feat(search): add subcommand to search across the packaging group

Search for an expression across the GitLab packaging group. To use a
filter, include it in your query. You may use wildcards (*) to use glob
matching. Available filters for the blobs scope: path, extension.

Every usage of the search command must be authenticated. Consult the
'pkgctl auth' command to authenticate with GitLab or view the
authentication status.

This command uses bats for pretty printing the results including line
numbers and syntax highlighting.

Component: pkgctl search
Co-authored-by: Christian Heusel <christian@heusel.eu>
Co-authored-by: Levente Polyak <anthraxx@archlinux.org>

3 files changed, 410 insertions, 7 deletions

diff --git a/src/lib/api/gitlab.sh b/src/lib/api/gitlab.sh
index e5f4237..e4b8a9d 100644
--- a/src/lib/api/gitlab.sh
+++ b/src/lib/api/gitlab.sh
@@ -13,13 +13,63 @@ source "${_DEVTOOLS_LIBRARY_DIR}"/lib/config.sh
 
 set -e
 
+graphql_api_call() {
+	local outfile=$1
+	local request=$2
+	local node_type=$3
+	local data=$4
+	local hasNextPage cursor
+
+	# empty token
+	if [[ -z "${GITLAB_TOKEN}" ]]; then
+		msg_error "  api call failed: No token provided"
+		return 1
+	fi
+
+	[[ -z ${WORKDIR:-} ]] && setup_workdir
+	api_workdir=$(mktemp --tmpdir="${WORKDIR}" --directory pkgctl-gitlab-api.XXXXXXXXXX)
+
+	# normalize graphql data and prepare query
+	data="${data//\"/\\\"}"
+	data='{
+		"query": "'"${data}"'"
+	}'
+	data="${data//$'\t'/ }"
+	data="${data//$'\n'/}"
+
+	cursor=""
+	hasNextPage=true
+	while [[ ${hasNextPage} == true ]]; do
+		data=$(sed -E 's|after: \\"[a-zA-Z0-9]*\\"|after: \\"'"${cursor}"'\\"|' <<< "${data}")
+		result="${api_workdir}/result.${cursor}"
+
+		if ! curl --request "${request}" \
+				--url "https://${GITLAB_HOST}/api/graphql" \
+				--header "Authorization: Bearer ${GITLAB_TOKEN}" \
+				--header "Content-Type: application/json" \
+				--data "${data}" \
+				--output "${result}" \
+				--silent; then
+			msg_error "  api call failed: $(cat "${outfile}")"
+			return 1
+		fi
+
+		hasNextPage=$(jq --raw-output ".data | .${node_type} | .pageInfo | .hasNextPage" < "${result}")
+		cursor=$(jq --raw-output ".data | .${node_type} | .pageInfo | .endCursor" < "${result}")
+
+		cp "${result}" "${api_workdir}/tmp"
+		jq ".data.${node_type}.nodes" "${api_workdir}/tmp" > "${result}"
+	done
+
+	jq --slurp add "${api_workdir}"/result.* > "${outfile}"
+	return 0
+}
 
 gitlab_api_call() {
 	local outfile=$1
 	local request=$2
 	local endpoint=$3
 	local data=${4:-}
-	local error
 
