rasa.core.training.structures
Checkpoint Objects
filter_trackers
Filters out all trackers that do not satisfy the conditions.
StoryStep Objects
A StoryStep is a section of a story block between two checkpoints.
NOTE: Checkpoints are not only limited to those manually written in the story file, but are also implicitly created at points where multiple intents are separated in one line by chaining them with "OR"s.
explicit_events
Returns events contained in the story step including implicit events.
Not all events are always listed in the story dsl. This includes listen actions as well as implicitly set slots. This functions makes these events explicit and returns them with the rest of the steps events.
Story Objects
from_events
Create a story from a list of events.
StoryGraph Objects
Graph of the story-steps pooled from all stories in the training data.
ordered_steps
Returns the story steps ordered by topological order of the DAG.
cyclic_edges
Returns the story steps ordered by topological order of the DAG.
overlapping_checkpoint_names
Find overlapping checkpoints names
with_cycles_removed
Create a graph with the cyclic edges removed from this graph.
get
Looks a story step up by its id.
as_story_string
Convert the graph into the story file format.
order_steps
Topological sort of the steps returning the ids of the steps.
topological_sort
Creates a top sort of a directed graph. This is an unstable sorting!
The function returns the sorted nodes as well as the edges that need to be removed from the graph to make it acyclic (and hence, sortable).
The graph should be represented as a dictionary, e.g.:
>>> example_graph = { ... "a": set("b", "c", "d"), ... "b": set(), ... "c": set("d"), ... "d": set(), ... "e": set("f"), ... "f": set()} >>> StoryGraph.topological_sort(example_graph) (deque([u'e', u'f', u'a', u'c', u'd', u'b']), [])
is_empty
Checks if StoryGraph
is empty.