Skip to content

Codebase intelligence for AI and humans: code health scores, auto-generated docs, git analytics, dead code detection, and architectural decisions via MCP.

refactor(core): move knowledge silo ranking and onboarding targets into core (#2517)

Raghav Chamadiya991a9ec3af4a
diff --git a/packages/core/src/repowise/core/analysis/knowledge_map.py b/packages/core/src/repowise/core/analysis/knowledge_map.py
new file mode 100644
index 000000000..42f72ada0
--- /dev/null
+++ b/packages/core/src/repowise/core/analysis/knowledge_map.py
@@ -0,0 +1,85 @@
+"""Knowledge silos and onboarding targets, folded from rows the caller fetched.
+
+Sync, no I/O, no clock. The server reads these rows with SQL; any other
+producer can hand in the same keys from its own store. Rows may be dicts,
+dataclasses or ORM/SQL rows (read through :func:`..health.rows.field`).
+"""
+
+from __future__ import annotations
+
+from collections.abc import Iterable, Mapping
+from typing import Any
+
+from repowise.core.analysis.health.rows import field
+
+#: A file is a silo when its primary owner holds more than this share of commits.
+SILO_OWNER_PCT = 0.8
+
+#: Onboarding targets returned per repository.
+ONBOARDING_LIMIT = 10
+
+
+def rank_silos(git_rows: Iterable[Any]) -> list[dict[str, Any]]:
+ """Files whose primary owner holds more than :data:`SILO_OWNER_PCT`, riskiest first.
+
+ Each row carries ``file_path``, ``primary_owner_email``,
+ ``primary_owner_commit_pct``, ``commit_count_90d`` and ``is_hotspot``; rows at
+ or under the threshold are dropped, so a caller may pass every file.
+ """
+ kept = [r for r in git_rows if (field(r, "primary_owner_commit_pct") or 0.0) > SILO_OWNER_PCT]
+ # Activity first, concentration second: sole ownership of a file nobody
+ # touches is a fact, not a risk, so previews should lead with live code.
+ kept.sort(
+ key=lambda r: (
+ not field(r, "is_hotspot"),
+ field(r, "commit_count_90d") is None,
+ -(field(r, "commit_count_90d") or 0),
+ -field(r, "primary_owner_commit_pct"),
+ )
+ )
+ return [
+ {
+ "file_path": field(r, "file_path"),
+ "owner_email": field(r, "primary_owner_email") or "",
+ "owner_pct": round(float(field(r, "primary_owner_commit_pct")), 3),
+ "commit_count_90d": int(field(r, "commit_count_90d") or 0),
+ "is_hotspot": bool(field(r, "is_hotspot")),
+ }
+ for r in kept
+ ]
+
+
+def onboarding_targets(
+ file_nodes: Iterable[Any],
+ doc_size: Mapping[str, int],
+ *,
+ limit: int = ONBOARDING_LIMIT,
+) -> list[dict[str, Any]]:
+ """High-centrality files carrying the least documentation.
+
+ ``file_nodes`` are non-test graph nodes with ``node_id`` and ``pagerank``;
+ nodes without a positive pagerank are skipped. ``doc_size`` maps a path to
+ its file page's word count, or any measure that ranks like one; 0 or absent
+ when undocumented. Rows report it as ``doc_words``, so a caller ranking by a
+ proxy must swap exact words in. Sorted by size, then pagerank descending;
+ ties keep input order.
