1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398#!/usr/bin/env bash
#
# Deploys real application code, database schema, and Hasura metadata into
# the AWS infrastructure Terraform already provisioned (infrastructure/).
# Run this after `terraform apply` — see QUICKSTART.md.
#
# Reads every target (function names, S3 bucket, RDS endpoint, secrets, ...)
# straight from `terraform output`, so nothing here is hardcoded to a
# particular AWS account or environment.
#
# Usage: scripts/deploy-app.sh [--skip-db] [--skip-hasura] [--skip-lambdas] [--skip-frontend] [--create-user EMAIL]
set -euo pipefail
SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
INFRA_DIR="$REPO_ROOT/infrastructure"
SKIP_DB=0
SKIP_HASURA=0
SKIP_LAMBDAS=0
SKIP_FRONTEND=0
CREATE_USER_EMAIL=""
usage() {
cat <<EOF
Usage: $0 [options]
Deploys application code/schema into infrastructure Terraform already
provisioned. Run 'terraform apply' in infrastructure/ first.
Options:
--skip-db Skip loading db/schema.sql + db/migrations/*.sql
--skip-hasura Skip applying db/hasura-metadata.json
--skip-lambdas Skip building/pushing the 6 Lambda functions
--skip-frontend Skip building/publishing the frontend
--create-user EMAIL Create a Cognito user with this email after deploying
-h, --help Show this help
EOF
}
while [ $# -gt 0 ]; do
case "$1" in
--skip-db) SKIP_DB=1; shift ;;
--skip-hasura) SKIP_HASURA=1; shift ;;
--skip-lambdas) SKIP_LAMBDAS=1; shift ;;
--skip-frontend) SKIP_FRONTEND=1; shift ;;
--create-user) CREATE_USER_EMAIL="$2"; shift 2 ;;
-h|--help) usage; exit 0 ;;
*) echo "Unknown option: $1" >&2; usage; exit 1 ;;
esac
done
log() { printf '\n\033[1;36m==> %s\033[0m\n' "$1"; }
die() { echo "Error: $1" >&2; exit 1; }
require() {
command -v "$1" >/dev/null 2>&1 || die "'$1' is required but not found in PATH. $2"
}
require terraform "Install: https://developer.hashicorp.com/terraform/install"
require aws "Install: https://docs.aws.amazon.com/cli/latest/userguide/getting-started-install.html"
require jq "Install: apt install jq / brew install jq"
require npm "Install Node.js 22.x: https://nodejs.org"
if [ "$SKIP_DB" = 0 ]; then
require psql "Install any recent PostgreSQL client, or pass --skip-db."
require session-manager-plugin "Install: https://docs.aws.amazon.com/systems-manager/latest/userguide/session-manager-working-with-install-plugin.html"
fi
if [ "$SKIP_LAMBDAS" = 0 ]; then
require mvn "Needed to build aws-lambda-verapdf-interface, or pass --skip-lambdas."
require zip "Install: apt install zip / brew install zip"
fi
[ -z "$CREATE_USER_EMAIL" ] || require openssl "Needed to generate the new user's password."
log "Reading Terraform outputs"
cd "$INFRA_DIR"
TF_OUT="$(terraform output -json)"
cd "$REPO_ROOT"
[ "$(echo "$TF_OUT" | jq -r '.lambda_function_names // empty')" != "" ] \
|| die "No Terraform outputs found. Run 'terraform apply' in infrastructure/ first."
