cli.py

"""hf-scanner CLI — command-line interface for hf-security-scanner v4.0.



Usage examples:

    hf-scanner scan https://huggingface.co/spaces/user/space

    hf-scanner scan ./my-project --format both --out ./results

    hf-scanner scan . --no-llm --severity-threshold WARNING

    hf-scanner scan . --create-baseline baseline.json

    hf-scanner scan . --baseline baseline.json --ignore-file .hfscanignore

    hf-scanner list-rules

    hf-scanner self-test

    hf-scanner version

"""
import json
import os
import sys
import tempfile
from pathlib import Path
from typing import Optional

# Ensure project root is on sys.path when run via `python cli.py`
sys.path.insert(0, str(Path(__file__).parent))

import typer
from typing_extensions import Annotated

from core import (
    apply_ignore_rules,
    bootstrap_binaries,
    filter_by_baseline,
    have_binary,
    load_baseline,
    parse_ignore_file,
    save_baseline,
)
from core.scanner import scan_repo
from report import generate_html_report, generate_sarif
from rules import ALL_LLM, ALL_PERFORMANCE, ALL_SECURITY

_VERSION = "4.0.0"

app = typer.Typer(
    name="hf-scanner",
    help="HF Security & Performance Scanner — scan HF Spaces and Git repos for vulnerabilities.",
    no_args_is_help=True,
)

# ── Exit codes ────────────────────────────────────────────────────────────────
EXIT_CLEAN = 0
EXIT_FINDINGS = 1
EXIT_ERROR = 2
EXIT_USAGE = 3

_SEV_RANK = {"ERROR": 4, "HIGH": 3, "WARNING": 2, "INFO": 1}


def _findings_exceed_threshold(findings: list, threshold: str) -> bool:
    thresh = _SEV_RANK.get(threshold.upper(), 2)
    return any(_SEV_RANK.get(f.get("severity", "INFO").upper(), 0) >= thresh
               for f in findings)


# ── Commands ──────────────────────────────────────────────────────────────────

@app.command()
def version() -> None:
    """Print the hf-scanner version."""
    typer.echo(f"hf-scanner {_VERSION}")


@app.command("list-rules")
def list_rules() -> None:
    """List all bundled Semgrep rule packs."""
    packs = ALL_SECURITY + ALL_PERFORMANCE + ALL_LLM
    typer.echo(f"{'Pack':<30}  {'Category':<14}  Path")
    typer.echo("-" * 70)
    for label, path, category in packs:
        typer.echo(f"{label:<30}  {category:<14}  {path}")


@app.command("self-test")
def self_test() -> None:
    """Check that all external tools are installed and executable."""
    tools = [
        ("semgrep",         "Static analysis (Python, JS, …)"),
        ("bandit",          "Python security linter"),
        ("detect-secrets",  "Secret detection"),
        ("pip-audit",       "Dependency CVE scanner"),
        ("ruff",            "Fast Python linter (perf rules)"),
        ("gitleaks",        "Git history secret scanner"),
        ("hadolint",        "Dockerfile linter"),
        ("agent-audit",     "OWASP Agentic Top 10 scanner"),
    ]
    ok = True
    typer.echo(f"{'Tool':<18}  {'Status':<8}  Description")
    typer.echo("-" * 60)
    for name, desc in tools:
        found = have_binary(name)
        status = "✓  ok" if found else "✗  MISSING"
        if not found:
            ok = False
        typer.echo(f"{name:<18}  {status:<8}  {desc}")

    typer.echo("")
    bootstrap_result = bootstrap_binaries()
    typer.echo(
        f"[bootstrap] gitleaks={bootstrap_result.get('gitleaks')},  "
        f"hadolint={bootstrap_result.get('hadolint')}"
    )

    if not ok:
        typer.echo("\nSome tools are missing. Run:  pip install hf-security-scanner[full]",
                   err=True)
        raise typer.Exit(code=EXIT_ERROR)


@app.command()
def scan(

    target: Annotated[str, typer.Argument(help="URL or local path to scan.")],  # noqa: AGENT-034

    format: Annotated[str, typer.Option(

        "--format", "-f",

        help="Output format: html | sarif | json | both (html+sarif).",

    )] = "both",

    out: Annotated[Optional[str], typer.Option(

        "--out", "-o",

        help="Output directory or file stem (without extension). Default: temp dir.",

    )] = None,

    security: Annotated[bool, typer.Option("--security/--no-security")] = True,

    llm: Annotated[bool, typer.Option("--llm/--no-llm")] = True,

    performance: Annotated[bool, typer.Option("--performance/--no-performance")] = True,

    deep_history: Annotated[bool, typer.Option("--deep-history")] = False,

    baseline: Annotated[Optional[str], typer.Option(

        "--baseline",

        help="Path to a baseline JSON file. Known findings will be suppressed.",

    )] = None,

    create_baseline: Annotated[Optional[str], typer.Option(

        "--create-baseline",

        help="Save current findings as a new baseline to this path.",

    )] = None,

    ignore_file: Annotated[str, typer.Option(

        "--ignore-file",

        help="Path to an .hfscanignore file.",

    )] = ".hfscanignore",

    severity_threshold: Annotated[str, typer.Option(

        "--severity-threshold",

        help="Exit 1 only when findings at or above this level exist. "

