Files
beavergram/backend/src/utils/search/models.py
T

81 lines
2.1 KiB
Python

from datetime import UTC, datetime, timedelta
from typing import Literal
from pydantic import BaseModel, field_validator
DEFAULT_LIMIT = 50
MAX_LIMIT = 500
_EPOCH = datetime(1970, 1, 1, tzinfo=UTC)
_MICROSECOND = timedelta(microseconds=1)
ChatKind = Literal["private", "group", "channel"]
HitSource = Literal["text", "stt", "version", "data"]
FoundWhere = Literal[
"text", "button", "link", "forward", "reply", "file", "preview", "data"
]
def encode_cursor(date: datetime, chat_id: int, message_id: int) -> str:
return f"{(date - _EPOCH) // _MICROSECOND}_{chat_id}_{message_id}"
def decode_cursor(cursor: str) -> tuple[datetime, int, int]:
micros, chat_id, message_id = (int(part) for part in cursor.split("_"))
return _EPOCH + micros * _MICROSECOND, chat_id, message_id
class SearchFilters(BaseModel):
account_id: int
query: str | None = None
chat_id: int | None = None
sender_id: int | None = None
kind: ChatKind | None = None
has_media: bool | None = None
deleted: bool | None = None
edited: bool | None = None
date_from: datetime | None = None
date_to: datetime | None = None
regex: str | None = None
everywhere: bool = False
cursor: str | None = None
limit: int = DEFAULT_LIMIT
@field_validator("cursor")
@classmethod
def _valid_cursor(cls, value: str | None) -> str | None:
if value is not None:
decode_cursor(value)
return value
class Found(BaseModel):
where: FoundWhere
path: str
value: str
button: str | None = None
class SearchHit(BaseModel):
chat_id: int
message_id: int
date: datetime
sender_id: int | None
text: str | None
has_media: bool
media_kind: str | None
edited_at: datetime | None
deleted_at: datetime | None
source: HitSource
extracted_text: str | None = None
version_text: str | None = None
version_at: datetime | None = None
found: Found | None = None
class SearchPage(BaseModel):
hits: list[SearchHit]
next_cursor: str | None
terms: list[str]
scanned_to: datetime | None = None