Coverage for astrocyte/conversations/__init__.py: 100%

1"""Astrocyte Conversation Engine — Hindsight-style turn-aware ingestion. 

2 

3A standalone subsystem for ingesting conversational input (chat 

4sessions, Slack threads, voice transcripts, multi-turn agent dialogs). 

5Conversations are inherently sequential — preserved as ordered turns, 

6chunked at turn boundaries (never mid-turn), with speaker context kept 

7intact. 

8 

9Independent of the Document Engine — different input shape, different 

10chunking strategy. Both feed the same Memory Engine downstream. 

11 

12To compose them with a memory backend, use ``astrocyte.ingest.ConversationIngestor`` 

13(Phase 3) which walks a conversation and calls memory.retain() per 

14chunk with opaque metadata. 

15 

16Public API: 

17 from astrocyte.conversations import ( 

18 Conversation, ConversationTurn, TurnRole, 

19 ConversationChunk, chunk_conversation, 

20 ConversationStore, InMemoryConversationStore, 

21 ) 

22""" 

23 

24from astrocyte.conversations.chunking import ( 

25 DEFAULT_MAX_CHARS_PER_CHUNK, 

26 ConversationChunk, 

27 chunk_conversation, 

28) 

29from astrocyte.conversations.ingestor import ConversationIngestor 

30from astrocyte.conversations.storage import ( 

31 ConversationNotFoundError, 

32 ConversationStore, 

33 InMemoryConversationStore, 

34) 

35from astrocyte.conversations.types import ( 

36 Conversation, 

37 ConversationTurn, 

38 TurnRole, 

39) 

40 

41__all__ = [ 

42 # types 

43 "Conversation", 

44 "ConversationTurn", 

45 "TurnRole", 

46 # chunking 

47 "ConversationChunk", 

48 "chunk_conversation", 

49 "DEFAULT_MAX_CHARS_PER_CHUNK", 

50 # storage 

51 "ConversationStore", 

52 "InMemoryConversationStore", 

53 "ConversationNotFoundError", 

54 # ingestor (Memory-Engine bridge) 

55 "ConversationIngestor", 

56]