#!/usr/bin/env bash
# nvidia/k8s-nim-operator 3.1.0 - base variant: aicr-eks-inference
# Path: fetch this entry's retained objects, check their SHA-256, then apply
# them with kubectl. It uses curl, shasum, awk and kubectl. It needs no cub
# CLI and no ConfigHub account.
# Generated by scripts/generate-public-site.mjs from this entry's listing:
# https://confighub.github.io/helm-expt/site/listings/nvidia-k8s-nim-operator-3-1-0-aicr-eks-inference.json
# Entry page: https://confighub.github.io/helm-expt/site/charts/nvidia-k8s-nim-operator-3-1-0.html
set -euo pipefail

say() { printf '\n>> %s\n' "$*"; }

for tool in curl awk kubectl; do
  if ! command -v "$tool" >/dev/null 2>&1; then
    printf '%s is required and was not found.\n' "$tool" >&2
    exit 1
  fi
done
if command -v shasum >/dev/null 2>&1; then
  sha256_of() { shasum -a 256 "$1" | awk '{print $1}'; }
elif command -v sha256sum >/dev/null 2>&1; then
  sha256_of() { sha256sum "$1" | awk '{print $1}'; }
else
  printf 'shasum is required and was not found.\n' >&2
  exit 1
fi

work=./nvidia-k8s-nim-operator-3-1-0-aicr-eks-inference-objects
objects_url=https://raw.githubusercontent.com/confighub/helm-expt/main/recipes/nvidia/k8s-nim-operator/3.1.0/revisions/aicr-eks-inference/r001/rendered/release-objects.yaml
objects_sha256=ef6956d9cec1d29fa5955d22d5e5df9db12fa1a122613c7da358e0edcd19dbf8

say "Fetch the retained objects into $work/objects.yaml"
mkdir -p "$work"
curl -fsSL -o "$work/objects.yaml" "$objects_url"

say "Check the file against the SHA-256 in the listing"
fetched_sha256="$(sha256_of "$work/objects.yaml")"
if [ "$fetched_sha256" != "$objects_sha256" ]; then
  printf 'The file differs from the listing, so the script stops. Nothing was applied.\n  listing: %s\n  fetched: %s\n' "$objects_sha256" "$fetched_sha256" >&2
  exit 1
fi
printf 'The file matches: sha256 %s\n' "$fetched_sha256"

say "Split the file into its 21 objects. The cluster is untouched so far"
rm -rf "$work/docs" "$work/crds" "$work/rest" "$work/unplaced"
mkdir -p "$work/docs" "$work/crds" "$work/rest" "$work/unplaced"
awk -v dir="$work/docs" 'BEGIN { n = 0; f = sprintf("%s/doc-%04d.yaml", dir, n) } /^---[ \t]*$/ { close(f); n++; f = sprintf("%s/doc-%04d.yaml", dir, n); next } { print > f }' "$work/objects.yaml"
for n in 0002 0004 0006 0008 0010 0012 0014 0016 0018; do mv "$work/docs/doc-$n.yaml" "$work/crds/"; done
for n in 0020 0021 0022 0025 0028 0029 0030; do mv "$work/docs/doc-$n.yaml" "$work/rest/"; done
for n in 0019 0023 0024 0026 0027; do mv "$work/docs/doc-$n.yaml" "$work/unplaced/"; done
ls "$work/crds" "$work/rest" "$work/unplaced"

if [ "${REQUIREMENTS_READY:-0}" != "1" ]; then
  say "Stop before applying anything. This entry's plan has 1 step that an install needs first"
  cat >&2 <<'EOF_REQUIREMENTS'
This script cannot carry out this step of the plan. It is yours to do, with your own values:
  - The base holds a webhook configuration with no CA bundle and the cert-manager objects that ask for its certificate. Nothing in the bundle issues that certificate. The listing gives the rest.
Nothing was applied. Complete it before you apply the objects.
When it is done, re-run with:
  REQUIREMENTS_READY=1 bash try-objects.sh
EOF_REQUIREMENTS
  exit 1
fi
say "REQUIREMENTS_READY=1 is set. The script takes your word for 1 step and does not check it"

say "Check that kubectl reaches a cluster"
if ! kubectl get --raw /version >/dev/null 2>&1; then
  printf 'kubectl cannot reach a cluster. Point it at a throwaway cluster and re-run this script.\n' >&2
  exit 1
fi
kubectl config current-context 2>/dev/null || true

