Whoosh test

whoosh_test.py

from whoosh import analysis, fields, index, qparser, query, searching, scoring
from whoosh.filedb.filestore import RamStorage
from whoosh.compat import b, u, text_type

def main():
    schema = fields.Schema(name=fields.TEXT(stored=True),
                           hobbies=fields.TEXT(stored=True))
    storage = RamStorage()
    ix = storage.create_index(schema)
    writer = ix.writer()
    writer.add_document(name=u('Frank'), hobbies=u('baseball, basketball'))
    writer.add_document(name=u('Joe'), hobbies=u('track, rugby'))
    writer.add_document(name=u('Adam'), hobbies=u('sportsball, Joe'))
    writer.add_document(name=u('Tom'), hobbies=u('whatever'))
    writer.commit()
    r = ix.reader()
    assert r.field_length("hobbies") == 7
    assert r.field_length("name") == 4
    r.close()

    writer = ix.writer()
    writer.add_document(name=u('Jonny'))
    writer.commit()

    with ix.searcher() as s:
        r = s.reader()
        assert r.field_length("hobbies") == 7
        assert r.field_length("name") == 5

        #################

        parser = qparser.MultifieldParser(['name', 'hobbies'], schema=schema)

        # this works because, by default, this is
        # searching both name and hobbies fields.
        q = parser.parse(u("baseball"))
        result = s.search(q)
        assert len(result) == 1 

        # this searches only the hobbies field
        q = parser.parse(u("hobbies:baseball"))
        result = s.search(q)
        assert len(result) == 1 

        # this searches only the name field
        q = parser.parse(u("name:Joe"))
        result = s.search(q)
        assert len(result) == 1 

        # one hobby, one name
        q = parser.parse(u("Joe"))
        result = s.search(q)
        assert len(result) == 2

        #################

        parser = qparser.QueryParser("name",schema=schema)

        q = parser.parse(u("baseball"))
        result = s.search(q)
        assert len(result) == 0

        # huh. this actually works.
        q = parser.parse(u("hobbies:baseball"))
        result = s.search(q)
        assert len(result) == 1 

        # only one result because only "name" is searched
        q = parser.parse(u("Joe"))
        result = s.search(q)
        assert len(result) == 1

        # here, all free search terms only search the name field
        # (returns none, even though there is a hobbies entry named Joe)
        q = parser.parse(u("hobbies:baseball Joe"))
        result = s.search(q)
        assert len(result) == 0

        q = parser.parse(u("hobbies:rugby Joe"))
        result = s.search(q)
        assert len(result) == 1



if __name__=="__main__":
    main()