def disjunctify(q):
## 'from="david carasso" to=amrit OR to=deep delay>4 AND delay<10 NOT subject=*erik* (rotsky OR rosensteel)'
## =====>
## '(from="david carasso") OR (to=amrit) OR (to=deep) OR (delay>4) OR (delay<10) OR (NOT subject=*erik*) OR ((rotsky OR rosensteel))'
return " OR ".join([
"(%s)" % term for term in tu.tokenize(q) if term not in ['OR', 'AND']
])
def getMostRestrictiveTerm(q, **kwargs):
tokens = tu.tokenize(q)
if len(tokens) < 2:
return q
mostRestrictive = None
smallestCount = 99999999999
for term in tokens:
if term not in ['OR','AND']:
count = termPopularity(term, **kwargs)
log("term: %s count: %s" % (term, count))
if count < smallestCount:
smallestCount = count
mostRestrictive = term
if count <= RARENESS_IS_GOOD_ENOUGH:
log("term %s is rare enough (%s) to use and break early." % (term, count))
break
return mostRestrictive
def getMostRestrictiveTerm(q, **kwargs):
tokens = tu.tokenize(q)
if len(tokens) < 2:
return q
mostRestrictive = None
smallestCount = 99999999999
for term in tokens:
if term not in ['OR', 'AND']:
count = termPopularity(term, **kwargs)
log("term: %s count: %s" % (term, count))
if count < smallestCount:
smallestCount = count
mostRestrictive = term
if count <= RARENESS_IS_GOOD_ENOUGH:
log("term %s is rare enough (%s) to use and break early." %
(term, count))
break
return mostRestrictive
def disjunctify(q):
## 'from="david carasso" to=amrit OR to=deep delay>4 AND delay<10 NOT subject=*erik* (rotsky OR rosensteel)'
## =====>
## '(from="david carasso") OR (to=amrit) OR (to=deep) OR (delay>4) OR (delay<10) OR (NOT subject=*erik*) OR ((rotsky OR rosensteel))'
return " OR ".join(["(%s)" % term for term in tu.tokenize(q) if term not in ['OR','AND']])