+ """
+ candidates = [
+ {
+ "path": field(n, "node_id"),
+ "pagerank": field(n, "pagerank"),
+ "doc_words": doc_size.get(field(n, "node_id"), 0),
+ }
+ for n in file_nodes
+ if (field(n, "pagerank") or 0.0) > 0.0
+ ]
+ candidates.sort(key=lambda c: (c["doc_words"], -c["pagerank"]))
+ return candidates[:limit]
+
+
+__all__ = [
+ "ONBOARDING_LIMIT",
+ "SILO_OWNER_PCT",
+ "onboarding_targets",
+ "rank_silos",
+]
diff --git a/packages/server/src/repowise/server/services/knowledge_map.py b/packages/server/src/repowise/server/services/knowledge_map.py
index 2c5c265d3..64588ee15 100644
--- a/packages/server/src/repowise/server/services/knowledge_map.py
+++ b/packages/server/src/repowise/server/services/knowledge_map.py
@@ -8,6 +8,11 @@ from typing import Any
from sqlalchemy import func, select
from sqlalchemy.ext.asyncio import AsyncSession
+from repowise.core.analysis.knowledge_map import (
+ SILO_OWNER_PCT,
+ onboarding_targets,
+ rank_silos,
+)
from repowise.core.persistence.models import GitMetadata, GraphNode, Page
@@ -20,7 +25,7 @@ async def compute_knowledge_silos(session: AsyncSession, repo_id: str) -> list[d
blob columns) and loads the full ``content`` of every file page to count
words for a top-10 onboarding list Overview never reads.
- Three columns, one query, no entity hydration.
+ Five columns, one query, no entity hydration; ranked by :func:`rank_silos`.
"""
rows = await session.execute(
select(
@@ -29,32 +34,12 @@ async def compute_knowledge_silos(session: AsyncSession, repo_id: str) -> list[d
GitMetadata.primary_owner_commit_pct,
GitMetadata.commit_count_90d,
GitMetadata.is_hotspot,
- )
- .where(
+ ).where(
GitMetadata.repository_id == repo_id,
- GitMetadata.primary_owner_commit_pct > 0.8,
- )
- # Activity first, concentration second. Sole ownership of a file nobody
- # has touched in a year is a fact, not a risk: the bus factor only
- # costs anything on code that still changes. Ordered here so every
- # caller previewing a few of these gets the ones that matter rather
- # than whatever the table returned first.
- .order_by(
- GitMetadata.is_hotspot.desc(),
- GitMetadata.commit_count_90d.desc().nulls_last(),
- GitMetadata.primary_owner_commit_pct.desc(),
+ GitMetadata.primary_owner_commit_pct > SILO_OWNER_PCT,
)
)
- return [
- {
- "file_path": file_path,
- "owner_email": owner_email or "",
- "owner_pct": round(float(owner_pct or 0.0), 3),
- "commit_count_90d": int(commits or 0),
- "is_hotspot": bool(is_hotspot),
- }
- for file_path, owner_email, owner_pct, commits, is_hotspot in rows
- ]
+ return rank_silos(rows)
async def compute_onboarding_targets(
@@ -67,7 +52,7 @@ async def compute_onboarding_targets(
feeds the first-index experience. Overview used to reach it through
:func:`compute_knowledge_map` and pay for the owner aggregation as well.
"""
- # `pagerank > 0` in SQL, not in the comprehension below: the filter
+ # `pagerank > 0` in SQL too, though the fold repeats it: the filter
# discards most rows and there is no reason to ship them.
node_result = await session.execute(
select(GraphNode.node_id, GraphNode.pagerank).where(
@@ -93,21 +78,11 @@ async def compute_onboarding_targets(
)
)
doc_chars: dict[str, int] = {path: int(n or 0) for path, n in size_result}
-
- candidates = [
- {
- "path": node_id,
- "pagerank": pagerank,
- "doc_chars": doc_chars.get(node_id, 0),
- }
- for node_id, pagerank in all_nodes
- ]
- candidates.sort(key=lambda x: (x["doc_chars"], -x["pagerank"]))
- shortlist = candidates[:10]
+ shortlist = onboarding_targets(all_nodes, doc_chars)
