From 4c693622e5a0a9a41f630a934f39a66b09bbcbe7 Mon Sep 17 00:00:00 2001
From: Ludger Sicking <sicking@comenius.de>
Date: Wed, 30 Oct 2024 09:25:43 +0100
Subject: [PATCH] Hilfsskripte, um Blog-Posts in Hugos "content" Ordner zu
 migrieren

---
 sb/scripts/asset-extensions-in-folder-Blog.sh | 23 +++++++++++
 .../move-blog-posts-to-hugo-content-dir.sh    | 39 +++++++++++++++++++
 2 files changed, 62 insertions(+)
 create mode 100644 sb/scripts/asset-extensions-in-folder-Blog.sh
 create mode 100755 sb/scripts/move-blog-posts-to-hugo-content-dir.sh

diff --git a/sb/scripts/asset-extensions-in-folder-Blog.sh b/sb/scripts/asset-extensions-in-folder-Blog.sh
new file mode 100644
index 0000000..d367ed3
--- /dev/null
+++ b/sb/scripts/asset-extensions-in-folder-Blog.sh
@@ -0,0 +1,23 @@
+#welche Dateiendungen sind in den Markdown-Dateien in Verweisen enthalten:
+grep -r -E '\.\w\w\w+\)' Blog/ | grep -v -E '\.html\)' \
+| grep -v -E '\.de\)' | grep -v -E '\.org\)' | grep -v -E '\.nrw\)' \
+| cut -d')' -f 1 | rev | cut -d'.' -f 1 | rev | sort -u
+
+# jpeg
+# jpg
+# JPG
+# mp4
+# pdf
+# png
+
+# 1970520 => AKRK.md
+# education => GwR-Tagung
+
+# $grep -r -E '\.\w\w\w+\)' Blog/ | grep -v -E '\.html\)' | grep -v -E '\.de\)' | grep -v -E '\.org\)' | grep -v -E '\.nrw\)' | cut -d')' -f 1 | grep educa
+# Blog/2024-09-17-GwR-Tagung.md:Unter dem spannenden Motto „...hier sollte eigentlich ein Titel stehen“ fand die diesjährige [GwR-Tagung](https://gwr.education
+
+# $grep -r -E '\.\w\w\w+\)' Blog/ | grep -v -E '\.html\)' | grep -v -E '\.de\)' | grep -v -E '\.org\)' | grep -v -E '\.nrw\)' | cut -d')' -f 1 | grep 1970
+# Blog/2024-10-02 AKRK.md:Wie OER insgesamt didaktische Prozesse verändern können und die Perspektive auf die Lernenden richten, beleuchtet auch die Forschung von [Eric Werth und Katherine Williams](https://doi.org/10.1080/02680513.2021.1970520
+
+#welche Zeilen in Markdown Dateien enthalten "Asset"-Links,, die nicht in den Wordpress-Uploads vorhanden sind?
+grep -r -E '\.\w\w\w+\)' Blog/ | grep -v -E '\.html\)' | grep -v -E '\.de\)' | grep -v -E '\.org\)' | grep -v -E '\.nrw\)' | cut -d')' -f 1 | grep -v 'oer\.comm'
diff --git a/sb/scripts/move-blog-posts-to-hugo-content-dir.sh b/sb/scripts/move-blog-posts-to-hugo-content-dir.sh
new file mode 100755
index 0000000..b889452
--- /dev/null
+++ b/sb/scripts/move-blog-posts-to-hugo-content-dir.sh
@@ -0,0 +1,39 @@
+#!/bin/bash
+set -e
+
+working_directory=$(basename $(pwd))
+if [ "$working_directory" != "FOERBICO" ]; then
+  echo "not in root dir 'FOERBICO'"
+  exit 1
+fi
+
+OURTMPDIR='tmp-8JmqbYUnu'
+
+# script im root dir ausführen
+find Blog -type f -name '*.md' | while read FILE ; do
+    echo 'erstelle Verzeichnis für Post' $FILE
+    post_directory=$(grep 'oerCommunityPermalink' "$FILE" | cut -d'/' -f 4) # "..." beachtet Leerzeichen im Dateinamen
+    
+    mkdir -p $OURTMPDIR
+    echo $post_directory >> $OURTMPDIR/permalink-all.txt
+
+    mkdir -p sb/content/$post_directory
+    
+    echo 'kopiere' $FILE 'ins Verzeichnis als index.md'
+    cp "$FILE" sb/content/$post_directory/index.md
+
+    #assets aus "wp-content" holen
+    asset_link_list=$(grep -Eo "(http|https)://[a-zA-Z0-9./?=_%:-]*" "$FILE" | grep wp-content | grep "oer\.community" | sed 's/http:/https:/g' | sort -u)
+
+    cd "sb/content/$post_directory/"
+
+    for wp_content_asset in $asset_link_list
+    do
+        asset_output_file_name=$(echo $wp_content_asset | rev | cut -d'/' -f 1 | rev)
+        curl -s --remote-name $wp_content_asset --output "$asset_output_file_name"
+        #rm $(basename $wp_content_asset)
+    done
+
+    cd ../../../
+
+done