Development notes and TODOs



WHAT'S THIS:
------------

        . With the increasing number of high capacity mass storages now
        accessible to computers or home appliances (TVs, set top boxes…),
        either directly (physically connected peripherals over USB…) or
        remotely (Cloud, Web, RSS, FTP, DLNA…), users face the problem of
        easy and rapid content browsing and sorting. The rule based browsing
        enables users to filter and organize his/her multimedia or any type
        of content easily and overall to see what he/she wants, where he/she
        expects to see it, independently of the content sources physical
        organization and location. In short, EzBrowser is a javascript rules
        based browsing solution. It enables the user to virtually organize
        content on a computer.

        . The client browser interacts with the server to create filters.
        A filter is a virtual folder, which will retain content based on
        javascript rules defined by the user (either manually or via a GUI)
        in the browser. Each filter, when created, is associated one
        or several plugins, responsible for extracting metadata from specific
        content (mp3 files, video files, etc). A filter is watching content
        from a physical location, and can optionnally scan recursively children
        directories. Everytime a new content (added or modified content) is
        detected, the content metadata is extracted by the associated plugin(s),
        and matched against the associated javascript rules (eg: keep only
        content which type is "mp3" and which "genre" is "rock").



HOW IT WORKS:
-------------

        . The solution is based on a client server architecture. The content
        browser is the client. The java based rules indexer is the server.
        The browser interacts with the server via a TCP/IP connection to
        manipulate and browse 'filters'. A filter can be considered as a
        'virtual folder'; It watches a location (url), and uses user defined
        rules to retain content. The filter relies on plugins to extract and
        apply rules to the content (accessible via the filter's url).

        . When creating a filter, the user selects the plugins associated with
        the filter (mp3, web, rss, etc) depending on the type of content he's
        interested in. The user will then, for each plugin associated with the
        filter, specify the rules to be applied to the metadata (eg: the mp3
        tag 'genre' must be equal to 'rock' or 'jazz). Every time new content
        is detected by the filter, the latter will request the plugins to
        extract  the meta data from the new content and to apply its rules to
        it. The plugin will either 'accept' or 'reject' the content based on
        the result of this operation. If any of the plugins associated with
        the filter accepts the content, the filter will retain it.

        . Filters can be encapsulated in a hierarchical manner, as would
        physical  directories be encapsulated. A child filter will only
        consider content which was retained by its parent filter.

        . When browsing a filters tree, the user will experience the same
        navigation paradigm as he would find in a traditional files browser,
        with the difference that the content is organized based on the rules
        he gave when creating the filters, rather than on the files physical
        location and naming. A filter tree, aka a 'Filter Set' can be saved
        (filters, associated plugins, rules, and indexing results) and reloaded
        very quickly. This enables the user to switch rapidly between two
        different virtual organization of his content.

        . The indexation results are saved in a database. The server uses the
        database to quickly retrieve the content (and its metadata) associated
        with a filter and pass it over to the client browser to show the user
        the filters tree.


WARNING:
--------

        . Data exchanged between the client and the server is NOT encrypted,
        including the user and password.

        . the wait_timeout option must be set when starting mysqld, else,
        the server will apparently stop reporting file modifications after
        several hours without file modifications on the watched directories,
        because the DB server will have dropped the connection.



DEV NOTES:
----------

        . Dependencies:

        Build Order:

                        PluginInterface
                                FilePlugin
                                        Mp3Plugin

                        ServerDatabase, ClientServerInterface, PluginInterface,
                         QFileExtensions
                                Server
                                Client

        Dev Tools/SDK: Qt Creator 2.0.1, Based on Qt 4.8.1 (64 bit)
        Database Mgmt: MySQL server (mysql-server 5.1),
         MySQL QT Driver (libqt4-sql-mysql 4.8.1)

        MP3 support: relies on libid3tag

        . Using mysql-admin, phpMyAdmin or CreateDatabase.sh (easier), create
         a "SIONDatabase" database, and add it a user/pwd SION/SION.

        . Copy QT's libqsqlmysql.so into the Server executable's directory

        . Create a SYMLINK to libFilePlugin.so.1.0.0, rename it "FilePlugin.so"
         and place into the Server executable's directory

        . Create a SYMLINK to libMp3Plugin.so.1.0.0, rename it "Mp3Plugin.so"
         and place into the Server executable's directory

        . Create a SYMLINK to libImdbPlugin.so.1.0.0, rename it "ImdbPlugin.so"
         and place into the Server executable's directory

        . Create a SYMLINK to libFacePlugin.so.1.0.0, rename it "FacePlugin.so"
         and place into the Server executable's directory

TODO:
-----

        . Add Time Out support in SyncHttp to prevent network outage from
         blocking the server indexing for ages

        . Support 'dynamic filters': when a filter is started, it prompts the
         client with the meta-data values. Thus, this type of filter can be
         used to do a search. If used as a child of another filter, only the
         content retained by the parent filter will be examined.

        . Implement a WEB plugin

        . Implement an RSS plugin

        . Implement an IMAP plugin

        . Implement an mpeg4 plugin

        . Create a help file

        . Create a real linux installer

        . Recompile/Test under windows