headquarter/apps/api/src/services/tool/instance_service.py

"""Tool instance service functions."""

import asyncio
import glob as glob_module
import logging
import os
import subprocess
import uuid
from datetime import datetime
from typing import Any

import httpx
from fastapi import APIRouter, HTTPException, status
from sqlalchemy import select
from sqlalchemy.ext.asyncio import AsyncSession

from src.models import ConfigProfile, GitRepository, SSHKey, ToolInstance, ToolType
from src.services.git.clone import check_dirty_state, clone_repository
from src.services.config.config_profile_resolver import (
    ConfigProfileCycleError,
    ResolvedProfile,
    apply_resolved_profile,
    expand_container_path,
    resolve_profile,
)
from src.services.docker import (
    connect_container_to_network,
    ensure_instance_directory,
    execute_compose_command,
    find_free_port,
    get_backend_network_name,
    get_container_id,
    get_container_ip_on_network,
    get_container_logs,
    get_container_status,
    is_container_on_network,
    render_compose_template,
    sort_volumes_by_specificity,
    wait_for_container_running,
    write_compose_file,
    write_config_files,
    write_env_file,
)
from src.services.shared.tunnel import (
    check_tunnel_health,
    recreate_tunnel,
    start_tunnel,
    stop_tunnel,
)
from src.services.build.docker_build import build_image
from src.services.build.manifest_compiler import (
    compile_compose,
    compile_dockerfile,
    compile_entrypoint,
    compute_image_tag,
    deep_merge,
    get_manifest_home_dir,
    merge_with_config,
    resolve_base,
)
from src.services.shared.permission_fixer import (
    apply_mount_permissions,
    apply_ssh_permissions,
)
from src.services.shared.readiness_probe import execute_probe
from src.services.shared.ssh_keys import cleanup_ssh_key_files, prepare_ssh_key_files
from src.services.instance.event_bus import InstanceEventBus
from src.services.instance.lifecycle_hooks import publish_lifecycle_event
from src.auth.dependencies import _get_owned_project, _get_user

logger = logging.getLogger(__name__)
_event_bus = InstanceEventBus()

async def resolve_git_mounts(
    session: AsyncSession,
    resolved: ResolvedProfile,
    instance_dir: str | None = None,
    working_directory: str | None = None,
    home_dir: str = "/root",
) -> list[dict]:
    """Convert git mounts from resolved profile to Docker volume mounts.

    Looks up repository paths, auto-clones if needed, handles branch checkout,
    expands glob patterns, and prepares bind mount entries.
    Logs warnings for missing repos or invalid paths (non-blocking).
    """
    if not resolved.git_mounts:
        return []

    # Process all git mounts concurrently
    tasks = []
    for git_mount in resolved.git_mounts:
        tasks.append(
            resolve_single_git_mount(
                session, git_mount, instance_dir, working_directory, home_dir
            )
        )

    results = await asyncio.gather(*tasks, return_exceptions=True)

    volume_mounts: list[dict[str, Any]] = []
    for result in results:
        if isinstance(result, Exception):
            logger.warning("Git mount failed: %s", result)
            continue
        if isinstance(result, list):
            volume_mounts.extend(result)

    return volume_mounts


def normalize_git_mount(entry: dict) -> dict:
    """Normalize a git mount entry to the unified mappings format.

    Converts legacy source_path + target_path into a single-entry mappings array.
    """
    entry = dict(entry)
    if "mappings" not in entry or not entry.get("mappings"):
        source = entry.get("source_path", ".")
        target = entry.get("target_path")
        if target is not None:
            entry["mappings"] = [{"source_path": source, "target_path": target}]
    entry.pop("source_path", None)
    entry.pop("target_path", None)
    return entry


def clone_git_repo(
    remote_url: str,
    branch: str | None,
    clone_parent: str,
) -> str:
    """Clone or pull a git repository.

