""" AUTHOR: Khushal P Soonderji DATE: Saturday, 7th Dec., 2024. OBJECTIVE: To define how messages will be stored in the database. REFERENCES: N/A DOWNLOADS: N/A """ # ***************************************************************************************************************** # ***** **** # *** IMPORT *** # ***** **** # ***************************************************************************************************************** # To make sibling directories accessible for imports: import sys sys.path.append(".") sys.path.append("..") # For making data behaviour_models: from pydantic import BaseModel, Field, field_validator, PastDatetime, AwareDatetime, constr from typing import Optional, Literal, Union, List, Any # My utils: from utils_v2.string import regex from utils_v2.date_time import date_time # To work with MongoDB: from bson.objectid import ObjectId # Data models: from models.core.ai.llm import LLMOutput # To work with date and time: import datetime # ***************************************************************************************************************** # ***** **** # *** MACROS / ONE-TIME INIT *** # ***** **** # ***************************************************************************************************************** # --- Nothing Yet # ***************************************************************************************************************** # ***** **** # *** VARIABLES *** # ***** **** # ***************************************************************************************************************** # --- Nothing Yet # ***************************************************************************************************************** # ***** **** # *** FUNCTIONS *** # ***** **** # ***************************************************************************************************************** class CoreMessageModel(BaseModel): messageId: ObjectId | None = Field( description = "the id of the document in mongodb that holds this information", frozen = True, default = None, alias = "_id", exclude = True ) ts: AwareDatetime = Field( description = "the time (utc) at which this message was sent by the sender", frozen = True ) syncTs: AwareDatetime = Field( description = "the time (utc) at which this message was pulled and stored in your server", frozen = True, default_factory = lambda: date_time.get_current_utc_date_time(as_string = False) ) tokenId: ObjectId = Field( description = "the id of the auth token that is associated with this message", frozen = True ) serviceType: Literal["email", "sms", "chat"] = Field( description = "the kind of service this message was sent/received from", frozen = True ) client: Literal[ "gmail", "outlook", # ........................................ Mail Clients "telegram", "whatsapp", "whatsappNimbus", # .................. Chat Clients "nimbusSmsIndia", "savvyBulkSmsKenya", # ..................... SMS Clients ] = Field( description = "the third-part client that was used", frozen = True ) clientMessageId: str | int | None = Field( description = "how the client identifies this message", frozen = True ) clientThreadId: str | int | None = Field( description = "how the client identifies the chat/thread in which this message was sent/received", frozen = True, default = None ) isSent: bool = Field( description = "to understand whether this message was an incoming message or outgoing message", frozen = False, default = False ) isBroadcast: bool = Field( description = "to understand if this message was broadcasted or sent one-to-one", frozen = True, default = False ) sentSuccessfully: bool | None = Field( description = "when a message is an outgoing message, this indicates if the message was send successfully", frozen = False, default = False ) sender: List[str] | None = Field( description = "the name of the sender; null if you are the sender", frozen = True ) recipient: List[str] | None = Field( description = "the name of the recipient; null if you are the recipient", frozen = True, default = None ) chat: str | None = Field( description = "the name of the chat where the message was exchanged; relevant in chat apps like telegram", frozen = True ) message: dict = Field( description = "the actual content(s) of the message; will differ for each service/client", frozen = True ) snippet: str = Field( description = "a truncated version of the actual textual content of the message", frozen = False ) aiSnippet: LLMOutput | dict | None = Field( description = "holds a short summary generated by an llm", frozen = False, default = None ) tags: List[Any] = Field( description = "a list of keywords to apply to this file/dir to filter it later", frozen = False, default = [], examples = ["urgent", "otp", "GST"] ) # ┏┓ ┏• # ┃ ┏┓┏┓╋┓┏┓ # ┗┛┗┛┛┗┛┗┗┫ # ┛ class Config: extra = "forbid" arbitrary_types_allowed = True def model_dump(self, *args, **kwargs): return super().model_dump(*args, by_alias = True, **kwargs) # ┏┓ • # ┃┃┏┓┏┓┏┓┏┓┏┓╋┓┏┓┏ # ┣┛┛ ┗┛┣┛┗ ┛ ┗┗┗ ┛ # ┛ @property def full(self): full_json = { "messageId": str(self.messageId), "ts": self.ts.isoformat(), "serviceType": self.serviceType, "client": self.client, "clientMessageId": self.clientMessageId, "clientThreadId": self.clientThreadId, "isSent": self.isSent, "isBroadcast": self.isBroadcast, "sentSuccessfully": self.sentSuccessfully, "sender": self.sender, "chat": self.chat, "message": self.message, "snippet": self.snippet, "aiSnippet": self.aiSnippet.to_json() if isinstance(self.aiSnippet, LLMOutput) else self.aiSnippet, "tags": self.tags } if full_json["aiSnippet"]: full_json["aiSnippet"]["invocationId"] = str(full_json["aiSnippet"]["invocationId"]) return full_json @property def preview(self): preview_json = { "messageId": str(self.messageId), "ts": self.ts.isoformat(), "serviceType": self.serviceType, "client": self.client, "clientMessageId": self.clientMessageId, "clientThreadId": self.clientThreadId, "isSent": self.isSent, "isBroadcast": self.isBroadcast, "sentSuccessfully": self.sentSuccessfully, "sender": self.sender, "chat": self.chat, "snippet": self.snippet, "aiSnippet": self.aiSnippet.to_json() if isinstance(self.aiSnippet, LLMOutput) else self.aiSnippet, "tags": self.tags } if preview_json["aiSnippet"]: preview_json["aiSnippet"]["invocationId"] = str(preview_json["aiSnippet"]["invocationId"]) return preview_json # ┓┏ ┓• ┓ • # ┃┃┏┓┃┓┏┫┏┓╋┓┏┓┏┓ # ┗┛┗┻┗┗┗┻┗┻┗┗┗┛┛┗ @field_validator("ts", "syncTs", mode = "before") def parse_date_time(cls, value): return date_time.parse_date_time(input_value = value, timezone = date_time.TIMEZONE_UTC) @field_validator("tokenId", mode = "before") def parse_oid(cls, value): try: if isinstance(value, str): value = ObjectId(value) except: pass return value @field_validator("sender", "recipient", mode = "before") def validate_participants(cls, value): if value is None: value = [] if isinstance(value, str): value = [value] return value @field_validator("tags", mode = "before") def validate_tags(cls, value): if value is None: value = [] return value # ***************************************************************************************************************** # ***** **** # *** MAIN PROGRAM *** # ***** **** # ***************************************************************************************************************** if __name__ == "__main__": from utils_v2.string import json message = CoreMessageModel( ts = date_time.get_current_utc_date_time(as_string = False), tokenId = "67519cf3a7804fcbc6f12452", serviceType = "email", client = "gmail", clientMessageId = 123, clientThreadId = 456, isSent = False, message = { "from": "bhopli@gmil.com", "to": "hello@thecaoffice.com", "message": "Hello, World!" }, sender = "Polki", chat = "T6 Cats", snippet = "Hi, there! How do you do?" ) print("MESSAGE MODEL:", json.to_string(message.model_dump(), default = str))