@haystackeditor/cli 0.25.1 → 0.26.0

package/dist/commands/cloud-verifier-data-store-census.js

@@ -1,539 +0,0 @@
-/**
- * Data-store census.
- *
- * The one job the Cloud Verifier needs from a repository map: what data stores
- * does this application use, so we can clone each one into a replica. Not
- * processes, not API calls, not code structure — the databases, caches, object
- * stores, search indexes, and queues that hold state a test must run against.
- *
- * Four source signals, strongest first:
- *   1. a service image in a compose/k8s manifest (`image: postgres:16`) — a
- *      store instance the app is wired to run beside, the surest signal;
- *   2. a connection URL scheme or conventional env var (`postgres://`,
- *      `REDIS_URL`) — an instance named in config;
- *   3. a config declaration that names a store without an image, library, or URL
- *      (Django `ENGINE`, PHP PDO, Laravel `DB_CONNECTION`, Rails `adapter:`);
- *   4. a client library in a dependency manifest (`pg`, `redis`, `mongoose`) —
- *      the app talks to that engine.
- *
- * Signals are folded per engine into one data store, with the evidence, a
- * confidence, and how to reproduce it. Everything here is a general convention,
- * never a repository-specific constant.
- */
-// How a replica of this category is made. Durable stores of record are copied
-// from a snapshot; caches and search indexes are usually derived and can be
-// rebuilt; queues are drained to empty.
-const CLONE_STRATEGY_BY_CATEGORY = {
-    'relational-database': 'snapshot-restore',
-    'document-database': 'snapshot-restore',
-    'key-value-store': 'snapshot-restore',
-    'wide-column-database': 'snapshot-restore',
-    'columnar-database': 'snapshot-restore',
-    'graph-database': 'snapshot-restore',
-    'vector-database': 'snapshot-restore',
-    'object-storage': 'copy-bucket-snapshot',
-    'search-index': 'rebuild-from-source',
-    'message-queue': 'reset-empty',
-    'time-series-database': 'snapshot-restore',
-    'embedded-database': 'copy-file',
-};
-// Client library (as written in a dependency manifest) -> engine. Covers the
-// common stores across Node, Python, Go, Ruby, Java, PHP. Keys are matched
-// case-insensitively against exact dependency names.
-// Null-prototype: keys are matched against arbitrary dependency tokens from a
-// manifest, so a token like `constructor`/`toString`/`hasOwnProperty` must NOT
-// resolve to an inherited Object.prototype member (that returned a truthy
-// non-profile and created a store with an undefined engine — crashed on a repo
-// whose manifest contained such a token).
-const LIBRARY_ENGINES = Object.create(null);
-function registerLibraries(profile, names) {
-    for (const name of names)
-        LIBRARY_ENGINES[name.toLowerCase()] = profile;
-}
-registerLibraries({ engine: 'postgres', category: 'relational-database' }, [
-    'pg', 'pg-promise', 'postgres', 'node-postgres', 'psycopg2', 'psycopg2-binary',
-    'psycopg', 'asyncpg', 'lib/pq', 'github.com/lib/pq', 'github.com/jackc/pgx',
-    'github.com/jackc/pgx/v5', 'pg gem', 'postgresql', 'org.postgresql:postgresql',
-]);
-registerLibraries({ engine: 'mysql', category: 'relational-database' }, [
-    'mysql', 'mysql2', 'mariadb', 'pymysql', 'mysqlclient', 'aiomysql',
-    'github.com/go-sql-driver/mysql', 'mysql-connector-python', 'mysql:mysql-connector-java',
-]);
-registerLibraries({ engine: 'sqlserver', category: 'relational-database' }, [
-    'mssql', 'tedious', 'pyodbc', 'pymssql', 'github.com/denisenkom/go-mssqldb',
-]);
-registerLibraries({ engine: 'oracle', category: 'relational-database' }, [
-    'oracledb', 'cx_oracle', 'cx-oracle',
-]);
-registerLibraries({ engine: 'mongodb', category: 'document-database' }, [
-    'mongodb', 'mongoose', 'pymongo', 'motor', 'mongoengine',
-    'go.mongodb.org/mongo-driver', 'mongo', 'mongoid',
-]);
-registerLibraries({ engine: 'couchdb', category: 'document-database' }, [
-    'nano', 'couchdb', 'pycouchdb', 'cloudant',
-]);