    Returns the path to the cloned repo (repo-clone directory).
    """
    import hashlib

    url_hash = hashlib.md5(remote_url.encode()).hexdigest()[:12]
    repo_name = remote_url.split("/")[-1].replace(".git", "") or "repo"
    clone_dir = os.path.join(clone_parent, "git-mounts", f"{repo_name}-{url_hash}")
    repo_path = os.path.join(clone_dir, "repo-clone")

    if not os.path.exists(repo_path):
        try:
            os.makedirs(clone_dir, exist_ok=True)
            repo_path = clone_repository(
                remote_url,
                None,  # No SSH key for now - can be added later
                clone_dir,
                branch or "main",
            )
            logger.debug("Cloned git mount repository %s to %s", remote_url, repo_path)
        except Exception as exc:
            logger.warning("Clone failed for git mount %s: %s", remote_url, exc)
            raise
    else:
        # Repo exists - pull latest updates
        try:
            pull_repository_updates(repo_path, remote_url)
            logger.debug("Pulled updates for git mount %s", remote_url)
        except Exception as exc:
            logger.warning("Failed to pull updates for %s: %s", remote_url, exc)

    # Handle branch checkout if specified
    if branch and repo_path:
        success = checkout_branch(repo_path, branch)
        if success:
            logger.debug("Checked out branch %s for %s", branch, remote_url)
        else:
            logger.warning(
                "Branch %s not found in %s, using current branch", branch, remote_url
            )

    return repo_path


def resolve_git_mount_mappings(
    repo_path: str,
    mappings: list[dict],
    working_directory: str | None,
    home_dir: str = "/root",
) -> list[dict]:
    """Resolve mappings from an already-cloned repo to volume mount entries.

    Returns a flat list of volume mount dicts.
    """
    volume_mounts = []

    for mapping in mappings:
        source_path = mapping.get("source_path", ".")
        target_path = mapping.get("target_path")

        if not target_path:
            logger.warning("Invalid mapping skipped: missing target_path")
            continue

        # Expand ~ and $HOME in target path
        target_path = expand_container_path(target_path, home_dir)

        # Resolve relative target paths against working directory
        final_target = target_path
        if not target_path.startswith("/"):
            if not working_directory:
                logger.warning(
                    "Git mount skipped: target_path '%s' is relative but no working_directory is configured. "
                    "Set working_directory in the tool config or use an absolute path.",
                    target_path,
                )
                continue
            final_target = os.path.join(working_directory, target_path)

        # Build source path and expand globs
        if source_path and source_path != ".":
            source_full = os.path.join(repo_path, source_path)
        else:
            source_full = repo_path

        # Expand glob patterns
        matched_paths = expand_glob_source(source_full, repo_path)

        if not matched_paths:
            logger.warning(
                "Git mount skipped: no files matched source path %s in repo",
                source_path,
            )
            continue

        for matched_path in matched_paths:
            if not os.path.exists(matched_path):
                continue

            # Determine target path for this match
            if len(matched_paths) == 1:
                # Single match: mount directly to target_path
                mount_target = final_target
            else:
                # Multiple matches: append relative path to target
                rel_path = os.path.relpath(matched_path, repo_path)
                mount_target = os.path.join(final_target, rel_path)

            volume_mounts.append(
                {
                    "source": matched_path,
                    "target": mount_target,
                    "type": "bind",
                }
            )
            logger.debug(
                "Added git mount: %s -> %s",
                matched_path,
                mount_target,
            )

    return volume_mounts


async def resolve_single_git_mount(
    session: AsyncSession,
    git_mount: dict,
    instance_dir: str | None = None,
    working_directory: str | None = None,
    home_dir: str = "/root",
) -> list[dict]:
    """Resolve a single git mount to volume mount entries.

    Clones directly from remote_url, no database lookup needed.
    Returns a list of volume mounts (one for each matched file/directory).
    """
    git_mount = normalize_git_mount(git_mount)
    remote_url = git_mount.get("remote_url")
    branch = git_mount.get("branch")
    mappings = git_mount.get("mappings", [])

    if not remote_url:
        logger.warning("Invalid git mount skipped: missing remote_url")
        return []

    if not mappings:
        logger.warning("Invalid git mount skipped: no mappings")
        return []

    if not instance_dir:
        logger.warning("Git mount skipped: no instance_dir provided for cloning")
        return []

    # Clone or pull the repository
    try:
        repo_path = await asyncio.to_thread(
            clone_git_repo, remote_url, branch, instance_dir
        )
    except Exception:
        return []

    # Resolve all mappings from the cloned repo
    return resolve_git_mount_mappings(repo_path, mappings, working_directory, home_dir)


def checkout_branch(repo_path: str, branch: str) -> bool:
    """Checkout a specific branch in a git repository.

