make-docs

#!/bin/sh
# The source of this file is https://raw.githubusercontent.com/grafana/writers-toolkit/main/docs/make-docs.
# # `make-docs` procedure changelog
#
# Updates should conform to the guidelines in https://keepachangelog.com/en/1.1.0/.
# [Semantic versioning](https://semver.org/) is used to help the reader identify the significance of changes.
# Changes are relevant to this script and the support docs.mk GNU Make interface.

# ## 4.2.0 (2023-09-01)

# ### Added

# - Retry the initial webserver request up to ten times to allow for the process to start.
#   If it is still failing after ten seconds, an error message is logged.

# ## 4.1.1 (2023-07-20)

# ### Fixed

# - Replaced use of `realpath` with POSIX compatible alternative to determine default value for REPOS_PATH.

# ## 4.1.0 (2023-06-16)

# ### Added

# - Mounts of `layouts` and `config` directories for the `website` project.
#   Ensures that local changes to mounts or shortcodes are reflected in the development server.

# ### Fixed

# - Version inference for versioned docs pages.
#   Pages in versioned projects now have the `versioned: true` front matter set to ensure that "version" in $.Page.Scratch is set on builds.

# ## 4.0.0 (2023-06-06)

# ### Removed

# - `doc-validator/%` target.
#   The behavior of the target was not as described.
#   Instead, to limit `doc-validator` to only specific files, refer to https://grafana.com/docs/writers-toolkit/writing-guide/tooling-and-workflows/validate-technical-documentation/#run-on-specific-files.

# ## 3.0.0 (2023-05-18)

# ### Fixed

# - Compatibility with the updated Make targets in the `website` repository.
#   `docs` now runs this script itself, `server-docs` builds the site with the `docs` Hugo environment.

# ## 2.0.0 (2023-05-18)

# ### Added

# - Support for the grafana-cloud/frontend-observability/faro-web-sdk project.
# - Use of `doc-validator` v2.0.x which includes breaking changes to command line options.

# ### Fixed

# - Source grafana-cloud project from website repository.

# ### Added

# - Support for running the Vale linter with `make vale`.

# ## 1.2.1 (2023-05-05)

# ### Fixed

# - Use `latest` tag of `grafana/vale` image by default instead of hardcoded older version.
# - Fix mounting multiple projects broken by the changes in 1.0.1

# ## 1.2.0 (2023-05-05)

# ### Added

# - Support for running the Vale linter with `make vale`.

# ### Fixed

# ## 1.1.0 (2023-05-05)

# ### Added

# - Rewrite error output so it can be followed by text editors.

# ### Fixed

# - Fix `docs-debug` container process port.

# ## 1.0.1 (2023-05-04)

# ### Fixed

# - Ensure complete section hierarchy so that all projects have a visible menu.

# ## 1.0.0 (2023-05-04)

# ### Added

# - Build multiple projects simultaneously if all projects are checked out locally.
# - Run [`doc-validator`](https://github.com/grafana/technical-documentation/tree/main/tools/cmd/doc-validator) over projects.
# - Redirect project root to mounted version.
#   For example redirect `/docs/grafana/` to `/docs/grafana/latest/`.
# - Support for Podman or Docker containers with `PODMAN` environment variable.
# - Support for projects:
#   - agent
#   - enterprise-logs
#   - enterprise-metrics
#   - enterprise-traces
#   - grafana
#   - grafana-cloud
#   - grafana-cloud/machine-learning
#   - helm-charts/mimir-distributed
#   - helm-charts/tempo-distributed
#   - incident
#   - loki
#   - mimir
#   - oncall
#   - opentelemetry
#   - phlare
#   - plugins
#   - slo
#   - tempo
#   - writers-toolkit


set -ef

readonly DOCS_CONTAINER="${DOCS_CONTAINER:-make-docs}"
readonly DOCS_HOST_PORT="${DOCS_HOST_PORT:-3002}"
readonly DOCS_IMAGE="${DOCS_IMAGE:-grafana/docs-base:latest}"

readonly DOC_VALIDATOR_INCLUDE="${DOC_VALIDATOR_INCLUDE:-.+\.md$}"
readonly DOC_VALIDATOR_SKIP_CHECKS="${DOC_VALIDATOR_SKIP_CHECKS:-^image-}"

readonly HUGO_REFLINKSERRORLEVEL="${HUGO_REFLINKSERRORLEVEL:-WARNING}"
readonly VALE_MINALERTLEVEL="${VALE_MINALERTLEVEL:-error}"
readonly WEBSITE_EXEC="${WEBSITE_EXEC:-make server-docs}"
# If set, the docs-base image will run a prebuild script that sets up Hugo mounts.
readonly WEBSITE_MOUNTS="${WEBSITE_MOUNTS:-}"

