Torando - Whoosh - MoreLike

tornadowhoosh_morelike.py

import whoosh,os
from whoosh import index
import whoosh.index
import whoosh.fields
import whoosh.qparser
import tornado.ioloop
import tornado.web

class Search(object):
    def __init__(self, indexdir, searchstr=None):
        self.indexdir = indexdir
        self.searchstr = searchstr

    def searcher(self):
        schema = whoosh.fields.Schema(
                path = whoosh.fields.ID(unique=True, stored=True),
                title = whoosh.fields.TEXT(stored=True, phrase=False),
                content = whoosh.fields.TEXT(stored=True),
                tag = whoosh.fields.TEXT(stored=True),
                category = whoosh.fields.TEXT(stored=True))

        if not os.path.exists(self.indexdir):
                os.mkdir(self.indexdir)

        ix = index.create_in(self.indexdir, schema)
        writer = ix.writer()
        writer.add_document(title=u"Welcome", content=u"This is welcome blog!",
                    path=u"/welcome", tag=u"Welcome", category=u"Welcome")
        writer.add_document(title=u"Python Whoosh", content=u"Whoosh search library in pure Python",
                    path=u"/whoosh", tag=u"whoosh", category=u"Search")
        writer.add_document(title=u"Python Tornado", content=u"Tornado Web Server for real-time web apps",
                    path=u"/tornado", tag=u"tornado", category=u"Web Server")
        writer.add_document(title=u"Python Tornado Async", content=u"Tornado Web Server provides async web requests",
                    path=u"/tornadoasync", tag=u"async", category=u"Web Server")
        writer.add_document(title=u"Python Tornado Templates", content=u"Tornado Web Server has template feature",
                    path=u"/tornadotemplates", tag=u"templates", category=u"Web Server")
        writer.add_document(title=u"Python Tornado", content=u"Tornado Web Server is awesome",
                    path=u"/tornado", tag=u"great", category=u"Web Server")

        writer.commit()
        _queryparser = whoosh.qparser.QueryParser('content', schema=schema)
        s = ix.searcher()

class Home(tornado.web.RequestHandler):
    def get(self):
        self.write('It Works!')

class MoreLikeThis(tornado.web.RequestHandler):
    def get(self):
        self.render('morelikethis.html')

    def post(self):
        path = self.get_argument('path')
        print path
        srch = Search('./indexer')
        s = srch.searcher()
        docnum = s.document_number(path=path)
        print docnum
        r = s.more_like(docnum, 'content')
        head = "<h3>Documents like more like %s</h3><br />" %path
        head1 = "<h3>Documents like more like this %s</h3><br />" %path
        hits = ''; subhits = ''
        for hit in r:
                hits += "Title: " + hit["title"] + " Tag: " + hit["tag"] + " Category: " + hit["category"] + "<br />"
                for subhit in hit.more_like_this("content", top=1):
                      subhits += "Title: " + subhit["title"] + " Tag: " + subhit["tag"] + "<br />"
        self.write(head+hits+head1+subhits)


application = tornado.web.Application([
    (r"/",Home ),
    (r"/morelikethis",MoreLikeThis ),
])

if __name__ == "__main__":
    application.listen(7777)
    tornado.ioloop.IOLoop.instance().start()
                                                           

How to use Whoosh in Tornado asynchronously？