From f5db6ddb21b6d7eb542c20e0c98bf8abe5332093 Mon Sep 17 00:00:00 2001
From: Jaidyn Ann <10477760+JadedCtrl@users.noreply.github.com>
Date: Sat, 6 May 2023 11:27:09 -0500
Subject: [PATCH] Init

---
 README.md          |   8 +++
 fedi-archive.sh    | 113 +++++++++++++++++++++++++++++++++++++++
 fedi-post.sh       | 129 +++++++++++++++++++++++++++++++++++++++++++++
 pleroma-migrate.sh |  54 +++++++++++++++++++
 pleroma-redate.sh  |  32 +++++++++++
 5 files changed, 336 insertions(+)
 create mode 100644 README.md
 create mode 100644 fedi-archive.sh
 create mode 100644 fedi-post.sh
 create mode 100755 pleroma-migrate.sh
 create mode 100644 pleroma-redate.sh

diff --git a/README.md b/README.md
new file mode 100644
index 0000000..c68ef01
--- /dev/null
+++ b/README.md
@@ -0,0 +1,8 @@
+# PLEROMA-MIGRATOR
+
+A horrific group of scripts that can help you copy your fedi posts to a new account. If the server's Pleroma and you have direct access to the database, you can even preserve post dates.
+
+See invididual script's --help messages, with pleroma-migrator.sh being the one that combines them all for you.
+
+
+License is CC0, ofc.
diff --git a/fedi-archive.sh b/fedi-archive.sh
new file mode 100644
index 0000000..80d539e
--- /dev/null
+++ b/fedi-archive.sh
@@ -0,0 +1,113 @@
+#!/bin/sh
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+# Name: fedi-archive.sh
+# Desc: Downloads (most) posts from a fedi account in parseable format.
+# Reqs: jq, curl
+# Date: 2023-05-06
+# Auth: @jadedctrl@jam.xwx.moe
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+
+
+# Given a JSON file containing /api/v1/accounts/$user/statuses output,
+# output a post at the given index from the file like so:
+#     FAVOURITE_COUNT DATE_POSTED POST_ID POST_URL
+#     [media: URL DESCRIPTION]
+#     [media: URL DESCRIPTION]
+#     [CONTENT…]
+# [] meaning "optional". There might be an arbitrary amount of Media: lines.
+output_post_of_index() {
+	local index="$1"
+	local file="$2"
+	jq -r --arg INDEX "$index" \
+'.[$INDEX|tonumber] | "\(.favourites_count) \(.created_at) \(.id) \(.url)
+\(.media_attachments[] | "media: " + .url + " " + .description)
+\(.content)"' \
+	< "$file"
+}
+
+
+# Fetch a list of a user's statuses, given their server and username.
+# `max_id` can be passed to return only messages older than said message.
+fetch_page() {
+	local server="$1"; local user="$2"; local max_id="$3"
+	local url="https://$server/api/v1/accounts/$user/statuses?exclude_replies=true&exclude_reblogs=true&limit=40"
+	if test -n "$max_id"; then
+		url="${url}&max_id=${max_id}"
+	fi
+	curl "$url"
+}
+
+
+# Given a JSON file containing /api/v1/accounts/$user/statuses output,
+# output each status into an individual file of the format of
+# output_post_of_index(); see its comment for more information.
+# Prints the ID of the last post of the file.
+archive_posts() {
+	local json_file="$1"
+	local prefix="$2"
+
+	local post_file="$prefix-$i"
+	local last_post_file=""
+	local i="0"
+
+	local output_ret=0
+	while test "$output_ret" -eq 0; do
+		post_file="$prefix-$i"
+		echo "$post_file" 1>&2
+		output_post_of_index "$i" "$json_file" \
+							 > "$post_file"
+		output_ret="$?"
+
+		if test -e "$post_file" -a -n "$(cat "$post_file")"; then
+			last_post_file="$post_file"
+        elif test -e "$post_file"; then
+			rm "$post_file"
+		fi
+		i="$(echo "$i + 1" | bc)"
+	done
+
+	head -1 "$last_post_file" \
+		| awk '{print $3}'
+}
+
+
+# Fetch all posts for the given user at given server.
