From 92226f9aea91ce0ebe3b7ede5981f9559f5c66e4 Mon Sep 17 00:00:00 2001
From: Bel LaPointe <breel@qualtrics.com>
Date: Thu, 10 Feb 2022 08:35:42 -0700
Subject: [PATCH] sparkles

---
 app/crawler/main.sh | 33 +++++++++++++++++++++++++++------
 1 file changed, 27 insertions(+), 6 deletions(-)

diff --git a/app/crawler/main.sh b/app/crawler/main.sh
index 7a0243f..a780b45 100644
--- a/app/crawler/main.sh
+++ b/app/crawler/main.sh
@@ -41,7 +41,7 @@ crawl() {
 }
 
 _crawl() {
-	log "crawling $*"
+	log "crawling? $*"
 	local id="$1"
 	local content="$(notes get "$id")"
 	local json="$(
@@ -77,21 +77,42 @@ crawl_with() {
 
 	local expanded=($($backend expand "$crawlable_source"))
 	log expand $crawlable_source:
+	notes_mkdir_p() {
+		local id="$1"
+		local subtitle="$2"
+		if ! notes get "$id" | grep -q '^404 page not found$'; then
+			return
+		fi
+		notes put "$id" "$subtitle" "autogenerated content"
+	}
 	one() {
+		encode() {
+			base64 | md5sum | cut -c 1-10 | awk '{print $1}' | tr -d '\n'
+		}
 		local i="$1"
 		export TITLE="$(
 			echo "$i" | base64 --decode
 		)"
+		TITLE="${TITLE##*/}"
 		export CONTENT="$(
 			$backend get "$crawlable_source" "$i"
 		)"
+		local pid="$(echo $json | jq -r .id)"
 		export ID="$(
-			echo "$crawlable_source/$TITLE" | base64 | md5sum | awk '{print $1}'
+			local sum="$pid/"
+			local title_so_far=""
+			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
+				fi
+				sum+="$(echo "$subtitle" | encode)/"
+				title_so_far+="$subtitle/"
+			done
+			echo "$sum"
 		)"
-		export PID="$(
-			echo $json | jq -r .id
-		)"
-		log "	$PID/$ID ($TITLE): ${#CONTENT}"
+		ID="${ID%/}"
+		log "	$ID ($TITLE): ${#CONTENT}"
 		push_crawled "$PID/$ID" "$TITLE" "$CONTENT"
 	}
 	for i in $(seq 1 $(("${#expanded[@]}"-1))); do