "reject"),
("xaaayz123aaabab", "accept"),
)

for string, result in testsuite:
m = re.search("a+b", string)
if m and len(m.group()) == 4:
print "accept",
else:
print "reject",
print result

$ python test.py

got expected
------ --------
accept accept
reject reject
accept accept
reject reject
accept accept

Thanks, but I'm looking for a solution in terms of a regular expression
only. In other words, "accept" means the regular expression matched, and
"reject" means the regular expression did not match. I want to see if I can
fulfill the requirements without additional code (such as checking
"len(m.group())").

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Tim Chase]

The below seems to pass all the tests you threw at it (taking the
modified 2nd test into consideration)

One other test that occurs to me would be

"xyz123aaabbaaabab"

where you have "aaab" in there twice.

-tkc

import re
tests = [
("xyz123aaabbab",True),
("xyz123aabbaaab", False),
("xayz123aaabab",True),
("xaaayz123abab", False),
("xaaayz123aaabab",True)
]

exp = '^([^b]|((?<!a)b))*aaab+[ab]*$'
r = re.compile(exp)
print "Using regexp: %s" % exp
for test,expectedResult in tests:
if r.match(test):
result = True
else:
result = False
if result == expectedResult:
print "%s passed" % test
else:
print "%s failed (expected %s, got %s)" % (
test,expectedResult,result)


--
--

[Michael Spencer]

Roger L. Cauvin wrote:

"Fredrik Lundh" <fr*****@pythonware.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

$ python test.py
got expected
---------------
accept accept
reject reject
accept accept
reject reject
accept accept
Thanks, but the second test case I listed contained a typo. It should
have
contained a sequence of three of the letter 'a'. The test cases should
be:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept
This passes your tests. I haven't closely followed the thread for other
requirements:
pattern = ".*?(?<![a+b])aaab" #look for aaab not preceded by any a+b test(pattern) got expected
------ --------
accept accept
reject reject
accept accept
reject reject
accept accept testsuite = ( ... ("xyz123aaabbab", "accept"),
... ("xyz123aabbaaab", "reject"),
... ("xayz123aaabab", "accept"),
... ("xaaayz123abab", "reject"),
... ("xaaayz123aaabab", "accept"),
... )
def test(pattern): ... print "got expected"
... print "------ --------"
... for string, result in testsuite:
... m = re.match(pattern, string)
... if m:
... print "accept",
... else:
... print "reject",
... print result
...

Michael

[Roger L. Cauvin]

"Michael Spencer" <ma**@telcopartners.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

"xyz123aaabbab" accept
"xyz123aabbaaab" reject
"xayz123aaabab" accept
"xaaayz123abab" reject
"xaaayz123aaabab" accept
This passes your tests. I haven't closely followed the thread for other
requirements:
>>> pattern = ".*?(?<![a+b])aaab" #look for aaab not preceded by any a+b

Very interesting. I think you may have solved the problem. The key seems
to be the "not preceded by" part. I'm unfamiliar with some of the notation.
Can you explain what "[a+b]" and the "(?<!" do?

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Roger L. Cauvin]

"Tim Chase" <py*********@tim.thechases.com> wrote in message
news:ma***************************************@pyt hon.org...

The below seems to pass all the tests you threw at it (taking the modified
2nd test into consideration)

One other test that occurs to me would be

"xyz123aaabbaaabab"

where you have "aaab" in there twice.
Good suggestion.
^([^b]|((?<!a)b))*aaab+[ab]*$

Looks good, although I've been unable to find a good explanation of the
"negative lookbehind" construct "(?<". How does it work?

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Tim Chase]

>> "xyz123aaabbaaabab"

where you have "aaab" in there twice.

Good suggestion.

I assumed that this would be a valid case. If not, the
expression would need tweaking.

^([^b]|((?<!a)b))*aaab+[ab]*$

Looks good, although I've been unable to find a good
explanation of the "negative lookbehind" construct "(?<". How
does it work?

The beginning part of the expression

([^b]|((?<!a)b))*

breaks down as

[^b] anything that isn't a "b"
| or
(...) this other thing

where "this other thing" is

(?<!a)b a "b" as long as it isn't immediately
preceeded by an "a"

The "(?<!...)" construct means that the "..." portion can't come
before the following token in the regexp...in this case, before a
"b".

