From 2c8abf7158f2751b8c9a1561b7fb5bd2ff5db14b Mon Sep 17 00:00:00 2001
From: Bel LaPointe <breel@qualtrics.com>
Date: Tue, 1 Feb 2022 11:30:15 -0700
Subject: [PATCH] gr

---
 app/crawler/gitlab.sh      | 13 ++++++++++---
 app/crawler/gitlab_test.sh | 24 +++++++++++++++++++++---
 app/crawler/main.sh        |  6 +++---
 3 files changed, 34 insertions(+), 9 deletions(-)

diff --git a/app/crawler/gitlab.sh b/app/crawler/gitlab.sh
index b193114..e69916a 100644
--- a/app/crawler/gitlab.sh
+++ b/app/crawler/gitlab.sh
@@ -6,10 +6,14 @@ gitlab() (
    }
 
    get() {
-      local project="$1"
+      local url="$1"
       local blob="$2"
+
+      local project="$(_url_to_project_root "$url" | head -n 1)"
       project="$(urlencode "$project")"
-      blob="$(urlencode "$blob")"
+      local root="$(_url_to_project_root "$url" | tail -n 1)"
+      blob="$(urlencode "$root/$blob")"
+
       local path="api/v4/projects/$project/repository/files/$blob/raw"
       _gcurl "https://gitlab-app.eng.qops.net/$path"
    }
@@ -75,7 +79,10 @@ gitlab() (
          i=$((i+1))
       done
       for b64_file in "${b64_files[@]}"; do
-         echo "$b64_file"
+         local file="$(echo "$b64_file" | base64 --decode)"
+         file="${file#$root}"
+         file="${file#/}"
+         echo "$file" | base64
       done
    }
 
diff --git a/app/crawler/gitlab_test.sh b/app/crawler/gitlab_test.sh
index 2c9eeb9..4d89dbf 100644
--- a/app/crawler/gitlab_test.sh
+++ b/app/crawler/gitlab_test.sh
@@ -20,9 +20,27 @@ test___expand() {
             ;;
          esac
       }
-      ! __expand project dir | grep -q ^$(echo project | base64)$ || return 1
-      __expand project dir | grep -q ^$(echo dir/blob | base64)$ || return 2
-      __expand project dir | grep -q ^$(echo dir/dir2/blob2 | base64)$ || return 3
+      local blob=false
+      local dir2blob2=false
+      local others=0
+      for result in \$(__expand project dir); do
+         if echo \$result | base64 --decode | grep -q ^blob$; then
+            blob=true
+         elif echo \$result | base64 --decode | grep -q ^dir2.blob2$; then
+            dir2blob2=true
+         else
+            others=\$((others+1))
+         fi
+      done
+      if [ \$others != 0 ]; then
+         return 101
+      fi
+      if ! \$blob; then
+         return 102
+      fi
+      if ! \$dir2blob2; then
+         return 103
+      fi
 EOF
 )"
 }
diff --git a/app/crawler/main.sh b/app/crawler/main.sh
index 912ad92..d1a1d8c 100644
--- a/app/crawler/main.sh
+++ b/app/crawler/main.sh
@@ -24,7 +24,7 @@ config() {
 }
 
 log() {
-   echo "$(date)> $*" >&2
+   echo "$(date +%H:%M:%S)> $*" >&2
 }
 
 ids() {
@@ -71,9 +71,9 @@ crawl_with() {
    local crawlable_source="$(extract_crawlable_source "$content")"
 
    local expanded=($($backend expand "$crawlable_source"))
-   local context="$expanded"
+   log expand $crawlable_source:
    for i in $(seq 1 "${#expanded[@]}"); do
-      log expand $(echo $context | base64 --decode), $(echo ${expanded[i]} | base64 --decode)
+      log "   $(echo ${expanded[i]} | base64 --decode)"
    done
 
    log not impl crawl with