notea-de-me/app/crawler/gitlab.sh

116 lines
2.9 KiB
Bash

#! /bin/bash
gitlab() (
is() {
echo "$*" | grep -q gitlab.app && ! echo "$*" | grep -q '/wikis/'
}
get() {
local project="$1"
local blob="$2"
project="$(urlencode "$project")"
blob="$(urlencode "$blob")"
local path="api/v4/projects/$project/repository/files/$blob/raw"
_gcurl "https://gitlab-app.eng.qops.net/$path"
}
expand() {
local cache_key="gitlab expand $*"
if cache get "$cache_key"; then
return 0
fi
_expand "$@" | sort | cache put "$cache_key"
}
_expand() {
local url="$1"
local project="$(_url_to_project_root "$url" | head -n 1)"
local root="$(_url_to_project_root "$url" | tail -n 1)"
__expand "$project" "$root"
}
_url_to_project_root() {
local url="$1"
local url_path="${url#http*://gitlab*.net/}"
local project="${url_path%%/-/*}"
local root="${url_path#*$project}"
local root="${root#*/-/}"
local root="${root#tree/}"
local root="${root#blob/}"
local root="${root#*/}"
echo "$project"
echo "$root"
}
__expand() {
local project="$1"
local root="${2:-"/"}"
local b64_files=()
local b64_trees=("$(echo "$root" | base64)")
local i=0
find_each() {
local type="$1"
shift
echo "$*" \
| jq -c .[] \
| grep "\"type\":\"$type\"" \
| jq -r .path \
| while read -r line; do echo "$line" | base64; done \
| grep .
}
while [ "$i" -lt "${#b64_trees[@]}" ]; do
got="$(_list_tree "$project" "$(echo "${b64_trees[i]}" | base64 --decode)")"
for b64_tree in $(find_each "tree" "$got"); do
if ! echo "${b64_trees[@]}" | grep -q "[ ^]$b64_tree[ $]"; then
b64_trees+=("$b64_tree")
fi
done
for b64_file in $(find_each "blob" "$got"); do
if ! echo "${b64_files[@]}" | grep -q "[ ^]$b64_file[ $]"; then
b64_files+=("$b64_file")
fi
done
i=$((i+1))
done
for b64_file in "${b64_files[@]}"; do
echo "$b64_file"
done
}
_list_tree() {
local project="$(urlencode "$1")"
local path="api/v4/projects/$project/repository/tree"
local query="recursive=true&path=$2"
_gcurl "https://gitlab-app.eng.qops.net/$path?$query"
}
_gcurl() {
local cache_key="gitlab _gcurl $*"
if cache get "$cache_key"; then
return 0
fi
__gcurl "$@" | cache put "$cache_key"
}
__gcurl() {
curl -sS -H "Authorization: Bearer $GITLAB_PAT" "$@"
}
"$@"
)
urlencode() (
LC_COLLATE=C
local length="${#1}"
for (( i = 0; i < length; i++ )); do
local c="${1:$i:1}"
case $c in
[a-zA-Z0-9.~_-]) printf '%s' "$c" ;;
*) printf '%%%02X' "'$c" ;;
esac
done
)