    Returns True if checkout succeeded, False if it failed.
    On failure, the repository remains on its current branch.
    """
    import subprocess

    # First try to checkout existing branch
    result = subprocess.run(
        ["git", "-C", repo_path, "checkout", branch],
        capture_output=True,
        text=True,
    )

    if result.returncode != 0:
        # Try fetching and checking out
        subprocess.run(
            ["git", "-C", repo_path, "fetch", "origin", branch],
            capture_output=True,
            text=True,
        )
        result = subprocess.run(
            ["git", "-C", repo_path, "checkout", "-b", branch, f"origin/{branch}"],
            capture_output=True,
            text=True,
        )

        if result.returncode != 0:
            logger.warning(
                "Failed to checkout branch %s in %s: %s",
                branch,
                repo_path,
                result.stderr.strip(),
            )
            return False

    return True


def pull_repository_updates(repo_path: str, remote_url: str) -> None:
    """Pull latest updates from remote repository.

    Used when starting a new container with an existing cloned repository
    to ensure the latest code is mounted.
    """
    import subprocess

    # Fetch latest changes
    result = subprocess.run(
        ["git", "-C", repo_path, "fetch", "origin"],
        capture_output=True,
        text=True,
    )

    if result.returncode != 0:
        raise RuntimeError(f"Failed to fetch updates: {result.stderr}")

    # Pull changes for current branch
    result = subprocess.run(
        ["git", "-C", repo_path, "pull", "origin"],
        capture_output=True,
        text=True,
    )

    if result.returncode != 0:
        raise RuntimeError(f"Failed to pull updates: {result.stderr}")


def expand_glob_source(source_path: str, repo_path: str) -> list[str]:
    """Expand glob patterns in source path.

    Returns a list of matched absolute paths.
    Limits results to prevent abuse.
    """
    MAX_GLOB_MATCHES = 100

    # Check if path contains glob characters
    if not any(c in source_path for c in "*?["):
        # No glob pattern: return single path if it exists
        return [source_path] if os.path.exists(source_path) else []

    # Expand glob pattern
    matched = glob_module.glob(source_path, recursive=True)
    total_matched = len(matched)

    # Filter to only paths within the repo and limit count
    results = []
    for path in matched:
        abs_path = os.path.abspath(path)
        if abs_path.startswith(os.path.abspath(repo_path)):
            results.append(abs_path)
        if len(results) >= MAX_GLOB_MATCHES:
            logger.warning(
                "Glob pattern matched %d files, limited to %d",
                total_matched,
                MAX_GLOB_MATCHES,
            )
            break

    return results


router = APIRouter(prefix="/projects", tags=["tool-instances"])


async def validate_config_profile(
    session: AsyncSession,
    profile_id: str | None,
    user_id: uuid.UUID,
    project_id: uuid.UUID,
    tool_type_id: uuid.UUID,
) -> uuid.UUID | None:
    """Validate a config profile selection.

    Args:
        session: Database session.
        profile_id: Profile ID string or None.
        user_id: Authenticated user ID.
        project_id: Project ID for compatibility check.
        tool_type_id: Tool type ID for compatibility check.

    Returns:
        Validated UUID or None.

    Raises:
        HTTPException: If profile is not found, not owned, or incompatible.
    """
    if not profile_id:
        return None

    try:
        profile_uuid = uuid.UUID(profile_id)
    except ValueError:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail=f"Invalid config profile ID: {profile_id}",
        )

    profile = await session.get(ConfigProfile, profile_uuid)
    if profile is None:
        raise HTTPException(
            status_code=status.HTTP_404_NOT_FOUND,
            detail=f"Config profile not found: {profile_id}",
        )

    if profile.user_id != user_id:
        raise HTTPException(
            status_code=status.HTTP_403_FORBIDDEN,
            detail="Not authorized to use this config profile",
        )

