Ejemplo n.º 1
0
def switch_indices(backing_index, object_type):
    """
    Switch the default index to point to the backing index, and delete the reindex alias

    Args:
        backing_index (str): The backing index of the reindex alias
        object_type (str): The object type for the index (post, comment, etc)
    """
    conn = get_conn()
    actions = []
    old_backing_indexes = []
    default_alias = get_default_alias_name(object_type)
    global_alias = get_default_alias_name(ALIAS_ALL_INDICES)
    if conn.indices.exists_alias(name=default_alias):
        # Should only be one backing index in normal circumstances
        old_backing_indexes = list(
            conn.indices.get_alias(name=default_alias).keys())
        for index in old_backing_indexes:
            actions.extend([
                {
                    "remove": {
                        "index": index,
                        "alias": default_alias
                    }
                },
                {
                    "remove": {
                        "index": index,
                        "alias": global_alias
                    }
                },
            ])
    actions.extend([
        {
            "add": {
                "index": backing_index,
                "alias": default_alias
            }
        },
        {
            "add": {
                "index": backing_index,
                "alias": global_alias
            }
        },
    ])
    conn.indices.update_aliases({"actions": actions})
    refresh_index(backing_index)
    for index in old_backing_indexes:
        conn.indices.delete(index)

    # Finally, remove the link to the reindexing alias
    conn.indices.delete_alias(name=get_reindexing_alias_name(object_type),
                              index=backing_index)
Ejemplo n.º 2
0
def find_similar_resources(*, user, value_doc):
    """
    Execute a "more like this" query to find learning resources that are similar to the one provided.

     Args:
        user (User): The user executing the search
        value_doc (dict):
            a document representing the data fields we want to search with

    Returns:
        dict: The Elasticsearch response dict
    """
    index = get_default_alias_name(ALIAS_ALL_INDICES)
    search = Search(index=index)
    search = _apply_general_query_filters(search, user)
    search = search.filter(Q("terms", object_type=LEARNING_RESOURCE_TYPES))
    search = search.query(
        MoreLikeThis(
            like={
                "doc": value_doc,
                "fields": list(value_doc.keys())
            },
            fields=SIMILAR_RESOURCE_RELEVANT_FIELDS,
            min_term_freq=settings.OPEN_RESOURCES_MIN_TERM_FREQ,
            min_doc_freq=settings.OPEN_RESOURCES_MIN_DOC_FREQ,
        ))
    response = search.execute()

    return [
        hit.to_dict() for hit in response.hits
        if (hit["id"] != value_doc.get("id", None)
            or hit["object_type"] != value_doc.get("object_type", None))
    ][0:settings.OPEN_DISCUSSIONS_SIMILAR_RESOURCES_COUNT]
Ejemplo n.º 3
0
def find_related_documents(*, user, post_id):
    """
    Execute a "more like this" query to find posts that are related to a specific post

     Args:
        user (User): The user executing the search
        post_id (str): The id of the post that you want to find related posts for

    Returns:
        dict: The Elasticsearch response dict
    """
    index = get_default_alias_name(ALIAS_ALL_INDICES)
    search = Search(index=index)
    search = _apply_general_query_filters(search, user)
    search = search.query(
        MoreLikeThis(
            like={
                "_id": gen_post_id(post_id),
                "_type": GLOBAL_DOC_TYPE
            },
            fields=RELATED_POST_RELEVANT_FIELDS,
            min_term_freq=1,
            min_doc_freq=1,
        ))
    # Limit results to the number indicated in settings
    search = search[0:settings.OPEN_DISCUSSIONS_RELATED_POST_COUNT]
    return search.execute().to_dict()
Ejemplo n.º 4
0
def execute_learn_search(*, user, query):
    """
    Execute a learning resources search based on the query

    Args:
        user (User): The user executing the search. Used to determine filters to enforce permissions.
        query (dict): The Elasticsearch query constructed in the frontend