-registerLibraries({ engine: 'redis', category: 'key-value-store' }, [
-    'redis', 'ioredis', 'redis-py', 'connect-redis', 'go-redis', 'github.com/redis/go-redis',
-    'github.com/go-redis/redis', 'hiredis', 'aioredis', 'redis-rb', 'predis/predis', 'jedis',
-]);
-registerLibraries({ engine: 'memcached', category: 'key-value-store' }, [
-    'memcached', 'memjs', 'pymemcache', 'python-memcached', 'dalli',
-]);
-registerLibraries({ engine: 'dynamodb', category: 'key-value-store' }, [
-    '@aws-sdk/client-dynamodb', 'dynamoose',
-]);
-registerLibraries({ engine: 'cassandra', category: 'wide-column-database' }, [
-    'cassandra-driver', 'cassandra', 'scylla-driver', 'datastax',
-]);
-registerLibraries({ engine: 'clickhouse', category: 'columnar-database' }, [
-    '@clickhouse/client', '@clickhouse/client-web', 'clickhouse-driver',
-    'clickhouse-connect', 'clickhouse', 'ecto_ch', 'clickhouse_ecto',
-    'github.com/clickhouse/clickhouse-go', 'clickhouse-sqlalchemy',
-]);
-registerLibraries({ engine: 'cockroachdb', category: 'relational-database' }, [
-    'cockroachdb', 'sqlalchemy-cockroachdb',
-]);
-registerLibraries({ engine: 'neo4j', category: 'graph-database' }, [
-    'neo4j-driver', 'neo4j', 'py2neo', 'neo4j-python-driver',
-]);
-registerLibraries({ engine: 'weaviate', category: 'vector-database' }, [
-    'weaviate-client', 'weaviate-ts-client', '@weaviate/client', 'weaviate',
-]);
-registerLibraries({ engine: 'qdrant', category: 'vector-database' }, [
-    'qdrant-client', '@qdrant/js-client-rest', 'qdrant',
-]);
-registerLibraries({ engine: 'milvus', category: 'vector-database' }, [
-    'pymilvus', '@zilliz/milvus2-sdk-node', 'milvus',
-]);
-registerLibraries({ engine: 'pinecone', category: 'vector-database' }, [
-    'pinecone-client', '@pinecone-database/pinecone', 'pinecone',
-]);
-registerLibraries({ engine: 'chroma', category: 'vector-database' }, [
-    'chromadb', 'chromadb-client',
-]);
-// Object-storage clients must be the specific S3 client, not a monolithic
-// cloud SDK: bare `aws-sdk`/`boto3`/`aws-sdk-go` are used for SES, SNS, SQS,
-// and dozens of non-storage services, so mapping them to S3 fabricates stores
-// (measured on medusa, saleor, n8n). Only the dedicated S3 client counts.
-registerLibraries({ engine: 's3', category: 'object-storage' }, [
-    '@aws-sdk/client-s3', 'minio', 'aws-sdk-s3',
-]);
-registerLibraries({ engine: 'gcs', category: 'object-storage' }, [
-    '@google-cloud/storage', 'google-cloud-storage', 'gcloud-storage',
-]);
-registerLibraries({ engine: 'azure-blob', category: 'object-storage' }, [
-    '@azure/storage-blob', 'azure-storage-blob', 'azure-storage',
-]);
-registerLibraries({ engine: 'elasticsearch', category: 'search-index' }, [
-    '@elastic/elasticsearch', 'elasticsearch', 'elasticsearch-py', 'elasticsearch-dsl',
-    '@opensearch-project/opensearch', 'opensearch-py', 'github.com/elastic/go-elasticsearch',
-]);
-registerLibraries({ engine: 'solr', category: 'search-index' }, [
-    'solr-client', 'pysolr', 'solr',
-]);
-registerLibraries({ engine: 'meilisearch', category: 'search-index' }, [
-    'meilisearch', 'meilisearch-python',
-]);
-registerLibraries({ engine: 'rabbitmq', category: 'message-queue' }, [
-    'amqplib', 'amqp', 'pika', 'bunny', 'github.com/streadway/amqp', 'github.com/rabbitmq/amqp091-go',
-]);
-registerLibraries({ engine: 'kafka', category: 'message-queue' }, [
-    'kafkajs', 'node-rdkafka', 'kafka-python', 'confluent-kafka', 'github.com/segmentio/kafka-go',
-    'sarama', 'github.com/ibm/sarama',
-]);
-registerLibraries({ engine: 'sqs', category: 'message-queue' }, [
-    '@aws-sdk/client-sqs', 'boto3-sqs',
-]);
-registerLibraries({ engine: 'nats', category: 'message-queue' }, [
-    'nats', 'nats.js', 'nats-py', 'github.com/nats-io/nats.go',
-]);
-// Distributed key-value / config stores that hold an app's own coordination
-// state (apisix keeps its routes in etcd; vault/apisix can use consul). Keys are
-// specific client coordinates — bare `consul`/`etcd` are avoided as too generic.
