Files
crawle-snapp/db/repositories/flow_configs.py
T
2026-06-20 17:07:48 +03:30

93 lines
3.2 KiB
Python

"""Repository for the `flow_configs` table."""
from __future__ import annotations
import json
from typing import Any
from db.connection import get_db
FlowConfigRow = dict[str, Any]
_FIELDS = [
"name",
"search_texts_json", "search_box_selector", "search_box_selector_type",
"search_submit_selector", "search_submit_selector_type",
"search_results_selector", "search_results_selector_type",
"scroll_container_selector", "scroll_container_selector_type",
"max_scrolls", "scroll_pause_ms", "no_new_content_timeout_ms",
"pagination_selector", "pagination_selector_type", "max_pages", "pagination_wait_ms",
"item_selector_template", "item_selector_type", "item_url_template", "target_item_ids_json",
"target_url",
]
def _deserialise(row: dict[str, Any]) -> FlowConfigRow:
for key in ("search_texts_json", "target_item_ids_json"):
if isinstance(row.get(key), str):
try:
row[key] = json.loads(row[key])
except (json.JSONDecodeError, TypeError):
row[key] = []
return row
def _serialise(data: dict[str, Any]) -> dict[str, Any]:
out = dict(data)
for key in ("search_texts_json", "target_item_ids_json"):
if isinstance(out.get(key), list):
out[key] = json.dumps(out[key])
return out
class FlowConfigRepository:
async def list(self) -> list[FlowConfigRow]:
async with get_db() as db:
rows = await (await db.execute(
"SELECT * FROM flow_configs ORDER BY id DESC"
)).fetchall()
return [_deserialise(dict(r)) for r in rows]
async def get(self, cfg_id: int) -> FlowConfigRow | None:
async with get_db() as db:
row = await (await db.execute(
"SELECT * FROM flow_configs WHERE id=?", (cfg_id,)
)).fetchone()
return _deserialise(dict(row)) if row else None
async def get_first(self) -> FlowConfigRow | None:
async with get_db() as db:
row = await (await db.execute(
"SELECT * FROM flow_configs ORDER BY id ASC LIMIT 1"
)).fetchone()
return _deserialise(dict(row)) if row else None
async def create(self, data: dict[str, Any]) -> int:
data = _serialise(data)
async with get_db() as db:
placeholders = ", ".join("?" for _ in _FIELDS)
cols = ", ".join(_FIELDS)
values = [data.get(f) for f in _FIELDS]
cur = await db.execute(
f"INSERT INTO flow_configs ({cols}) VALUES ({placeholders})", values
)
await db.commit()
return int(cur.lastrowid or 0)
async def update(self, cfg_id: int, data: dict[str, Any]) -> None:
data = _serialise(data)
async with get_db() as db:
set_clause = ", ".join(f"{f}=?" for f in _FIELDS)
values = [data.get(f) for f in _FIELDS] + [cfg_id]
await db.execute(
f"UPDATE flow_configs SET {set_clause} WHERE id=?", values
)
await db.commit()
async def delete(self, cfg_id: int) -> None:
async with get_db() as db:
await db.execute("DELETE FROM flow_configs WHERE id=?", (cfg_id,))
await db.commit()
flow_config_repo = FlowConfigRepository()