taskclf 0.2.3__tar.gz → 0.3.0__tar.gz

{taskclf-0.2.3 → taskclf-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: taskclf
-Version: 0.2.3
+Version: 0.3.0
 Summary: Local-first task classifier that infers your work type from computer activity signals
 Author: fruitiecutiepie
 Author-email: fruitiecutiepie <104437268+fruitiecutiepie@users.noreply.github.com>
@@ -26,7 +26,6 @@ Requires-Dist: uvicorn[standard]>=0.34.0
 Requires-Dist: pyyaml>=6.0
 Requires-Dist: pystray>=0.19.5
 Requires-Dist: pillow>=11.0.0
-Requires-Dist: plyer>=2.1.0
 Requires-Python: >=3.14
 Project-URL: Homepage, https://github.com/fruitiecutiepie/taskclf
 Project-URL: Documentation, https://fruitiecutiepie.github.io/taskclf

{taskclf-0.2.3 → taskclf-0.3.0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "taskclf"
-version = "0.2.3"
+version = "0.3.0"
 description = "Local-first task classifier that infers your work type from computer activity signals"
 readme = "README.md"
 license = {text = "MIT"}
@@ -32,7 +32,6 @@ dependencies = [
     "pyyaml>=6.0",
     "pystray>=0.19.5",
     "Pillow>=11.0.0",
-    "plyer>=2.1.0",
 ]
 
 [project.urls]

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/cli/main.py

@@ -1,6 +1,7 @@
 """Typer CLI entrypoint and command definitions for taskclf."""
 
 import datetime as dt
+import logging
 from importlib.metadata import version as _pkg_version
 from pathlib import Path
 
@@ -48,8 +49,16 @@ def main(
         False, "--version", "-v", callback=_version_callback, is_eager=True,
         help="Show version and exit.",
     ),
+    verbose: bool = typer.Option(
+        False, "--verbose", help="Enable DEBUG logging with file/line info.",
+    ),
 ) -> None:
     """Local-first task classifier CLI."""
+    level = logging.DEBUG if verbose else logging.WARNING
+    logging.basicConfig(
+        level=level,
+        format="%(levelname)s %(name)s %(pathname)s:%(lineno)d — %(message)s",
+    )
 
 
 # -- ingest -------------------------------------------------------------------
@@ -470,6 +479,26 @@ train_app = typer.Typer()
 app.add_typer(train_app, name="train")
 
 
+def _load_real_labels(
+    data_dir: str,
+    start: dt.date,
+    end: dt.date,
+) -> list:
+    """Load label spans from disk, filtered to the given date range."""
+    from taskclf.labels.store import read_label_spans
+
+    labels_path = Path(data_dir) / "labels_v1" / "labels.parquet"
+    if not labels_path.exists():
+        return []
+
+    all_spans = read_label_spans(labels_path)
+    start_dt = dt.datetime(start.year, start.month, start.day, tzinfo=dt.timezone.utc)
+    end_dt = dt.datetime(
+        end.year, end.month, end.day, 23, 59, 59, tzinfo=dt.timezone.utc
+    )
+    return [s for s in all_spans if s.end_ts >= start_dt and s.start_ts <= end_dt]
+
+
 @train_app.command("build-dataset")
 def train_build_dataset_cmd(
     date_from: str = typer.Option(..., "--from", help="Start date (YYYY-MM-DD)"),
@@ -496,11 +525,18 @@ def train_build_dataset_cmd(
     all_labels: list = []
     current = start
 
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -508,10 +544,8 @@ def train_build_dataset_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
 
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -562,14 +596,21 @@ def train_lgbm_cmd(
     end = dt.date.fromisoformat(date_to)
 
     all_features: list[pd.DataFrame] = []
-    all_labels = []
+    all_labels: list = []
     current = start
 
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -577,10 +618,8 @@ def train_lgbm_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
 
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -671,11 +710,19 @@ def train_evaluate_cmd(
     all_features: list[pd.DataFrame] = []
     all_labels: list = []
     current = start
+
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -683,9 +730,7 @@ def train_evaluate_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -817,11 +862,19 @@ def train_tune_reject_cmd(
     all_features: list[pd.DataFrame] = []
     all_labels: list = []
     current = start
+
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -829,9 +882,7 @@ def train_tune_reject_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -846,7 +897,7 @@ def train_tune_reject_cmd(
         raise typer.Exit(code=1)
 
     splits = split_by_time(labeled_df)
-    val_df = splits["val"]
+    val_df = labeled_df.iloc[splits["val"]].reset_index(drop=True)
     if val_df.empty:
         typer.echo("Validation split is empty — using full dataset.", err=True)
         val_df = labeled_df
@@ -920,14 +971,19 @@ def train_calibrate_cmd(
     all_features: list[pd.DataFrame] = []
     all_labels: list = []
     current = start
+
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
-            from taskclf.features.build import generate_dummy_features
-            from taskclf.labels.store import generate_dummy_labels
-
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -935,10 +991,8 @@ def train_calibrate_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
 
