Anna

Anna offers an abstraction layer which helps separate the analytic code from other code.

There are two parts in Anna:

1. Anna.iOS provides a class named Analyzer, which plays like a view model in MVVM. It hooks event callbacks from UIResponder (also its subclasses) then exposes what it receives to Anna.Core.
2. Anna.Core is in JavaScript. It identities every exposed UIResponder with a unique path, which consists of Nodes. When events happen on Nodes, it runs the registered tasks to "dig" what happened on the Nodes (and up along the path).

Finally results are sent back to Anna.iOS, where the results could be uploaded to a remote server or some analytic service providers.

How to Use

Basic Usage

For example, if we want to track did-select event from a cell in a table view controller, we need to register such a task:

/* in MasterViewController.js */

match(
  'master/tableView/cell/did-select',
  (node) => { return { action: 'selected', id: node.path }; }
);

Expose MasterViewController and its table view to Anna.Core:

import Anna

class MasterViewController: UITableViewController, AnalyzableObject {
    lazy var analyzer: Analyzing = { Analyzer.analyzer(with: self) }()
    static let subAnalyzableKeys: Set<String> = [#keyPath(tableView)]
    deinit {
        self.analyzer.detach()
    }
    override func viewDidLoad() {
        super.viewDidLoad()
        self.analyzer.enable(naming: "master")
    }
}

class AnalyzableTableView : UITableView, Analyzable {
    lazy var analyzer: Analyzing = { Analyzer.analyzer(with: self) }()
    deinit { 
        self.analyzer.detach() 
    }
}

Receive the result:

class AppDelegate: UIResponder, UIApplicationDelegate, Delegate {
    func manager(_ manager :Manager, didSend result :Any) {
        print(result)
        //
        // Supposed to output
        // { //     action = selected;
        //     id = "/master/tableView/cell";
        // }
    }
}

Responder Chain & Focus Path

A classic application structure, which contains master & detail view controller navigated within a navigation controller, may look like this:

AppDelegate/
├── Anna.RootAnalyzer/
└── UIWindow/
    └── UINavigationController/
        ├── MasterViewController/
        │   ├── Anna.Analyzer("master")/
        │   └── UITableView/
        │       ├── Anna.Analyzer("table")/
        │       └── TableViewCell/
        │           └── Anna.Analyzer("cell", #)/
        └── DetailViewController/
            ├── Anna.Analyzer("detail")/
            └── UIButton/
                └── Anna.Analyzer("button")/

We can notice that some Nodes in the tree above have a Analyzer as instance member. With the names given to the Analyzers, in Anna.Core, this tree results in another tree which is slightly different:

/
└── master/
    └── table/
        └── cell/
            └── delail/
                └── button/

The major different starts from Node detail which belongs to the DetailViewController. In the Responder Chain provided by UIKit, the next responder (parent in the tree) of DetailViewController, is the UINavigationController. However, in Anna.Core, the parent of detail is cell, which means that user's focus moves from the cell to the detail. This behavior is because, from the aspect of analytics, the useful information for every single view usually belongs to the views that user paid attention on before. Thus, in Anna.Core every path from the root to a Node is actually a Focus Path.

Analyzer

Analyzers are interface for objects to record events on and perform other interactions with Nodes in Anna.Core.

Super-Sub Analyzer & Parent-Child Node

In most cases, an Analyzer has a name to identify a keyed relationship between the Node it binds and the Node's parent. The name is usually given by its super Analyzer. However, sometimes, a Analyzer's super Analyzer doesn't have direct access to it, in this case a standalone name works as well.

An array of sub Analyzers are given indexes respectively to represent indexed relationships.

Owner of a super Analyzer can add sub Analyzers to it by confirming protocol Anna.AnalyzableObject and implementing method subAnalyzableKeys. In this way, sub Analyzers create sub Nodes in focus path.

If an Analyzer is not added by any other Analyzer as sub Analyzer, it looks up the responder chain for a super Analyzer. In the looking up process, Anna.FocusPathConstituting.parentConstitutor() and Anna.FocusPathConstitutionRedirecting.redirectedConstitutor() are called to derive detailed information about how the super-sub relationship is like. UIResponder and most of its subclasses have default implementation for these methods, but they can be override to support custom behavior.

Root Analyzer & Manager

UIViews, UIViewControllers, UIControls and all the other objects in Responder Chain may own a Analyzer if they are supposed to be analyzed, including the root responder - UIApplicaiontDelegate. However, the Analyzer of UIApplicaiontDelegate is slightly different. It is of class RootAnalyzer and initialized with a Manager. Manager acts like port between Anna.iOS and Anna.Core. It receives events from Anna.iOS and returns calculated results into delegate methods.

Manager starts by running the module which the first parameter moduleURL of its initializer points to. Its behavior can be altered via attributes of dependency (the second parameter).