-registerLibraries({ engine: 'etcd', category: 'key-value-store' }, [
-    'etcd3', 'python-etcd3', 'go.etcd.io/etcd', 'go.etcd.io/etcd/client/v3',
-    'github.com/coreos/etcd', 'coreos-etcd',
-]);
-registerLibraries({ engine: 'consul', category: 'key-value-store' }, [
-    'python-consul', 'python-consul2', 'github.com/hashicorp/consul/api', 'node-consul',
-]);
-// Time-series databases (flagsmith stores analytics in InfluxDB).
-registerLibraries({ engine: 'influxdb', category: 'time-series-database' }, [
-    'influxdb-client', 'influxdb', '@influxdata/influxdb-client', 'influxdb-python',
-    // Go clients are imported at their versioned module path.
-    'github.com/influxdata/influxdb-client-go', 'github.com/influxdata/influxdb-client-go/v2',
-    'github.com/influxdata/influxdb1-client', 'github.com/influxdata/influxdb1-client/v2',
-]);
-// Redis-backed job/queue libraries are NOT a separate store — their queue
-// lives inside Redis. Mapping them to `redis` folds them into the broker
-// (measured: they created a phantom `redis-queue` alongside redis in 8 repos).
-// `celery` is omitted because its broker is configurable (Redis or RabbitMQ),
-// and the real broker is detected from its own client or a connection URL.
-registerLibraries({ engine: 'redis', category: 'key-value-store' }, [
-    'bull', 'bullmq', 'bee-queue', 'sidekiq', 'resque', 'rq',
-]);
-registerLibraries({ engine: 'sqlite', category: 'embedded-database' }, [
-    'sqlite3', 'better-sqlite3', 'sqlite', 'aiosqlite',
-    'github.com/mattn/go-sqlite3', 'modernc.org/sqlite', 'github.com/glebarez/sqlite',
-    'gorm.io/driver/sqlite',
-    // .NET
-    'microsoft.data.sqlite', 'microsoft.entityframeworkcore.sqlite', 'system.data.sqlite',
-]);
-// .NET data-access packages (from .csproj / Directory.Packages.props)
-registerLibraries({ engine: 'postgres', category: 'relational-database' }, [
-    'npgsql', 'npgsql.entityframeworkcore.postgresql',
-]);
-registerLibraries({ engine: 'mysql', category: 'relational-database' }, [
-    'mysqlconnector', 'mysql.data', 'pomelo.entityframeworkcore.mysql',
-]);
-registerLibraries({ engine: 'sqlserver', category: 'relational-database' }, [
-    'microsoft.data.sqlclient', 'system.data.sqlclient',
-    'microsoft.entityframeworkcore.sqlserver',
-]);
-registerLibraries({ engine: 'oracle', category: 'relational-database' }, [
-    'oracle.manageddataaccess', 'oracle.manageddataaccess.core',
-]);
-registerLibraries({ engine: 'mongodb', category: 'document-database' }, [
-    'mongodb.driver',
-]);
-registerLibraries({ engine: 'redis', category: 'key-value-store' }, [
-    'stackexchange.redis',
-]);
-// ORM/driver packages that pin a specific SQL engine.
-registerLibraries({ engine: 'postgres', category: 'relational-database' }, [
-    'gorm.io/driver/postgres', 'github.com/uptrace/bun/dialect/pgdialect',
-]);
-registerLibraries({ engine: 'mysql', category: 'relational-database' }, [
-    'gorm.io/driver/mysql',
-]);
-// Elixir Hex drivers (from mix.exs / mix.lock). Each is engine-specific; the
-// adapter-agnostic `ecto`/`ecto_sql` are deliberately omitted (they name no
-// engine on their own). `ecto_ch`/`clickhouse_ecto` are already in the
-// ClickHouse set above.