out() { echo "$TF_OUT" | jq -r "$1"; }
RDS_ENDPOINT="$(out '.rds_endpoint.value')"
DB_HOST="${RDS_ENDPOINT%%:*}"
DB_PORT="${RDS_ENDPOINT##*:}"
DB_NAME="$(out '.db_name.value')"
DB_USERNAME="$(out '.db_username.value')"
DB_PASSWORD_SECRET_ARN="$(out '.db_password_secret_arn.value')"
BASTION_ID="$(out '.bastion_instance_id.value')"
DEPLOY_ARTIFACTS_BUCKET="$(out '.deploy_artifacts_bucket_name.value')"
HASURA_ADMIN_SECRET_ARN="$(out '.hasura_admin_secret_arn.value')"
GRAPHQL_URL="$(out '.graphql_url.value')"
GRAPHQL_BASE_URL="${GRAPHQL_URL%/v1/graphql}"
FRONTEND_BUCKET="$(out '.frontend_bucket_name.value')"
FRONTEND_DIST_ID="$(out '.frontend_cloudfront_distribution_id.value')"
FRONTEND_URL="$(out '.frontend_url.value')"
COGNITO_USER_POOL_ID="$(out '.cognito_user_pool_id.value')"
fn_name() { out ".lambda_function_names.value.$1"; }
# ---------------------------------------------------------------------------
if [ "$SKIP_DB" = 0 ]; then
log "Loading database schema (db/schema.sql + db/migrations/*.sql)"
# RDS is deliberately private (no public IP, security group only allows the
# backend Lambda / Hasura task / this bastion) — tunnel through the SSM
# bastion rather than connecting to $DB_HOST directly.
LOCAL_PORT=15432
STARTED_BASTION=0
TUNNEL_PID=""
cleanup_tunnel() {
[ -n "$TUNNEL_PID" ] && kill "$TUNNEL_PID" >/dev/null 2>&1 || true
[ -n "$TUNNEL_PID" ] && wait "$TUNNEL_PID" 2>/dev/null || true
if [ "$STARTED_BASTION" = 1 ]; then
echo "Stopping bastion instance ($BASTION_ID)..."
aws ec2 stop-instances --instance-ids "$BASTION_ID" >/dev/null 2>&1 || true
fi
}
trap cleanup_tunnel EXIT
BASTION_STATE="$(aws ec2 describe-instances --instance-ids "$BASTION_ID" \
--query 'Reservations[0].Instances[0].State.Name' --output text)"
if [ "$BASTION_STATE" != "running" ]; then
echo "Starting bastion instance ($BASTION_ID)..."
aws ec2 start-instances --instance-ids "$BASTION_ID" >/dev/null
aws ec2 wait instance-running --instance-ids "$BASTION_ID"
STARTED_BASTION=1
fi
echo "Waiting for the bastion's SSM agent to register..."
SSM_ONLINE=0
for _ in $(seq 1 30); do
PING="$(aws ssm describe-instance-information \
--filters "Key=InstanceIds,Values=$BASTION_ID" \
--query 'InstanceInformationList[0].PingStatus' --output text 2>/dev/null || echo "")"
if [ "$PING" = "Online" ]; then
SSM_ONLINE=1
break
fi
sleep 5
done
[ "$SSM_ONLINE" = 1 ] || die "Bastion SSM agent never came online ($BASTION_ID). Check the instance and try again."
echo "Opening SSM tunnel: localhost:$LOCAL_PORT -> $DB_HOST:$DB_PORT (via $BASTION_ID)"
aws ssm start-session --target "$BASTION_ID" \
--document-name AWS-StartPortForwardingSessionToRemoteHost \
--parameters "{\"host\":[\"$DB_HOST\"],\"portNumber\":[\"$DB_PORT\"],\"localPortNumber\":[\"$LOCAL_PORT\"]}" \
>/tmp/equalify-ssm-tunnel.log 2>&1 &
TUNNEL_PID=$!
echo "Waiting for tunnel to accept connections..."
TUNNEL_READY=0
for _ in $(seq 1 30); do
if (exec 3<>"/dev/tcp/127.0.0.1/$LOCAL_PORT") 2>/dev/null; then
TUNNEL_READY=1
break
fi
sleep 1
done
[ "$TUNNEL_READY" = 1 ] || die "SSM tunnel never became ready. See /tmp/equalify-ssm-tunnel.log."