    Returns:
        dict: The Elasticsearch response dict
    """
    index = get_default_alias_name(ALIAS_ALL_INDICES)
    search = Search(index=index)
    search.update_from_dict(query)
    search = _apply_learning_query_filters(search, user)
    return transform_results(search.execute().to_dict(), user)
def mocked_es(mocker, settings):
    """Mocked ES client objects/functions"""
    index_name = "test"
    settings.ELASTICSEARCH_INDEX = index_name
    conn = mocker.Mock()
    get_conn_patch = mocker.patch("search.indexing_api.get_conn",
                                  autospec=True,
                                  return_value=conn)
    mocker.patch("search.connection.get_conn", autospec=True)
    default_alias = get_default_alias_name(POST_TYPE)
    reindex_alias = get_reindexing_alias_name(POST_TYPE)
    yield SimpleNamespace(
        get_conn=get_conn_patch,
        conn=conn,
        index_name=index_name,
        default_alias=default_alias,
        reindex_alias=reindex_alias,
        active_aliases=[default_alias, reindex_alias],
    )
Ejemplo n.º 6
0
def get_similar_topics(value_doc, num_topics, min_term_freq, min_doc_freq):
    """
    Get a list of similar topics based on text values

    Args:
        value_doc (dict):
            a document representing the data fields we want to search with
        num_topics (int):
            number of topics to return
        min_term_freq (int):
            minimum times a term needs to show up in input
        min_doc_freq (int):
            minimum times a term needs to show up in docs

    Returns:
        list of str:
            list of topic values
    """
    index = get_default_alias_name(ALIAS_ALL_INDICES)
    search = Search(index=index)
    search = search.filter(Q("terms", object_type=[COURSE_TYPE]))
    search = search.query(
        MoreLikeThis(
            like=[{
                "doc": value_doc,
                "fields": list(value_doc.keys())
            }],
            fields=[
                "course_id", "title", "short_description", "full_description"
            ],
            min_term_freq=min_term_freq,
            min_doc_freq=min_doc_freq,
        ))
    search = search.source(includes="topics")

    response = search.execute()

    topics = [topic for hit in response.hits for topic in hit.topics]

    counter = Counter(topics)

    return list(dict(counter.most_common(num_topics)).keys())
def test_switch_indices(mocked_es, mocker, default_exists, object_type):
    """
    switch_indices should atomically remove the old backing index
    for the default alias and replace it with the new one
    """
    refresh_mock = mocker.patch("search.indexing_api.refresh_index",
                                autospec=True)
    conn_mock = mocked_es.conn
    conn_mock.indices.exists_alias.return_value = default_exists
    old_backing_index = "old_backing"
    conn_mock.indices.get_alias.return_value.keys.return_value = [
        old_backing_index
    ]

    backing_index = "backing"
    switch_indices(backing_index, object_type)

    conn_mock.indices.delete_alias.assert_any_call(
        name=get_reindexing_alias_name(object_type), index=backing_index)
    default_alias = get_default_alias_name(object_type)
    all_alias = get_default_alias_name(ALIAS_ALL_INDICES)
    conn_mock.indices.exists_alias.assert_called_once_with(name=default_alias)

    actions = []
    if default_exists:
        actions.extend([
            {
                "remove": {
                    "index": old_backing_index,
                    "alias": default_alias
                }
            },
            {
                "remove": {
                    "index": old_backing_index,
                    "alias": all_alias
                }
            },
        ])
    actions.extend([
        {
            "add": {
                "index": backing_index,
                "alias": default_alias
            }
        },
        {
            "add": {
                "index": backing_index,
                "alias": all_alias
            }
        },
    ])
    conn_mock.indices.update_aliases.assert_called_once_with(
        {"actions": actions})
    refresh_mock.assert_called_once_with(backing_index)
    if default_exists:
        conn_mock.indices.delete.assert_called_once_with(old_backing_index)
    else:
        assert conn_mock.indices.delete.called is False

    conn_mock.indices.delete_alias.assert_called_once_with(
        name=get_reindexing_alias_name(object_type), index=backing_index)