talks_tweeter

Table of Contents

• talks_tweeter
  • Configuring
  • Building and Running
  • API
    • Architecture
    • Talks Importer
    • Talks Database
    • Scheduling Publishing Talks
    • Publisher
  • References

Configuring

Create the config/sys.config file with the following content:

[
  %% talks_tweeter options are optional, you have only to provide 'etweet'
  %% options
  {talks_tweeter, [
    {action_interval, {0,1,0}}, %% check every 1 minute for talk
    {time_window,     {0,5,0}}, %% time period which will be used for
                                %% querying for talks. 'talks_tweeter' will
                                %% check for talks within
                                %% (now + before_talk, now + before_talk + time_window) minutes
    {schedule_start_time, {{2016,2,18}, {6,0,0}}}, %% start time from which
                                                   %% all talks has begin
    {schedule_end_time, {{2016,2,19}, {18,0,0}}}, %% end time for all talks
    {before_talk, {0,15,0}} %% tweet 15 minutes before talk
  ]},
  {etweet, [
           {screen_name, "TO_FILL"},
           {consumer_key, "TO_FILL"},
           {consumer_secret, "TO_FILL"},
           {access_token, "TO_FILL"},
           {access_token_secret, "TO_FILL"}
          ]
  }].

Building and Running

You need to have Erlang 18.3

make compile
make test
make shell

The config/sys.config file has to be filled with appropriate values.

Running as a release

Ensure you have valid config/sys.config file

make rel
cd _build/default/rel/talks_tweeter
. /bin/talks_tweeter start #run application
. /bin/talks_tweeter attach #attach to console

In console run:

1> talks_tweeter_app:run().

and exit by pressing Ctrl + D.

Architecture

API

Talks Importer

tt_importer:start_link() -> Result.

tt_importer:import_file(Filename) -> ok.
tt_importer:import_file("priv/talks") -> ok.

tt_importer:import_csv_file(Filename, StartDate) -> ok | {error ,file_not_exists}.
tt_importer:import_csv_file("talks.csv", {2016,02,04}).

The tt_importer is responsible for importing talks from file. The tt_importer:import_file/1 function takes a file name as an argument and imports its content into the database using its API (tt_store:add/4).

The file with the talks should have the following format:

[
  {"School of Erlang", {{2015,12,15}, {10,00,00}}, {{2015,12,15}, {11,00,00}}, "ESL Office"},
  {"School of Elixir", {{2015,12,15}, {11,00,00}}, {{2015,12,15}, {12,00,00}}, "ESL Office"}
]

The tt_importer:import_csv_file/2 functions takes two arguments: a CSV file and a date that talks stored in the file start on. An example file content is:

"Keynote: Haskell",room 3,2,185
I love teacher?,room 1,1,330
"Bad news: teacher?",room 3,2,240
"Joe, do you know Cooking",room 3,1,360

The format is: Title, Location, Day No, Start Time in Minutes from 7:00

There're 3 locations which map as follows:

room 1 -> Aula średnia A
room 2 -> Aula średnia B
room 3 -> Aula mała

Talks Database

tt_store:start_link() -> Result.

tt_store:add(Title, StartTime, EndTime, Where) -> ok.
tt_store:add("School of Erlang", {{2015,12,17}, {17,00,00}}, {{2015,12,17}, {19,00,00}}) -> ok.

tt_store:find_by_time(StartTime, EndTime) -> 
tt_store:find_by_time({{2015,12,17}, {17,00,00}}, {{2015,12,17}, {20,00,00}}) -> [{TalkTitle, StartTime, EndTime, Where}]

tt_store:list() -> [{TalkTitle, StartTime, EndTime, Where}]

The tt_store module is responsible for storing talks in ETS tables. The tt_store:add/3 function is for adding new talks to the database. tt_store:find_by_time/2 is for getting a list of talks that are planned to start in a given time range. The tt_store:list/0 returns the list of all the talks in the store.

Scheduling Publishing Talks

tt_scheduler:start_link() -> Result.

tt_scheduler:schedule(StartTime, EndTime, PublishInterval).
tt_scheduler:schedule({{2015, 12, 17}, {15, 00, 00}}, {{2015, 12, 17}, {20, 00, 00}}, {01,00,00}) -> Ref.

tt_scheduler:cancel_schedule(Ref) -> ok.

The tt_scheduler is intended for setting up scheduling of publishing information on talks. For example calling tt_scheduler:schedule({{2015, 12, 17}, {15, 00, 00}}, {{2015, 12, 17}, {20, 00, 00}}, {00,20,00}). starts a process that will check every hour what are the planned talks for the next hour and pass the result to the tt_publisher.

Publisher

tt_publisher:start_link() -> Result.

tt_publisher:publish([{Title, StartTime, EndTime, Where}], TimeInterval) -> ok
tt_publisher:publish([
                      {"School of Erlang", {{2015,12,17}, {17,00,00}}, {{2015,12,17}, {17,30,00}}, "ESL Office"}
                      {"School of Elixir", {{2015,12,17}, {18,00,00}}, {{2015,12,17}, {18,30,00}}, "ESL Office"},
                      {01,00,00}
                     ) -> ok.

tt_publisher:publish/1 publishes the talks on Twitter. The example invocation above result in something like following in the Twitter account:

The plan for the following hour is:

• 17:00 - 17:30 "School of Erlang", ESL Office
• 17:30 - 18:00 "School of Elixir", ESL Office

References

• rebar3 for building
• Erlang Applications
• Erlang Supervisors
• Erlang Generic Servers