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -1047,11 +1101,19 @@ def train_retrain_cmd(
     all_features: list[pd.DataFrame] = []
     all_labels: list = []
     current = start
+
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -1059,9 +1121,8 @@ def train_retrain_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
+
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -1743,11 +1804,19 @@ def infer_compare_cmd(
     all_features: list[pd.DataFrame] = []
     all_labels: list = []
     current = start
+
+    if not synthetic:
+        all_labels = _load_real_labels(data_dir, start, end)
+        typer.echo(f"Loaded {len(all_labels)} label spans from disk")
+        if not all_labels:
+            typer.echo("WARNING: no label spans found — labeled dataset will be empty", err=True)
+
     while current <= end:
         if synthetic:
             rows = generate_dummy_features(current, n_rows=60)
             df = pd.DataFrame([r.model_dump() for r in rows])
             labels = generate_dummy_labels(current, n_rows=60)
+            all_labels.extend(labels)
         else:
             parquet_path = Path(data_dir) / f"features_v1/date={current.isoformat()}" / "features.parquet"
             if not parquet_path.exists():
@@ -1755,9 +1824,7 @@ def infer_compare_cmd(
                 current += dt.timedelta(days=1)
                 continue
             df = read_parquet(parquet_path)
-            labels = generate_dummy_labels(current, n_rows=len(df))
         all_features.append(df)
-        all_labels.extend(labels)
         current += dt.timedelta(days=1)
 
     if not all_features:
@@ -2185,6 +2252,9 @@ def tray_cmd(
 
     from taskclf.ui.tray import run_tray
 
+    if dev:
+        logging.getLogger("taskclf").setLevel(logging.DEBUG)
+
     tmp_dir: str | None = None
     if data_dir is not None:
         resolved_data_dir = data_dir
@@ -2244,6 +2314,9 @@ def ui_serve_cmd(
     from taskclf.ui.tray import ActivityMonitor, _LabelSuggester
     from taskclf.ui.window import WindowAPI, run_window
 
+    if dev:
+        logging.getLogger("taskclf").setLevel(logging.DEBUG)
+
     tmp_dir: str | None = None
     if data_dir is not None:
         resolved_data_dir = data_dir

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/core/model_io.py

@@ -3,12 +3,15 @@
 from __future__ import annotations
 
 import json
+import logging
 import random
 import subprocess
 from datetime import UTC, date, datetime
 from pathlib import Path
 from typing import Any, Literal
 
+logger = logging.getLogger(__name__)
+
 import lightgbm as lgb
 import pandas as pd
 from pydantic import BaseModel, Field
@@ -49,6 +52,7 @@ def _current_git_commit() -> str:
         )
         return result.stdout.strip() if result.returncode == 0 else "unknown"
     except Exception:
+        logger.debug("Could not determine git commit", exc_info=True)
         return "unknown"
 
 

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/infer/resolve.py

@@ -124,6 +124,7 @@ class ActiveModelReloader:
         try:
             return self._active_path.stat().st_mtime
         except OSError:
+            logger.debug("Could not stat %s", self._active_path, exc_info=True)
             return None
 
     def check_reload(

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/model_registry.py

@@ -752,7 +752,7 @@ def should_switch_active(
         try:
             current_f1 = float(current.reason["macro_f1"])  # type: ignore[arg-type]
         except (TypeError, ValueError):
-            pass
+            logger.debug("Could not parse macro_f1 from active pointer reason", exc_info=True)
 
     if current_f1 is None:
         return True

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/train/evaluate.py

@@ -3,6 +3,7 @@
 from __future__ import annotations
 
 import json
+import logging
 from pathlib import Path
 from typing import Any, Sequence
 
@@ -13,6 +14,8 @@ from pydantic import BaseModel
 from sklearn.metrics import accuracy_score, f1_score
 from sklearn.preprocessing import LabelEncoder
 
