gimme-job/gimme_job/cli.py

"""gimme-job CLI entry point."""
from __future__ import annotations

import sys
from pathlib import Path
from typing import Optional

import typer
from rich.console import Console
from rich.panel import Panel
from rich.table import Table

app = typer.Typer(
    name="gimme-job",
    help="Local job posting aggregator — AI-assisted learning, non-AI runtime.",
    add_completion=False,
)
console = Console()


# ── init ──────────────────────────────────────────────────────────────────────


@app.command()
def init(
    force: bool = typer.Option(False, "--force", help="Re-initialize even if already set up"),
) -> None:
    """Initialize project directories, database, and config."""
    from gimme_job.db.engine import init_db
    from gimme_job.utils.paths import ensure_workspace_dirs, project_root, sites_dir

    console.print(Panel("[bold cyan]gimme-job init[/bold cyan]", expand=False))

    # Create workspace directories
    ensure_workspace_dirs()
    console.print("[green]✓[/green] Workspace directories created")

    # Create sites/global.yaml if absent
    global_yaml = sites_dir() / "global.yaml"
    if not global_yaml.exists() or force:
        _write_default_global_yaml(global_yaml)
        console.print(f"[green]✓[/green] Created {global_yaml}")
    else:
        console.print(f"[dim]  {global_yaml} already exists[/dim]")

    # Copy .env.example -> .env if absent
    env_example = project_root() / ".env.example"
    env_file = project_root() / ".env"
    if env_example.exists() and not env_file.exists():
        import shutil
        shutil.copy(env_example, env_file)
        console.print(f"[green]✓[/green] Copied .env.example → .env (fill in your tokens)")
    else:
        console.print("[dim]  .env already exists[/dim]")

    # Initialize SQLite DB
    init_db()
    console.print("[green]✓[/green] SQLite database initialized")

    # Preflight checks
    _check_playwright()
    _check_ollama()
    _check_claude()

    console.print(Panel("[bold green]Initialization complete.[/bold green]", expand=False))


def _write_default_global_yaml(path: Path) -> None:
    from gimme_job.utils.json_io import write_yaml
    data = {
        "runtime": {
            "timezone": "America/Phoenix",
            "headless": True,
            "profile_name": "JobAgent",
            "slow_mo_ms": 0,
            "default_timeout_ms": 15000,
            "navigation_timeout_ms": 30000,
            "max_pages_per_site": 3,
            "min_delay_ms": 1200,
            "max_delay_ms": 3500,
        },
        "search_defaults": {
            "keywords": ["orthodontist"],
            "location": "",
            "remote": False,
            "date_mode": "today_or_last_24h",
            "sort": "relevance",
            "max_items_per_site": 30,
        },
        "summarization": {
            "ollama_base_url": "http://127.0.0.1:11434",
            "model": "qwen3.5:9b",
            "temperature": 0.1,
            "max_input_items": 200,
        },
        "notification": {
            "provider": "kakaotalk",
            "fallback_markdown": True,
        },
    }
    write_yaml(path, data)


def _check_playwright() -> None:
    try:
        from playwright.sync_api import sync_playwright
        with sync_playwright() as p:
            _ = p.chromium
        console.print("[green]✓[/green] Playwright available")
    except Exception as e:
        console.print(f"[yellow]![/yellow] Playwright check failed: {e}")
        console.print("  Run: [bold]uv run playwright install chromium[/bold]")


def _check_ollama() -> None:
    try:
        import httpx
        from gimme_job.config import load_global_config
        cfg = load_global_config()
        r = httpx.get(f"{cfg.summarization.ollama_base_url}/api/tags", timeout=3.0)
        if r.status_code == 200:
            console.print("[green]✓[/green] Ollama reachable")
        else:
            console.print(f"[yellow]![/yellow] Ollama returned {r.status_code}")
    except Exception as e:
        console.print(f"[yellow]![/yellow] Ollama not reachable: {e}")
        console.print("  Make sure Ollama is running: [bold]ollama serve[/bold]")


def _check_claude() -> None:
    import subprocess
    try:
        result = subprocess.run(
            ["claude", "--version"], capture_output=True, text=True, timeout=10
        )
        if result.returncode == 0:
            console.print(f"[green]✓[/green] Claude Code CLI: {result.stdout.strip()}")
        else:
            console.print("[yellow]![/yellow] Claude Code CLI not found (needed for learn/repair)")
    except FileNotFoundError:
        console.print("[yellow]![/yellow] Claude Code CLI not found (needed for learn/repair)")