# Exact words for the ten rows that are actually returned. Most of them are
# undocumented files with no page at all, so this usually fetches nothing.
- documented = [c["path"] for c in shortlist if c["doc_chars"] > 0]
+ documented = [c["path"] for c in shortlist if c["doc_words"] > 0]
exact_words: dict[str, int] = {}
if documented:
word_rows = await session.execute(
@@ -119,16 +94,7 @@ async def compute_onboarding_targets(
)
exact_words = {path: len((content or "").split()) for path, content in word_rows}
- onboarding_targets = [
- {
- "path": c["path"],
- "pagerank": c["pagerank"],
- "doc_words": exact_words.get(c["path"], 0),
- }
- for c in shortlist
- ]
-
- return onboarding_targets
+ return [{**c, "doc_words": exact_words.get(c["path"], 0)} for c in shortlist]
async def compute_knowledge_map(session: AsyncSession, repo_id: str) -> dict[str, Any]:
@@ -191,13 +157,11 @@ async def compute_knowledge_map(session: AsyncSession, repo_id: str) -> dict[str
"owner_pct": round(float(pct or 0.0), 3),
}
for file_path, owner_email, _owner_name, pct in all_git
- if (pct or 0.0) > 0.8
+ if (pct or 0.0) > SILO_OWNER_PCT
]
- onboarding_targets = await compute_onboarding_targets(session, repo_id)
-
return {
"top_owners": top_owners,
"knowledge_silos": knowledge_silos,
- "onboarding_targets": onboarding_targets,
+ "onboarding_targets": await compute_onboarding_targets(session, repo_id),
}
diff --git a/tests/unit/analysis/test_knowledge_map_folds.py b/tests/unit/analysis/test_knowledge_map_folds.py
new file mode 100644
index 000000000..2c35e8c94
--- /dev/null
+++ b/tests/unit/analysis/test_knowledge_map_folds.py
@@ -0,0 +1,60 @@
+"""The knowledge-map folds over plain dicts, as a non-SQL producer passes them."""
+
+from __future__ import annotations
+
+from repowise.core.analysis.knowledge_map import onboarding_targets, rank_silos
+
+
+def _git(path, pct, commits=0, hotspot=False, email="[email protected]"):
+ return {
+ "file_path": path,
+ "primary_owner_email": email,
+ "primary_owner_commit_pct": pct,
+ "commit_count_90d": commits,
+ "is_hotspot": hotspot,
+ }
+
+
+def test_rank_silos_filters_then_ranks_hotspot_activity_concentration() -> None:
+ rows = [
+ _git("quiet.py", 1.0),
+ _git("shared.py", 0.8, commits=99, hotspot=True),
+ _git("unknown.py", 0.9, commits=None),
+ _git("busy.py", 0.85, commits=20),
+ _git("hot.py", 0.81, commits=3, hotspot=True),
+ _git("hot_tie.py", 0.95, commits=3, hotspot=True, email=None),
+ _git("no_pct.py", None),
+ # Equal once rounded to 3 places; ranked by the raw share.