And the results returned by Manager can be received in Manager.delegate. The methods in Manager.delegate are called asynchronously, so if they are not expected to be called on main thread, configure that via Manager.delegateQueue:

Focus Marking

To construct a proper Focus Path, Anna.iOS must know which Analyzer (and the Node it binds) is focused. Anna.iOS derives focus-relative information from Responder Chain and Touch Event. Cases with UIButton and UITableView are automatically handled. However, when touch events are detected by UITapRecognizer, touchesEnded(with event) and in other custom cases, focus marking need to be handled manually. To mark an Analyzer focused, we need to call markFocused() on it.

Methods on Analyzer

There are several kinds of Analyzers which all conform to protocol Analyzing.

Available methods on Analyzing:

Analytics Module

The parameter moduleURL in Manager.init points to entrance (a node module) of our analytic module. It finds where the Anna.Core module, passes in the location of the task module, and returns a configured constructor to Anna.iOS. A classic implementation of an analytic module looks like this:

/* In
 * analytic.bundle/
 * └── index.js
 */
module.exports = require('../anna.bundle').configured({
  task: (__dirname + '/task')
});

Task Registration

The tasks to be registered are in the module where the parameter task points to. The file index.js in the module is guaranteed to be require-ed before any Nodes of focus path created. Other files are require-ed according to the name space of the object create the focus path Node. For example, when DetailViewController creates a focus path Node, DetailViewController.js will be require-ed. Hence all the tasks contained in DetailViewController.js will be registered.

To register a task when a specific event triggered on a Node, simply call

match(
  `focus/path/to/node/event`,
  (node) => { return node.path; }
);

match is a global function injected by Anna.Core before tasks loading process. The first parameter contains two part. The components before the last one identify the kind of Nodes we care about. The last component refers to the event we care about. The digging function (the second parameter) returns the result dug out from the Node.

CoreJS

Inside Anna.Manager there is a tiny NodeJS environment called CoreJS. It is so tiny that only the basic require function (includes cache) is implemented. CoreJS starts by requrie-ing the entrance module, of which the URL is passed in via the first parameter of Manager.init. Anna.Core is just a normal module require-ed in the entrance module, of which the constructor is assigned to the exports and finally returned to Anna.iOS.

Focus Path Node

A Node (in Anna.Core) keeps references to get its context, mainly speaking its parent Node and all other ancestor Nodes, so it is easy to know from where the user move focus to the current Node.

From a more abstract aspect, focus paths represent user's interaction with the application in the past. And past itself is the object of analytics.

Available attributes and functions on Node:

Special Events

Visibility

Anna.Core records every Node's visibility be receiving event appeared and disappeared. These two events are generated by hooking UIKit objects. For example,

• an UIView reports appeared when it has a super view, is in a window and not hidden.
• an UITableViewCell reports appeared the delegate's tableView(_ tableView, willDisplay cell, forRowAt indexPath) is called.
• an UIViewController reports appeared when its viewDidAppeared is called.

Base the visibility, tasks like analyzing data exposure can be done.

Update & Observe

Value changes on a Node can be tracked by calling update(_ value, for keyPath). An series of value changes on an attribute can be distinguished from one another with the help of key path.

If we want to keep tracking an attribute whenever its value changes, we can call observe(_ observee, for keyPath).

Debug

Sometimes it gets frustrated when Anna doesn't behave as expected, even thought everything is believed to be properly configured. In this case, we can call Manager.logSnapshot to log out a marked up text to the Manager.dependency.standardOutput to have all the details of the underlying state. The snapshot contains:

• all the currently registered focus path Node
• the tasks registered on the Nodes
• the most recent ten events happened on the Nodes

An sample of snapshot:

<__root__ id="105553117049152" class="ana-node" createdAt="1530341056867">
  <master id="105827995931584" class="ana-node" createdAt="1530341056870">
    <ana-appeared class="ana-event" time="1530341056875" />
    <ana-disappeared class="ana-event" time="1530341063401" />
    <tableView id="105827995931808" class="ana-node" createdAt="1530341056872">
      <cell id="105553117261952/0" class="ana-node" createdAt="1530341062226" index="0">
        <match>
          <branches length="1">
            <did-select />
          </branches>
        </match>
        <ana-updated key-path="text" value="2018-06-30 06:44:22 +0000" class="ana-event" time="1530341062227" />
        <ana-appeared class="ana-event" time="1530341062227" />
        <did-select class="ana-event" time="1530341062861" />

To call Manager.logSnapshot without inserting extra lines into the code base, pause the application and type in lldb command:

e -l swift -- import MyApp; ((UIApplication.shared.delegate as! AppDelegate).analyzer as! RootAnalyzer).manager.logSnapshot()