New table: forgejo_commit_activity

This commit is contained in:
null 2026-05-24 20:24:43 -05:00
parent 1fc1df5aeb
commit 7919c29bbd
4 changed files with 248 additions and 12 deletions

View File

@ -16,12 +16,13 @@ from sqlmodel import func, select
from app.api.deps import ORG_MEMBER_DEP, OrganizationContext from app.api.deps import ORG_MEMBER_DEP, OrganizationContext
from app.core.time import utcnow from app.core.time import utcnow
from app.db.session import get_session from app.db.session import async_session_maker, get_session
from app.models.board_repository_links import BoardRepositoryLink from app.models.board_repository_links import BoardRepositoryLink
from app.models.forgejo_commit_activity import ForgejoCommitDay
from app.models.forgejo_connections import ForgejoConnection from app.models.forgejo_connections import ForgejoConnection
from app.models.forgejo_issues import ForgejoIssue from app.models.forgejo_issues import ForgejoIssue
from app.models.forgejo_repositories import ForgejoRepository from app.models.forgejo_repositories import ForgejoRepository
from app.schemas.metrics import HeatmapDay, HeatmapResponse, LastPushRead, MetricsResponse from app.schemas.metrics import CommitActivityDay, HeatmapDay, HeatmapResponse, LastPushRead, MetricsResponse
from app.services.forgejo_client import ForgejoAPIClient from app.services.forgejo_client import ForgejoAPIClient
if TYPE_CHECKING: if TYPE_CHECKING:
@ -43,10 +44,11 @@ _LINE_STATS_TTL_MISS = 30 # 30 s — retry cadence while Forgejo is still com
async def _bg_fetch_line_stats( async def _bg_fetch_line_stats(
cache_key: str, cache_key: str,
repos: list[tuple[str, str, str, str | None]], # (owner, repo, base_url, token) # (owner, repo, base_url, token, repository_id, organization_id)
since_iso: str, # ISO-8601 date string for the commits?since= filter repos: list[tuple[str, str, str, str | None, UUID, UUID]],
since_iso: str,
) -> None: ) -> None:
"""Background task: sum commit line stats across all tracked repos and cache.""" """Background task: sum commit line stats, persist per-repo day counts to DB."""
async def _one(owner: str, repo: str, base_url: str, token: str | None) -> tuple[int, int, dict[str, int]]: async def _one(owner: str, repo: str, base_url: str, token: str | None) -> tuple[int, int, dict[str, int]]:
try: try:
@ -56,14 +58,52 @@ async def _bg_fetch_line_stats(
return 0, 0, {} return 0, 0, {}
try: try:
results = await asyncio.gather(*[_one(o, r, bu, tok) for o, r, bu, tok in repos]) results = await asyncio.gather(*[_one(o, r, bu, tok) for o, r, bu, tok, _, _ in repos])
total_adds = sum(a for a, _, _ in results) total_adds = sum(a for a, _, _ in results)
total_dels = sum(d for _, d, _ in results) total_dels = sum(d for _, d, _ in results)
merged_days: dict[str, int] = {} merged_days: dict[str, int] = {}
# Persist per-repo per-day counts to DB (upsert).
try:
from datetime import date as _date
async with async_session_maker() as session:
for (_, _, _, _, repo_id, org_id), (_, _, day_counts) in zip(repos, results):
for day_str, cnt in day_counts.items():
try:
day_obj = _date.fromisoformat(day_str)
except ValueError:
continue
existing = (
await session.exec(
select(ForgejoCommitDay).where(
ForgejoCommitDay.repository_id == repo_id,
ForgejoCommitDay.date == day_obj,
)
)
).first()
if existing is None:
session.add(
ForgejoCommitDay(
organization_id=org_id,
repository_id=repo_id,
date=day_obj,
commit_count=cnt,
)
)
elif existing.commit_count != cnt:
existing.commit_count = cnt
existing.updated_at = utcnow()
await session.commit()
except Exception as db_exc:
from app.core.logging import get_logger as _get_logger
_get_logger(__name__).warning(
"commit_activity_db_write_failed", extra={"error": str(db_exc)}
)
for _, _, day_counts in results: for _, _, day_counts in results:
for day, cnt in day_counts.items(): for day, cnt in day_counts.items():
merged_days[day] = merged_days.get(day, 0) + cnt merged_days[day] = merged_days.get(day, 0) + cnt
# Always mark has_data=True — the commits endpoint is reliable
_line_stats_cache[cache_key] = (_time.monotonic(), total_adds, total_dels, True, merged_days) _line_stats_cache[cache_key] = (_time.monotonic(), total_adds, total_dels, True, merged_days)
finally: finally:
_line_stats_fetching.discard(cache_key) _line_stats_fetching.discard(cache_key)
@ -419,13 +459,13 @@ async def get_forgejo_heatmap(
# Normalise base_url the same way get_forgejo_client() does, eagerly, # Normalise base_url the same way get_forgejo_client() does, eagerly,
# so the background task never touches a potentially-closed session. # so the background task never touches a potentially-closed session.
import re as _re import re as _re
repo_tuples: list[tuple[str, str, str, str | None]] = [] repo_tuples: list[tuple[str, str, str, str | None, UUID, UUID]] = []