             "Choices: ERROR | WARNING | INFO.",

    )] = "WARNING",

    hf_token: Annotated[Optional[str], typer.Option(

        "--hf-token", envvar="HF_TOKEN",

        help="HuggingFace API token for private repos.",

    )] = None,

    quiet: Annotated[bool, typer.Option("--quiet/--verbose")] = False,

) -> None:
    """Scan a Git repo, HF Space, or local directory for security issues."""

    fmt = format.lower()
    if fmt not in ("html", "sarif", "json", "both"):
        typer.echo(f"Unknown format '{format}'. Choose html, sarif, json, or both.",
                   err=True)
        raise typer.Exit(code=EXIT_USAGE)

    if not quiet:
        typer.echo(f"[hf-scanner] scanning {target} …")

    try:
        findings, log_lines = scan_repo(
            target,
            hf_token=hf_token,
            deep_history=deep_history,
            run_security=security,
            run_performance=performance,
            run_llm=llm,
        )
    except Exception as exc:  # noqa: BLE001
        typer.echo(f"Scan error: {exc}", err=True)
        raise typer.Exit(code=EXIT_ERROR) from exc

    if not quiet:
        for line in log_lines:
            typer.echo(f"  {line}")

    # ── Baseline suppression ──────────────────────────────────────────────────
    suppressed_count = 0
    if baseline and os.path.isfile(baseline):
        known = load_baseline(baseline)
        findings, suppressed = filter_by_baseline(findings, known)
        suppressed_count = len(suppressed)
        if not quiet:
            typer.echo(f"[baseline] suppressed {suppressed_count} known findings")

    # ── .hfscanignore suppression ─────────────────────────────────────────────
    ignore_rules = parse_ignore_file(ignore_file)
    if ignore_rules:
        findings, ignored_count = apply_ignore_rules(findings, ignore_rules)
        if not quiet:
            typer.echo(f"[ignore] {ignored_count} findings suppressed by {ignore_file}")

    # ── Create baseline ───────────────────────────────────────────────────────
    if create_baseline:
        save_baseline(findings, create_baseline)
        if not quiet:
            typer.echo(f"[baseline] saved {len(findings)} fingerprints → {create_baseline}")

    # ── Determine output stem ─────────────────────────────────────────────────
    if out:
        out_path = Path(out)
        if out_path.is_dir():
            stem = out_path / "scan_report"
        else:
            stem = out_path
    else:
        stem = Path(tempfile.gettempdir()) / "scan_report"

    scan_meta = {
        "target": target,
        "n_targets": 1,
        "timestamp": __import__("datetime").datetime.now(
            __import__("datetime").timezone.utc
        ).isoformat() + "Z",
        "footer": f"hf-scanner {_VERSION}",
    }

    written = []

    if fmt in ("html", "both"):
        html_path = str(stem) + ".html"
        with open(html_path, "w", encoding="utf-8") as fh:
            fh.write(generate_html_report(findings, scan_meta))
        written.append(html_path)

    if fmt in ("sarif", "both"):
        sarif_path = str(stem) + ".sarif"
        with open(sarif_path, "w", encoding="utf-8") as fh:
            json.dump(generate_sarif(findings, scan_meta), fh, indent=2)
        written.append(sarif_path)

    if fmt == "json":
        json_path = str(stem) + ".json"
        with open(json_path, "w", encoding="utf-8") as fh:
            json.dump(findings, fh, indent=2, default=str)
        written.append(json_path)
        # Also print to stdout for piping
        typer.echo(json.dumps(findings, indent=2, default=str))

    if not quiet:
        typer.echo(f"\n[hf-scanner] {len(findings)} findings")
        for p in written:
            typer.echo(f"  → {p}")

    # ── Exit code ─────────────────────────────────────────────────────────────
    if _findings_exceed_threshold(findings, severity_threshold):
        raise typer.Exit(code=EXIT_FINDINGS)
    raise typer.Exit(code=EXIT_CLEAN)


def main() -> None:
    app()


if __name__ == "__main__":
    main()