+ _git("close_low.py", 0.8496, commits=7),
+ _git("close_high.py", 0.8504, commits=7),
+ ]
+ out = rank_silos(rows)
+ assert [s["file_path"] for s in out] == [
+ "hot_tie.py",
+ "hot.py",
+ "busy.py",
+ "close_high.py",
+ "close_low.py",
+ "quiet.py",
+ "unknown.py",
+ ]
+ assert out[0] == {
+ "file_path": "hot_tie.py",
+ "owner_email": "",
+ "owner_pct": 0.95,
+ "commit_count_90d": 3,
+ "is_hotspot": True,
+ }
+ assert out[-1]["commit_count_90d"] == 0
+
+
+def test_onboarding_targets_skip_unranked_nodes_and_cap() -> None:
+ nodes = [{"node_id": f"f{i}.py", "pagerank": 0.1 * (i + 1)} for i in range(12)]
+ nodes.append({"node_id": "zero.py", "pagerank": 0.0})
+ out = onboarding_targets(nodes, {"f11.py": 5, "f10.py": 2}, limit=10)
+ assert len(out) == 10
+ assert out[0] == {"path": "f9.py", "pagerank": 0.1 * 10, "doc_words": 0}
+ everything = onboarding_targets(nodes, {"f11.py": 5, "f10.py": 2}, limit=20)
+ assert len(everything) == 12
+ tail = everything[-2:]
+ assert [t["path"] for t in tail] == ["f10.py", "f11.py"]
diff --git a/tests/unit/health/test_lightweight_health_imports.py b/tests/unit/health/test_lightweight_health_imports.py
index 5f3dfc1fa..e5420243f 100644
--- a/tests/unit/health/test_lightweight_health_imports.py
+++ b/tests/unit/health/test_lightweight_health_imports.py
@@ -27,6 +27,7 @@ _LIGHTWEIGHT = (
"repowise.core.analysis.health.ranking",
"repowise.core.analysis.health.aggregation",
"repowise.core.analysis.health.trends",
+ "repowise.core.analysis.knowledge_map",
"repowise.core.analysis.health.grading",
"repowise.core.analysis.health.suggestions",
"repowise.core.analysis.health.defect_accuracy",
diff --git a/tests/unit/server/golden/knowledge_map.json b/tests/unit/server/golden/knowledge_map.json
new file mode 100644
index 000000000..f9f5af13e
--- /dev/null
+++ b/tests/unit/server/golden/knowledge_map.json
@@ -0,0 +1,305 @@
+{
+ "silos": [
+ {
+ "file_path": "src/db.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.99,
+ "commit_count_90d": 30,
+ "is_hotspot": true
+ },
+ {
+ "file_path": "src/core.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.85,
+ "commit_count_90d": 30,
+ "is_hotspot": true
+ },
+ {
+ "file_path": "src/cli.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.81,
+ "commit_count_90d": 12,
+ "is_hotspot": false
+ },
+ {
+ "file_path": "src/api.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.95,
+ "commit_count_90d": 4,
+ "is_hotspot": false
+ },
+ {
+ "file_path": "src/old.py",
+ "owner_email": "",
+ "owner_pct": 1.0,
+ "commit_count_90d": 0,
+ "is_hotspot": false
+ }
+ ],
+ "onboarding_targets": [
+ {
+ "path": "src/m07.py",
+ "pagerank": 0.08,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m06.py",
+ "pagerank": 0.07,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m05.py",
+ "pagerank": 0.06,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m04.py",
+ "pagerank": 0.05,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m03.py",
+ "pagerank": 0.04,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m02.py",
+ "pagerank": 0.03,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m01.py",
+ "pagerank": 0.02,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m00.py",
+ "pagerank": 0.01,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m11.py",
+ "pagerank": 0.3,
+ "doc_words": 6
+ },
+ {
+ "path": "src/m10.py",
+ "pagerank": 0.2,
+ "doc_words": 1
+ }
+ ],
+ "knowledge_map": {
+ "top_owners": [
+ {
+ "email": "[email protected]",
+ "name": "Alice",
+ "files_owned": 2,
+ "percentage": 25.0
+ },
+ {
+ "email": "[email protected]",
+ "name": "Bob",
+ "files_owned": 2,
+ "percentage": 25.0
+ },
+ {
+ "email": "[email protected]",
+ "name": "Carol",
