def save(self, **kwargs):
self.last_name = self.last_name.replace(u'\u2019', "'") # I know, but store it like this anyway.
self.slug = slugify(self.name())
first_name_metaphone = dm(self.first_name)
last_name_metaphone = dm(self.last_name)
self.first_name_metaphone = first_name_metaphone[0]
self.first_name_metaphone_alt = first_name_metaphone[1] or ''
self.last_name_metaphone = last_name_metaphone[0]
self.last_name_metaphone_alt = last_name_metaphone[1] or ''
super(Person, self).save(**kwargs)
def save(self, **kwargs):
self.last_name = self.last_name.replace(
u'\u2019', "'") # I know, but store it like this anyway.
self.slug = slugify(self.name())
first_name_metaphone = dm(self.first_name)
last_name_metaphone = dm(self.last_name)
self.first_name_metaphone = first_name_metaphone[0]
self.first_name_metaphone_alt = first_name_metaphone[1] or ''
self.last_name_metaphone = last_name_metaphone[0]
self.last_name_metaphone_alt = last_name_metaphone[1] or ''
super(Person, self).save(**kwargs)
def compare(s1f, s1l, s2f, s2l): s1f = unicode(s1f.lower()) s1l = unicode(s1l.lower()) s2f = unicode(s2f.lower()) s2l = unicode(s2l.lower()) s1 = "%s %s" % (s1f, s1l) s2 = "%s %s" % (s2f, s2l) soundex1 = soundex(s1) soundex2 = soundex(s2) soundex1f = soundex(s1f) soundex2f = soundex(s2f) soundex1l = soundex(s1l) soundex2l = soundex(s2l) soundexMatch = (soundex1 == soundex2) soundexFMatch = (soundex1f == soundex2f) soundexLMatch = (soundex1l == soundex2l) dm1 = dm(s1) dm2 = dm(s2) dm1f = dm(s1f) dm2f = dm(s2f) dm1l = dm(s1l) dm2l = dm(s2l) dmMatch = (dm1 == dm2 or dm1[0] == dm2[0] or dm1[0] == dm2[1] or dm1[1] == dm2[0]) dmFMatch = (dm1f == dm2f or dm1[0] == dm2[0] or dm1[0] == dm2[1] or dm1[1] == dm2[0]) dmLMatch = (dm1l == dm2l or dm1[0] == dm2[0] or dm1[0] == dm2[1] or dm1[1] == dm2[0]) jarowN = jarow(s1, s2) jarowNf = jarow(s1f, s2f) jarowNl = jarow(s1l, s2l) damerauN = damerau(s1, s2) damerauNf = damerau(s1f, s2f) damerauNl = damerau(s1l, s2l) qnumN = qnum(s1, s2) qnumNf = qnum(s1f, s2f) qnumNl = qnum(s1l, s2l) print "\n%s entered, %s wanted" % (s2, s1) #print "Soundex\t\tFull:%s/%s %s\tFirst:%s/%s %s\tLast:%s/%s %s" % (soundex1, soundex2, soundexMatch, soundex1f, soundex2f, soundexFMatch, soundex1l, soundex2l, soundexLMatch) print "Metaphone\tFull:%s/%s %s\tFirst:%s/%s %s\tLast:%s/%s %s" % (dm1, dm2, dmMatch, dm1f, dm2f, dmFMatch, dm1l, dm2l, dmLMatch) print "Algorithm\tFull name\tFirst name\tLast name" print "Jarow\t\t%.4f\t\t%.4f\t\t%.4f\nDamerau\t\t%.4f\t\t%.4f\t\t%.4f" % (jarowN, jarowNf, jarowNl, damerauN, damerauNf, damerauNl)
def search_people(search, force_similar=False, use_distance=True):
people = []
sounds_people = 0
names = search.split(None, 3)
if len(names) == 1:
names[0] = names[0].replace(u'\u2019', "'")
if force_similar:
people = Person.objects.exclude(
first_name__icontains=names[0]).exclude(
last_name__icontains=names[0])
else:
people = Person.objects.filter(
Q(first_name__icontains=names[0])
| Q(last_name__icontains=names[0]))
if force_similar:
sounds_people = 2
dm_, dm_alt = dm(names[0])
people = people.filter(
Q(first_name_metaphone=dm_) | Q(last_name_metaphone=dm_))
elif not people and re.match(r'(?i)[a-z\s\'-]+$', names[0]):
sounds_people = 1
dm_, dm_alt = dm(names[0])
people = Person.objects.filter(
Q(first_name_metaphone=dm_) | Q(last_name_metaphone=dm_)
# Q(first_name_metaphone=dm_alt) | #Q(first_name_metaphone_alt=dm_) |
# Q(last_name_metaphone_alt=dm_) | #Q(last_name_metaphone=dm_alt)
)
# if not people:
# allnames = []
# for p in Person.objects.all():
# allnames.extend((p.first_name, p.last_name))
# people = difflib.get_close_matches(names[0], allnames)
# people = Person.objects.filter(Q(first_name__in=people) | Q(last_name__in=people))
if not people and use_distance:
people = []
for p in Person.objects.all():