    # Check compatibility: profile must be portable or match project/tool
    is_compatible = (
        (profile.project_id is None and profile.tool_type_id is None)
        or (profile.project_id == project_id)
        or (profile.tool_type_id == tool_type_id)
        or (profile.project_id == project_id and profile.tool_type_id == tool_type_id)
    )

    if not is_compatible:
        raise HTTPException(
            status_code=status.HTTP_400_BAD_REQUEST,
            detail="Selected config profile is not compatible with this project and tool type",
        )

    return profile_uuid


def sanitize_compose_file(compose_path: str) -> None:
    """Remove invalid port mappings (target port 0) from compose file."""
    import yaml
    from pathlib import Path

    compose_file = Path(compose_path)
    if not compose_file.exists():
        return

    content = compose_file.read_text()
    compose_data = yaml.safe_load(content)

    if not compose_data or "services" not in compose_data:
        return

    modified = False
    for service_name, service_config in compose_data["services"].items():
        if "ports" in service_config:
            valid_ports = []
            for port_mapping in service_config["ports"]:
                if isinstance(port_mapping, str) and ":" in port_mapping:
                    parts = port_mapping.split(":")
                    if len(parts) == 2:
                        host_port, container_port = parts
                        # Skip invalid mappings (target port 0 or empty)
                        if container_port == "0" or not container_port:
                            modified = True
                            continue
                valid_ports.append(port_mapping)

            if valid_ports:
                service_config["ports"] = valid_ports
            else:
                del service_config["ports"]
                modified = True
        break  # Only check first service

    if modified:
        compose_file.write_text(yaml.dump(compose_data, default_flow_style=False))


def modify_compose_file(
    compose_path: str,
    port_override: int | None = None,
    start_command: str | None = None,
    working_directory: str | None = None,
    extra_volumes: list[dict] | None = None,
    home_dir: str = "/root",
) -> None:
    """Modify compose file with runtime overrides."""
    import yaml
    from pathlib import Path

    compose_file = Path(compose_path)
    content = compose_file.read_text()
    compose_data = yaml.safe_load(content)

    if not compose_data or "services" not in compose_data:
        return

    # Apply modifications to the first service
    for service_name, service_config in compose_data["services"].items():
        if port_override and "ports" in service_config:
            # Update port mapping
            for i, port_mapping in enumerate(service_config["ports"]):
                if isinstance(port_mapping, str) and ":" in port_mapping:
                    host_port, container_port = port_mapping.split(":", 1)
                    service_config["ports"][i] = f"{port_override}:{container_port}"
                    break

        if start_command:
            service_config["command"] = start_command

        if working_directory:
            service_config["working_dir"] = expand_container_path(
                working_directory, home_dir
            )

        if extra_volumes:
            if "volumes" not in service_config:
                service_config["volumes"] = []
            for vol in extra_volumes:
                source = vol.get("source", "")
                target = expand_container_path(vol.get("target", ""), home_dir)
                vol_type = vol.get("type", "bind")
                if vol_type == "bind":
                    service_config["volumes"].append(f"{source}:{target}")
                else:
                    service_config["volumes"].append(f"{source}:{target}:{vol_type}")

        # Sort volumes so parent paths come before child paths
        if service_config.get("volumes"):
            service_config["volumes"] = sort_volumes_by_specificity(
                service_config["volumes"]
            )

        break  # Only modify the first service

    # Write back
    compose_file.write_text(yaml.dump(compose_data, default_flow_style=False))


def ensure_container_name_in_compose(compose_path: str, container_name: str) -> None:
    """Ensure compose file has explicit container_name for predictable naming.

    Docker Compose auto-generates container names from the project directory
    when container_name is absent. This breaks tunnel connectivity because
    get_container_name(instance.name) cannot find the container. We inject
    container_name into every service so the container has a predictable name.
    """
    import yaml
    from pathlib import Path

    compose_file = Path(compose_path)
    if not compose_file.exists():
        return

    content = compose_file.read_text()
    compose_data = yaml.safe_load(content)

    if not compose_data or "services" not in compose_data:
        return

    modified = False
    for svc_name, svc_config in compose_data["services"].items():
        if "container_name" not in svc_config:
            svc_config["container_name"] = container_name.lower()
            modified = True

    if modified:
        compose_file.write_text(yaml.dump(compose_data, default_flow_style=False))
        logger.info(
            "Injected container_name '%s' into compose file",
            container_name.lower(),
        )


def ensure_web_bind_address(
    compose_path: str, tool_type_name: str, default_port: int
) -> None:
    """Auto-inject bind address for known web tools that default to 127.0.0.1.