 	# empty token
 	if [[ -z "${GITLAB_TOKEN}" ]]; then
@@ -38,27 +88,102 @@ gitlab_api_call() {
 		return 1
 	fi
 
+	if ! gitlab_check_api_errors "${outfile}"; then
+		return 1
+	fi
+
+	return 0
+}
+
+gitlab_api_call_paged() {
+	local outfile=$1
+	local request=$2
+	local endpoint=$3
+	local data=${4:-}
+	local result header
+
+	# empty token
+	if [[ -z "${GITLAB_TOKEN}" ]]; then
+		msg_error "  api call failed: No token provided"
+		return 1
+	fi
+
+	[[ -z ${WORKDIR:-} ]] && setup_workdir
+	api_workdir=$(mktemp --tmpdir="${WORKDIR}" --directory pkgctl-gitlab-api.XXXXXXXXXX)
+
+	next_page=1
+	while [[ -n "${next_page}" ]]; do
+		result="${api_workdir}/result.${next_page}"
+		header="${api_workdir}/header"
+		if ! curl --request "${request}" \
+				--get \
+				--url "https://${GITLAB_HOST}/api/v4/${endpoint}&per_page=100&page=${next_page}" \
+				--header "PRIVATE-TOKEN: ${GITLAB_TOKEN}" \
+				--header "Content-Type: application/json" \
+				--data-urlencode "${data}" \
+				--dump-header "${header}" \
+				--output "${result}" \
+				--silent; then
+			msg_error "  api call failed: $(cat "${result}")"
+			return 1
+		fi
+
+		if ! gitlab_check_api_errors "${result}"; then
+			return 1
+		fi
+
+		next_page=$(grep "x-next-page" "${header}" | tr -d '\r' | awk '{ print $2 }')
+	done
+
+	jq --slurp add "${api_workdir}"/result.* > "${outfile}"
+	return 0
+}
+
+gitlab_check_api_errors() {
+	local file=$1
+	local error
+
+	# search API only returns an array, no errors
+	if [[ $(jq --raw-output 'type' < "${file}") == "array" ]]; then
+		return 0
+	fi
+
 	# check for general purpose api error
-	if error=$(jq --raw-output --exit-status '.error' < "${outfile}"); then
+	if error=$(jq --raw-output --exit-status '.error' < "${file}"); then
 		msg_error "  api call failed: ${error}"
 		return 1
 	fi
 
 	# check for api specific error messages
-	if ! jq --raw-output --exit-status '.id' < "${outfile}" >/dev/null; then
-		if jq --raw-output --exit-status '.message | keys[]' < "${outfile}" &>/dev/null; then
+	if ! jq --raw-output --exit-status '.id' < "${file}" >/dev/null; then
+		if jq --raw-output --exit-status '.message | keys[]' < "${file}" &>/dev/null; then
 			while read -r error; do
 				msg_error "  api call failed: ${error}"
-			done < <(jq --raw-output --exit-status '.message|to_entries|map("\(.key) \(.value[])")[]' < "${outfile}")
-		elif error=$(jq --raw-output --exit-status '.message' < "${outfile}"); then
+			done < <(jq --raw-output --exit-status '.message|to_entries|map("\(.key) \(.value[])")[]' < "${file}")
+		elif error=$(jq --raw-output --exit-status '.message' < "${file}"); then
 			msg_error "  api call failed: ${error}"
 		fi
 		return 1
 	fi
-
 	return 0
 }
 
+graphql_check_api_errors() {
+	local file=$1
+	local error
+
+	# early exit if we do not have errors
+	if ! jq --raw-output --exit-status '.errors[]' < "${file}" &>/dev/null; then
+		return 0
+	fi
+
+	# check for api specific error messages
+	while read -r error; do
+		msg_error "  api call failed: ${error}"
+	done < <(jq --raw-output --exit-status '.errors[].message' < "${file}")
+	return 1
+}
+
 gitlab_api_get_user() {
 	local outfile username
 
