Dialogs description format

Dialog system

Initialization

Inicialization of the module.

from dialog import Dialog

if __name__ == "__main__":
	DLG = Dialog(globals(), debug=True)
	DLG.load("example.dlg")
	DLG.interprete()

We need to share global scope of program to dialog system. It's needed to call functions and use variables from the dialog.

Dialog description format

.dlg files describes dialog model.

Identation

Identation defines input or output of the dialog system.

# this is a comment
# white-symbol lines ignored

# identation structure
Question
	Answer
		Question
			Answer
			
# few questions at the top level
Question1
	Answer1
Question2
	Answer2	

# both of these Questions triggers Answer
Question1
Question2
	Answer
	
# this question calls two Answers
Question
	Answer1
		...
	Answer2
		...

# Question calls Answer1 and Answer2
# ... is Answer2 child, Answer1 has no childs
Question
	Answer1 
	Answer2
		...

No matter:

# Question calls Answer1 and Answer2
# ... is child of Answer1 and Answer2
Question
	Answer1 &
	Answer2
		...



# lower idenation levels are possible
Question
	Answer
		Question
			Answer
				Question
					Answer
Question
	Answer
		Question
Question
	Answer

Answer isn't obligatory at the last level. Take a look at these identation errors:

# identation error
Question
	Answer
			Question

# valid structure, but AnswerX is Question1 child
Question1
	Answer
		Question
			Answer
#Question2
	AnswerX
		Question

Random answers

Used to add different possible phrases for the same context.

Not available at the question-level. Question-level is zero, even identation level.

# Question calls Answer3 and one of (Answer1, Answer2, Answer3)
# with the same probability 33.3%
Question
	% Answer1
	% Answer2
	Answer3
	% Answer4

# here is possibility of collecting posibility groups
# 50% between (Answer1, Answer3) and 50% between (Answer2, Answer4)
Question
	%1 Answer1
	%2 Answer2
	%1 Answer3
	%2 Answer4
	Answer5
	
# if no identifcator available - it is an empty group
# 50% between (Answer2, Answer3) and 33.3% between (Answer1, Answer4, Answer5)
Question
	%1 Answer1
	% Answer2
	% Answer3
	%1 Answer4
	%1 Answer5

# custom probability values
# calls Answer2, 30% Answer1, 20% Answer4 and 25% between (Answer4, Answer5)
Question
	30% Answer1
	Answer2
	20% Answer3
	% Answer4	
	% Answer5

TODO possibility between groups

Usage exmaple:

Hi
Hello
Good morning!
	10% Hi
	% Hello

State flow

Dialog has few states simultaneously. At initial state acceptable questions without identation.

This examples show possible dialog states:

-> Is pizza delievered?
	Not yet # function binding below
		Call to the courier
			I'm goinng to call the courier
-> What's the wheater today?
	Shiny, but 15 degrees
		Which clothes do you recommend?
			Sweater and jacket

# possible inputs:
What's the wheater today?
Is pizza delievered?

# after phrase Is pizza delievered?:
# and answer: Not yet
Is pizza delievered?
	Not yet # function binding below
		-> Call to the courier
			I'm goinng to call the courier
-> What's the wheater today?
	Shiny, but 15 degrees
		Which clothes do you recommend?
			Sweater and jacket

??? losing state

Inline data

You can use variables and function defined in code before sharing the global scope. Or varaibles updated during the interpretation.

Python code, somewhere at the global scope:

def btc_rate():
	rate_eur = get_url_request()
	return rate_eur

cpu_temp = 54

Dialog code:

What's the current bitcoin rate?
	It's `btc_rate` euros for one bitcoin

What is your CPU temperature?
	`cpu_temp` degrees
	unfortunately not below zero

Set data

Fixed data

`variable_name:Python_literal` 

Can be used anywhere, sets if dialog going thru this phrase. Used in question-leve and answer-level.

Python code, somewhere at the global scope:

def look_report():
	if not friendly:
		return "You look great, as always"
	else:
		return joke("look")

Dialog code:

You are annoying `friendly:False`
	Sorry
	
How i looks like?
	`look_report`

Flexible data

`variable_name~Python_literal` 

Can be used anywhere, sets if dialog going thru this phrase. Used in question-leve and answer-level.

Dialog code:

...
	What is your name?
		My name is `collocutor~Mark`
			Nice to meet you, `collocutor`

Routines call

`function_name?` 

Call is only element at the line at answer-level. Values are at question-level. Activity is at answer-level. ANswers are Python literal.

`function_name`? # it will generate inline call and append ? to the end

This calls aren't locking proceess. They creates Queue, which processed at the same time as user input.

Python code, somewhere at the global scope:

def make_tea():
	go_to_kitchen()
	check_tea()
	check_water()
	do_manipulations()

Dialog code:

Make me a tea.
	No problems, i'm gooing to kitchen
	`make_tea?`
		'no tea'
			Sorry there is no tea, can i add tea to your shopping list?
		'no water'
			I can't find water.
			
# after Make me a tea phrase and response
# robot calls make_tea function in new subprocess
# at the input stage checks queue

??? end keyword - needed if we would like use probability to end a dialog