DB_PASSWORD="$(aws secretsmanager get-secret-value --secret-id "$DB_PASSWORD_SECRET_ARN" --query SecretString --output text)"
export PGPASSWORD="$DB_PASSWORD"
ALREADY_LOADED="$(psql -h 127.0.0.1 -p "$LOCAL_PORT" -U "$DB_USERNAME" -d "$DB_NAME" -tAc \
"select 1 from information_schema.tables where table_schema='public' limit 1" 2>/dev/null || echo "")"
if [ "$ALREADY_LOADED" = "1" ]; then
echo "Schema already present — skipping db/schema.sql (pass --skip-db to silence this check)."
else
# --single-transaction: if schema.sql fails partway, nothing commits, so
# a rerun's ALREADY_LOADED check above can't see a half-applied schema
# and mistake it for "already loaded" (the failure mode that bit an
# earlier version of this script).
psql -h 127.0.0.1 -p "$LOCAL_PORT" -U "$DB_USERNAME" -d "$DB_NAME" --single-transaction -v ON_ERROR_STOP=1 -f db/schema.sql
fi
for f in db/migrations/*.sql; do
[ -e "$f" ] || continue
echo "Applying migration: $f"
psql -h 127.0.0.1 -p "$LOCAL_PORT" -U "$DB_USERNAME" -d "$DB_NAME" --single-transaction -v ON_ERROR_STOP=1 -f "$f"
done
unset PGPASSWORD
cleanup_tunnel
trap - EXIT
else
log "Skipping database schema (--skip-db)"
fi
# ---------------------------------------------------------------------------
if [ "$SKIP_HASURA" = 0 ]; then
log "Applying Hasura metadata (db/hasura-metadata.json) to $GRAPHQL_BASE_URL"
HASURA_ADMIN_SECRET="$(aws secretsmanager get-secret-value --secret-id "$HASURA_ADMIN_SECRET_ARN" --query SecretString --output text)"
METADATA_ARGS="$(jq -c '{metadata: .metadata, allow_inconsistent_metadata: true}' db/hasura-metadata.json)"
RESPONSE="$(curl -sf -X POST "$GRAPHQL_BASE_URL/v1/metadata" \
-H "X-Hasura-Admin-Secret: $HASURA_ADMIN_SECRET" \
-H "Content-Type: application/json" \
-d "{\"type\":\"replace_metadata\",\"version\":2,\"args\":$METADATA_ARGS}")"
# With allow_inconsistent_metadata:true (set above so one bad object
# doesn't hard-fail the whole apply), a successful call returns
# {"is_consistent": ..., "inconsistent_objects": [...]} instead of the
# plain {"message":"success"} it'd return without that flag. Only an
# actual API error has an "error"/"code" field, so check for that instead
# of requiring one specific success shape.
echo "$RESPONSE" | jq -e '(.error == null) and (.code == null)' >/dev/null \
|| die "Hasura metadata apply failed: $RESPONSE"
if echo "$RESPONSE" | jq -e '.is_consistent == false' >/dev/null 2>&1; then
echo "Warning: metadata applied but reported inconsistent objects:"
echo "$RESPONSE" | jq '.inconsistent_objects'
fi
# Hasura only re-introspects tracked tables' actual Postgres columns on a
# metadata reload — not automatically, and not just because the ECS task
# restarted. Without this, columns added by db/migrations/*.sql (which
# just ran, in the DB step above) stay invisible to GraphQL indefinitely
# even though they're really in the database.
log "Reloading Hasura metadata (picks up any schema/migration changes)"
RELOAD_RESPONSE="$(curl -sf -X POST "$GRAPHQL_BASE_URL/v1/metadata" \
-H "X-Hasura-Admin-Secret: $HASURA_ADMIN_SECRET" \
-H "Content-Type: application/json" \
-d '{"type":"reload_metadata","args":{"reload_remote_schemas":false}}')"
echo "$RELOAD_RESPONSE" | jq -e '(.error == null) and (.code == null)' >/dev/null \
|| die "Hasura metadata reload failed: $RELOAD_RESPONSE"
else
log "Skipping Hasura metadata (--skip-hasura)"
fi
# ---------------------------------------------------------------------------
if [ "$SKIP_LAMBDAS" = 0 ]; then
deploy_node_lambda() {
local svc_dir="$1" fn="$2"
log "Building & deploying $svc_dir -> $fn"
(
cd "services/$svc_dir"