    Many web tools (code-server, jupyter) bind to localhost by default,
    making them inaccessible from the Docker network. This function detects
    known tool images and injects the correct --bind-addr or --ip flag.
    """
    import yaml
    from pathlib import Path

    if default_port <= 0:
        return

    KNOWN_BIND_FIXES: dict[str, str] = {
        "code-server": f"--bind-addr 0.0.0.0:{default_port}",
        "jupyter-notebook": f"start-notebook.sh --ip=0.0.0.0 --port={default_port} --no-browser",
    }

    bind_command = KNOWN_BIND_FIXES.get(tool_type_name)
    if not bind_command:
        return

    compose_file = Path(compose_path)
    if not compose_file.exists():
        return

    content = compose_file.read_text()
    compose_data = yaml.safe_load(content)

    if not compose_data or "services" not in compose_data:
        return

    for service_config in compose_data["services"].values():
        image = service_config.get("image", "")
        if not image:
            continue

        # LSIO images already bind to 0.0.0.0 — command override breaks s6 init
        if "linuxserver" in image:
            existing_command = service_config.get("command", "")
            if "--bind-addr" in existing_command or "--host" in existing_command:
                del service_config["command"]
                compose_file.write_text(
                    yaml.dump(compose_data, default_flow_style=False)
                )
                logger.warning(
                    "Removed broken command override from LSIO image: %s",
                    existing_command,
                )
                return
            return

        # Check if the image matches a known tool
        is_code_server = tool_type_name == "code-server" and (
            "code-server" in image or "coder" in image
        )
        is_jupyter = tool_type_name == "jupyter-notebook" and (
            "jupyter" in image or "notebook" in image
        )
        if not is_code_server and not is_jupyter:
            continue

        existing_command = service_config.get("command", "")
        if existing_command:
            # Already correct — nothing to do
            if bind_command in existing_command:
                return
            # Fix broken or outdated bind flags
            if (
                "--bind-addr" in existing_command
                or "--host" in existing_command
                or "--ip=" in existing_command
            ):
                service_config["command"] = bind_command
                compose_file.write_text(
                    yaml.dump(compose_data, default_flow_style=False)
                )
                logger.warning(
                    "Replaced broken bind address for %s: %s → %s",
                    tool_type_name,
                    existing_command,
                    bind_command,
                )
                return
            # Some other command override exists — don't touch it
            return

        # No command yet — inject the correct bind address
        service_config["command"] = bind_command
        compose_file.write_text(yaml.dump(compose_data, default_flow_style=False))
        logger.info("Injected bind address for %s: %s", tool_type_name, bind_command)
        return


def ensure_backend_network_in_compose(compose_path: str) -> None:
    """Inject the backend network into the compose file so compose up attaches it.

    Instead of running 'docker network connect' after container creation (which
    is prone to race conditions and silent failures), we declare the network in
    the compose file itself. Docker Compose then connects the container to the
    network atomically during 'docker compose up'.
    """
    import yaml
    from pathlib import Path

    compose_file = Path(compose_path)
    if not compose_file.exists():
        return

    content = compose_file.read_text()
    compose_data = yaml.safe_load(content)

    if not compose_data or "services" not in compose_data:
        return

    network_name = get_backend_network_name()
    modified = False

    for svc_config in compose_data["services"].values():
        existing = svc_config.get("networks", [])
        if network_name not in existing:
            svc_config["networks"] = existing + [network_name]
            modified = True
        break  # Only modify first service

    # Declare the network as external at the top level
    if "networks" not in compose_data:
        compose_data["networks"] = {}
    if network_name not in compose_data["networks"]:
        compose_data["networks"][network_name] = {"external": True}
        modified = True

    if modified:
        compose_file.write_text(yaml.dump(compose_data, default_flow_style=False))
        logger.info("Injected backend network '%s' into compose file", network_name)



async def prepare_manifest_instance(
    session: AsyncSession,
    instance: ToolInstance,
    instance_dir: str,
    repo_path: str,
    env_vars: dict,
    extra_volumes: list,
    working_directory: str | None,
) -> tuple[str, str, dict, str]:
    """Build image and generate compose from a manifest-based tool type.