+logger = logging.getLogger(__name__)
+
 from taskclf.core.defaults import DEFAULT_REJECT_THRESHOLD, MIXED_UNKNOWN
 from taskclf.core.metrics import (
     calibration_curve_data,
@@ -389,6 +392,6 @@ def write_evaluation_artifacts(
         plt.close(fig)
         paths["calibration_plot"] = plot_path
     except Exception:
-        pass
+        logger.debug("Calibration plot generation failed", exc_info=True)
 
     return paths

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/train/retrain.py

@@ -178,6 +178,7 @@ def find_latest_model(models_dir: Path) -> Path | None:
                 best_ts = created
                 best_path = candidate
         except (json.JSONDecodeError, OSError):
+            logger.debug("Skipping unreadable metadata at %s", meta_path, exc_info=True)
             continue
 
     return best_path
@@ -212,6 +213,7 @@ def check_retrain_due(
         age = datetime.now(UTC) - model_ts
         return age >= timedelta(days=cadence_days)
     except (ValueError, TypeError):
+        logger.debug("Could not parse model created_at timestamp", exc_info=True)
         return True
 
 
@@ -243,6 +245,7 @@ def check_calibrator_update_due(
             store_ts = store_ts.replace(tzinfo=UTC)
         return datetime.now(UTC) - store_ts >= timedelta(days=cadence_days)
     except (json.JSONDecodeError, ValueError, TypeError, OSError):
+        logger.debug("Could not parse calibrator store.json", exc_info=True)
         return True
 
 

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/ui/frontend/src/App.tsx

@@ -1,7 +1,17 @@
-import { createSignal, Show, type Component } from "solid-js";
+import {
+  createEffect,
+  createSignal,
+  onMount,
+  Show,
+  type Component,
+} from "solid-js";
 import { LabelGrid } from "./components/LabelGrid";
 import { LiveBadge } from "./components/LiveBadge";
 import { StatePanel } from "./components/StatePanel";
+import {
+  requestPermission,
+  showTransitionNotification,
+} from "./lib/notifications";
 import { useWebSocket } from "./lib/ws";
 import { host } from "./lib/host";
 
@@ -145,6 +155,27 @@ const App: Component = () => {
 
   const ws = useWebSocket();
 
+  onMount(() => {
+    requestPermission();
+  });
+
+  createEffect(() => {
+    const count = ws.labelGridRequested();
+    if (count > 0) {
+      setHovering(true);
+      if (!inBrowser) host.invoke({ cmd: "showLabelGrid" });
+    }
+  });
+
+  createEffect(() => {
+    const prompt = ws.latestPrompt();
+    if (!prompt) return;
+    showTransitionNotification(prompt, () => {
+      setHovering(true);
+      if (!inBrowser) host.invoke({ cmd: "showLabelGrid" });
+    });
+  });
+
   function showLabel() {
     setHovering(true);
     host.invoke({ cmd: "showLabelGrid" });

taskclf-0.3.0/src/taskclf/ui/frontend/src/lib/notifications.ts

@@ -0,0 +1,40 @@
+import type { PromptLabelEvent } from "./ws";
+
+let permissionGranted = false;
+
+export async function requestPermission(): Promise<boolean> {
+  if (!("Notification" in window)) return false;
+  if (Notification.permission === "granted") {
+    permissionGranted = true;
+    return true;
+  }
+  if (Notification.permission === "denied") return false;
+  const result = await Notification.requestPermission();
+  permissionGranted = result === "granted";
+  return permissionGranted;
+}
+
+export function showTransitionNotification(
+  prompt: PromptLabelEvent,
+  onClick: () => void,
+): Notification | null {
+  if (!permissionGranted || !("Notification" in window)) return null;
+
+  const body = prompt.suggested_label
+    ? `${prompt.prev_app} → ${prompt.new_app} (${prompt.duration_min} min)\nSuggested: ${prompt.suggested_label}`
+    : `${prompt.prev_app} → ${prompt.new_app} (${prompt.duration_min} min)`;
+
+  const n = new Notification("taskclf — Activity changed", {
+    body,
+    tag: "taskclf-transition",
+    renotify: true,
+  });
+
+  n.onclick = () => {
+    window.focus();
+    onClick();
+    n.close();
+  };
+
+  return n;
+}