sim = distance(names[0].lower(), p.first_name.lower())
sim2 = distance(names[0].lower(), p.last_name.lower())
if sim >= threshold or sim2 >= threshold:
people.append((1 - max(sim, sim2), p))
people.sort()
people = [person for _, person in people]
elif len(names) == 2:
names[1] = names[1].replace(u'\u2019', "'")
people = Person.objects.filter(first_name__icontains=names[0],
last_name__icontains=names[1])
if (not people
and re.match(r'(?i)[a-z\s\'-]+$', search)) or force_similar:
sounds_people = 1
dm_first, dm_first_alt = dm(names[0])
dm_last, dm_last_alt = dm(names[1])
qs = Q()
if dm_first:
# Both names homophones
if dm_last:
qs |= Q(first_name_metaphone=dm_first, last_name_metaphone=dm_last) \
| Q(first_name_metaphone=dm_first, last_name_metaphone_alt=dm_last) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone=dm_last) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name_metaphone=dm_first, last_name_metaphone=dm_last_alt) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone=dm_last_alt)
# First name homophone, Last name match
qs |= Q(first_name_metaphone=dm_first, last_name__icontains=names[1]) \
| Q(first_name_metaphone_alt=dm_first, last_name__icontains=names[1])
if dm_first_alt:
qs |= Q(first_name_metaphone=dm_first_alt,
last_name__icontains=names[1])
if dm_last:
qs |= Q(first_name_metaphone=dm_first_alt, last_name_metaphone=dm_last) \
| Q(first_name_metaphone=dm_first_alt, last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name_metaphone=dm_first_alt,
last_name_metaphone=dm_last_alt)
if dm_last:
# First name match, last name homophone
qs |= Q(first_name__icontains=names[0], last_name_metaphone=dm_last) \
| Q(first_name__icontains=names[0], last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name__icontains=names[0],
last_name_metaphone=dm_last_alt)
people = Person.objects.filter(qs)
if not people and use_distance:
people = []
people2 = []
people3 = []
for p in Person.objects.all():
sim = distance(names[0].lower(), p.first_name.lower())
sim2 = distance(names[1].lower(), p.last_name.lower())
simB = distance(' '.join(names).lower(),
('%s %s' %
(p.first_name, p.last_name)).lower())
if names[1].lower() == p.last_name.lower(
) and sim >= threshold:
people.append((1 - sim, p))
elif re.search(names[0], p.first_name,
re.I) and sim2 >= threshold:
people2.append((1 - sim2, p))
elif simB >= threshold:
people3.append((1 - simB, p))
elif sim >= threshold and sim2 >= threshold:
people3.append((1 - max(sim, sim2), p))
people.sort()
people2.sort()
people3.sort()
people = people + people2 + people3
people = [person for _, person in people]
elif len(names) == 3:
names[1] = names[1].replace(u'\u2019', "'")
names[2] = names[2].replace(u'\u2019', "'")
people = Person.objects.filter(
Q(first_name__icontains=' '.join(names[0:2]),
last_name__icontains=names[2])
| Q(first_name__icontains=names[0],
last_name__icontains=' '.join(names[1:3])))
elif len(names) == 4:
names[3] = names[3].replace(u'\u2019', "'")
people = Person.objects.filter(
Q(first_name__icontains=' '.join(names[0:3]),
last_name__icontains=names[3])
| Q(first_name__icontains=names[0],
last_name__icontains=' '.join(names[1:4])))
return people, sounds_people
def search_people(search, force_similar=False, use_distance=True):
people = []
sounds_people = 0
names = search.split(None, 3)
if len(names)==1:
names[0] = names[0].replace(u'\u2019', "'")
if force_similar:
people = Person.objects.exclude(first_name__icontains=names[0]).exclude(last_name__icontains=names[0])
else:
people = Person.objects.filter(Q(first_name__icontains=names[0]) | Q(last_name__icontains=names[0]))
if force_similar:
sounds_people = 2
dm_, dm_alt = dm(names[0])
people = people.filter(
Q(first_name_metaphone=dm_) | Q(last_name_metaphone=dm_)
)
elif not people and re.match('[a-z\s\'-]+$(?i)', names[0]):