@@ -81,6 +206,23 @@ gitlab_api_get_user() {
 	return 0
 }
 
+gitlab_api_get_project_name_mapping() {
+	local query=$1
+	local outfile
+
+	[[ -z ${WORKDIR:-} ]] && setup_workdir
+	outfile=$(mktemp --tmpdir="${WORKDIR}" pkgctl-gitlab-api.XXXXXXXXXX)
+
+	# query user details
+	if ! graphql_api_call "${outfile}" POST projects "${query}"; then
+		msg_warn "  Invalid token provided?"
+		exit 1
+	fi
+
+	cat "${outfile}"
+	return 0
+}
+
 # Convert arbitrary project names to GitLab valid path names.
 #
 # GitLab has several limitations on project and group names and also maintains
@@ -130,3 +272,21 @@ gitlab_api_create_project() {
 	printf "%s" "${path}"
 	return 0
 }
+
+# TODO: parallelize
+# https://docs.gitlab.com/ee/api/search.html#scope-blobs
+gitlab_api_search() {
+	local search=$1
+	local outfile
+
+	[[ -z ${WORKDIR:-} ]] && setup_workdir
+	outfile=$(mktemp --tmpdir="${WORKDIR}" pkgctl-gitlab-api.XXXXXXXXXX)
+
+	if ! gitlab_api_call_paged "${outfile}" GET "/groups/archlinux%2fpackaging%2fpackages/search?scope=blobs" "search=${search}"; then
+		return 1
+	fi
+
+	cat "${outfile}"
+
+	return 0
+}
diff --git a/src/lib/cache.sh b/src/lib/cache.sh
new file mode 100644
index 0000000..24056fa
--- /dev/null
+++ b/src/lib/cache.sh
@@ -0,0 +1,22 @@
+#!/hint/bash
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+[[ -z ${DEVTOOLS_INCLUDE_CACHE_SH:-} ]] || return 0
+DEVTOOLS_INCLUDE_CACHE_SH=1
+
+set -e
+
+readonly XDG_DEVTOOLS_CACHE_DIR="${XDG_CACHE_HOME:-$HOME/.cache}/devtools"
+
+get_cache_file() {
+	local filename=$1
+	local path="${XDG_DEVTOOLS_CACHE_DIR}/${filename}"
+
+	mkdir --parents -- "$(dirname -- "$path")"
+	if [[ ! -f ${path} ]]; then
+		touch -- "${path}"
+	fi
+
+	printf '%s' "${path}"
+}
diff --git a/src/lib/search.sh b/src/lib/search.sh
new file mode 100644
index 0000000..cf64db3
--- /dev/null
+++ b/src/lib/search.sh
@@ -0,0 +1,221 @@
+#!/bin/bash
+#
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+[[ -z ${DEVTOOLS_INCLUDE_SEARCH_SH:-} ]] || return 0
+DEVTOOLS_INCLUDE_SEARCH_SH=1
+
+_DEVTOOLS_LIBRARY_DIR=${_DEVTOOLS_LIBRARY_DIR:-@pkgdatadir@}
+# shellcheck source=src/lib/common.sh
+source "${_DEVTOOLS_LIBRARY_DIR}"/lib/common.sh
+# shellcheck source=src/lib/cache.sh
+source "${_DEVTOOLS_LIBRARY_DIR}"/lib/cache.sh
+# shellcheck source=src/lib/api/gitlab.sh
+source "${_DEVTOOLS_LIBRARY_DIR}"/lib/api/gitlab.sh
+
+source /usr/share/makepkg/util/message.sh
+
+set -eo pipefail
+
+
+pkgctl_search_usage() {
+	local -r COMMAND=${_DEVTOOLS_COMMAND:-${BASH_SOURCE[0]##*/}}
+	cat <<- _EOF_
+		Usage: ${COMMAND} [OPTIONS] QUERY
+
+		Search for an expression across the GitLab packaging group.
+
+		To use a filter, include it in your query. You may use wildcards (*) to
+		use glob matching.
+
+		Available filters for the blobs scope: path, extension
+
+		Every usage of the search command must be authenticated. Consult the
+		'pkgctl auth' command to authenticate with GitLab or view the
+		authentication status.
+
+		SEARCH TIPS
+		    Syntax  Description    Example
+		    ───────────────────────────────────────
+		    "       Exact search   "gem sidekiq"
+		    ~       Fuzzy search   J~ Doe
+		    |       Or             display | banner
+		    +       And            display +banner
+		    -       Exclude        display -banner
+		    *       Partial        bug error 50*
+		    \\       Escape         \\*md
+		    #       Issue ID       #23456
+		    !       Merge request  !23456
+
+		OPTIONS
+		    --json               Enable printing results in JSON
+		    --no-default-filter  Do not apply default filter (like -path:keys/pgp/*.asc)
+		    -h, --help           Show this help text
+
+		EXAMPLES
+		    $ ${COMMAND} linux
+		    $ ${COMMAND} '"pytest -v" +PYTHONPATH'
+_EOF_
+}
+
+pkgctl_search() {
+	if (( $# < 1 )); then
+		pkgctl_search_usage
+		exit 0
+	fi
+
+	# options
+	local search
+	local formatter=pretty
+	local use_default_filter=1
+
+	# variables
+	local default_filter="-path:keys/pgp/*.asc"
+	local graphql_lookup_batch=200
