michael-balint / sms-todo Goto Github PK

Will need to update or remove from GitHub, committed in a1cbe9d

• add options on reminder to engage user to remind again, complete, or dismiss (e.g. Google Calendar)
• calendar integration
• email integration
• custom NLP / ML engine
• slack integration
• allow others to assign todos to each other

Need to identify the source, but guessing this is a Twilio response, the response should be

1. create a todo, start a text with 'Remind me to'
2. retrieve the list of all your todos, text 'list'
3. edit a todo, text 'Edit N' (where N is the todo number)
4. delete a todo, text 'Delete N' (where N is the todo number)
5. view your settings type 'settings'

• add in condition that ensures only admins can invite new users
• evaluate the vulnerability and openness of the GET call

For future features (need unique mobile pages, linked via text message > monetization):

• secure credit card information (Venmo / Stripe)?
• ordering pages / checkout?

• Ran every minute
• Query and Search all todos in DynamoDB, checking date / time
  • remind a user on a specific date/time
  • remind a user daily up and until that date/time
• send SMS to phone number if match
• delete todos who have an expired date / time

Twilio has documentation using their XML, nothing from Plivo. I sent an email to Plivo's support.

spaces turn to "+" and no capitalization past the first word.

Receiving this error when texting Woodhouse (works locally)

• identify the structure of the todo, current recommendation:
  • keywords
  • relations (actions)
  • taxonomy
  • date (optional)
  • time (optional)
  • qty (optional)
  • importance?
• identify the best way to parse this information (NLP vs tiered questions)
• how to list all todos (requires Query and Search against DynamoDB and indexing)
• figure out how to Edit a todo (using an index of some kind?)
• figure out how to Delete a todo (using an index of some kind?)

NLP API options: (http://www.programmableweb.com/news/how-5-natural-language-processing-apis-stack/analysis/2014/07/28)

• http://www.alchemyapi.com/developers
• http://aylien.com/text-api
• https://www.meaningcloud.com/products
• https://www.textrazor.com/docs/rest

...otherwise random people can send texts with the api with minimal hacking.

For todos with specific due date/times, auto-delete them (CRON job) when they have expired.

• quick description (svg animation)
• signup for waiting list
• counter

Works fine locally (pinging production DynamoDB)

Current weekday: Tuesday

Ex: Monday and Wednesday at 5PM

It'll set Monday and Wednesday for the following week because both weekdays have to be in the future.

Scenarios:

• EVERY: Wednesday's start date should be the next day (not a week + 1 day)
• EVERY OTHER: not sure yet

Prompt the user to see if they want to merge or cancel?

• track geo data to send text message notifications
• check phone data to see if you actually call someone
• check email to see if you actually send an email (may not need an app to accomplish this)
• visualization of todos?

http://www.timeanddate.com/time/zones/

When passing UTC values to DynamoDB, they are being removed and not saved as Elements.

params sent:

{ StartDate: Fri May 20 2016 17:00:00 GMT-0400 (EDT),
     StopDate: Fri May 20 2016 19:00:00 GMT-0400 (EDT),
     Repeat: 'bi-weekly',
     DateCreated: Tue May 17 2016 16:27:14 GMT-0400 (EDT),
     Input: 'Remind me to take my pills every other Friday from 5 to 7pm',
     Phone: '19186918755' }

saved values:

Repeat: 'bi-weekly',
Input: 'some text',
Phone: '19186918755'

Saving the date as ISO works.

• multiple sentences? (validated by nlp.text(TEXT).sentences.length)
• repeat back the todo + reminder format

Date / Time

• make sure dates are always in the future (AND, TO, THROUGH, OR)
• multiple days / reminder times (chrono breaks it into multiple objects)
  • Monday AND Friday or Monday, Wednesday, AND Friday
  • Wednesday OR Friday
  • differentiate between AND and OR
    • OR is a much rarer use case (may revisit in next sprint)
• differentiate between prepositions and determiners
  • ON / THIS (specific day)
  • BY (daily reminder until that day)
  • EVERY (should also apply to multiple days / reminder times)
  • EVERY OTHER (maybe wait for this one?)
  • NEXT (the week after the current one) >> temporarily resolved
    • chrono and nlp_compromise references the current week (Mon-Sun)
      • if on Sun and "next Thursday", identifies the first coming up Thursday
      • if on Mon and "next Thursday", idenifies the next week's Thursday
    • possibly ask the user which week they are referring to
• ranges
  • TO (e.g. Friday from 5 to 7PM)
  • THROUGH/THRU (e.g. Friday from 5 through 7PM)
  • TO (e.g. Tuesday to Thursday)
  • THROUGH/THRU (e.g. Tuesday through Thursday)
• other conditions
  • in 30 minutes
  • in an hour
  • in a day
  • in a week
  • next week/month
  • in a couple/few days
  • later today
  • every QUALIFIER (e.g. third) UNIT_OF_TIME
• EVERY condition
  • day
  • hour
  • minutes
  • second
• can't handle phrases like "this weekend"
• nlp.text(TEXT).date() is currently broken, the returned values in the documentation doesn't accurately detect the date or multiple dates
• can't handle multiple subjects (nouns) and date/times (e.g. pick up the groceries on Friday and call the babysitter on Monday) > leads to higher order question of breaking up a todo into segments

Quantity

• no idea how to handle at this point
  • tags don't identify a quantity
  • separate quantity from date

currently using https://github.com/wanasit/chrono and https://github.com/nlp-compromise/nlp_compromise

Setup an archival DB table for all todos to perform analytics and support future NLP / ML work.