+ "files_owned": 1,
+ "percentage": 12.5
+ },
+ {
+ "email": "[email protected]",
+ "name": "Dan",
+ "files_owned": 1,
+ "percentage": 12.5
+ }
+ ],
+ "knowledge_silos": [
+ {
+ "file_path": "src/api.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.95
+ },
+ {
+ "file_path": "src/cli.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.81
+ },
+ {
+ "file_path": "src/core.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.85
+ },
+ {
+ "file_path": "src/db.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.99
+ },
+ {
+ "file_path": "src/old.py",
+ "owner_email": "",
+ "owner_pct": 1.0
+ }
+ ],
+ "onboarding_targets": [
+ {
+ "path": "src/m07.py",
+ "pagerank": 0.08,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m06.py",
+ "pagerank": 0.07,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m05.py",
+ "pagerank": 0.06,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m04.py",
+ "pagerank": 0.05,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m03.py",
+ "pagerank": 0.04,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m02.py",
+ "pagerank": 0.03,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m01.py",
+ "pagerank": 0.02,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m00.py",
+ "pagerank": 0.01,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m11.py",
+ "pagerank": 0.3,
+ "doc_words": 6
+ },
+ {
+ "path": "src/m10.py",
+ "pagerank": 0.2,
+ "doc_words": 1
+ }
+ ]
+ },
+ "endpoint": {
+ "top_owners": [
+ {
+ "email": "[email protected]",
+ "name": "Alice",
+ "files_owned": 2,
+ "percentage": 25.0
+ },
+ {
+ "email": "[email protected]",
+ "name": "Bob",
+ "files_owned": 2,
+ "percentage": 25.0
+ },
+ {
+ "email": "[email protected]",
+ "name": "Carol",
+ "files_owned": 1,
+ "percentage": 12.5
+ },
+ {
+ "email": "[email protected]",
+ "name": "Dan",
+ "files_owned": 1,
+ "percentage": 12.5
+ }
+ ],
+ "knowledge_silos": [
+ {
+ "file_path": "src/api.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.95
+ },
+ {
+ "file_path": "src/cli.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.81
+ },
+ {
+ "file_path": "src/core.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.85
+ },
+ {
+ "file_path": "src/db.py",
+ "owner_email": "[email protected]",
+ "owner_pct": 0.99
+ },
+ {
+ "file_path": "src/old.py",
+ "owner_email": "",
+ "owner_pct": 1.0
+ }
+ ],
+ "onboarding_targets": [
+ {
+ "path": "src/m07.py",
+ "pagerank": 0.08,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m06.py",
+ "pagerank": 0.07,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m05.py",
+ "pagerank": 0.06,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m04.py",
+ "pagerank": 0.05,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m03.py",
+ "pagerank": 0.04,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m02.py",
+ "pagerank": 0.03,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m01.py",
+ "pagerank": 0.02,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m00.py",
+ "pagerank": 0.01,
+ "doc_words": 0
+ },
+ {
+ "path": "src/m11.py",
+ "pagerank": 0.3,
+ "doc_words": 6
+ },
+ {
+ "path": "src/m10.py",
+ "pagerank": 0.2,
+ "doc_words": 1
+ }
+ ]
+ }
+}
diff --git a/tests/unit/server/test_knowledge_map_golden.py b/tests/unit/server/test_knowledge_map_golden.py
new file mode 100644
index 000000000..b96d79ca5
--- /dev/null
+++ b/tests/unit/server/test_knowledge_map_golden.py
@@ -0,0 +1,132 @@
+"""Golden output of the knowledge-map services and endpoint.
+
+Pins silos, onboarding targets and the full map exactly, so moving the folds
+between packages cannot change what Overview or ``/knowledge-map`` serves.
+Rewrite with ``REPOWISE_REWRITE_GOLDEN=1`` only for an intended behaviour change.