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/ui/frontend/src/lib/ws.ts

@@ -62,7 +62,28 @@ export interface TrayState {
   dev_mode: boolean;
 }
 
-export type WSEvent = Prediction | LabelSuggestion | StatusEvent | TrayState;
+export interface ShowLabelGridEvent {
+  type: "show_label_grid";
+}
+
+export interface PromptLabelEvent {
+  type: "prompt_label";
+  prev_app: string;
+  new_app: string;
+  block_start: string;
+  block_end: string;
+  duration_min: number;
+  suggested_label: string | null;
+  suggested_confidence: number | null;
+}
+
+export type WSEvent =
+  | Prediction
+  | LabelSuggestion
+  | StatusEvent
+  | TrayState
+  | ShowLabelGridEvent
+  | PromptLabelEvent;
 
 export type ConnectionStatus = "connecting" | "connected" | "disconnected";
 
@@ -88,6 +109,9 @@ export function useWebSocket() {
   );
   const [activeSuggestion, setActiveSuggestion] =
     createSignal<LabelSuggestion | null>(null);
+  const [latestPrompt, setLatestPrompt] =
+    createSignal<PromptLabelEvent | null>(null);
+  const [labelGridRequested, setLabelGridRequested] = createSignal(0);
   const [connectionStatus, setConnectionStatus] =
     createSignal<ConnectionStatus>("connecting");
   const [wsStats, setWsStats] = createSignal<WSStats>({
@@ -162,6 +186,22 @@ export function useWebSocket() {
               lastMessageAt: now,
             }));
             break;
+          case "show_label_grid":
+            setLabelGridRequested((n) => n + 1);
+            setWsStats((prev) => ({
+              ...prev,
+              messageCount: prev.messageCount + 1,
+              lastMessageAt: now,
+            }));
+            break;
+          case "prompt_label":
+            setLatestPrompt(data);
+            setWsStats((prev) => ({
+              ...prev,
+              messageCount: prev.messageCount + 1,
+              lastMessageAt: now,
+            }));
+            break;
         }
       } catch {
         // ignore malformed messages
@@ -208,6 +248,8 @@ export function useWebSocket() {
     latestPrediction,
     latestTrayState,
     activeSuggestion,
+    latestPrompt,
+    labelGridRequested,
     connectionStatus,
     wsStats,
     dismissSuggestion,

{taskclf-0.2.3 → taskclf-0.3.0}/src/taskclf/ui/server.py

@@ -115,6 +115,12 @@ class UserConfigUpdateRequest(BaseModel):
     username: str | None = None
 
 
+class NotificationAcceptRequest(BaseModel):
+    block_start: str = Field(description="ISO-8601 start of the activity block")
+    block_end: str = Field(description="ISO-8601 end of the activity block")
+    label: str = Field(description="Suggested label to accept")
+
+
 # ---------------------------------------------------------------------------
 # App factory
 # ---------------------------------------------------------------------------
@@ -378,6 +384,46 @@ def create_app(
             return {"available": False, "visible": False}
         return {"available": True, "visible": window_api.visible}
 
+    @app.post("/api/window/show-label-grid")
+    async def window_show_label_grid() -> dict[str, str]:
+        if window_api is not None:
+            window_api.show_label_grid()
+        await bus.publish({"type": "show_label_grid"})
+        return {"status": "ok"}
+
+    # -- REST: notification actions -------------------------------------------
+
+    @app.post("/api/notification/skip")
+    async def notification_skip() -> dict[str, str]:
+        logger.info("Notification skipped by user (no label change needed)")
+        return {"status": "skipped"}
+
+    @app.post("/api/notification/accept")
+    async def notification_accept(body: NotificationAcceptRequest) -> LabelResponse:
+        uid = user_config.user_id
+        try:
+            span = LabelSpan(
+                start_ts=dt.datetime.fromisoformat(body.block_start),
+                end_ts=dt.datetime.fromisoformat(body.block_end),
+                label=body.label,
+                provenance="suggestion",
+                user_id=uid,
+            )
+        except (ValueError, Exception) as exc:
+            raise HTTPException(status_code=422, detail=str(exc)) from exc
+        try:
+            append_label_span(span, labels_path)
+        except ValueError as exc:
+            raise HTTPException(status_code=409, detail=str(exc)) from exc
+        logger.info("Accepted suggested label: %s (%s → %s)", body.label, body.block_start, body.block_end)
+        return LabelResponse(
+            start_ts=span.start_ts.isoformat(),
+            end_ts=span.end_ts.isoformat(),
+            label=span.label,
+            provenance=span.provenance,
+            user_id=span.user_id,
+        )
+
     # -- WebSocket ------------------------------------------------------------
 
     @app.websocket("/ws/predictions")