From 165f85a443a245ec739ac377fa1b16e091a4d098 Mon Sep 17 00:00:00 2001
From: Bel LaPointe <breel@qualtrics.com>
Date: Tue, 1 Feb 2022 14:12:06 -0700
Subject: [PATCH] mvp

---
 app/crawler/main.sh  | 18 +++++++++---------
 app/crawler/notea.sh | 12 ++++++------
 2 files changed, 15 insertions(+), 15 deletions(-)

diff --git a/app/crawler/main.sh b/app/crawler/main.sh
index 185e89d..48593c1 100644
--- a/app/crawler/main.sh
+++ b/app/crawler/main.sh
@@ -40,12 +40,12 @@ crawl() {
 }
 
 _crawl() {
-   log crawling $*
+   log "crawling $*"
    local id="$1"
    local json="$(notea get "$id")"
    local content="$(echo "$json" | jq -r .content)"
    if ! is_crawlable "$content"; then
-      log $content is not crawlable
+      log "$content is not crawlable"
       return 0
    fi
    local crawlable_source="$(extract_crawlable_source "$content")"
@@ -60,7 +60,7 @@ _crawl() {
 }
 
 extract_crawlable_source() {
-   echo "$*" | head -n 1 | awk '{print $NF}'
+   echo "$*" | head -n 1 | awk '{print $NF}' | sed 's/^<//' | sed 's/>$//'
 }
 
 crawl_with() {
@@ -73,12 +73,12 @@ crawl_with() {
    local expanded=($($backend expand "$crawlable_source"))
    log expand $crawlable_source:
    for i in $(seq 1 $(("${#expanded[@]}"-1))); do
-      local title="$(echo "${expanded[i]}" | base64 --decode)"
-      CONTENT="$($backend get "$crawlable_source" "${expanded[i]}")" \
-         ID="$(echo "$crawlable_source/$title" | base64)" \
-         PID="$(echo $json | jq -r .id)" \
-         TITLE="$title" \
-         notea put
+      export TITLE="$(echo "${expanded[i]}" | base64 --decode)"
+      export CONTENT="$($backend get "$crawlable_source" "${expanded[i]}")"
+      export ID="$(echo "$crawlable_source/$TITLE" | base64 | md5sum | awk '{print $1}')"
+      export PID="$(echo $json | jq -r .id)"
+      log "   $PID/$ID ($TITLE): ${#CONTENT}"
+      notea put
    done
 
    log not impl crawl with
diff --git a/app/crawler/notea.sh b/app/crawler/notea.sh
index 4ec5ebe..474ea35 100644
--- a/app/crawler/notea.sh
+++ b/app/crawler/notea.sh
@@ -41,20 +41,20 @@ notea() (
       local xsrf_token="$(echo "$contains_tokens" | grep -o '"csrfToken":[^,]*' | tr ':' '\n' | jq -r . | tail -n 1)"
       local xsrf_cookie="$(echo "$contains_tokens" | grep ^set.cookie: | sed 's/^set.cookie: //' | tr ';' '\n' | head -n 1)"
       local request="$(echo '{
-         "content": '$(echo "$CONTENT" | jq -Rs)',
+         "content": '"$(printf "%s\n" "$CONTENT" | jq -Rs)"',
          "deleted": 0,
-         "id": '$(echo "$ID" | jq -R)',
-         "pid": '$(echo "$PID" | jq -R)',
+         "id": '"$(echo "$ID" | jq -R)"',
+         "pid": '"$(echo "$PID" | jq -R)"',
          "pinned": 0,
          "shared": 0,
-         "title": '$(echo "$TITLE" | jq -R)'
+         "title": '"$(echo "$TITLE" | jq -R)"'
       }' | jq -c .)"
-      ncurl \
+      echo "$request" | ncurl \
          -X POST \
          -H "$xsrf_key: $xsrf_token" \
          -b "$xsrf_cookie" \
          -H "Content-Type: application/json" \
-         -d "$request" \
+         -d @- \
          $NOTEA_ADDR/api/notes \
          | grep -q "$ID"
    }