def boolSearch(ands=[], ors=[], nots=[]):
"""Build a simple boolean query.
each word in B{ands} is equiv to +word
each word in B{ors} is equiv to word
each word in B{nots} is equiv to -word
e.g. boolSearch(['spam'], ['eggs'], ['parrot', 'cheese'])
is equiv to +spam eggs -parrot -cheese in Google/Lucene syntax"""
q = BooleanQuery()
for a in ands:
t = Term('text', a)
tq = TermQuery(t)
q.add(tq, True, False)
for a in ors:
t = Term('text', a)
tq = TermQuery(t)
q.add(tq, False, False)
for a in nots:
t = Term('text', a)
tq = TermQuery(t)
q.add(tq, False, True)
return q
def boolSearch(self, field, ands=[], ors=[], nots=[]):
"""Build a simple boolean query.
Each word in C{ands} is equiv to +word
Each word in C{ors} is equiv to word
Each word in C{nots} is equiv to -word
E.g. C{boolSearch(['spam'], ['eggs'], ['parrot', 'cheese'])} is
equiv to C{+spam eggs -parrot -cheese} in Google/Lucene syntax.
"""
q = BooleanQuery()
for a in ands:
t = Term(field, a)
tq = TermQuery(t)
q.add(tq, True, False)
for a in ors:
t = Term(field, a)
tq = TermQuery(t)
q.add(tq, False, False)
for a in nots:
t = Term(field, a)
tq = TermQuery(t)
q.add(tq, False, True)
return q
def phraseSearch(self, field, words):
"Search for a phrase (given as a list of words) in C{field}."
q = PhraseQuery()
for word in words:
t = Term(field, word)
q.add(t)
return q
def phraseSearch(qStr, field='body'):
"""Find all docs containing the phrase C{qStr}."""
parts = qStr.split()
q = PhraseQuery()
for p in parts:
t = Term(field, p)
q.add(t)
return q
def delete(self, **kw):
"Delete the first document containing the specified term. See also L{deleteAll}."
# Not very efficient for bulk deletes
# Use deleteAll for bulk deletes
self._setupSearcher()
if len(kw) != 1:
raise RuntimeError, 'one and only one field for the moment'
field, value = kw.items()[0]
t = Term(field, value)
self.searcher.reader.deleteTerm(t)
def deleteAll(self, **kw):
"Remove all documents containing this field and value."
self.close()
reader = indexsearcher.open(self.name)
if len(kw) != 1:
raise RuntimeError, 'one and only one field for the moment'
field, values = kw.items()[0]
for value in values:
t = Term(field, value)
reader.deleteTerm(t)
# commit the deletes
reader.close()
def _formats_query(main_query, formats):
if formats is None:
return main_query
q = BooleanQuery()
fq = BooleanQuery()
for f in formats:
t = Term("formats", f.lower())
tq = TermQuery(t)
fq.add(tq, required = False, prohibited = False)
q.add(fq, required = True, prohibited = False)
q.add(main_query, required = True, prohibited = False)
return q
def titleSearch(qStr):
t = Term('title', qStr)
q = TermQuery(t)
return q
def termSearch(qStr):
t = Term('text', qStr)
q = TermQuery(t)
return q
def termSearch(qStr):
t = Term('body', qStr)
q = TermQuery(t)
return q
def _phrase_query(field, phrase):
q = BooleanQuery()
for p in phrase.split():
t = Term(field, p.lower())
q.add(TermQuery(t), required = True, prohibited = False)
return q
def termSearch(self, field, term):
"Search for a single C{term} in a C{field}."
t = Term(field, term)
q = TermQuery(t)
return q