+	local output result query entries from until length
+	local project_name_cache_file project_name_lookup project_ids project_id project_name project_slice
+	local mapping_output path startline data
+
+	while (( $# )); do
+		case $1 in
+			-h|--help)
+				pkgctl_search_usage
+				exit 0
+				;;
+			--json)
+				formatter=json
+				shift
+				;;
+			--no-default-filter)
+				use_default_filter=0
+				shift
+				;;
+			--)
+				shift
+				break
+				;;
+			-*)
+				die "invalid argument: %s" "$1"
+				;;
+			*)
+				break
+				;;
+		esac
+	done
+
+	if (( $# == 0 )); then
+		pkgctl_search_usage
+		exit 1
+	fi
+
+	# assign search parameter
+	search="${*}"
+	if (( use_default_filter )); then
+		search+=" ${default_filter}"
+	fi
+
+	stat_busy "Querying gitlab search api"
+	output=$(gitlab_api_search "${search}")
+	stat_done
+
+	project_name_cache_file=$(get_cache_file gitlab/project_id_to_name)
+	lock 11 "${project_name_cache_file}" "Locking project name cache"
+	mapfile -t project_ids < <(
+		jq --raw-output '[.[].project_id] | unique[]' <<< "${output}" | \
+			grep --invert-match --file <(awk '{ print $1 }' < "${project_name_cache_file}" ))
+
+	stat_busy "Querying project names"
+	local entries="${#project_ids[@]}"
+	local until=0
+	while (( until < entries )); do
+		from=${until}
+		until=$(( until + graphql_lookup_batch ))
+		if (( until > entries )); then
+			until=${entries}
+		fi
+		length=$(( until - from ))
+
+		project_slice=("${project_ids[@]:${from}:${length}}")
+		printf -v projects '"gid://gitlab/Project/%s",' "${project_slice[@]}"
+		query='{
+			projects(after: "" ids: ['"${projects}"']) {
+				pageInfo {
+					startCursor
+					endCursor
+					hasNextPage
+				}
+				nodes {
+					id
+					name
+				}
+			}
+		}'
+		mapping_output=$(gitlab_api_get_project_name_mapping "${query}")
+
+		# update cache
+		while read -r project_id project_name; do
+			printf "%s %s\n" "${project_id}" "${project_name}" >> "${project_name_cache_file}"
+		done < <(jq --raw-output \
+			'.[] | "\(.id | rindex("/") as $lastSlash | .[$lastSlash+1:]) \(.name)"' \
+			<<< "${mapping_output}")
+	done
+	stat_done
+
+	# read project_id to name mapping from cache
+	declare -A project_name_lookup=()
+	while read -r project_id project_name; do
+		project_name_lookup[${project_id}]=${project_name}
+	done < "${project_name_cache_file}"
+
+	# close project name cache lock
+	lock_close 11
+
+	# output mode JSON
+	if [[ ${formatter} == json ]]; then
+		jq --from-file <(
+			for project_id in $(jq '.[].project_id' <<< "${output}"); do
+				project_name=${project_name_lookup[${project_id}]}
+				printf 'map(if .project_id == %s then . + {"project_name": "%s"} else . end) | ' \
+					"${project_id}" "${project_name}"
+			done
+			printf .
+		) <<< "${output}"
+		exit 0
+	fi
+
+	# pretty print each result
+	while read -r result; do
+		# read properties from search result
+		mapfile -t data < <(jq --raw-output ".data" <<< "${result}")
+		{ read -r project_id; read -r path; read -r startline; } < <(
+			jq --raw-output ".project_id, .path, .startline" <<< "${result}"
+		)
+		project_name=${project_name_lookup[${project_id}]}
+
+		# remove trailing newline for multiline results
+		if (( ${#data[@]} > 1 )) && [[ ${data[-1]} == "" ]]; then
+			unset "data[${#data[@]}-1]"
+		fi
+
+		# prepend empty lines to match startline
+		if (( startline > 1 )); then
+			mapfile -t data < <(
+				printf '%.0s\n' $(seq 1 "$(( startline - 1 ))")
+				printf "%s\n" "${data[@]}"
+			)
+		fi
+
+		bat \
+			--file-name="${project_name}/${path}" \
+			--line-range "${startline}:" \
+			--paging=never \
+			--force-colorization \
+			--map-syntax "PKGBUILD:Bourne Again Shell (bash)" \
+			--map-syntax ".SRCINFO:INI" \
+			--map-syntax "*install:Bourne Again Shell (bash)" \
+			--map-syntax "*sysusers*:Bourne Again Shell (bash)" \
+			--map-syntax "*tmpfiles*:Bourne Again Shell (bash)" \
+			--map-syntax "*.hook:INI" \
+			<(printf "%s\n" "${data[@]}")
+	done < <(jq --compact-output '.[]' <<< "${output}")
+}