PODMAN="$(if command -v podman >/dev/null 2>&1; then echo podman; else echo docker; fi)"

if ! command -v curl >/dev/null 2>&1; then
  if ! command -v wget >/dev/null 2>&1; then
    errr 'either `curl` or `wget` must be installed for this script to work.'

    exit 1
  fi
fi

if ! command -v "${PODMAN}" >/dev/null 2>&1; then
  errr 'either `podman` or `docker` must be installed for this script to work.'

  exit 1
fi


about() {
  cat <<EOF
Test documentation locally with multiple source repositories.

The REPOS_PATH environment variable is a colon (:) separated list of paths in which to look for project repositories.
EOF
}

usage() {
  cat <<EOF
Usage:
  REPOS_PATH=<PATH[:<PATH>...]> $0 [<PROJECT>[:<VERSION>[:<REPO>[:<DIR>]]]...]

Examples:
  REPOS_PATH=~/ext/grafana/ $0 writers-toolkit tempo:latest helm-charts/mimir-distributed:latest:mimir:docs/sources/mimir-distributed
EOF
}

if [ $# -lt 1 ]; then
  cat <<EOF >&2
ERRR: arguments required but not supplied.

$(about)

$(usage)
EOF
  exit 1
fi

readonly REPOS_PATH="${REPOS_PATH:-$(cd "$(git rev-parse --show-toplevel)/.." && echo "${PWD}")}"

if [ -z "${REPOS_PATH}" ]; then
  cat <<EOF >&2
ERRR: REPOS_PATH environment variable is required but has not been provided.

$(usage)
EOF
  exit 1
fi

SOURCES_as_code='as-code-docs'
SOURCES_enterprise_metrics='backend-enterprise'
SOURCES_enterprise_metrics_='backend-enterprise'
SOURCES_grafana_cloud='website'
SOURCES_grafana_cloud_alerting_and_irm_machine_learning='machine-learning'
SOURCES_grafana_cloud_alerting_and_irm_slo='slo'
SOURCES_grafana_cloud_k6='k6-docs'
SOURCES_grafana_cloud_data_configuration_integrations='cloud-onboarding'
SOURCES_grafana_cloud_frontend_observability_faro_web_sdk='faro-web-sdk'
SOURCES_helm_charts_mimir_distributed='mimir'
SOURCES_helm_charts_tempo_distributed='tempo'
SOURCES_opentelemetry='opentelemetry-docs'
SOURCES_plugins_grafana_splunk_datasource='splunk-datasource'

VERSIONS_as_code='UNVERSIONED'
VERSIONS_grafana_cloud='UNVERSIONED'
VERSIONS_grafana_cloud_alerting_and_irm_machine_learning='UNVERSIONED'
VERSIONS_grafana_cloud_alerting_and_irm_slo='UNVERSIONED'
VERSIONS_grafana_cloud_k6='UNVERSIONED'
VERSIONS_grafana_cloud_data_configuration_integrations='UNVERSIONED'
VERSIONS_grafana_cloud_frontend_observability_faro_web_sdk='UNVERSIONED'
VERSIONS_opentelemetry='UNVERSIONED'
VERSIONS_technical_documentation='UNVERSIONED'
VERSIONS_website='UNVERSIONED'
VERSIONS_writers_toolkit='UNVERSIONED'

PATHS_grafana_cloud='content/docs/grafana-cloud'
PATHS_helm_charts_mimir_distributed='docs/sources/helm-charts/mimir-distributed'
PATHS_helm_charts_tempo_distributed='docs/sources/helm-charts/tempo-distributed'
PATHS_mimir='docs/sources/mimir'
PATHS_tempo='docs/sources/tempo'
PATHS_website='content/docs'

# identifier STR
# Replace characters that are not valid in an identifier with underscores.
identifier() {
  echo "$1" | tr -C '[:alnum:]_\n' '_'
}