-registerLibraries({ engine: 'postgres', category: 'relational-database' }, ['postgrex']);
-registerLibraries({ engine: 'mysql', category: 'relational-database' }, ['myxql']);
-registerLibraries({ engine: 'redis', category: 'key-value-store' }, ['redix']);
-registerLibraries({ engine: 'mongodb', category: 'document-database' }, ['mongodb_driver']);
-registerLibraries({ engine: 'cassandra', category: 'wide-column-database' }, ['xandra']);
-// Rust crates (from Cargo.toml). Only engine-specific drivers are named; the
-// adapter-agnostic `sqlx`/`diesel`/`sea-orm` are omitted (they name no engine on
-// their own — their enabled feature, e.g. "postgres", carries that signal and is
-// matched as its own token). The generic `postgres`/`mysql`/`redis`/`mongodb`/
-// `clickhouse`/`elasticsearch` crate names already match existing library keys.
-registerLibraries({ engine: 'postgres', category: 'relational-database' }, [
-    'tokio-postgres', 'deadpool-postgres', 'bb8-postgres', 'sqlx-postgres',
-]);
-registerLibraries({ engine: 'mysql', category: 'relational-database' }, ['mysql_async', 'sqlx-mysql']);
-registerLibraries({ engine: 'sqlite', category: 'embedded-database' }, ['rusqlite']);
-// `fred` is deliberately omitted — it is a common word (author names, comments)
-// and the coarse token scan is global across all manifests, so it would mint
-// phantom Redis stores. The `-redis`-suffixed crates are unambiguous.
-registerLibraries({ engine: 'redis', category: 'key-value-store' }, ['deadpool-redis', 'bb8-redis']);
-registerLibraries({ engine: 'kafka', category: 'message-queue' }, ['rdkafka', 'rskafka']);
-registerLibraries({ engine: 'rabbitmq', category: 'message-queue' }, ['lapin']);
-registerLibraries({ engine: 'cassandra', category: 'wide-column-database' }, ['cdrs-tokio', 'scylla']);
-// Container image base name -> engine. Matched against the image before its
-// tag and registry path (`postgres:16`, `bitnami/redis`).
-// Null-prototype (see LIBRARY_ENGINES): image path segments are arbitrary text.
-const IMAGE_ENGINES = Object.assign(Object.create(null), {
-    postgres: { engine: 'postgres', category: 'relational-database' },
-    postgresql: { engine: 'postgres', category: 'relational-database' },
-    postgis: { engine: 'postgres', category: 'relational-database' },
-    // Postgres distributions/extensions: the store to clone is still Postgres.
-    pgvector: { engine: 'postgres', category: 'relational-database' },
-    'pgvecto-rs': { engine: 'postgres', category: 'relational-database' },
-    timescaledb: { engine: 'postgres', category: 'relational-database' },
-    supabase: { engine: 'postgres', category: 'relational-database' },
-    // Vector databases
-    weaviate: { engine: 'weaviate', category: 'vector-database' },
-    qdrant: { engine: 'qdrant', category: 'vector-database' },
-    milvus: { engine: 'milvus', category: 'vector-database' },
-    chroma: { engine: 'chroma', category: 'vector-database' },
-    chromadb: { engine: 'chroma', category: 'vector-database' },
-    mysql: { engine: 'mysql', category: 'relational-database' },
-    mariadb: { engine: 'mysql', category: 'relational-database' },
-    mssql: { engine: 'sqlserver', category: 'relational-database' },
-    'mssql-server': { engine: 'sqlserver', category: 'relational-database' },
-    'azure-sql-edge': { engine: 'sqlserver', category: 'relational-database' },
-    mongo: { engine: 'mongodb', category: 'document-database' },
-    mongodb: { engine: 'mongodb', category: 'document-database' },
-    couchdb: { engine: 'couchdb', category: 'document-database' },
-    redis: { engine: 'redis', category: 'key-value-store' },
-    valkey: { engine: 'redis', category: 'key-value-store' },
-    clickhouse: { engine: 'clickhouse', category: 'columnar-database' },
-    'clickhouse-server': { engine: 'clickhouse', category: 'columnar-database' },
-    cockroachdb: { engine: 'cockroachdb', category: 'relational-database' },
-    cockroach: { engine: 'cockroachdb', category: 'relational-database' },
-    azurite: { engine: 'azure-blob', category: 'object-storage' },
-    'servicebus-emulator': { engine: 'azure-servicebus', category: 'message-queue' },
-    memcached: { engine: 'memcached', category: 'key-value-store' },
-    cassandra: { engine: 'cassandra', category: 'wide-column-database' },
-    scylladb: { engine: 'cassandra', category: 'wide-column-database' },
-    neo4j: { engine: 'neo4j', category: 'graph-database' },
-    minio: { engine: 's3', category: 'object-storage' },
-    elasticsearch: { engine: 'elasticsearch', category: 'search-index' },
-    opensearchproject: { engine: 'elasticsearch', category: 'search-index' },
-    solr: { engine: 'solr', category: 'search-index' },
-    meilisearch: { engine: 'meilisearch', category: 'search-index' },
-    rabbitmq: { engine: 'rabbitmq', category: 'message-queue' },
-    kafka: { engine: 'kafka', category: 'message-queue' },
-    nats: { engine: 'nats', category: 'message-queue' },
-    // Distributed KV / config stores and time-series DBs run as their own service.