npm install
npm run dist
# @axe-core/puppeteer locates axe-core via a runtime require.resolve()
# + fs.readFileSync (it injects axe's actual source into the browser
# page) — a dynamic filesystem lookup esbuild can't inline away like a
# normal import, so the real package files must ship alongside the
# bundled code. Resolve via Node itself rather than assuming a local
# node_modules/axe-core path: this is an npm workspaces monorepo, so
# the package is often hoisted to the repo root instead. No-op for
# services (scan-pdf, scan-sqs-router, crawler) that don't depend on
# axe-core at all.
AXE_CORE_DIR="$(node -e "try { console.log(require.resolve('axe-core/package.json').replace(/[\\/\\\\]package\.json$/, '')) } catch (e) {}")"
if [ -n "$AXE_CORE_DIR" ] && [ -d "$AXE_CORE_DIR" ]; then
mkdir -p dist/node_modules
cp -r "$AXE_CORE_DIR" dist/node_modules/axe-core
fi
cd dist
rm -f lambda.zip
zip -rq lambda.zip lambda.* $( [ -d node_modules ] && echo node_modules )
aws lambda update-function-code --function-name "$fn" --zip-file "fileb://lambda.zip" >/dev/null
rm -f lambda.zip
)
}
deploy_node_lambda "aws-lambda-scan-sqs-router" "$(fn_name scan_sqs_router)"
deploy_node_lambda "aws-lambda-scan-html" "$(fn_name scan_html)"
deploy_node_lambda "aws-lambda-scan-pdf" "$(fn_name scan_pdf)"
deploy_node_lambda "aws-lambda-crawler" "$(fn_name crawler)"
# aws-lambda-scan-html uses @sparticuz/chromium-min, which expects the
# actual (large, brotli-compressed) Chromium binary to already be present
# at /opt/nodejs/node_modules/@sparticuz/chromium/bin — i.e. provided by a
# Lambda layer, not bundled in the function's own zip. Skip rebuilding if a
# layer with the expected name is already attached (it's a slow ~50MB
# npm-install-and-zip, not worth repeating on every deploy).
SCAN_HTML_FN="$(fn_name scan_html)"
CHROMIUM_LAYER_NAME="${SCAN_HTML_FN}-chromium"
# jq against raw JSON, not --query/--output text: when Layers is absent
# (null) on a fresh function, --query "Layers[?...]" --output text prints
# the literal string "None" — non-empty, so a naive [ -n ... ] check
# thinks a layer is already attached and skips the build every time.
EXISTING_LAYER="$(aws lambda get-function-configuration --function-name "$SCAN_HTML_FN" --output json \
| jq -r --arg name "$CHROMIUM_LAYER_NAME" '(.Layers // [])[] | select(.Arn | contains($name)) | .Arn' | head -1)"
if [ -n "$EXISTING_LAYER" ]; then
log "Chromium layer already attached to $SCAN_HTML_FN — skipping rebuild"
else
log "Building & publishing Chromium layer for $SCAN_HTML_FN"
CHROMIUM_RANGE="$(jq -r '.dependencies["@sparticuz/chromium-min"]' services/aws-lambda-scan-html/package.json)"
LAYER_TMPDIR="$(mktemp -d)"
mkdir -p "$LAYER_TMPDIR/nodejs"
npm install "@sparticuz/chromium@${CHROMIUM_RANGE}" --prefix "$LAYER_TMPDIR/nodejs" --no-save --omit=dev >/dev/null
(cd "$LAYER_TMPDIR" && zip -rq chromium-layer.zip nodejs)