There's also a "negative lookahead" (rather than "lookbehind")
which prevents items from following. This should be usable in
this scenario as wall and works with the aforementioned tests, using

"^([^a]|(a(?!b)))*aaab+[ab]*$"

which would be "anything that's not an 'a'; or an 'a' as long as
it's not followed by a 'b'"

The gospel is at:
http://docs.python.org/lib/re-syntax.html

but is a bit terse. O'reily has a fairly good book on regexps if
you want to dig a bit deeper.

-tkc

[Peter Hansen]

Roger L. Cauvin wrote:

"Michael Spencer" <ma**@telcopartners.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

>"xyz123aaabbab" accept
>"xyz123aabbaaab" reject
>"xayz123aaabab" accept
>"xaaayz123abab" reject
>"xaaayz123aaabab" accept
>

This passes your tests. I haven't closely followed the thread for other
requirements:

>>> pattern = ".*?(?<![a+b])aaab" #look for aaab not preceded by any a+b

Very interesting. I think you may have solved the problem. The key seems
to be the "not preceded by" part. I'm unfamiliar with some of the notation.
Can you explain what "[a+b]" and the "(?<!" do?

I think you might need to add a test case involving a pattern of aaaab
prior to another aaab. From what I gather (not reading too closely),
you would want this to be rejected. Is that true?

xyz123aaaababaaabab

-Peter

[Roger L. Cauvin]

"Peter Hansen" <pe***@engcorp.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:

"Michael Spencer" <ma**@telcopartners.com> wrote in message
news:ma***************************************@pyt hon.org...

Roger L. Cauvin wrote:
>>"xyz123aaabbab" accept
>>"xyz123aabbaaab" reject
>>"xayz123aaabab" accept
>>"xaaayz123abab" reject
>>"xaaayz123aaabab" accept
>>

This passes your tests. I haven't closely followed the thread for other
requirements:

>>> pattern = ".*?(?<![a+b])aaab" #look for aaab not preceded by any a+b

Very interesting. I think you may have solved the problem. The key
seems to be the "not preceded by" part. I'm unfamiliar with some of the
notation. Can you explain what "[a+b]" and the "(?<!" do?

I think you might need to add a test case involving a pattern of aaaab
prior to another aaab. From what I gather (not reading too closely), you
would want this to be rejected. Is that true?

xyz123aaaababaaabab

Adding that test would be a good idea. You're right; I would want that
string to be rejected, since in that string the first sequence of 'a'
directly preceding a 'b' is of length 4 instead of 3.

Thanks for the solution!

--
Roger L. Cauvin
no**********@cauvin.org (omit the "nospam_" part)
Cauvin, Inc.
Product Management / Market Research
http://www.cauvin-inc.com

[Alex Martelli]

Christoph Conrad <no****@spamgourmet.com> wrote:

Hallo Alex,

r = re.compile("[^a]*a{3}b+(a+b*)*") matches = [s for s in
listOfStringsToTest if r.match(s)]

Unfortunately, the OP's spec is even more complex than this, if we are
to take to the letter what you just quoted; e.g. aazaaab SHOULD match,

Then it's again "a{3}b", isn't it?

Except that this one would also match aazaaaaab, which it shouldn't.
Alex

[Scott David Daniels]

How about:
pattern = re.compile('^([^a]|(a+[^ab]))*aaab')

Which basically says, "precede with arbitrarily many non-a's
or a sequences ending in non-b, then must have 3 as followed by a b."

cases = ["xyz123aaabbab", "xayz123aaabab", "xaaayz123aaabab",
"xyz123aaaababaaabab", "xyz123aabbaaab", "xaaayz123abab"]
[re.search(pattern, case) is not None for case in cases]
[True, True, True, False, False, False]

--Scott David Daniels
sc***********@acm.org

[Armin Steinhoff]

Alex Martelli wrote:

Christoph Conrad <no****@spamgourmet.com> wrote:

Hello Roger,

since the length of the first sequence of the letter 'a' is 2. Yours
accepts it, right?
Yes, i misunderstood your requirements. So it must be modified
essentially to that what Tim Chase wrote:

m = re.search('^[^a]*a{3}b', 'xyz123aabbaaab')

...but that rejects 'aazaaab' which should apparently be accepted.

... and that is OK. That was the request:
I'm looking for a regular expression that matches the first, and only
the first, sequence of the letter 'a', and only if the length of the
sequence is exactly 3.
--Armin


Alex