-    etcd: { engine: 'etcd', category: 'key-value-store' },
-    consul: { engine: 'consul', category: 'key-value-store' },
-    influxdb: { engine: 'influxdb', category: 'time-series-database' },
-});
-// URL scheme -> engine, for connection strings in config or source.
-// Null-prototype (see LIBRARY_ENGINES): schemes come from arbitrary text.
-const URL_SCHEME_ENGINES = Object.assign(Object.create(null), {
-    'postgres': { engine: 'postgres', category: 'relational-database' },
-    'postgresql': { engine: 'postgres', category: 'relational-database' },
-    'mysql': { engine: 'mysql', category: 'relational-database' },
-    'mongodb': { engine: 'mongodb', category: 'document-database' },
-    'mongodb+srv': { engine: 'mongodb', category: 'document-database' },
-    'redis': { engine: 'redis', category: 'key-value-store' },
-    'rediss': { engine: 'redis', category: 'key-value-store' },
-    'amqp': { engine: 'rabbitmq', category: 'message-queue' },
-    'amqps': { engine: 'rabbitmq', category: 'message-queue' },
-    'elasticsearch': { engine: 'elasticsearch', category: 'search-index' },
-    'clickhouse': { engine: 'clickhouse', category: 'columnar-database' },
-    'clickhouses': { engine: 'clickhouse', category: 'columnar-database' },
-    'cockroachdb': { engine: 'cockroachdb', category: 'relational-database' },
-});
-const CONFIDENCE_RANK = {
-    running: 3, configured: 2, library: 1,
-};
-function manifestDependencyNames(input) {
-    const { lowerName, text } = input;
-    const names = new Set();
-    if (lowerName === 'package.json') {
-        try {
-            const parsed = JSON.parse(text);
-            for (const field of ['dependencies', 'optionalDependencies', 'peerDependencies']) {
-                const deps = parsed[field];
-                if (deps && typeof deps === 'object') {
-                    for (const name of Object.keys(deps))
-                        names.add(name);
-                }
-            }
-        }
-        catch { /* ignore malformed manifest */ }
-        return [...names];
-    }
-    if (lowerName === 'requirements.txt' || lowerName.startsWith('requirements')) {
-        for (const line of text.split('\n')) {
-            const match = line.match(/^\s*([A-Za-z0-9_.-]+)/);
-            if (match)
-                names.add(match[1]);
-        }
-        return [...names];
-    }
-    if (lowerName === 'pyproject.toml' || lowerName === 'go.mod' || lowerName === 'gemfile'
-        || lowerName === 'composer.json' || lowerName === 'pom.xml'
-        || lowerName === 'build.gradle' || lowerName === 'build.gradle.kts'
-        || lowerName === 'mix.exs' || lowerName === 'cargo.toml') {
-        // Coarse token scan: a store library name appearing anywhere in these
-        // manifests is a dependency signal. The library map keys are specific
-        // enough (pg, ioredis, amqplib, cassandra-driver, postgrex, ecto_ch) that
-        // false matches are rare, and go.mod/pom paths are matched as full module
-        // coordinates. Elixir deps are `{:postgrex, ...}` atoms, so `{`/`}` split
-        // too, leaving the bare package name as a token. Only `mix.exs` (direct
-        // deps) is read, never `mix.lock` — a lockfile is the full transitive tree
-        // (dev/test deps and their deps), the same reason package-lock.json is
-        // skipped in favor of package.json. `Cargo.toml` (Rust) is read the same
-        // way; a store crate's engine-carrying feature (e.g. `features=["postgres"]`)
-        // matches as its own token, so the adapter-agnostic sqlx/diesel still resolve.