# PublishLayerVersion's direct --zip-file upload caps at ~70MB (the
# request body is base64-inline); the real Chromium binary pushes past
# that, so stage it in S3 and reference it by location instead.
LAYER_S3_KEY="chromium-layer/${CHROMIUM_LAYER_NAME}-$(date +%s).zip"
aws s3 cp "$LAYER_TMPDIR/chromium-layer.zip" "s3://${DEPLOY_ARTIFACTS_BUCKET}/${LAYER_S3_KEY}" >/dev/null
LAYER_ARN="$(aws lambda publish-layer-version \
--layer-name "$CHROMIUM_LAYER_NAME" \
--description "@sparticuz/chromium ${CHROMIUM_RANGE}" \
--content "S3Bucket=${DEPLOY_ARTIFACTS_BUCKET},S3Key=${LAYER_S3_KEY}" \
--compatible-runtimes nodejs22.x \
--compatible-architectures x86_64 \
--query LayerVersionArn --output text)"
aws s3 rm "s3://${DEPLOY_ARTIFACTS_BUCKET}/${LAYER_S3_KEY}" >/dev/null 2>&1 || true
rm -rf "$LAYER_TMPDIR"
aws lambda update-function-configuration --function-name "$SCAN_HTML_FN" --layers "$LAYER_ARN" >/dev/null
aws lambda wait function-updated --function-name "$SCAN_HTML_FN"
fi
log "Building & deploying aws-lambda-verapdf-interface -> $(fn_name verapdf_interface)"
(
cd services/aws-lambda-verapdf-interface
mvn -q clean package
JAR="$(find target -maxdepth 1 -name '*.jar' ! -name 'original-*' | head -1)"
aws lambda update-function-code --function-name "$(fn_name verapdf_interface)" --zip-file "fileb://$JAR" >/dev/null
)
log "Building & deploying apps/backend -> $(fn_name backend)"
(
cd apps/backend
npm install
npx esbuild index.ts --bundle --platform=node --outdir=dist --external:@aws-sdk --loader:.node=file
cd dist
rm -f lambda.zip
zip -rq lambda.zip index.js
aws lambda update-function-code --function-name "$(fn_name backend)" --zip-file "fileb://lambda.zip" >/dev/null
rm -f lambda.zip
)
else
log "Skipping Lambda deploys (--skip-lambdas)"
fi
# ---------------------------------------------------------------------------
if [ "$SKIP_FRONTEND" = 0 ]; then
log "Building & publishing frontend -> s3://$FRONTEND_BUCKET"
(
cd apps/frontend
# .local, not .env.production directly: Vite loads .env.production.local
# with higher precedence, and it's already covered by the repo's
# .gitignore (`.env.*`) — this deployment's own values never touch the
# tracked .env.production, which holds the real org's production config
# for the existing (non-Terraform) GitHub Actions deploy in
# .github/workflows/deploy-apps.yml.
echo "$TF_OUT" | jq -r '.frontend_env_hints.value | to_entries[] | "\(.key)=\"\(.value)\""' > .env.production.local
npm install
npx vite build --mode production
aws s3 sync --delete ./dist "s3://$FRONTEND_BUCKET"
aws cloudfront create-invalidation --distribution-id "$FRONTEND_DIST_ID" --paths "/*" >/dev/null
)
else
log "Skipping frontend deploy (--skip-frontend)"
fi
# ---------------------------------------------------------------------------
if [ -n "$CREATE_USER_EMAIL" ]; then
log "Creating Cognito user: $CREATE_USER_EMAIL"
# Set a permanent password directly rather than emailing a temporary one:
# Cognito's own account-verification email (separate from the app's own
# SES setup) works without any SES configuration, but a fresh install
# shouldn't have its first login depend on any email deliverability at all.
aws cognito-idp admin-create-user \
--user-pool-id "$COGNITO_USER_POOL_ID" \
--username "$CREATE_USER_EMAIL" \
--user-attributes Name=email,Value="$CREATE_USER_EMAIL" Name=email_verified,Value=true \
--message-action SUPPRESS >/dev/null
# Guaranteed to satisfy the pool's password policy (min 12 chars, upper +
# lower + number; symbols not required) regardless of what the random
# portion happens to contain.
CREATE_USER_PASSWORD="$(openssl rand -hex 9)Aa1!"
aws cognito-idp admin-set-user-password \
--user-pool-id "$COGNITO_USER_POOL_ID" \
--username "$CREATE_USER_EMAIL" \
--password "$CREATE_USER_PASSWORD" \
--permanent >/dev/null
echo ""
echo " Email: $CREATE_USER_EMAIL"
echo " Password: $CREATE_USER_PASSWORD"
echo ""
echo "Save this now — it is not stored anywhere and won't be shown again."
echo "(This becomes the first admin account automatically, via the"
echo " Cognito PostConfirmation trigger's first-user bootstrap.)"
fi
log "Done. Frontend: $FRONTEND_URL"