+archive_all_posts() {
+	local server="$1"
+	local username="$2"
+	local temp="$(mktemp)"
+
+	fetch_page "$server" "$username" \
+			   > "$temp"
+
+	local page="1"
+	local next_id="$(archive_posts "$temp" "$page")"
+	while test -n "$next_id"; do
+		page="$(echo "$page + 1" | bc)"
+		echo "$next_id - $page…"
+		fetch_page "$server" "$username" "$next_id" \
+				   > "$temp"
+		next_id="$(archive_posts "$temp" "$page")"
+	done
+
+	rm "$temp"
+}
+
+
+usage() {
+	echo "usage: $(basename $0) username server" 1>&2
+	echo "" 1>&2
+	echo "$(basename $0) is a script that fetches all of a user's Mastodon/Pleroma" 1>&2
+	echo "posts for archival purposes." 1>&2
+	echo "Mainly for use with fedi-post.sh or pleroma-migrate.sh." 1>&2
+	exit 2;
+}
+
+
+USERNAME="$1"
+SERVER="$2"
+if test -z "$USERNAME" -o -z "$SERVER" -o "$1" = "-h" -o "$1" = "--help"; then
+   usage
+fi
+
+archive_all_posts "$SERVER" "$USERNAME"
diff --git a/fedi-post.sh b/fedi-post.sh
new file mode 100644
index 0000000..a412422
--- /dev/null
+++ b/fedi-post.sh
@@ -0,0 +1,129 @@
+#!/bin/sh
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+# Name: fedi-post.sh
+# Desc: Makes a new post using an post archived with fedi-archive.sh.
+# Reqs: curl, jq
+# Date: 2023-05-06
+# Auth: @jadedctrl@jam.xwx.moe
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+
+
+# Outputs only the status text of a user's post, with a link to the original
+# appended to the bottom.
+file_status() {
+	local file="$1"
+	local id="$(head -1 "$file" | awk '{print $3}')"
+	local url="$(head -1 "$file" | awk '{print $4}')"
+	tail +2 "$file" \
+		| grep -v ^media: \
+		| grep -v "$id" \
+		| sed 's%\\%\\\\%g' \
+		| sed 's%"%\\"%g' \
+		| sed -z 's%\n%\\n%g'
+	if test -n "$url"; then
+		printf '<br/>[<a href=\\"%s\\">Originala afiŝo</a>]\\n' "$url"
+	fi
+}
+
+
+# Turns a space-delimited list of uploaded media-IDs into a JSON array.
+media_json() {
+	local ids="$1"
+	echo "$ids" \
+		| sed 's%^ %%' \
+		| sed 's% $%%' \
+		| sed 's%^%["%' \
+		| sed 's% %","%g' \
+		| sed 's%$%"]%'
+}
+
+
+# Takes a post message and JSON-array of uploaded media-IDs, outputting
+# the post's appropriate JSON.
+post_json() {
+	local message="$1"
+	local media_ids="$2"
+	printf '{ "content_type": "text/html", "visibility": "unlisted"'
+	if test -n "$media_ids"; then
+		printf ', "media_ids": %s, ' "$media_ids"
+	fi
+	printf '"status": "%s" }\n' "$message"
+}
+
+
+# Upload a file to the fedi server with the given description.
+post_media() {
+	local media_file="$1"
+	local description="$2"
+
+	curl --request POST \
+		 --header "Authorization: Bearer $FEDI_AUTH" \
+		 --header "Content-Type: multipart/form-data" \
+		 --form "file=@$media_file" \
+		 --form "description=$description" \
+		 "https://jam.xwx.moe/api/v1/media"
+}
+
+
+# Post a status of the given message and JSON-array of uploaded media-IDs.
+post_status() {
+	local message="$1"
+	local media_ids="$2"
+
+	curl --request POST \
+		 --header "Authorization: Bearer $FEDI_AUTH" \
+		 --header "Content-Type: application/json" \
+		 --data "$(post_json "$message" "$media_ids")" \
+		 "https://jam.xwx.moe/api/v1/statuses"
+}
+
+
+# Take a post file generated by fedi-archive.sh, and post it.
+# Just *do it*. Why not? What're you scared of? Huh, huh? Huh?!