# ── run ───────────────────────────────────────────────────────────────────────


@app.command()
def run(
    site: Optional[str] = typer.Option(None, "--site", help="Run a single site only"),
    dry_run: bool = typer.Option(False, "--dry-run", help="Extract but do not save or notify"),
    skip_notify: bool = typer.Option(False, "--skip-notify", help="Skip KakaoTalk notification"),
    today_only: bool = typer.Option(False, "--today-only", help="Only process today's new items"),
    limit_sites: Optional[int] = typer.Option(None, "--limit-sites", help="Max sites to run"),
) -> None:
    """Collect job postings from all enabled sites."""
    from gimme_job.config import load_global_config
    from gimme_job.db.engine import get_engine, get_session_factory
    from gimme_job.runtime.orchestrator import RunOrchestrator

    cfg = load_global_config()
    engine = get_engine()
    factory = get_session_factory(engine)

    orchestrator = RunOrchestrator(global_config=cfg, session_factory=factory)
    summary = orchestrator.run_all(
        site_filter=site,
        dry_run=dry_run,
        skip_notify=skip_notify,
        limit_sites=limit_sites,
    )

    # Print summary table
    table = Table(title="Run Summary", show_header=True)
    table.add_column("Site", style="cyan")
    table.add_column("Status")
    table.add_column("Found", justify="right")
    table.add_column("New", justify="right")

    for r in summary.site_results:
        status_color = {
            "success": "green",
            "partial": "yellow",
            "failed": "red",
            "repair_needed": "red",
        }.get(r.status.value, "white")
        table.add_row(
            r.site_id,
            f"[{status_color}]{r.status.value}[/{status_color}]",
            str(r.items_found),
            str(r.new_items),
        )

    console.print(table)
    console.print(f"\nTotal: [bold]{summary.total_found}[/bold] found, [bold green]{summary.total_new}[/bold green] new")

    if summary.summary_text:
        console.print("\n[bold]Summary sent.[/bold]")


# ── learn ─────────────────────────────────────────────────────────────────────


@app.command()
def learn(
    site_id: str = typer.Option(..., "--site-id", help="Site identifier (e.g. linkedin)"),
    url: str = typer.Option(..., "--url", help="Job search results URL to learn from"),
    profile_name: str = typer.Option("JobAgent", "--profile-name"),
    headed: bool = typer.Option(True, "--headed/--headless"),
    timeout_seconds: int = typer.Option(180, "--timeout-seconds"),
) -> None:
    """Learn a new job site adapter using Claude Code."""
    from gimme_job.config import load_global_config
    from gimme_job.runtime.claude_cli import ClaudeCodeClient
    from gimme_job.runtime.learn import LearnMode

    cfg = load_global_config()
    client = ClaudeCodeClient()
    mode = LearnMode(global_config=cfg, claude_client=client)

    console.print(Panel(f"[bold cyan]Learning site: {site_id}[/bold cyan]", expand=False))
    success = mode.learn_site(
        site_id=site_id,
        url=url,
        profile_name=profile_name,
        headed=headed,
        timeout=timeout_seconds,
    )

    if success:
        console.print(f"[green]✓[/green] Successfully learned [bold]{site_id}[/bold]")
    else:
        console.print(f"[red]✗[/red] Failed to learn [bold]{site_id}[/bold]")
        raise typer.Exit(1)


# ── repair ────────────────────────────────────────────────────────────────────


@app.command()
def repair(
    site: Optional[str] = typer.Argument(None, help="Site ID to repair"),
    all_: bool = typer.Option(False, "--all", help="Repair all repair_needed sites"),
) -> None:
    """Repair a broken site adapter using Claude Code."""
    from gimme_job.config import load_global_config, list_all_sites
    from gimme_job.runtime.claude_cli import ClaudeCodeClient
    from gimme_job.runtime.repair import RepairMode

    if not site and not all_:
        console.print("[red]Provide a site ID or --all[/red]")
        raise typer.Exit(1)

    cfg = load_global_config()
    client = ClaudeCodeClient()
    mode = RepairMode(global_config=cfg, claude_client=client)

    sites_to_repair = list_all_sites() if all_ else [site]

    for s in sites_to_repair:
        console.print(f"[cyan]Repairing {s}...[/cyan]")
        success = mode.repair_site(s)
        if success:
            console.print(f"[green]✓[/green] {s} repaired")
        else:
            console.print(f"[red]✗[/red] {s} repair failed")