-        for (const token of text.split(/[\s"',=:(){}[\]]+/)) {
-            if (token)
-                names.add(token);
-        }
-        return [...names];
-    }
-    // .NET project files: PackageReference / PackageVersion Include names. Covers
-    // .csproj, .fsproj, .vbproj, Directory.Packages.props, Directory.Build.props,
-    // and packages.config (the whole .NET dependency-manifest surface).
-    if (lowerName.endsWith('.csproj') || lowerName.endsWith('.fsproj')
-        || lowerName.endsWith('.vbproj') || lowerName.endsWith('.props')
-        || lowerName === 'packages.config') {
-        for (const match of text.matchAll(/<(?:PackageReference|PackageVersion|package)\s+[^>]*\b(?:Include|id)\s*=\s*"([^"]+)"/gi)) {
-            names.add(match[1]);
-        }
-        return [...names];
-    }
-    return [];
-}
-// CI pipeline config paths. A `services:` container declared in a CI workflow
-// (GitHub Actions, GitLab, CircleCI, ...) is ephemeral test infrastructure, not
-// the application's production topology — a Postgres spun up to run the test
-// suite is not a store the app ships with. Scanning these for `image:` invents
-// running stores (measured: twenty's ClickHouse came from .github/workflows,
-// not its compose). Matched on the normalized repo-relative path, not content.
-function isCiConfigPath(path) {
-    const p = path.toLowerCase().replace(/\\/g, '/');
-    return p.includes('.github/workflows/')
-        || p.includes('.gitea/workflows/')
-        || p.includes('.forgejo/workflows/')
-        || p.includes('.circleci/')
-        || p.includes('.buildkite/')
-        || /(^|\/)\.gitlab-ci\.ya?ml$/.test(p)
-        || /(^|\/)\.drone\.ya?ml$/.test(p)
-        || /(^|\/)\.woodpecker\.ya?ml$/.test(p)
-        || p.includes('/.woodpecker/')
-        || /(^|\/)bitbucket-pipelines\.ya?ml$/.test(p)
-        || /(^|\/)azure-pipelines\.ya?ml$/.test(p);
-}
-function serviceImages(input) {
-    const { lowerName, text } = input;
-    const isCompose = /(^|\/)(docker-)?compose\.ya?ml$/.test(lowerName)
-        || lowerName === 'compose.yml' || lowerName === 'compose.yaml'
-        || lowerName === 'docker-compose.yml' || lowerName === 'docker-compose.yaml';
-    const isYaml = lowerName.endsWith('.yml') || lowerName.endsWith('.yaml');
-    if (!isCompose && !isYaml)
-        return [];
-    const images = [];
-    for (const line of text.split('\n')) {
-        const match = line.match(/^\s*(?:-\s*)?image\s*:\s*["']?([^"'\s]+)["']?/);
-        if (match)
-            images.push(match[1]);
-    }
-    return images;
-}
-function connectionUrlSchemes(input) {
-    const schemes = [];
-    for (const match of input.text.matchAll(/\b([a-z][a-z0-9+.-]*)\s*:\/\//gi)) {
-        schemes.push(match[1].toLowerCase());
-    }
-    return schemes;
-}
-// High-signal config declarations that name a store without an image, a client
-// library, or a URL. Two whole ecosystems configure their stores this way —
-// Python/Django (`DATABASES … ENGINE`), and PHP (built-in PDO/mysqli named only
-// in an .ini/.php config) — plus the conventional `<ENGINE>_HOST` env keys. On
-// the survey these were the dominant recall gap (zulip's four stores, matomo's
-// database). Kept deliberately specific to avoid matching prose or an example.