# aget ARRAY KEY
# Get the value of KEY from associative array ARRAY.
# Characters that are not valid in an identifier are replaced with underscores.
aget() {
  eval echo '$'"$(identifier "$1")_$(identifier "$2")"
}

# new_proj populates a new project structure.
new_proj() {
  _project="$1"
  _version="$2"
  _repo="$3"
  _path="$4"

  # If version is not set, use the script mapping of project to default versions if it exists.
  # Fallback to 'latest'.
  if [ -z "${_version}" ]; then
    if [ -z "$(aget VERSIONS "${_project}")" ]; then
      _version=latest
    else
      _version="$(aget VERSIONS "${_project}")"
    fi
  fi

  # If repo is not set, use the script mapping of project to repo name if it exists.
  # Fallback to using the project name.
  if [ -z "${_repo}" ]; then
    if [ -z "$(aget SOURCES "${_project}")" ]; then
      _repo="${_project}"
    else
      _repo="$(aget SOURCES "${_project}")"
    fi
  fi

  # If path is not set, use the script mapping of project to docs sources path if it exists.
  # Fallback to using 'docs/sources'.
  if [ -z "${_path}" ]; then
    if [ -z "$(aget PATHS "${_project}")" ]; then
      _path="docs/sources"
    else
      _path="$(aget PATHS "${_project}")"
    fi
  fi

  echo "${_project}:${_version}:${_repo}:${_path}"
  unset _project _version _repo _path
}

# proj_url returns the webserver URL for a project.
# It expects a complete project structure as input.
proj_url() {
  IFS=: read -r _project _version _ _ <<POSIX_HERESTRING
$1
POSIX_HERESTRING

  if [ "${_project}" = 'website' ]; then
    echo "http://localhost:${DOCS_HOST_PORT}/docs/"

    unset _project _version
    return
  fi

  if [ -z "${_version}" ] || [ "${_version}" = 'UNVERSIONED' ]; then
    echo "http://localhost:${DOCS_HOST_PORT}/docs/${_project}/"
  else
    echo "http://localhost:${DOCS_HOST_PORT}/docs/${_project}/${_version}/"
  fi

  unset _project _version
}

# proj_ver returns the version for a project.
# It expects a complete project structure as input.
proj_ver() {
  IFS=: read -r _ _ver _ _ <<POSIX_HERESTRING
$1
POSIX_HERESTRING

  echo "${_ver}"
  unset _ver
}

# proj_dst returns the container path to content source for a project.
# It expects a complete project structure as input.
proj_dst() {
  IFS=: read -r _project _version _ _ <<POSIX_HERESTRING
$1
POSIX_HERESTRING

  if [ "${_project}" = 'website' ]; then
    echo '/hugo/content/docs'

    unset _project _version
    return
  fi

  if [ -z "${_version}" ] || [ "${_version}" = 'UNVERSIONED' ]; then
    echo "/hugo/content/docs/${_project}"
  else
    echo "/hugo/content/docs/${_project}/${_version}"
  fi

  unset _project _version
}

# repo_path returns the host path to the project repository.
# It looks for the provided repository name in each of the paths specified in the REPOS_PATH environment variable.
repo_path() {
  _repo="$1"
  IFS=:
  for lookup in ${REPOS_PATH}; do
    if [ -d "${lookup}/${_repo}" ]; then
      echo "${lookup}/${_repo}"
      unset _path _repo
      return
    fi
  done
  unset IFS

  errr "could not find project '${_repo}' in any of the paths in REPOS_PATH '${REPOS_PATH}'."
  note "you must have a checkout of the project '${_repo}' at '${REPOS_PATH##:*}/${_repo}'."
  note "if you have cloned the repository into a directory with a different name, consider changing it to ${_repo}."

  unset _repo
  exit 1
}

# proj_src returns the host path to content source for a project.
# It expects a complete project structure as input.
# It looks for the provided repository name in each of the paths specified in the REPOS_PATH environment variable.
proj_src() {
  IFS=: read -r _ _ _repo _path <<POSIX_HERESTRING
$1
POSIX_HERESTRING

