From 6c3229eeb310eff5d229c0bd7ddb83c0bfdf7852 Mon Sep 17 00:00:00 2001
From: Bel LaPointe <breel@qualtrics.com>
Date: Tue, 1 Feb 2022 08:33:55 -0700
Subject: [PATCH] ok i need tests now

---
 app/crawler/gitlab.sh | 32 +++++++++++++++++++++-----------
 app/crawler/main.sh   | 16 +++++++++++++++-
 2 files changed, 36 insertions(+), 12 deletions(-)

diff --git a/app/crawler/gitlab.sh b/app/crawler/gitlab.sh
index d36815a..854da59 100644
--- a/app/crawler/gitlab.sh
+++ b/app/crawler/gitlab.sh
@@ -5,16 +5,13 @@ gitlab() (
       echo "$*" | grep -q gitlab.app && ! echo "$*" | grep -q '/wiki/'
    }
 
-   _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 $GITLAB_PAT" "$@"
+   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() {
@@ -76,8 +73,9 @@ gitlab() (
          done
          i=$((i+1))
       done
+      echo "$project" | base64
       for b64_file in "${b64_files[@]}"; do
-         echo "$b64_file" | base64 --decode
+         echo "$b64_file"
       done
    }
 
@@ -88,6 +86,18 @@ gitlab() (
       _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 $GITLAB_PAT" "$@"
+   }
+
    "$@"
 )
 
diff --git a/app/crawler/main.sh b/app/crawler/main.sh
index 78546bf..ea232df 100644
--- a/app/crawler/main.sh
+++ b/app/crawler/main.sh
@@ -49,7 +49,7 @@ crawl() {
    if ! is_crawlable "$content"; then
       return 0
    fi
-   local crawlable_source="$(echo "$content" | head -n 1 | awk '{print $NF}')"
+   local crawlable_source="$(extract_crawlable_source "$content")"
    for backend in gitlab; do
       if $backend is "$crawlable_source"; then
          crawl_with $backend "$json"
@@ -60,9 +60,23 @@ crawl() {
    return 1
 }
 
+extract_crawlable_source() {
+   echo "$*" | head -n 1 | awk '{print $NF}'
+}
+
 crawl_with() {
    local backend="$1"
    local json="$2"
+
+   local content="$(echo "$json" | jq -r .content)"
+   local crawlable_source="$(extract_crawlable_source "$content")"
+
+   local expanded=("$($backend expand "$crawlable_source")")
+   local context="$expanded"
+   for i in $(seq 1 "${#expanded[@]}"); do
+      log expand $context, ${expanded[i]}
+   done
+
    log not impl crawl with
    return 1
 }