-const CONFIG_ENGINE_PATTERNS = [
-    // Django database backends: `django.db.backends.postgresql`
-    { re: /django\.db\.backends\.postgresql/i, profile: { engine: 'postgres', category: 'relational-database' } },
-    { re: /django\.db\.backends\.mysql/i, profile: { engine: 'mysql', category: 'relational-database' } },
-    { re: /django\.db\.backends\.oracle/i, profile: { engine: 'oracle', category: 'relational-database' } },
-    { re: /django\.db\.backends\.sqlite3/i, profile: { engine: 'sqlite', category: 'embedded-database' } },
-    // Django cache backends
-    { re: /django(?:\.core\.cache\.backends\.memcached|_bmemcached|\.core\.cache\.backends\.pymemcache)/i, profile: { engine: 'memcached', category: 'key-value-store' } },
-    { re: /(?:django_redis|django\.core\.cache\.backends\.redis)/i, profile: { engine: 'redis', category: 'key-value-store' } },
-    // PHP built-in DB drivers named only in config (matomo, many PHP apps)
-    { re: /\b(?:adapter\s*=\s*)?PDO\\?_?MYSQL\b/i, profile: { engine: 'mysql', category: 'relational-database' } },
-    { re: /\b(?:adapter\s*=\s*)?PDO\\?_?PGSQL\b/i, profile: { engine: 'postgres', category: 'relational-database' } },
-    // Laravel .env: DB_CONNECTION names the built-in PDO driver (firefly, koel)
-    { re: /\bDB_CONNECTION\s*=\s*["']?(?:mysql|mariadb)\b/i, profile: { engine: 'mysql', category: 'relational-database' } },
-    { re: /\bDB_CONNECTION\s*=\s*["']?(?:pgsql|postgres)\b/i, profile: { engine: 'postgres', category: 'relational-database' } },
-    { re: /\bDB_CONNECTION\s*=\s*["']?sqlite\b/i, profile: { engine: 'sqlite', category: 'embedded-database' } },
-    { re: /\bDB_CONNECTION\s*=\s*["']?sqlsrv\b/i, profile: { engine: 'sqlserver', category: 'relational-database' } },
-    // Rails config/database.yml: adapter names the driver (chatwoot, many Rails)
-    { re: /\badapter\s*:\s*postgresql\b/i, profile: { engine: 'postgres', category: 'relational-database' } },
-    { re: /\badapter\s*:\s*mysql2?\b/i, profile: { engine: 'mysql', category: 'relational-database' } },
-    { re: /\badapter\s*:\s*sqlite3\b/i, profile: { engine: 'sqlite', category: 'embedded-database' } },
-    { re: /\badapter\s*:\s*sqlserver\b/i, profile: { engine: 'sqlserver', category: 'relational-database' } },
-    // Conventional <ENGINE>_HOST / _DATABASE env keys for the SQL engines too
-    { re: /\bPOSTGRES(?:QL)?_(?:HOST|DB|DATABASE)\b/i, profile: { engine: 'postgres', category: 'relational-database' } },
-    { re: /\bPG(?:HOST|DATABASE)\b/, profile: { engine: 'postgres', category: 'relational-database' } },
-    { re: /\bMYSQL_(?:HOST|DATABASE|DB)\b/i, profile: { engine: 'mysql', category: 'relational-database' } },
-    // Conventional <ENGINE>_HOST / _LOCATION / _SERVERS / _CONTACT_POINTS env keys
-    { re: /\bREDIS_(?:HOST|HOSTS|SERVER|URL)\b/i, profile: { engine: 'redis', category: 'key-value-store' } },
-    { re: /\bMEMCACHED_(?:HOST|HOSTS|LOCATION|SERVERS)\b/i, profile: { engine: 'memcached', category: 'key-value-store' } },
-    { re: /\bRABBITMQ_(?:HOST|HOSTS|URL)\b/i, profile: { engine: 'rabbitmq', category: 'message-queue' } },
-    { re: /\bKAFKA_(?:BROKERS|BOOTSTRAP|BOOTSTRAP_SERVERS|HOSTS?|SERVERS)\b/i, profile: { engine: 'kafka', category: 'message-queue' } },
-    { re: /\bCLICKHOUSE_(?:HOST|URL|DSN|ADDR)\b/i, profile: { engine: 'clickhouse', category: 'columnar-database' } },
-    { re: /\bCASSANDRA_(?:HOST|HOSTS|CONTACT_POINTS|SEEDS)\b/i, profile: { engine: 'cassandra', category: 'wide-column-database' } },
-    { re: /\bMONGO(?:DB)?_(?:HOST|URL|URI)\b/i, profile: { engine: 'mongodb', category: 'document-database' } },
-    { re: /\bELASTICSEARCH_(?:HOST|HOSTS|URL)\b/i, profile: { engine: 'elasticsearch', category: 'search-index' } },
-    { re: /\bETCD_(?:HOST|HOSTS|ENDPOINT|ENDPOINTS)\b/i, profile: { engine: 'etcd', category: 'key-value-store' } },
-    { re: /\bINFLUX(?:DB)?_(?:HOST|URL|BUCKET|ORG)\b/i, profile: { engine: 'influxdb', category: 'time-series-database' } },
-    { re: /\bCONSUL_(?:HTTP_ADDR|HOST|URL)\b/i, profile: { engine: 'consul', category: 'key-value-store' } },
-];
-function configEngineSignals(input) {
-    const found = [];
-    for (const { re, profile } of CONFIG_ENGINE_PATTERNS) {
-        if (re.test(input.text))
-            found.push(profile);
-    }
-    return found;
-}
-function imageEngineFor(image) {
-    // Drop the digest, then test each path segment against the engine map, last
-    // segment first. `bitnami/postgresql:16` matches on `postgresql`;
-    // `mcr.microsoft.com/mssql/server` matches on the `mssql` segment (the engine
-    // is not always the final segment behind a registry path).