+"""
+
+from __future__ import annotations
+
+import json
+import os
+from datetime import UTC, datetime
+from pathlib import Path
+
+import pytest
+from httpx import AsyncClient
+
+from repowise.core.persistence import crud
+from repowise.core.persistence.database import get_session
+from repowise.core.persistence.models import GraphNode, Page
+from repowise.server.routers import knowledge_map
+from repowise.server.services.knowledge_map import (
+ compute_knowledge_map,
+ compute_knowledge_silos,
+ compute_onboarding_targets,
+)
+from tests.unit.server.conftest import create_test_repo
+
+GOLDEN = Path(__file__).parent / "golden" / "knowledge_map.json"
+_AT = datetime(2026, 9, 1, tzinfo=UTC)
+
+# (path, owner, email, pct, commits_90d, hotspot)
+_GIT = [
+ ("src/api.py", "Alice", "[email protected]", 0.95, 4, False),
+ ("src/core.py", "Bob", "[email protected]", 0.85, 30, True),
+ ("src/db.py", "Alice", "[email protected]", 0.99, 30, True),
+ ("src/cli.py", "Carol", "[email protected]", 0.81, 12, False),
+ ("src/old.py", "Carol", None, 1.0, 0, False),
+ ("src/shared.py", "Bob", "[email protected]", 0.8, 50, True),
+ ("src/team.py", "Dan", "[email protected]", 0.4, 9, False),
+ ("src/none.py", None, None, None, 0, False),
+]
+
+# (node_id, pagerank, is_test)
+_NODES = [(f"src/m{i:02d}.py", 0.01 * (i + 1), False) for i in range(8)] + [
+ ("src/m10.py", 0.2, False),
+ ("src/m11.py", 0.3, False),
+ ("src/api.py", 0.5, False),
+ ("src/core.py", 0.9, False),
+ ("tests/test_core.py", 0.7, True),
+ ("src/zero.py", 0.0, False),
+]
+
+# Shortlisted by character length (m11 before m10), reported in words.
+_PAGES = {
+ "src/m11.py": "a b c d e f",
+ "src/m10.py": "supercalifragilisticexpialidocious",
+ "src/api.py": "a " * 400,
+ "src/core.py": "documented " * 50,
+}
+
+
+async def _seed(session_factory, repo_id: str) -> None:
+ async with get_session(session_factory) as session:
+ for path, owner, email, pct, commits, hotspot in _GIT:
+ await crud.upsert_git_metadata(
+ session,
+ repository_id=repo_id,
+ file_path=path,
+ primary_owner_name=owner,
+ primary_owner_email=email,
+ primary_owner_commit_pct=pct,
+ commit_count_90d=commits,
+ is_hotspot=hotspot,
+ )
+ session.add_all(
+ [
+ GraphNode(
+ id=f"n{i:02d}",
+ repository_id=repo_id,
+ node_id=node_id,
+ pagerank=pagerank,
+ is_test=is_test,
+ )
+ for i, (node_id, pagerank, is_test) in enumerate(_NODES)
+ ]
+ )
+ session.add_all(
+ [
+ Page(
+ id=f"file_page:{path}",
+ repository_id=repo_id,
+ page_type="file_page",
+ title=path,
+ content=content,
+ target_path=path,
+ source_hash="h",
+ model_name="m",
+ provider_name="p",
+ created_at=_AT,
+ updated_at=_AT,
+ )
+ for path, content in _PAGES.items()
+ ]
+ )
+ await session.commit()
+
+
+async def test_knowledge_map_matches_the_golden(
+ client: AsyncClient, app, session_factory
+) -> None:
+ # The shared test app does not mount this router.
+ app.include_router(knowledge_map.router)
+ repo = await create_test_repo(client)
+ await _seed(session_factory, repo["id"])
+
+ async with get_session(session_factory) as session:
+ result = {
+ "silos": await compute_knowledge_silos(session, repo["id"]),
+ "onboarding_targets": await compute_onboarding_targets(session, repo["id"]),
+ "knowledge_map": await compute_knowledge_map(session, repo["id"]),
+ }
+ resp = await client.get(f"/api/repos/{repo['id']}/knowledge-map")
+ assert resp.status_code == 200
+ result["endpoint"] = resp.json()
+
+ if os.environ.get("REPOWISE_REWRITE_GOLDEN"):
+ GOLDEN.parent.mkdir(exist_ok=True)
+ GOLDEN.write_text(json.dumps(result, indent=2) + "\n", encoding="utf-8", newline="\n")
+ pytest.skip("golden rewritten")
+ assert result == json.loads(GOLDEN.read_text(encoding="utf-8"))