Extend run stats DSL with statistical and temporal analysis

app/routes.py

@@ -1,4 +1,6 @@
+import csv
 import hashlib
+import io
 import re
 import secrets
 from datetime import UTC, datetime
@@ -7,6 +9,7 @@ from zoneinfo import available_timezones
 
 from flask import (
     Blueprint,
+    Response,
     abort,
     current_app,
     flash,
@@ -30,6 +33,8 @@ from .history import activity_page, latest_mapped_run, overview, records_for
 from .models import Activity, Friendship, GarminConnection, User
 from .services import activity_data, import_cache, import_upload, route_data
 from .stats import DEFAULT, METRICS, explore, parse_query, period_bounds, query_text
+from .stats_analysis import EXTRA_UNITS, FUNCTIONS, GROUPS
+from .stats_dsl import UNITS
 
 web = Blueprint("web", __name__)
 DUMMY_HASH = generate_password_hash("not-a-real-account-password")
@@ -214,6 +219,19 @@ def stats():
         results=results,
         error=error,
         submitted_query=request.args.get("query", ""),
+        language={
+            "functions": sorted(FUNCTIONS),
+            "groups": sorted(GROUPS),
+            "fields": {
+                key: {
+                    "label": info[0],
+                    "unit": info[1],
+                    "description": info[2],
+                    "units": list(({**UNITS, **EXTRA_UNITS}).get(key, {})),
+                }
+                for key, info in METRICS.items()
+            },
+        },
     ), 400 if error else 200
 
 
@@ -222,15 +240,14 @@ def stats():
 def stats_runs():
     try:
         query = parse_query(request.args)
-        bucket = (
-            datetime.strptime(request.args.get("bucket", ""), "%d %b %Y").replace(tzinfo=UTC).date()
-        )
+        bucket = request.args.get("bucket", "")
         results = explore(
             current_user.id,
             current_user.timezone,
             query,
             int(request.args.get("page", "1")),
             selected_bucket=bucket,
+            raw_group=request.args.get("scope") == "group",
         )
     except (ValueError, OverflowError) as exc:
         return jsonify(error=str(exc)), 400
@@ -252,13 +269,69 @@ def stats_validate():
     try:
         query = parse_query({"query": payload["query"]})
         period_bounds(query, current_user.timezone)
-        return jsonify(valid=True, canonical=query_text(query), diagnostics=[])
+        return jsonify(
+            valid=True, canonical=query_text(query), diagnostics=[], derived=query["derived"]
+        )
     except (ValueError, OverflowError) as exc:
         return jsonify(
             valid=False, diagnostics=[{"line": getattr(exc, "line", 1), "message": str(exc)}]
         )
 
 
+@web.get("/stats/export")
+@login_required
+@limiter.limit("20 per minute")
+def stats_export():
+    try:
+        query = parse_query(request.args)
+        result = explore(current_user.id, current_user.timezone, query, export=True)
+        kind = request.args.get("kind", "groups")
+        if kind not in {"runs", "groups"}:
+            raise ValueError("Export runs or groups.")
+    except (ValueError, OverflowError) as exc:
+        return jsonify(error=str(exc)), 400
+    output = io.StringIO(newline="")
+    writer = csv.writer(output)
+
+    def safe(value):
+        # Keep user-authored run titles inert when opened in a spreadsheet.
+        text = str(value)
+        return "'" + text if text.lstrip().startswith(("=", "+", "-", "@")) else text
+
+    columns = result["cards"] if kind == "groups" else result["evidence_columns"]
+    headers = [
+        f"{card['key']} ({'sec/km' if card['unit'] == '/km' else card['unit']})" for card in columns
+    ]
+    if kind == "groups":
+        writer.writerow(["Group", "Contributing runs", *headers])
+        for group in result["groups"]:
+            writer.writerow([safe(group["label"]), group["count"], *group["raw_values"]])
+    else:
+        writer.writerow(
+            ["Run", "Local date", "Distance (km)", "Previous run", "Previous run date", *headers]
+        )
+        for row in result["rows"]:
+            previous = row["previous_run"] or {}
+            writer.writerow(
+                [
+                    safe(row["title"]),
+                    row["date"],
+                    row["distance"],
+                    safe(previous.get("title", "")),
+                    previous.get("date", ""),
+                    *row["raw_values"],
+                ]
+            )
+    return Response(
+        output.getvalue(),
+        mimetype="text/csv",
+        headers={
+            "Content-Disposition": f'attachment; filename="stride-{kind}.csv"',
+            "Cache-Control": "private, no-store",
+        },
+    )
+
+
 @web.route("/friends", methods=["GET", "POST"])
 @login_required
 @limiter.limit("20 per hour", methods=["POST"])