  _repo="$(repo_path "${_repo}")"
  echo "${_repo}/${_path}"

  unset _path _repo
}

# proj_canonical returns the canonical absolute path partial URI for a project.
# It expects a complete project structure as input.
proj_canonical() {
  IFS=: read -r _project _version _ _ <<POSIX_HERESTRING
$1
POSIX_HERESTRING

  if [ "${_project}" = 'website' ]; then
    echo '/docs'

    unset _project _version
    return
  fi

  if [ -z "${_version}" ] || [ "${_version}" = 'UNVERSIONED' ]; then
    echo "/docs/${_project}"
  else
    echo "/docs/${_project}/${_version}"
  fi

  unset _project _version
}

proj_to_url_src_dst_ver() {
  _url="$(proj_url "$1")"
  _src="$(proj_src "$1")"
  _dst="$(proj_dst "$1")"
  _ver="$(proj_ver "$1")"

  echo "${_url}^${_src}^${_dst}^${_ver}"
  unset _url _src _dst _ver
}

url_src_dst_vers() {
  for arg in "$@"; do
    IFS=: read -r _project _version _repo _path <<POSIX_HERESTRING
$arg
POSIX_HERESTRING

    case "${_project}" in
      # Workaround for arbitrary mounts where the version field is expected to be the local directory
      # and the repo field is expected to be the container directory.
      arbitrary)
        echo "${_project}^${_version}^${_repo}^" # TODO
        ;;
      logs)
        proj_to_url_src_dst_ver "$(new_proj loki "${_version}")"
        proj_to_url_src_dst_ver "$(new_proj enterprise-logs "${_version}")"
        ;;
      metrics)
        proj_to_url_src_dst_ver "$(new_proj mimir "${_version}")"
        proj_to_url_src_dst_ver "$(new_proj helm-charts/mimir-distributed "${_version}")"
        proj_to_url_src_dst_ver "$(new_proj enterprise-metrics "${_version}")"
        ;;
      traces)
        proj_to_url_src_dst_ver "$(new_proj tempo "${_version}")"
        proj_to_url_src_dst_ver "$(new_proj enterprise-traces "${_version}")"
        ;;
      *)
        proj_to_url_src_dst_ver "$(new_proj "${_project}" "${_version}" "${_repo}" "${_path}")"
        ;;
    esac
  done

  unset _project _version _repo _path
}

await_build() {
  url="$1"
  req="$(if command -v curl >/dev/null 2>&1; then echo 'curl -s -o /dev/null'; else echo 'wget -q'; fi)"

  i=1
  max=10
  while [ "${i}" -ne "${max}" ]
  do
    sleep 1
    debg "Retrying request to webserver assuming the process is still starting up."
    i=$((i + 1))

    if ${req} "${url}"; then
      echo
      echo "View documentation locally:"
      for x in ${url_src_dst_vers}; do
        IFS='^' read -r url _ _ <<POSIX_HERESTRING
$x
POSIX_HERESTRING

        if [ -n "${url}" ]; then
          if [ "${_url}" != "arbitrary" ]; then
            echo "  ${url}"
          fi
        fi
      done
      echo
      echo 'Press Ctrl+C to stop the server'

      unset i max req url
      return
    fi
  done

  echo
  errr 'The build was interrupted or a build error occurred, check the previous logs for possible causes.'
  unset i max req url
}

debg() {
  if [ -n "${DEBUG}" ]; then
    echo "DEBG: $1" >&2
  fi
}

errr() {
  echo "ERRR: $1" >&2
}

note() {
  echo "NOTE: $1" >&2
}

url_src_dst_vers="$(url_src_dst_vers "$@")"

volumes=""
redirects=""

for arg in "$@"; do
  IFS=: read -r _project _ _repo _ <<POSIX_HERESTRING
${arg}
POSIX_HERESTRING
  if [ "${_project}" = website ]; then
    _repo="$(repo_path website)"
    volumes="--volume=${_repo}/config:/hugo/config"
    volumes="${volumes} --volume=${_repo}/layouts/partials:/hugo/layouts/partials"
    volumes="${volumes} --volume=${_repo}/layouts/shortcodes:/hugo/layouts/shortcodes"
  fi
  unset _project _repo
done

for x in ${url_src_dst_vers}; do
  IFS='^' read -r _url _src _dst _ver <<POSIX_HERESTRING
$x
POSIX_HERESTRING

  if [ "${_url}" != "arbitrary" ]; then
    if [ ! -f "${_src}/_index.md" ]; then
      errr "Index file '${_src}/_index.md' does not exist."
      note "Is '${_src}' the correct source directory?"
      exit 1
    fi
  fi

  debg "DEBG: Mounting '${_src}' at container path '${_dst}'"

  if [ -z "${volumes}" ]; then
    volumes="--volume=${_src}:${_dst}"
  else
    volumes="${volumes} --volume=${_src}:${_dst}"
  fi

  if [ -n "${_ver}" ] && [ "${_ver}" != 'UNVERSIONED' ]; then
    if [ -z "${redirects}" ]; then
      redirects="${_dst}^${_ver}"
    else
      redirects="${redirects} ${_dst}^${_ver}"
    fi
  fi
  unset _url _src _dst _ver
done

