lychee.toml

include_fragments = "full"
include_verbatim = false
root_dir = "."
no_progress = true
format = "compact"
fallback_extensions = ["mdx", "md"]

# Only the Mintlify site under docs/ is checked. These paths are legacy
# Docusaurus content, build output, or generated translations that are not part
# of the Mintlify navigation, so exclude them as lychee inputs.
exclude_path = [
  # Legacy Docusaurus content, pending removal
  "^\\./en/",
  "^\\./changelogs/",
  "^\\./_templates/",
  "^\\./_includes/",
  "^\\./_migration/",
  # Site source assets (redirects.json, favicon, logos, customization JS) that
  # are referenced by docs.json, not doc pages to link-check.
  "^\\./_site/",
  # Local scratch directory (untracked; the repo convention for temp files),
  # never present in CI.
  "^\\./tmp/",
  # Generated translations of the English site. These live both at the top
  # level (./zh/) and under snippets/ (./snippets/zh/), so match the locale
  # directory at any depth. The codes only ever name locale dirs, so this is
  # safe -- unlike the legacy dirs above, which must stay top-level-anchored.
  "/ar/",
  "/es/",
  "/fr/",
  "/ja/",
  "/ko/",
  "/pt-BR/",
  "/ru/",
  "/zh/",
  # Repo housekeeping files, not doc pages
  "^\\./README\\.md$",
  "^\\./AGENTS\\.md$",
  "^\\./changelog_entry_guidelines\\.md$",
  # Authoring templates, not site pages: their links are placeholders
  # (e.g. /get-started/quickstarts/example), so don't check them.
  "/_TEMPLATE\\.mdx?$",
  # The pg_clickhouse reference is a known-stale upstream copy whose outbound
  # links use the wrong anchor casing; exclude it as an input so its links are
  # not checked (the `pg_clickhouse` entry in `exclude` below only covers links
  # pointing *into* it, not the outbound links it contains). Fix separately.
  "/pg_clickhouse/",
]

exclude = [
  # Links into the legacy Docusaurus English tree, which is excluded from the
  # site (see exclude_path above). Skip them here too so they are not silently
  # satisfied by a placeholder rather than genuinely checked.
  "/en/",
  # OpenAPI-generated routes — no .mdx files on disk, runtime-only
  "api-reference/",
  "academic-overview\\.mdx#page-",
  "github\\.com",
  "youtube\\.com",
  # pg_clickhouse reference has many stale upstream anchors — fix separately
  "pg_clickhouse",
  # ClickStack/managed-postgres dashboard download buttons use a JSX expression
  # href: `<TrackedLink href={'https://.../examples/<name>-dashboard.json'}>`.
  # lychee cannot parse the `{'...'}` JSX wrapper and mangles it into a bogus
  # relative link (`{'https://...'}`), so it false-positives regardless of the
  # URL. The URLs are valid external S3 links (verified separately); exclude the
  # pattern so the JSX wrapper is not misread as a broken local path.
  "-dashboard\\.json",
  # Connection strings in docker-compose / config examples, not links.
  "mongodb://",
  "PLAINTEXT://",
  # chDB connection-string examples in docs tables (e.g.
  # `file:test.db?param1=value1&param2=value2`) that lychee reads as file:// URLs.
  # `test.db` is only ever an example database name, never a real link target.
  "test\\.db",
  # Concrete placeholder filenames listed to explain S3 wildcard expansion.
  # Keep the patterns narrow so real objects under these prefixes are checked.
  "clickhouse-public-datasets\\.s3\\.amazonaws\\.com/my-bucket/(?:some|another)_folder/some_file_[1-3]\\.csv$",
  "clickhouse-public-datasets\\.s3\\.amazonaws\\.com/my-test-bucket-768/(?:some|another)_prefix/some_file_[1-4]\\.csv$",
  # The base-aware homepage MCP routes are computed at runtime in JSX. Lychee
  # reads the identifier in `href={href}` as a literal `{href}` path.
  "(?:%7B|\\{)href(?:%7D|\\})$",
  # Local and placeholder hosts used in setup instructions and connection
  # examples. They are not expected to resolve from the link checker.
  "^https?://localhost(?::[0-9]+)?(?:/|$)",
  "^https?://hostname(?::[0-9]+)?(?:/|$)",
  "^https?://proxy(?::[0-9]+)?(?:/|$)",
  "^https://abc\\.def\\.clickhouse\\.cloud(?:/|$)",
]