-    const segments = image.split('@')[0].split('/').map((segment) => segment.toLowerCase());
-    // Strip the tag from the final segment ONLY — a colon in an earlier segment is
-    // a registry port (`localhost:5000/postgres:16`), not a tag separator.
-    segments[segments.length - 1] = segments[segments.length - 1].split(':')[0];
-    for (let index = segments.length - 1; index >= 0; index -= 1) {
-        const profile = IMAGE_ENGINES[segments[index]];
-        if (profile)
-            return profile;
-    }
-    return undefined;
-}
-/**
- * Fold a stream of manifest/compose/config files into the distinct data stores
- * the application uses. `inputs` is every scanned text file with its lowercased
- * basename; the detector reads only the manifests and configs it recognizes.
- */
-export function detectDataStores(inputs) {
-    const byEngine = new Map();
-    const upsert = (profile, confidence, signal) => {
-        const existing = byEngine.get(profile.engine);
-        if (existing) {
-            existing.signals.push(signal);
-            if (CONFIDENCE_RANK[confidence] > CONFIDENCE_RANK[existing.confidence]) {
-                existing.confidence = confidence;
-            }
-            return;
-        }
-        byEngine.set(profile.engine, {
-            engine: profile.engine,
-            category: profile.category,
-            clone_strategy: CLONE_STRATEGY_BY_CATEGORY[profile.category],
-            confidence,
-            signals: [signal],
-        });
-    };
-    for (const input of inputs) {
-        // A CI pipeline file is test infrastructure end to end: its service images,
-        // its `POSTGRES_HOST`/`REDIS_URL` env, and its connection strings all describe
-        // the test harness, not the app's production stores. Skip it for every signal
-        // (it is never a dependency manifest, so nothing of value is lost).
-        if (isCiConfigPath(input.path))
-            continue;
-        for (const image of serviceImages(input)) {
-            const profile = imageEngineFor(image);
-            if (profile)
-                upsert(profile, 'running', { kind: 'service-image', value: image, path: input.path });
-        }
-        for (const dependency of manifestDependencyNames(input)) {
-            const profile = LIBRARY_ENGINES[dependency.toLowerCase()];
-            if (profile)
-                upsert(profile, 'library', { kind: 'client-library', value: dependency, path: input.path });
-        }
-        for (const scheme of connectionUrlSchemes(input)) {
-            const profile = URL_SCHEME_ENGINES[scheme];
-            if (profile)
-                upsert(profile, 'configured', { kind: 'connection-url', value: `${scheme}://`, path: input.path });
-        }
-        for (const profile of configEngineSignals(input)) {
-            upsert(profile, 'configured', { kind: 'config-declaration', value: profile.engine, path: input.path });
-        }
-    }
-    return [...byEngine.values()]
-        .map((store) => ({
-        ...store,
-        // Cap and stabilize evidence for a deterministic, bounded record.
-        signals: store.signals
-            .slice()
-            .sort((left, right) => left.kind.localeCompare(right.kind) ||
-            left.path.localeCompare(right.path) ||
-            left.value.localeCompare(right.value))
-            .slice(0, 24),
-    }))
-        .sort((left, right) => left.engine.localeCompare(right.engine));
-}