IFS=':' read -r image _ <<POSIX_HERESTRING
${DOCS_IMAGE}
POSIX_HERESTRING

case "${image}" in
  'grafana/doc-validator')
    proj="$(new_proj "$1")"
    echo
    "${PODMAN}" run \
                --init \
                --interactive \
                --name "${DOCS_CONTAINER}" \
                --platform linux/amd64 \
                --rm \
                --tty \
                ${volumes} \
                "${DOCS_IMAGE}" \
                "--include=${DOC_VALIDATOR_INCLUDE}" \
                "--skip-checks=${DOC_VALIDATOR_SKIP_CHECKS}" \
                /hugo/content/docs \
                "$(proj_canonical "${proj}")" | sed "s#$(proj_dst "${proj}")#sources#"
    ;;
  'grafana/vale')
    proj="$(new_proj "$1")"
    echo
    "${PODMAN}" run \
                --init \
                --interactive \
                --name "${DOCS_CONTAINER}" \
                --platform linux/amd64 \
                --rm \
                --tty \
                ${volumes} \
                "${DOCS_IMAGE}" \
                "--minAlertLevel=${VALE_MINALERTLEVEL}" \
                --config=/etc/vale/.vale.ini \
                --output=line \
                /hugo/content/docs | sed "s#$(proj_dst "${proj}")#sources#"
    ;;
  *)
    tempfile="$(mktemp -t make-docs.XXX)"
    cat <<EOF >"${tempfile}"
#!/usr/bin/env bash
for redirect in ${redirects}; do
  IFS='^' read -r path ver <<<"\${redirect}"
  echo -e "---\\nredirectURL: \"\${path/\/hugo\/content/}\"\\ntype: redirect\\nversioned: true\\n---\\n" > "\${path/\${ver}/_index.md}"
done

for x in "${url_src_dst_vers}"; do
  IFS='^' read -r _ _ dst _ <<<"\${x}"

  while [[ -n "\${dst}" ]]; do
    touch "\${dst}/_index.md"
    dst="\${dst%/*}"
  done
done

if [[ -n "${WEBSITE_MOUNTS}" ]]; then
  unset WEBSITE_SKIP_MOUNTS
fi

${WEBSITE_EXEC}
EOF
    chmod +x "${tempfile}"
    volumes="${volumes} --volume=${tempfile}:/entrypoint"
    readonly volumes

    IFS='' read -r cmd <<EOF
${PODMAN} run \
  --env=HUGO_REFLINKSERRORLEVEL=${HUGO_REFLINKSERRORLEVEL} \
  --init \
  --interactive \
  --name=${DOCS_CONTAINER} \
  --platform=linux/amd64 \
  --publish=${DOCS_HOST_PORT}:3002 \
  --publish=3003:3003 \
  --rm \
  --tty \
  ${volumes} \
  ${DOCS_IMAGE} \
  /entrypoint
EOF
    await_build http://localhost:3002 &

    if [ -n "${DEBUG}" ]; then
      ${cmd}
    else
      ${cmd} 2>&1| sed \
                     -e '/Web Server is available at http:\/\/localhost:3003\/ (bind address 0.0.0.0)/ d' \
                     -e '/^hugo server/ d' \
                     -e '/fatal: not a git repository (or any parent up to mount point \/)/ d' \
                     -e '/Stopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set)./ d' \
                     -e "/Makefile:[0-9]*: warning: overriding recipe for target 'docs'/ d" \
                     -e "/docs.mk:[0-9]*: warning: ignoring old recipe for target 'docs'/ d" \
                     -e '/\/usr\/bin\/make -j 2 proxy hserver-docs HUGO_PORT=3003/ d' \
                     -e '/website-proxy/ d' \
                     -e '/rm -rf dist*/ d' \
                     -e '/Press Ctrl+C to stop/ d' \
                     -e '/make/ d' || echo
    fi
    ;;
esac