politics-tracking/backend/politrack/cli.py

"""`politrack` command line. Run `politrack --help` for all commands."""

from __future__ import annotations

import json
import logging
from datetime import datetime
from pathlib import Path

import typer

app = typer.Typer(help="Politrack command line", no_args_is_help=True)
sources_app = typer.Typer(help="Source specs and adapters")
app.add_typer(sources_app, name="sources")
logging.basicConfig(level=logging.INFO, format="%(asctime)s %(levelname)s %(name)s: %(message)s")


def _params(values: list[str]) -> dict:
    """--param key=value (value parsed as JSON when possible)."""
    out = {}
    for item in values:
        key, _, raw = item.partition("=")
        try:
            out[key] = json.loads(raw)
        except json.JSONDecodeError:
            out[key] = raw
    return out


@app.command()
def migrate() -> None:
    """Apply SQL migrations, sync source specs and install the job-queue schema."""
    from politrack.core.registry import sync_sources
    from politrack.db import connection
    from politrack.migrate import migrate as run_migrations

    run_migrations()
    with connection() as conn:
        typer.echo(f"synced {sync_sources(conn)} sources")
        installed = conn.execute("SELECT to_regclass('procrastinate_jobs') AS t").fetchone()["t"]
    if not installed:
        from politrack.worker import app as jobs

        with jobs.open():
            jobs.schema_manager.apply_schema()


@sources_app.command("list")
def sources_list() -> None:
    """List adapters, enrichers and their schedules."""
    from politrack.core import registry

    registry.load_builtin()
    for name, cls in sorted({**registry.ADAPTERS, **registry.ENRICHERS}.items()):
        typer.echo(f"{name:34} {cls.queue:7} {cls.schedule or 'manual':16} {cls.spec.slug:22} {cls.description}")


@sources_app.command("sync")
def sources_sync() -> None:
    """Upsert SourceSpecs into the database."""
    from politrack.core.registry import sync_sources
    from politrack.db import connection

    with connection() as conn:
        typer.echo(f"synced {sync_sources(conn)} sources")


@app.command("run")
def run_cmd(
    name: str,
    param: list[str] = typer.Option([], "--param", "-p", help="key=value, e.g. -p periods='[\"XXVIII\"]'"),
    since: str = typer.Option(None, help="ISO date for incremental adapters"),
    max_requests: int = typer.Option(None, help="Stop after N requests (testing)"),
) -> None:
    """Run one adapter or enricher now, e.g. `politrack run parlament.persons`."""
    from politrack.core.runner import run

    result = run(name, params=_params(param), since=datetime.fromisoformat(since) if since else None,
                 max_requests=max_requests)
    typer.echo(json.dumps(result, indent=2, default=str))


@app.command()
def reparse(name: str, param: list[str] = typer.Option([], "--param", "-p")) -> None:
    """Re-run parse() on the archived responses of an adapter (after a parser fix, no network)."""
    from politrack.core.runner import run

    typer.echo(json.dumps(run(name, mode="reparse", params=_params(param)), indent=2, default=str))


@app.command()
def enrich() -> None:
    """Run all enrichers once (linking, events, legal, topics, roll calls, corroboration, embeddings)."""
    from politrack.core import registry
    from politrack.core.runner import run

    registry.load_builtin()
    for name in ("enrich.linker", "enrich.events", "enrich.legal", "enrich.topics", "enrich.roles",
                 "enrich.rollcall", "enrich.corroborate", "enrich.embed"):
        typer.echo(f"{name}: {run(name)}")


@app.command()
def status(limit: int = 30) -> None:
    """Show recent runs."""
    from politrack.db import connection

    with connection() as conn:
        for r in conn.execute("SELECT adapter, mode, status, started_at, finished_at, stats FROM run "
                              "ORDER BY id DESC LIMIT %s", (limit,)):
            took = (r["finished_at"] - r["started_at"]).seconds if r["finished_at"] else "…"
            typer.echo(f"{r['started_at']:%Y-%m-%d %H:%M} {r['adapter']:32} {r['status']:8} {took}s "
                       f"{json.dumps(r['stats'])[:120]}")


@app.command()
def review(
    approve: list[int] = typer.Option([], help="Claim ids to publish"),
    reject: list[int] = typer.Option([], help="Claim ids to reject"),
    reviewer: str = typer.Option("cli", help="Your name, stored with the decision"),
) -> None:
    """List claims waiting for review (legal matters detected in headlines) or approve/reject them."""
    from politrack import queries
    from politrack.db import connection

    with connection() as conn:
        for claim_id in approve:
            queries.set_review(conn, claim_id, "published", reviewer)
        for claim_id in reject:
            queries.set_review(conn, claim_id, "rejected", reviewer)
        if approve or reject:
            typer.echo(f"published {len(approve)}, rejected {len(reject)}")
            return
        for c in queries.pending_claims(conn, limit=100):
            typer.echo(f"[{c['id']}] {c['occurred_at']:%Y-%m-%d} {c['full_name']}: {c['summary_de']}")
            for ev in c["evidence"] or []:
                typer.echo(f"      {ev['title']}  {ev['url']}")


@app.command()
def fixture(name: str, url: str, case: str, kind: str = "default", meta: str = "{}", method: str = "GET",
            body: str = typer.Option(None, help="JSON request body for POST")) -> None:
    """Capture a live response as a golden-test fixture (tests/fixtures/<adapter>/<case>.*)."""
    import httpx

    from politrack.config import get_settings

    out = Path(__file__).resolve().parent.parent / "tests" / "fixtures" / name
    out.mkdir(parents=True, exist_ok=True)
    resp = httpx.request(method, url, json=json.loads(body) if body else None, timeout=120, follow_redirects=True,
                         headers={"User-Agent": get_settings().user_agent})
    resp.raise_for_status()
    suffix = {"application/json": "json", "text/html": "html", "text/xml": "xml", "application/xml": "xml"}.get(
        resp.headers.get("content-type", "").split(";")[0], "bin")
    (out / f"{case}.{suffix}").write_bytes(resp.content)
    (out / f"{case}.request.json").write_text(json.dumps(
        {"url": url, "kind": kind, "meta": json.loads(meta), "method": method, "json": json.loads(body) if body else None,
         "content_type": resp.headers.get("content-type", ""), "body_file": f"{case}.{suffix}"}, indent=2), encoding="utf-8")
    typer.echo(f"saved {out / case}.* - now run: UPDATE_GOLDEN=1 pytest tests/test_golden.py -k {case}")


@app.command()
def openapi(json_out: Path = Path("openapi.json"), ts_out: Path = typer.Option(None, "--ts")) -> None:
    """Export the OpenAPI schema and (optionally) TypeScript types for the web app."""
    from politrack.api.main import app as api_app
    from politrack.openapi_ts import to_typescript

    schema = api_app.openapi()
    json_out.write_text(json.dumps(schema, indent=2, ensure_ascii=False), encoding="utf-8")
    if ts_out:
        ts_out.write_text(to_typescript(schema), encoding="utf-8")
    typer.echo(f"wrote {json_out}" + (f" and {ts_out}" if ts_out else ""))


@app.command()
def api(host: str = "0.0.0.0", port: int = 8000) -> None:
    import uvicorn

    uvicorn.run("politrack.api.main:app", host=host, port=port, proxy_headers=True)


@app.command()
def mcp(host: str = "0.0.0.0", port: int = 8001) -> None:
    """Serve the MCP endpoint for AI agents (streamable HTTP at /mcp)."""
    from politrack.mcp_server import serve

    serve(host, port)


if __name__ == "__main__":
    app()