From 529abd37e9ca8f9ef094ae748ce7c9bfad769b89 Mon Sep 17 00:00:00 2001 From: Bel LaPointe Date: Thu, 10 Feb 2022 10:49:35 -0700 Subject: [PATCH] neverends --- app/crawler/cache.sh | 2 +- app/crawler/gitlab.sh | 5 ++++- app/crawler/main.sh | 10 ++++++---- spike/review/reinvent/ezmded/server/server.go | 2 +- 4 files changed, 12 insertions(+), 7 deletions(-) diff --git a/app/crawler/cache.sh b/app/crawler/cache.sh index 52e6dc9..7c89f62 100644 --- a/app/crawler/cache.sh +++ b/app/crawler/cache.sh @@ -2,7 +2,7 @@ cache() ( path() { - echo "$CACHE/$(echo "$*" | base64)" + echo "$CACHE/$(echo "$*" | base64 | md5sum | awk '{print $1}')" } get() { local path="$(path "$*")" diff --git a/app/crawler/gitlab.sh b/app/crawler/gitlab.sh index bcb6ef6..f058a72 100644 --- a/app/crawler/gitlab.sh +++ b/app/crawler/gitlab.sh @@ -12,7 +12,10 @@ gitlab() ( local project="$(_url_to_project_root "$url" | head -n 1)" project="$(urlencode "$project")" local root="$(_url_to_project_root "$url" | tail -n 1)" - blob="$(urlencode "$root/$blob")" + blob="$(urlencode "$blob")" + if [ -n "$root" ]; then + blob="$(urlencode "$root/$blob")" + fi local path="api/v4/projects/$project/repository/files/$blob/raw" _gcurl "https://gitlab-app.eng.qops.net/$path" diff --git a/app/crawler/main.sh b/app/crawler/main.sh index ad6991b..4786338 100644 --- a/app/crawler/main.sh +++ b/app/crawler/main.sh @@ -65,7 +65,7 @@ _crawl() { } extract_crawlable_source() { - echo "$*" | head -n 1 | awk '{print $NF}' | sed 's/^$//' + echo "$*" | head -n 1 | awk '{print $NF}' | sed 's/^$//' | sed 's/^\///' | sed 's/\/$//' } crawl_with() { @@ -86,7 +86,7 @@ crawl_with() { log expand $crawlable_source: notes_mkdir_p() { local id="$1" - local subtitle="$2" + local subtitle="${2%/}" notes put "$id" "$subtitle" "autogenerated content" } one() { @@ -107,7 +107,9 @@ crawl_with() { for subtitle in $(echo "$i" | base64 --decode | tr '/' '\n' | while read -r subtitle; do echo "$subtitle" | base64; done); do local subtitle="$(echo "$subtitle" | base64 --decode)" if [ -n "$title_so_far" ]; then - notes_mkdir_p "${sum%/}" "${title_so_far%/}" >&2 + local mkdir_p_title="${title_so_far%/}" + mkdir_p_title="${mkdir_p_title##*/}" + notes_mkdir_p "${sum%/}" "${mkdir_p_title}" >&2 fi sum+="$(echo "$subtitle" | encode)/" title_so_far+="$subtitle/" @@ -118,7 +120,7 @@ crawl_with() { log " $ID ($TITLE): ${#CONTENT}" push_crawled "$PID/$ID" "$TITLE" "$CONTENT" } - for i in $(seq 1 $(("${#expanded[@]}"-1))); do + for i in $(seq 0 $(("${#expanded[@]}"-1))); do one "${expanded[i]}" done } diff --git a/spike/review/reinvent/ezmded/server/server.go b/spike/review/reinvent/ezmded/server/server.go index 45e4e14..1feb702 100644 --- a/spike/review/reinvent/ezmded/server/server.go +++ b/spike/review/reinvent/ezmded/server/server.go @@ -296,7 +296,7 @@ func (server *Server) urlFileId(id []string) string { func (server *Server) fileId(r *http.Request) []string { return strings.Split( - strings.TrimPrefix( + strings.Trim( strings.TrimPrefix( strings.Trim(r.URL.Path, "/"), "api/v0/files",