def process_hits(self, filter_nb):
last_hits = SearchHit.objects.all()
processed_hits = []
for hit in last_hits:
query = hit.query
# blacklist
if query in config.HISTORY_BLACKLIST:
continue
if hit.nb_results < filter_nb:
SearchHitHistoric(query=hit.query,
nb_results=hit.nb_results,
date=hit.date).save()
hit.delete()
continue
# manual get_or_create
try:
search_query = SearchQuery.objects.get(query=query)
created = False
except SearchQuery.DoesNotExist:
search_query = SearchQuery(query=query)
created = True
# if it's a new one, initialize it
if created:
search_query.phonex = phonex(query)
# clean the query, the '_' char cause bugy clean_query
query = query.replace('_', '')
lems = lemmatize(query.split())
clean_query = [lem for lem in lems if lem]
clean_query = ' '.join(clean_query)
clean_phonex = phonex(clean_query)
search_query.clean_query = clean_query
search_query.clean_phonex = clean_phonex
search_query.nb_total_search = 0
search_query.pondered_search_nb = 0
search_query.nb_recent_search = 0
search_query.nb_results = hit.nb_results
search_query.nb_total_search += 1
search_query.pondered_search_nb += 1
search_query.nb_recent_search += 1
weight = (search_query.pondered_search_nb * config.HISTORY_BETA +
search_query.nb_results * config.HISTORY_GAMMA)
search_query.weight = weight
search_query.save()
# we can now create SearchHitHistoric
SearchHitHistoric(query=hit.query,
nb_results=hit.nb_results,
date=hit.date).save()
hit.delete()
def process_hits(self, filter_nb):
last_hits = SearchHit.objects.all()
processed_hits = []
for hit in last_hits:
query = hit.query
# blacklist
if query in config.HISTORY_BLACKLIST:
continue
if hit.nb_results < filter_nb:
SearchHitHistoric(query=hit.query,
nb_results=hit.nb_results,
date=hit.date).save()
hit.delete()
continue
# manual get_or_create
try:
search_query = SearchQuery.objects.get(query=query)
created = False
except SearchQuery.DoesNotExist:
search_query = SearchQuery(query=query)
created = True
# if it's a new one, initialize it
if created:
search_query.phonex = phonex(query)
# clean the query, the '_' char cause bugy clean_query
query = query.replace('_', '')
lems = lemmatize(query.split())
clean_query = [lem for lem in lems if lem]
clean_query = ' '.join(clean_query)
clean_phonex = phonex(clean_query)
search_query.clean_query = clean_query
search_query.clean_phonex = clean_phonex
search_query.nb_total_search = 0
search_query.pondered_search_nb = 0
search_query.nb_recent_search = 0
search_query.nb_results = hit.nb_results
search_query.nb_total_search += 1
search_query.pondered_search_nb += 1
search_query.nb_recent_search += 1
weight = (search_query.pondered_search_nb * config.HISTORY_BETA +
search_query.nb_results * config.HISTORY_GAMMA)
search_query.weight = weight
search_query.save()
# we can now create SearchHitHistoric
SearchHitHistoric(query=hit.query,
nb_results=hit.nb_results,
date=hit.date).save()
hit.delete()
def highlight(text, words, index = None):
"""
Give the position of words in a text, cleaning everything as sesql does
That can be used to highlight the words, for example
The index will be use to lemmatize, if none, it'll use the default one
"""
if not text:
return []
if index is None:
index = fieldmap.primary
if index is None:
raise ValueError, "Not index given and no primary one"
size = len(text)
letters = set(string.ascii_letters)
# Lemmatize the words
lems = lemmatize(words, index)
# Marshall everything
text = index.marshall(text, use_cleanup = False)
# Now find the lemmatized words inside the text
found = []
foundwords = set()
for i, lem in enumerate(lems):
if not lem:
continue
wordsize = len(lem)
pos = 0
while True:
begin = text.find(lem, pos)
if begin < 0:
break
end = begin + wordsize
# We found something, ensure it's a normal word
if begin and text[begin - 1] in letters:
pos = end
continue
# Now find the end of the word
while end < size and text[end] in letters:
end += 1
found.append((begin, end, i))
foundwords.add(text[begin:end])
pos = end
# Lemmatize all found words
foundwords = list(foundwords)
foundlems = lemmatize(foundwords, index)
foundlems = dict(zip(foundwords, foundlems))
# And now, second pass, ensure lemmatized version of word is word
results = []
for begin, end, i in found:
word = text[begin:end]
lem = foundlems[word]
wanted_lem = lems[i]
if lem == wanted_lem:
results.append((begin, end, i))
return results
