| import sqlite3 |
| from pathlib import Path |
| |
| import pandas as pd |
| |
| from db.utils.json_utils import _safe_parse |
| |
| |
| def load_results( |
| db_path: str | Path = "mcp.db", |
| model_id: str | None = None, |
| ) -> pd.DataFrame: |
| """ |
| Load the `results` table, expand JSON columns, and optionally |
| filter to a single model_id. |
| |
| Returns a DataFrame with columns: |
| model_id, ts, <all params fields>, <all output fields> |
| """ |
| # 1) fetch raw rows |
| con = sqlite3.connect(str(db_path)) |
| raw_df = pd.read_sql("SELECT model_id, params, outputs, ts FROM results", con) |
| con.close() |
| |
| # print(raw_df.head()) |
| # 2) optionally filter to only the given model_id |
| print("===========",model_id,"==========") |
| if model_id is not None: |
| raw_df = raw_df[raw_df["model_id"] == model_id] |
| # print(raw_df.head()) |
| # 3) parse the JSON columns safely |
| raw_df["params"] = raw_df["params"].apply(_safe_parse) |
| raw_df["outputs"] = raw_df["outputs"].apply(_safe_parse) |
| # print(raw_df.head()) |
| # 4) drop any rows where parsing failed (empty dict) |
| filtered = raw_df[ |
| raw_df["params"].apply(bool) & raw_df["outputs"].apply(bool) |
| ].reset_index(drop=True) |
| |
| # 5) expand the dict columns into separate DataFrame columns |
| params_df = pd.json_normalize(filtered["params"]) |
| outputs_df = pd.json_normalize(filtered["outputs"]) |
| |
| # 6) concatenate model_id, ts, parameters, and outputs |
| print(params_df.head()) |
| print(outputs_df.head()) |
| final = pd.concat( |
| [ |
| filtered[["model_id", "ts"]], |
| params_df, |
| outputs_df |
| ], |
| axis=1 |
| ) |
| |
| return final |
