Merge pull request optuna#723 from eukaryo/code-fix/download_csv

Code fix/download csv

optuna_dashboard/_app.py

@@ -1,8 +1,12 @@
 from __future__ import annotations
 
+import csv
 import functools
+import io
+from itertools import chain
 import logging
 import os
+import re
 import typing
 from typing import Any
 from typing import Optional
@@ -449,6 +453,48 @@ def save_trial_note(study_id: int, trial_id: int) -> dict[str, Any]:
         response.status = 204  # No content
         return {}
 
+    @app.get("/csv/<study_id:int>")
+    def download_csv(study_id: int) -> BottleViewReturn:
+        # Create a CSV file
+        try:
+            study_name = storage.get_study_name_from_id(study_id)
+            study = optuna.load_study(storage=storage, study_name=study_name)
+        except KeyError:
+            response.status = 404  # Not found
+            return {"reason": f"study_id={study_id} is not found"}
+        trials = study.trials
+        param_names = sorted(set(chain.from_iterable([t.params.keys() for t in trials])))
+        user_attr_names = sorted(set(chain.from_iterable([t.user_attrs.keys() for t in trials])))
+        param_names_header = [f"Param {x}" for x in param_names]
+        user_attr_names_header = [f"UserAttribute {x}" for x in user_attr_names]
+        n_objs = len(study.directions)
+        if study.metric_names is not None:
+            value_header = study.metric_names
+        else:
+            value_header = ["Value"] if n_objs == 1 else [f"Objective {x}" for x in range(n_objs)]
+        column_names = (
+            ["Number", "State"] + value_header + param_names_header + user_attr_names_header
+        )
+
+        buf = io.StringIO("")
+        writer = csv.writer(buf)
+        writer.writerow(column_names)
+        for frozen_trial in trials:
+            row = [frozen_trial.number, frozen_trial.state.name]
+            row.extend(frozen_trial.values if frozen_trial.values is not None else [None] * n_objs)
+            row.extend([frozen_trial.params.get(name, None) for name in param_names])
+            row.extend([frozen_trial.user_attrs.get(name, None) for name in user_attr_names])
+            writer.writerow(row)
+
+        # Set response headers
+        output_name = "-".join(re.sub(r'[\\/:*?"<>|]+', "", study_name).split(" "))
+        response.headers["Content-Type"] = "text/csv; chatset=cp932"
+        response.headers["Content-Disposition"] = f"attachment; filename={output_name}.csv"
+
+        # Response body
+        buf.seek(0)
+        return buf.read()
+
     @app.get("/favicon.ico")
     def favicon() -> BottleViewReturn:
         use_gzip = "gzip" in request.headers["Accept-Encoding"]

optuna_dashboard/ts/components/StudyDetail.tsx

@@ -12,6 +12,7 @@ import {
 import Grid2 from "@mui/material/Unstable_Grid2"
 import ChevronRightIcon from "@mui/icons-material/ChevronRight"
 import HomeIcon from "@mui/icons-material/Home"
+import DownloadIcon from "@mui/icons-material/Download"
 
 import { StudyNote } from "./Note"
 import { actionCreator } from "../action"
@@ -149,11 +150,39 @@ export const StudyDetail: FC<{
     content = <TrialList studyDetail={studyDetail} />
   } else if (page === "trialTable") {
     content = (
-      <Card sx={{ margin: theme.spacing(2) }}>
-        <CardContent>
-          <TrialTable studyDetail={studyDetail} initialRowsPerPage={50} />
-        </CardContent>
-      </Card>
+      <Box sx={{ display: "flex", width: "100%", flexDirection: "column" }}>
+        <Card
+          sx={{
+            margin: theme.spacing(2),
+            width: "auto",
+            height: "auto",
+            display: "flex",
+            justifyContent: "left",
+            alignItems: "left",
+          }}
+        >
+          <CardContent>
+            <IconButton
+              aria-label="download csv"
+              size="small"
+              color="inherit"
+              download
+              sx={{ margin: "auto 0" }}
+              href={`/csv/${studyDetail?.id}`}
+            >
+              <DownloadIcon />
+              <Typography variant="button" sx={{ margin: theme.spacing(2) }}>
+                Download CSV File
+              </Typography>
+            </IconButton>
+          </CardContent>
+        </Card>
+        <Card sx={{ margin: theme.spacing(2) }}>
+          <CardContent>
+            <TrialTable studyDetail={studyDetail} initialRowsPerPage={50} />
+          </CardContent>
+        </Card>
+      </Box>
     )
   } else if (page === "note" && studyDetail !== null) {
     content = (