# ── test ──────────────────────────────────────────────────────────────────────


@app.command()
def test(
    site: str = typer.Argument(..., help="Site ID to test"),
) -> None:
    """Run smoke test for a site adapter."""
    import subprocess

    console.print(Panel(f"[bold cyan]Testing: {site}[/bold cyan]", expand=False))
    result = subprocess.run(
        ["uv", "run", "pytest", f"tests/adapters/test_{site}.py", "-v"],
        cwd=str(Path(__file__).parent.parent),
    )
    raise typer.Exit(result.returncode)


# ── notify ────────────────────────────────────────────────────────────────────


@app.command()
def notify(
    today: bool = typer.Option(True, "--today/--no-today", help="Re-send today's digest"),
) -> None:
    """Re-send today's job summary via KakaoTalk."""
    from datetime import date

    from gimme_job.config import load_global_config
    from gimme_job.db.engine import get_engine, get_session_factory
    from gimme_job.db.repo import JobPostingRepo, SummaryRepo
    from gimme_job.runtime.notifier import NotificationDispatcher

    cfg = load_global_config()
    engine = get_engine()
    factory = get_session_factory(engine)
    run_date = date.today()

    with factory() as session:
        summary_repo = SummaryRepo()
        summary = summary_repo.get_latest(session, run_date)

        if summary:
            text = summary.content
        else:
            # Build summary from today's postings
            postings = JobPostingRepo().get_today_new(session, run_date)
            if not postings:
                console.print("[yellow]No new postings found for today.[/yellow]")
                raise typer.Exit(0)
            from gimme_job.runtime.summarizer import OllamaSummarizer
            summarizer = OllamaSummarizer(
                base_url=cfg.summarization.ollama_base_url,
                model=cfg.summarization.model,
                temperature=cfg.summarization.temperature,
            )
            text = summarizer.summarize(postings)

    dispatcher = NotificationDispatcher(global_config=cfg, session_factory=factory)
    ok = dispatcher.send(text, run_date)
    if ok:
        console.print("[green]✓[/green] Notification sent")
    else:
        console.print("[yellow]![/yellow] Notification failed — saved as markdown fallback")


# ── status ────────────────────────────────────────────────────────────────────


@app.command()
def status() -> None:
    """Show health status of all configured sites."""
    from gimme_job.config import list_all_sites, load_site_manifest
    from gimme_job.db.engine import get_engine, get_session_factory
    from gimme_job.db.repo import SiteConfigRepo, SiteRunRepo

    engine = get_engine()
    factory = get_session_factory(engine)

    table = Table(title="Site Status", show_header=True)
    table.add_column("Site", style="cyan")
    table.add_column("Enabled")
    table.add_column("Repair?")
    table.add_column("Last Run")
    table.add_column("Last Status")
    table.add_column("Failures", justify="right")

    run_repo = SiteRunRepo()
    config_repo = SiteConfigRepo()

    with factory() as session:
        for site_id in list_all_sites():
            try:
                manifest = load_site_manifest(site_id)
            except Exception:
                continue

            config = session.get(__import__("gimme_job.models.db", fromlist=["SiteConfig"]).SiteConfig, site_id)
            recent = run_repo.get_recent_runs(session, site_id, limit=1)

            enabled = "[green]yes[/green]" if manifest.enabled else "[dim]no[/dim]"
            repair = "[red]YES[/red]" if manifest.repair_needed else "[green]no[/green]"
            last_run = recent[0].started_at.strftime("%m-%d %H:%M") if recent else "[dim]never[/dim]"
            last_status = recent[0].status if recent else "[dim]-[/dim]"
            failures = str(config.consecutive_failures) if config else "0"

            table.add_row(site_id, enabled, repair, last_run, last_status, failures)

    console.print(table)


if __name__ == "__main__":
    app()