for repo, conn in repos_with_conns: for repo, conn in repos_with_conns:
bu = (conn.base_url or "").rstrip("/") bu = (conn.base_url or "").rstrip("/")
if "/api/v1" in bu: if "/api/v1" in bu:
m = _re.match(r"(https?://[^/]+)", bu) m = _re.match(r"(https?://[^/]+)", bu)
bu = m.group(1).rstrip("/") if m else bu bu = m.group(1).rstrip("/") if m else bu
repo_tuples.append((repo.owner, repo.repo, bu, getattr(conn, "token", None))) repo_tuples.append((repo.owner, repo.repo, bu, getattr(conn, "token", None), repo.id, repo.organization_id))
_line_stats_fetching.add(cache_key) _line_stats_fetching.add(cache_key)
asyncio.create_task( asyncio.create_task(
@ -440,12 +480,30 @@ async def get_forgejo_heatmap(
has_line_stats = False has_line_stats = False
commit_day_counts = {} commit_day_counts = {}
# Heatmap grid: use commit-per-day counts when available; fall back to # Heatmap grid: prefer commit-per-day counts from DB (persistent, no restarts);
# issue-event counts (from DB) while the background fetch is in progress. # fall back to in-memory cache, then to issue-event counts while the first
if has_line_stats and commit_day_counts: # background fetch is in progress.
commit_db_rows = (
await session.exec(
select(ForgejoCommitDay.date, func.sum(ForgejoCommitDay.commit_count).label("cnt"))
.where(
ForgejoCommitDay.repository_id.in_(repo_ids),
ForgejoCommitDay.date >= line_stats_since.date(),
)
.group_by(ForgejoCommitDay.date)
.order_by(ForgejoCommitDay.date)
)
).all()
if commit_db_rows:
days_list = [HeatmapDay(date=str(day), count=int(cnt)) for day, cnt in commit_db_rows]
max_count = max((d.count for d in days_list), default=0)
elif has_line_stats and commit_day_counts:
# In-memory cache hit but DB write hasn't landed yet
days_list = [HeatmapDay(date=k, count=v) for k, v in sorted(commit_day_counts.items())] days_list = [HeatmapDay(date=k, count=v) for k, v in sorted(commit_day_counts.items())]
max_count = max((d.count for d in days_list), default=0) max_count = max((d.count for d in days_list), default=0)
else: else:
# First-ever load — fall back to issue-event counts from DB while background fetch runs
counts: dict[str, int] = {} counts: dict[str, int] = {}
created_rows = ( created_rows = (
@ -545,6 +603,81 @@ async def get_last_push(
return cached[1] if cached is not None else None return cached[1] if cached is not None else None
@router.get(
"/commit-activity",
response_model=list[CommitActivityDay],
summary="Per-day commit counts from persistent DB store",
description=(
"Returns commit counts per calendar day from the forgejo_commit_activity table. "
"Populated by the background line-stats fetch; accurate across restarts. "
"Scope by organization_id or repository_id; defaults to the caller's full org."
),
)
async def get_commit_activity(
organization_id: UUID | None = Query(None, description="Filter by organisation ID"),
repository_id: UUID | None = Query(None, description="Filter by a single repository ID"),
days: int = Query(default=90, ge=1, le=365, description="How many days back to return"),
session: AsyncSession = SESSION_DEP,
ctx: OrganizationContext = ORG_MEMBER_DEP,
) -> list[CommitActivityDay]:
"""Return aggregated commit-per-day counts from the DB."""
if organization_id and organization_id != ctx.organization.id:
raise HTTPException(status_code=status.HTTP_403_FORBIDDEN)
since = (utcnow() - timedelta(days=days)).date()
if repository_id:
# Verify repo belongs to this org
repo = (
await session.exec(
select(ForgejoRepository).where(
ForgejoRepository.id == repository_id,
ForgejoRepository.organization_id == ctx.organization.id,
)
)
).first()
if repo is None:
raise HTTPException(status_code=status.HTTP_404_NOT_FOUND)
rows = (
await session.exec(
select(ForgejoCommitDay.date, ForgejoCommitDay.commit_count)
.where(
ForgejoCommitDay.repository_id == repository_id,
ForgejoCommitDay.date >= since,
)
.order_by(ForgejoCommitDay.date)
)
).all()
else:
# Org-wide: aggregate across all repos
repo_ids_rows = (
await session.exec(
select(ForgejoRepository.id).where(
ForgejoRepository.organization_id == ctx.organization.id
)
)
).all()
repo_ids = list(repo_ids_rows)
if not repo_ids:
return []
rows = (
await session.exec(
select(
ForgejoCommitDay.date,
func.sum(ForgejoCommitDay.commit_count).label("commit_count"),
)
.where(
ForgejoCommitDay.repository_id.in_(repo_ids),
ForgejoCommitDay.date >= since,
)
.group_by(ForgejoCommitDay.date)
.order_by(ForgejoCommitDay.date)
)
).all()
return [CommitActivityDay(date=str(day), commit_count=int(cnt)) for day, cnt in rows]
def _zeroed_metrics() -> MetricsResponse: def _zeroed_metrics() -> MetricsResponse:
"""Return zeroed metrics for empty scopes.""" """Return zeroed metrics for empty scopes."""
return MetricsResponse( return MetricsResponse(