serves_kind() {
  kubectl api-resources --api-group="$1" --no-headers 2>/dev/null | awk -v kind="$2" '$NF == kind { found = 1 } END { exit found ? 0 : 1 }'
}

say "Check that the cluster serves the 2 kinds this file uses and does not define"
missing_kinds=""
if ! serves_kind cert-manager.io Certificate; then
  missing_kinds="$missing_kinds Certificate.cert-manager.io"
fi
if ! serves_kind cert-manager.io Issuer; then
  missing_kinds="$missing_kinds Issuer.cert-manager.io"
fi
if [ -n "$missing_kinds" ]; then
  printf 'The cluster does not serve:%s\n' "$missing_kinds" >&2
  printf 'This file uses these kinds and does not define them. Install what defines them, then re-run this script. Nothing was applied.\n' >&2
  exit 1
fi

say "Create the nvidia-nim namespace if it does not exist"
if ! kubectl get namespace nvidia-nim >/dev/null 2>&1; then
  kubectl create namespace nvidia-nim
fi

wait_for_crd() {
  crd_name="$1"
  deadline=$(( $(date +%s) + 180 ))
  until kubectl get "crd/${crd_name}" >/dev/null 2>&1; do
    if [ "$(date +%s)" -ge "$deadline" ]; then
      printf "CRD %s did not appear within 180 seconds.\n" "$crd_name" >&2
      return 1
    fi
    sleep 2
  done
  kubectl wait --for=condition=Established --timeout=120s "crd/${crd_name}"
}

say "Check the 9 CRDs this file holds"
missing_crds=0
if ! kubectl get crd/nemocustomizers.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nemodatastores.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nemoentitystores.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nemoevaluators.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nemoguardrails.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nimbuilds.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nimcaches.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nimpipelines.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if ! kubectl get crd/nimservices.apps.nvidia.com >/dev/null 2>&1; then
  missing_crds=1
fi
if [ "$missing_crds" -eq 1 ]; then
  say "A CRD is missing, so apply the file's CRDs first"
  kubectl apply --server-side -f "$work/crds"
  crd_outcome="were applied first"
elif kubectl diff --server-side -f "$work/crds" >/dev/null 2>&1; then
  say "The CRDs already exist and match this file. The script leaves them under their current owner"
  crd_outcome="already existed and matched, so the script left them alone"
elif [ "${APPLY_CRDS:-0}" = "1" ]; then
  say "The existing CRDs differ from this file. APPLY_CRDS=1 is set, so the script applies the file's CRDs"
  kubectl apply --server-side -f "$work/crds"
  crd_outcome="differed and were applied, because APPLY_CRDS=1 was set"
else
  say "The CRDs already exist and differ from this file, or another tool owns them"
  cat <<'EOF_CRDS'
This script does not change CRDs it did not just create.
If this is an upgrade, the objects that follow may need the newer CRDs.
To let this script apply the file's CRDs, re-run with:
  APPLY_CRDS=1 bash try-objects.sh
EOF_CRDS
  exit 1
fi
wait_for_crd nemocustomizers.apps.nvidia.com
wait_for_crd nemodatastores.apps.nvidia.com
wait_for_crd nemoentitystores.apps.nvidia.com
wait_for_crd nemoevaluators.apps.nvidia.com
wait_for_crd nemoguardrails.apps.nvidia.com
wait_for_crd nimbuilds.apps.nvidia.com
wait_for_crd nimcaches.apps.nvidia.com
wait_for_crd nimpipelines.apps.nvidia.com
wait_for_crd nimservices.apps.nvidia.com

say "Apply the other 7 objects"
kubectl apply -f "$work/rest"

say "Apply 5 objects that name no namespace into nvidia-nim, the namespace this entry was built for"
kubectl apply -n nvidia-nim -f "$work/unplaced"

say "Done. The 9 CRDs $crd_outcome. kubectl applied the other 12 objects of this entry's retained file."
say "The script stops here. It did not wait for any workload to start"
say "Not run by this script: 2 steps of this entry's plan. The entry page says who runs each one."
cat <<'EOF_NOT_RUN'
   - The base holds a webhook configuration with no CA bundle and the cert-manager objects that ask for its certificate. Nothing in the bundle issues that certificate. The listing gives the rest.
   - Before an upgrade, move the CRDs to the new version. Helm does not upgrade CRDs from a chart's crds directory, so the chart ships this Job. The listing gives the rest.
EOF_NOT_RUN