Allow person entities (people, groups, organisations) to save a search and be notified when the results change.

• Developer Guidelines
• Save Search Project
• Subscription Project

Overview

In an event of disaster, this functionality would help a lot in terms of providing information (take the example of dissemination of information related to availability of food, shelter, hospitals etc.). The project's essential capability is to disseminate information about various aspects through various means such as email/twitter/sms. Integrating this service into a Sahana Eden system will add to the platforms' capability of tackling various issues faced in emergency/disaster management.

Models

A single model for holding information about the search query and the parameters that are saved by the user. This will be in /modules/eden/pr.py as a component of pr_entity.

name

Name of the saved search as specified by the user.

pe_id

The person entity to be notified when the search results change. Defaults to the pe_id record associated with the current user account.

prefix

The module prefix where the search was performed.

function

The controller function where the search was performed. This is used with the prefix to build requests that go through the same controllers.

resource_name

The module resource that the search is being performed on.

query

The query string for the search.

notification_format

Format for the notification, either a list (default), map, report, graph (this is a report).

notification_method

Subscription mode, e.g., email (default), Twitter, SMS.

notification_frequency

How often the search is checked for changes, e.g., never (default), every 15 minutes (not hourly?), daily, weekly and monthly.

last_executed

The date and time the search was last executed by the subscription method. Set to match modified_on when record is first created.

public

Is this saved search viewable to everyone? This should trigger an authorisation process. Boolean, default is False.

comments

Reusable comments field for the user to write in.

Views

A standard REST controller will be used to manage the saved searches, rather than custom controllers and methods. One custom method, "load", will be used to execute the search (redirect the user to the query URL).

Search

Each search page will have a widget the user can activate to save the current search. They will be directed to a new page where they can type a name for the search.

Saved Search List

Each person entity has a component for saved searches. This page will list all searches saved for this person entity, displaying the name of the search (as defined by the user when it was saved) and the query parameters.

Given the following search query string

organisation.country__belongs=NZ&organisation.organisation_type_id__belongs=2&organisation.sector_id__anyof=1%2C2&parent.acronym%7Cparent.name%7Cacronym%7Cname__like=%2Ared%2A

The user should see

Search=red AND Type=Red Cross / Red Crescent AND Sector=DRR,Food Security AND Home Country=New Zealand

Any field that has no matching representation should be omitted.

The user will have three options for each search record:

• run executes the search
• edit allows the user to edit the saved search record
• delete deletes the saved search record

Saved Search Edit

A page to edit a saved search. The name, pe_id, notification_format, notification_method and notification_frequency can be changed here.

Some fields, like format and method, will be hidden until non-default values are supported.

Process

Saving a search

1. User searches a resource
2. If the search has no errors, a "Save this search" button is shown next to the submit ("Search") button in the search form.
   • The filters, added to the current resource using add_filter(), are extracted from the resource and stored as a pickled string.
   • This button POSTs an s3json request to a save_search REST controller.
   • The s3json contains the prefix, resource_name, and the filters applied to the resource.

Notifications

1. Scheduled tasks are used to fetch the saved searches for a particular frequency, e.g., daily, monthly, etc.
2. The notification function fetches the saved searches with that frequency.
3. An HTTP request is made for each search so that the search results are filtered properly.
4. Records with a modified_on value greater than last_executed are added to the message.
5. Format the response for special media, e.g., Twitter, SMS.
6. The message is sent via the method specified.
7. last_executed field updated to when the request was made.