View File

@ -0,0 +1,34 @@
"""Per-repository per-day commit activity for heatmap and trend queries."""
from __future__ import annotations
from datetime import date, datetime
from uuid import UUID, uuid4
from sqlmodel import Field, Index, SQLModel
from app.core.time import utcnow
class ForgejoCommitDay(SQLModel, table=True):
"""One row per repository per calendar day recording commit count."""
__tablename__ = "forgejo_commit_activity" # pyright: ignore[reportAssignmentType]
id: UUID = Field(default_factory=uuid4, primary_key=True)
organization_id: UUID = Field(foreign_key="organizations.id", index=True)
repository_id: UUID = Field(foreign_key="forgejo_repositories.id", index=True)
date: date = Field(index=True)
commit_count: int = Field(default=0)
created_at: datetime = Field(default_factory=utcnow)
updated_at: datetime = Field(default_factory=utcnow)
__table_args__ = (
Index(
"ix_forgejo_commit_activity_repo_date",
"repository_id",
"date",
unique=True,
),
)

View File

@ -149,6 +149,13 @@ class HeatmapResponse(SQLModel):
has_line_stats: bool = False # False when Forgejo is still computing stats (HTTP 202) has_line_stats: bool = False # False when Forgejo is still computing stats (HTTP 202)
class CommitActivityDay(SQLModel):
"""Per-day commit count for a single repository or aggregated across an org."""
date: str # "YYYY-MM-DD"
commit_count: int
class MetricsResponse(SQLModel): class MetricsResponse(SQLModel):
"""Generic metrics response wrapper.""" """Generic metrics response wrapper."""

View File

@ -0,0 +1,62 @@
"""Add forgejo_commit_activity table for per-repository per-day commit counts.
Revision ID: b1c2d3e4f5a6
Revises: a1b2c3d4e5f9, a3c5e7f9b1d2
Create Date: 2026-05-24 00:00:00.000000
"""
from __future__ import annotations
import sqlalchemy as sa
from alembic import op
revision = "b1c2d3e4f5a6"
down_revision = ("a1b2c3d4e5f9", "a3c5e7f9b1d2")
branch_labels = None
depends_on = None
def upgrade() -> None:
op.create_table(
"forgejo_commit_activity",
sa.Column("id", sa.Uuid(), nullable=False),
sa.Column("organization_id", sa.Uuid(), nullable=False),
sa.Column("repository_id", sa.Uuid(), nullable=False),
sa.Column("date", sa.Date(), nullable=False),
sa.Column("commit_count", sa.Integer(), nullable=False, server_default="0"),
sa.Column("created_at", sa.DateTime(), nullable=False),
sa.Column("updated_at", sa.DateTime(), nullable=False),
sa.ForeignKeyConstraint(["organization_id"], ["organizations.id"]),
sa.ForeignKeyConstraint(["repository_id"], ["forgejo_repositories.id"]),
sa.PrimaryKeyConstraint("id"),
)
op.create_index(
"ix_forgejo_commit_activity_organization_id",
"forgejo_commit_activity",
["organization_id"],
)
op.create_index(
"ix_forgejo_commit_activity_repository_id",
"forgejo_commit_activity",
["repository_id"],
)
op.create_index(
"ix_forgejo_commit_activity_date",
"forgejo_commit_activity",
["date"],
)
op.create_index(
"ix_forgejo_commit_activity_repo_date",
"forgejo_commit_activity",
["repository_id", "date"],
unique=True,
)
def downgrade() -> None:
op.drop_index("ix_forgejo_commit_activity_repo_date", "forgejo_commit_activity")
op.drop_index("ix_forgejo_commit_activity_date", "forgejo_commit_activity")
op.drop_index("ix_forgejo_commit_activity_repository_id", "forgejo_commit_activity")
op.drop_index("ix_forgejo_commit_activity_organization_id", "forgejo_commit_activity")
op.drop_table("forgejo_commit_activity")