+post_archived_post() {
+	local file="$1"
+	IFS="
+"
+	local ids=""
+	for media in $(grep "^media: " "$file"); do
+		local url="$(echo "$media" | awk '{print $2}')"
+		local desc="$(echo "$media" | awk '{ $1=$2=""; print}' | sed 's%^ %%')"
+
+		curl -o "$(basename "$url")" "$url"
+		ids="$ids $(post_media "$(basename "$url")" "$desc" | jq -r '.id')"
+		rm "$(basename "$url")"
+	done
+
+	printf '%s ' "$(head -1 "$file" | awk '{print $1, $2}')"
+	post_status "$(file_status "$file")" "$(media_json "$ids")" \
+		| jq -r .uri
+}
+
+
+usage() {
+	echo "usage: $(basename "$0") ARCHIVED-POST" 1>&2
+	echo "" 1>&2
+	echo "Will post a new status with the same text and attachments as one" 1>&2
+	echo "from an archived post (in fedi-archive.sh format)." 1>&2
+	echo "Your authorization key must be borrowed from your web-browser and" 1>&2
+	echo 'placed in the $FEDI_AUTH environment variable.' 1>&2
+	exit 2
+}
+
+
+if test -z "$FEDI_AUTH"; then
+	echo 'You need to set the environment variable $FEDI_AUTH!' 1>&2
+	echo 'You can find your auth key by examining the "Authentication: Bearer" header' 1>&2
+	echo "used in requests by your server's web-client." 1>&2
+    echo 'In Firefox, F12→Network.' 1>&2
+	echo "" 1>&2
+	usage
+fi
+
+FILE="$1"
+if test -z "$FILE" -o "$1" = "-h" -o "$1" = "--help"; then
+	usage
+fi
+
+
+post_archived_post "$FILE"
diff --git a/pleroma-migrate.sh b/pleroma-migrate.sh
new file mode 100755
index 0000000..db83048
--- /dev/null
+++ b/pleroma-migrate.sh
@@ -0,0 +1,54 @@
+#!/bin/sh
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+# Name: pleroma-migrate.sh
+# Desc: Downloads posts from a pleroma account on a foreign server, to post them
+#       on your new server. Then directly edits Pleroma's database to match the
+#       new posts' dates with the original posts' dates.
+# Reqs: fedi-post.sh, fedi-archive.sh, pleroma-redate.sh
+# Date: 2023-05-06
+# Auth: @jadedctrl@jam.xwx.moe
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+
+usage() {
+	echo "usage: $(basename "$0") USERNAME OLD-SERVER" 1>&2
+	echo "" 1>&2
+	echo "Will archive all posts from your old account (fedi-archive.sh)," 1>&2
+	echo "post them to your new one (fedi-post.sh), and then modify Pleroma's" 1>&2
+	echo "database to match the copy-posts' creation dates with the original posts." 1>&2
+	echo 'The env variable $FEDI_AUTH must contain your authentication key from your browser.' 1>&2
+	exit 2
+}
+
+
+USERNAME="$1"
+SERVER="$2"
+if test -z "$USERNAME" -o -z "$SERVER" -o "$1" = "-h" -o "$1" = "--help"; then
+	usage
+fi
+
+
+mkdir archive
+cd archive/
+sh ../fedi-archive.sh "$USERNAME" "$SERVER"
+
+
+for file in ./*; do
+	sh ../fedi-post.sh "$file" \
+	   >> imports-data.txt
+done
+
+
+IFS="
+"
+
+echo "It's time to re-date your posts!"
+echo "Are you suuuuuuuuuuuuure you wanna risk your database? Do a backup, first!"
+echo "^C now, before you risk it! Hit ENTER, if you're sure."
+read
+sleep 5
+echo "Alright, then, you brave fellow! I'm touched you trust me so much, though :o"
+
+for line in $(cat imports-data.txt); do
+	sh ../pleroma-redate.sh "$(echo "$line" | awk '{print $3}')" \
+	   "$(echo "$line" | awk '{print $2}')"
+done
diff --git a/pleroma-redate.sh b/pleroma-redate.sh
new file mode 100644
index 0000000..c4dd10d
--- /dev/null
+++ b/pleroma-redate.sh
@@ -0,0 +1,32 @@
+#!/bin/sh
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+# Name: pleroma-redate.sh
+# Desc: Changes the creation-date of a post in Pleroma's database.
+# Reqs: psql, sudo
+# Date: 2023-05-06
+# Auth: @jadedctrl@jam.xwx.moe
+#―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
+
+usage() {
+	echo "usage: $(basename "$0") URL NEW-DATE" 1>&2
+	echo "" 1>&2
+	echo "Will change the stored date of an archived fedi post" 1>&2
+	echo "in fedi-archive.sh format, by editing directly Pleroma's database." 1>&2
+	echo "URL ought be the direct /object/ URL, and NEW-DATE ought be ISO-8601" 1>&2
+	echo "up to the milisecond." 1>&2
+	echo "Assumes you can sudo as 'postgres' and the database is called 'pleroma'." 1>&2
+	exit 2
+}
+
+
+URL="$1"
+NEWDATE="$2"
+if test -z "$URL" -o -z "$NEWDATE" -o "$1" = "-h" -o "$1" = "--help"; then
+	usage
+fi
+
+
+sudo -u postgres psql --dbname=pleroma \
+	 -c "UPDATE objects
+SET data = jsonb_set(data, '{published}', '\"$NEWDATE\"'::jsonb, false)
+WHERE CAST(data::json->'id' AS TEXT) = '\"$URL\"' ;"