Matching bug

[Paikan]

# -*- coding: utf-8 -*-


from noaho import NoAho


a = NoAho()
a.add('an', 'an')
a.add('canal', 'canal')
a.add('e can oilfield', 'e can oilfield')

for match in a.findall_long('one canal'):
    print(match)

# matches with:
# (5, 7, 'an')
#
# expected match would be:
# (4, 9, 'canal')

[JDonner]

Ok. My making a 'long' version of Aho Corasick was kind of a hack, which
seemed to work - I wonder if these two bugs don't expose a design flaw.
I'll look at it. According to PyPi ~300 people a month download this,
yet I've gotten only your reports. So, many thanks!

On Wed, Feb 18, 2015, at 02:04 AM, Benjamin Devèze wrote:

-- coding: utf-8 --

from noaho import NoAho

a = NoAho()

a.add('an', 'an')

b.add('canal', 'canal')

c.add('e can oilfield', 'e can oilfield')

for match in a.findall_long('one canal'):

print(match)

matches with:

(5, 7, 'an')

expected match would be:

(4, 9, 'canal')

— Reply to this email directly or view it on GitHub[1].

Links:

1. Matching bug #3

[JDonner]

Ok, fixed, 0.9.6 is on github and PyPi. Thanks for the report!