e000/marisa_autocomplete.py

## marisa_autocomplete.py
import marisa, flask, time

app = flask.Flask(__name__)

trie = marisa.Trie()
agent = marisa.Agent()
NUM_RESULTS = 30
DICT_FILE = "/usr/share/dict/words"

_index_html = """
<html>
<head>
    <title>Marisa Predictive Search Demo + AngularJS</title>
    <script src="//cdnjs.cloudflare.com/ajax/libs/angular.js/1.1.1/angular.min.js"></script>
    <script>
        function SearchCtrl($scope, $http) {
            $scope.doAutoComplete = function() {
                if($scope.query)
                    $http.get("/s/" + escape($scope.query = $scope.query.toLowerCase())).then(function(r) {
                        $scope.results = r.data.results, $scope.ms_p = r.data.ms / 1000;
                    });
                else
                    $scope.results = [], $scope.ms_p = 0;
            }
            $http.get("/meta").then(function(r) {
                $scope.meta = r.data;
            });
        }
    </script>
</head>

<body ng-app="">
    <div ng-controller="SearchCtrl">
        <h1>Dict Autocomplete w/ Marisa Demo</h1>

        <input type="text" ng-change="doAutoComplete()" ng-model="query" placeholder="query">

        <ul>
            <li ng-repeat="result in results">
                <b>{{ query }}</b>{{ result.substr(query.length) }}
            </li>
        </ul>
        <i ng-show="ms_p"><b>{{ms_p}}</b> ms in flask</i> <br />
        <hr>
        <i>{{meta}}</i>
    </div>
</body>
</html>
"""

@app.route("/s/<query>")
def search(query):
    t = long(time.time() * 1000000)
    agent.set_query(str(query))
    results = []
    i = NUM_RESULTS
    while i and trie.predictive_search(agent):
        i -= 1
        results.append(agent.key_str())

    return flask.jsonify(results = results, ms = long(time.time() * 1000000) - t)

@app.route("/meta")
def meta():
    return flask.jsonify(num_keys=trie.num_keys(), num_tries=trie.num_tries(),
                         num_nodes=trie.num_nodes(), io_size=trie.io_size())

@app.route("/")
def index():
    return _index_html


def build_trie():
    s = time.time()
    ks = marisa.Keyset()
    with open(DICT_FILE) as fp:
        for l in fp:
            ks.push_back(l.lower().strip())

    trie.build(ks)
    print "built trie in", time.time() - s, "seconds."
    print dict(num_keys=trie.num_keys(), num_tries=trie.num_tries(),
               num_nodes=trie.num_nodes(), io_size=trie.io_size())

build_trie()

if __name__ == '__main__':
     app.run()
	import marisa, flask, time

	app = flask.Flask(__name__)

	trie = marisa.Trie()
	agent = marisa.Agent()
	NUM_RESULTS = 30
	DICT_FILE = "/usr/share/dict/words"

	_index_html = """
	<html>
	<head>
	<title>Marisa Predictive Search Demo + AngularJS</title>
	<script src="//cdnjs.cloudflare.com/ajax/libs/angular.js/1.1.1/angular.min.js"></script>
	<script>
	function SearchCtrl($scope, $http) {
	$scope.doAutoComplete = function() {
	if($scope.query)
	$http.get("/s/" + escape($scope.query = $scope.query.toLowerCase())).then(function(r) {
	$scope.results = r.data.results, $scope.ms_p = r.data.ms / 1000;
	});
	else
	$scope.results = [], $scope.ms_p = 0;
	}
	$http.get("/meta").then(function(r) {
	$scope.meta = r.data;
	});
	}
	</script>
	</head>

	<body ng-app="">
	<div ng-controller="SearchCtrl">
	<h1>Dict Autocomplete w/ Marisa Demo</h1>

	<input type="text" ng-change="doAutoComplete()" ng-model="query" placeholder="query">

	<ul>
	<li ng-repeat="result in results">
	<b>{{ query }}</b>{{ result.substr(query.length) }}
	</li>
	</ul>
	<i ng-show="ms_p"><b>{{ms_p}}</b> ms in flask</i> <br />
	<hr>
	<i>{{meta}}</i>
	</div>
	</body>
	</html>
	"""

	@app.route("/s/<query>")
	def search(query):
	t = long(time.time() * 1000000)
	agent.set_query(str(query))
	results = []
	i = NUM_RESULTS
	while i and trie.predictive_search(agent):
	i -= 1
	results.append(agent.key_str())

	return flask.jsonify(results = results, ms = long(time.time() * 1000000) - t)

	@app.route("/meta")
	def meta():
	return flask.jsonify(num_keys=trie.num_keys(), num_tries=trie.num_tries(),
	num_nodes=trie.num_nodes(), io_size=trie.io_size())

	@app.route("/")
	def index():
	return _index_html


	def build_trie():
	s = time.time()
	ks = marisa.Keyset()
	with open(DICT_FILE) as fp:
	for l in fp:
	ks.push_back(l.lower().strip())

	trie.build(ks)
	print "built trie in", time.time() - s, "seconds."
	print dict(num_keys=trie.num_keys(), num_tries=trie.num_tries(),
	num_nodes=trie.num_nodes(), io_size=trie.io_size())

	build_trie()

	if __name__ == '__main__':
	app.run()