python_tests/test_csv_download.py

@@ -0,0 +1,109 @@
+from __future__ import annotations
+
+from typing import Any
+
+import optuna
+from optuna.trial import TrialState
+from optuna_dashboard._app import create_app
+import pytest
+
+from .wsgi_client import send_request
+
+
+def _validate_output(
+    storage: optuna.storages.BaseStorage,
+    correct_status: int,
+    study_id: int,
+    expect_no_result: bool = False,
+    extra_col_names: list[str] | None = None,
+) -> None:
+    app = create_app(storage)
+    status, _, body = send_request(
+        app,
+        f"/csv/{study_id}",
+        "GET",
+        content_type="application/json",
+    )
+    assert status == correct_status
+    decoded_csv = str(body.decode("utf-8"))
+    if expect_no_result:
+        assert "is not found" in decoded_csv
+    else:
+        col_names = ["Number", "State"] + ([] if extra_col_names is None else extra_col_names)
+        assert all(col_name in decoded_csv for col_name in col_names)
+
+
+def test_download_csv_no_trial() -> None:
+    def objective(trial: optuna.Trial) -> float:
+        x = trial.suggest_float("x", -100, 100)
+        y = trial.suggest_categorical("y", [-1, 0, 1])
+        return x**2 + y
+
+    storage = optuna.storages.InMemoryStorage()
+    study = optuna.create_study(storage=storage)
+    study.optimize(objective, n_trials=0)
+    _validate_output(storage, 200, 0)
+
+
+def test_download_csv_all_waiting() -> None:
+    storage = optuna.storages.InMemoryStorage()
+    study = optuna.create_study(storage=storage)
+    study.add_trial(optuna.trial.create_trial(state=TrialState.WAITING))
+    _validate_output(storage, 200, 0)
+
+
+def test_download_csv_all_running() -> None:
+    storage = optuna.storages.InMemoryStorage()
+    study = optuna.create_study(storage=storage)
+    study.add_trial(optuna.trial.create_trial(state=TrialState.RUNNING))
+    _validate_output(storage, 200, 0)
+
+
+@pytest.mark.parametrize("study_id", [0, 1])
+def test_download_csv_fail(study_id: int) -> None:
+    def objective(trial: optuna.Trial) -> float:
+        x = trial.suggest_float("x", -100, 100)
+        y = trial.suggest_categorical("y", [-1, 0, 1])
+        return x**2 + y
+
+    storage = optuna.storages.InMemoryStorage()
+    study = optuna.create_study(storage=storage)
+    optuna.logging.set_verbosity(optuna.logging.ERROR)
+    study.optimize(objective, n_trials=10)
+    expect_no_result = study_id != 0
+    cols = ["Param x", "Param y", "Value"]
+    _validate_output(storage, 404 if expect_no_result else 200, study_id, expect_no_result, cols)
+
+
+@pytest.mark.parametrize("is_multi_obj", [True, False])
+def test_download_csv_multi_obj(is_multi_obj: bool) -> None:
+    def objective(trial: optuna.Trial) -> Any:
+        x = trial.suggest_float("x", -100, 100)
+        y = trial.suggest_categorical("y", [-1, 0, 1])
+        if is_multi_obj:
+            return x**2, y
+        return x**2 + y
+
+    storage = optuna.storages.InMemoryStorage()
+    directions = ["minimize", "minimize"] if is_multi_obj else ["minimize"]
+    study = optuna.create_study(storage=storage, directions=directions)
+    optuna.logging.set_verbosity(optuna.logging.ERROR)
+    study.optimize(objective, n_trials=10)
+    cols = ["Param x", "Param y"]
+    cols += ["Objective 0", "Objective 1"] if is_multi_obj else ["Value"]
+    _validate_output(storage, 200, 0, extra_col_names=cols)
+
+
+def test_download_csv_user_attr() -> None:
+    def objective(trial: optuna.Trial) -> float:
+        x = trial.suggest_float("x", -100, 100)
+        y = trial.suggest_categorical("y", [-1, 0, 1])
+        trial.set_user_attr("abs_y", abs(y))
+        return x**2 + y
+
+    storage = optuna.storages.InMemoryStorage()
+    study = optuna.create_study(storage=storage)
+    optuna.logging.set_verbosity(optuna.logging.ERROR)
+    study.optimize(objective, n_trials=10)
+    cols = ["Param x", "Param y", "Value", "UserAttribute abs_y"]
+    _validate_output(storage, 200, 0, extra_col_names=cols)