sounds_people = 1
dm_, dm_alt = dm(names[0])
people = Person.objects.filter(
Q(first_name_metaphone=dm_) | Q(last_name_metaphone=dm_)
#Q(first_name_metaphone=dm_alt) | #Q(first_name_metaphone_alt=dm_) |
#Q(last_name_metaphone_alt=dm_) | #Q(last_name_metaphone=dm_alt)
)
#if not people:
# allnames = []
# for p in Person.objects.all():
# allnames.extend((p.first_name, p.last_name))
# people = difflib.get_close_matches(names[0], allnames)
# people = Person.objects.filter(Q(first_name__in=people) | Q(last_name__in=people))
if not people and use_distance:
people = []
for p in Person.objects.all():
sim = distance(names[0].lower(), p.first_name.lower())
sim2 = distance(names[0].lower(), p.last_name.lower())
if sim >= threshold or sim2 >= threshold:
people.append((1-max(sim, sim2), p))
people.sort()
people = [ person for _, person in people ]
elif len(names)==2:
names[1] = names[1].replace(u'\u2019', "'")
people = Person.objects.filter(first_name__icontains=names[0], last_name__icontains=names[1])
if (not people and re.match('[a-z\s\'-]+$(?i)', search)) or force_similar:
sounds_people = 1
dm_first, dm_first_alt = dm(names[0])
dm_last, dm_last_alt = dm(names[1])
qs = Q()
if dm_first:
# # Both names homophones
if dm_last:
qs |= Q(first_name_metaphone=dm_first, last_name_metaphone=dm_last) \
| Q(first_name_metaphone=dm_first, last_name_metaphone_alt=dm_last) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone=dm_last) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name_metaphone=dm_first, last_name_metaphone=dm_last_alt) \
| Q(first_name_metaphone_alt=dm_first, last_name_metaphone=dm_last_alt)
# First name homophone, Last name match
qs |= Q(first_name_metaphone=dm_first, last_name__icontains=names[1]) \
| Q(first_name_metaphone_alt=dm_first, last_name__icontains=names[1])
if dm_first_alt:
qs |= Q(first_name_metaphone=dm_first_alt, last_name__icontains=names[1])
if dm_last:
qs |= Q(first_name_metaphone=dm_first_alt, last_name_metaphone=dm_last) \
| Q(first_name_metaphone=dm_first_alt, last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name_metaphone=dm_first_alt, last_name_metaphone=dm_last_alt)
if dm_last:
# First name match, last name homophone
qs |= Q(first_name__icontains=names[0], last_name_metaphone=dm_last) \
| Q(first_name__icontains=names[0], last_name_metaphone_alt=dm_last)
if dm_last_alt:
qs |= Q(first_name__icontains=names[0], last_name_metaphone=dm_last_alt)
people = Person.objects.filter( qs )
if not people and use_distance:
people = []
people2 = []
people3 = []
for p in Person.objects.all():
sim = distance(names[0].lower(), p.first_name.lower())
sim2 = distance(names[1].lower(), p.last_name.lower())
simB = distance(' '.join(names).lower(), ('%s %s' % (p.first_name, p.last_name)).lower())
if names[1].lower() == p.last_name.lower() and sim >= threshold:
people.append((1-sim, p))
elif re.search(names[0], p.first_name, re.I) and sim2 >= threshold:
people2.append((1-sim2, p))
elif simB >= threshold:
people3.append((1-simB, p))
elif sim >= threshold and sim2 >= threshold:
people3.append((1-max(sim, sim2), p))
people.sort()
people2.sort()
people3.sort()
people = people + people2 + people3
people = [ person for _, person in people ]
elif len(names)==3:
names[1] = names[1].replace(u'\u2019', "'")
names[2] = names[2].replace(u'\u2019', "'")
people = Person.objects.filter(
Q(first_name__icontains=' '.join(names[0:2]), last_name__icontains=names[2]) |
Q(first_name__icontains=names[0], last_name__icontains=' '.join(names[1:3]))
)
elif len(names)==4:
names[3] = names[3].replace(u'\u2019', "'")
people = Person.objects.filter(
Q(first_name__icontains=' '.join(names[0:3]), last_name__icontains=names[3]) |
Q(first_name__icontains=names[0], last_name__icontains=' '.join(names[1:4]))
)
return people, sounds_people