    Returns:
        Tuple of (image_tag, compose_content, resolved_manifest, home_dir)
    """
    from src.models import ToolDefinitionManifest

    tool_type = await session.get(ToolType, instance.tool_type_id)
    if not tool_type:
        raise RuntimeError(f"Tool type {instance.tool_type_id} not found")

    manifest_def = await session.get(ToolDefinitionManifest, tool_type.manifest_id)

    if not manifest_def:
        raise RuntimeError(f"Manifest not found for tool type {tool_type.id}")

    manifest = dict(manifest_def.manifest)

    # Resolve base if referenced
    if manifest_def.base_definition_id:
        base_def = await session.get(
            ToolDefinitionManifest, manifest_def.base_definition_id
        )
        if base_def:
            base_manifest = dict(base_def.manifest)
            manifest = resolve_base(deep_merge(base_manifest, manifest))
        else:
            logger.warning(
                "Base definition %s not found for manifest %s",
                manifest_def.base_definition_id,
                manifest_def.id,
            )

    manifest = merge_with_config(manifest)

    # Resolve extra env and volumes from merge_with_config
    extra_env = manifest.pop("_extra_env", {})
    extra_cfg_volumes = manifest.pop("_extra_volumes", [])
    env_vars.update(extra_env)
    extra_volumes.extend(extra_cfg_volumes)

    # Compute image tag
    image_tag = compute_image_tag(tool_type.name, manifest)

    # Check if image already exists
    check = subprocess.run(
        ["docker", "images", "-q", image_tag],
        capture_output=True,
        text=True,
    )
    image_exists = check.returncode == 0 and check.stdout.strip()

    if not image_exists:
        # Compile and build
        dockerfile = compile_dockerfile(manifest)
        entrypoint = compile_entrypoint(manifest)

        logger.debug(
            "Compiled Dockerfile for instance %s (%d chars)",
            instance.id,
            len(dockerfile),
        )

        build_ctx = {
            "Dockerfile": dockerfile,
            ".headquarter/entrypoint.sh": entrypoint,
        }

        returncode, stdout, stderr = await asyncio.to_thread(
            build_image,
            instance_dir=instance_dir,
            dockerfile=dockerfile,
            tag=image_tag,
            build_context=build_ctx,
        )

        if returncode != 0:
            raise RuntimeError(f"Docker build failed: {stderr}")

        logger.info("Built image %s for instance %s", image_tag, instance.id)
    else:
        logger.info("Reusing existing image %s for instance %s", image_tag, instance.id)

    # Prepare SSH path for mount resolution
    ssh_path = ""
    if instance.clone_mode == "clone":
        ssh_path = os.path.join(instance_dir, ".ssh")

    # Resolve git mount variables from config profile
    git_mount_vars = {}
    if instance.selected_config_profile_id:
        resolved_profile = await resolve_profile(
            session, instance.selected_config_profile_id
        )
        for gm in resolved_profile.git_mounts or []:
            ref = gm.get("git_mount_ref", "default")
            # The actual resolution happens in resolve_git_mounts; we store placeholder
            git_mount_vars[f"GIT_MOUNT_{ref}"] = ""

    variables = {
        "IMAGE_TAG": image_tag,
        "INSTANCE_NAME": instance.name.lower(),
        "INSTANCE_DIR": instance_dir,
        "REPO_PATH": repo_path,
        "SSH_PATH": ssh_path,
        "TOOL_PORT": instance.port or 0,
        "EXTRA_ENV": env_vars,
        "EXTRA_VOLUMES": extra_volumes,
        **git_mount_vars,
    }

    compose_content = compile_compose(manifest, variables)

    logger.debug(
        "_prepare_manifest_instance for %s: repo_path=%s compose_volumes=%s",
        instance.id,
        repo_path or "<empty>",
        manifest.get("mounts", []),
    )
    logger.debug(
        "Generated compose for %s:\n%s",
        instance.id,
        compose_content,
    )

    # Cache
    instance.image_tag = image_tag
    instance.manifest_compiled_at = datetime.now()

    home_dir = get_manifest_home_dir(manifest)
    return image_tag, compose_content, manifest, home_dir