Fix broken exporter - bookdarts - Extract KOReader generated highlights and publish them in a variety of formats↵
DIR Log
DIR Files
DIR Refs
---
DIR commit 411e4b7296ef59b11d8cf8cbc4985bc680ef91cb
DIR parent 2cb9ba25d0eb8d25575fa0b5c8912acf08f5db4d
HTML Author: Scarlett McAllister <no+reply@roygbyte.com>
Date: Tue, 19 Sep 2023 22:45:10 -0300
Fix broken exporter
Diffstat:
M bookdarts.sh | 122 +++++++++++++++++++++++++++++--
M export-from-device.sh | 16 ++++++++++------
2 files changed, 124 insertions(+), 14 deletions(-)
---
DIR diff --git a/bookdarts.sh b/bookdarts.sh
@@ -1,11 +1,33 @@
#!/bin/sh
+# Configuration
+HIGHLIGHTS_FILE="highlights-converted.json"
+PUBLISH_DIR="publish"
+GPH_MENU_FILE="$PUBLISH_DIR/index.gph"
+MARKDOWN_DOCUMENTS_DIR="$PUBLISH_DIR/documents"
+
+# Functions
log() {
printf "$1\n"
}
+standard_input_to_file() {
+ # Helper function, simply prints standard
+ # input into a given file.
+ # $1: Filename for menu
+ data=$(cat)
+ printf "$data" >> "$1"
+ return 1
+}
+
nth_element() {
- cat "$HIGHLIGHTS_FILE" | jq .documents.[$1]
+ cat "$HIGHLIGHTS_FILE" \
+ | tr -d '—' \
+ | tr -t '“”' '\"' \
+ | tr -t '—' '-' \
+ | tr -d '[:cntrl:]' \
+ | sed -E "s/[^a-zA-Z0-9?=&\-\[\]\{\}]//g" \
+ | jq .documents.[$1]
return 1
}
@@ -15,11 +37,13 @@ get_document_attribute() {
# $1: document as JSON object
# $2: name of the attribute.
# $3: index of attribute.
- case $2 in
- title) printf "$1" | jq .title ;;
- author) cat "$1" | jq .author ;;
- file) cat "$1" | jq .file ;;
- entry) cat "$1" | jq .entries[$3] ;;
+ data=$(cat)
+ case $1 in
+ title) printf "$data" | jq .title ;;
+ author) printf "$data" | jq .author ;;
+ file) printf "$data" | jq .file ;;
+ entries) printf "$data" | jq .entries ;;
+ entry) printf "$data" | jq .entries[$3] ;;
*) ;;
esac
}
@@ -34,12 +58,94 @@ get_entry_attribute() {
esac
}
-HIGHLIGHTS_FILE="highlights.json"
+create_entries_file() {
+ # $1: Filename
+ data=$(cat)
+ touch "$1"
+ TITLE=$(printf "$data" | get_document_attribute title)
+ AUTHOR=$(printf "$data" | get_document_attribute author)
+ printf "${TITLE}\n${AUTHOR}\n" >> "$1"
+ return 1
+}
+
+publish_entry() {
+ # $1: Filename to add entry into
+ data=$(cat)
+ log "Publishing to: $1"
+ TEXT=$(printf "$data" | get_entry_attribute text | fold -w 20)
+ CHAPTER=$(printf "$data" | get_entry_attribute chapter)
+ LINE=$(awk -v len=30 -v line=- "BEGIN{i=0;while(i<len){printf line;i++}}")
+ printf "\n$LINE\n$TEXT\n$CHAPTER\n" >> "$1"
+ return 1
+}
+
+create_gph_menu_item() {
+ # $1: type
+ # $2: desc
+ # $3: path
+ printf "[$1|$2|$3|||]\n"
+ return 1
+}
+
+# Main
DOCUMENTS_COUNT=$(cat "$HIGHLIGHTS_FILE" | jq .documents.[].title | wc -l)
log "Found $DOCUMENTS_COUNT documents."
+# Refresh the publishing directory.
+if [ -d "$PUBLISH_DIR" ]; then
+ log "Removing previously published files from $PUBLISH_DIR"
+ rm -r "$PUBLISH_DIR"
+fi
+mkdir -p "$MARKDOWN_DOCUMENTS_DIR"
+
+# Publish the new files, starting with gopher menu
+standard_input_to_file "$GPH_MENU_FILE" <<EOF
+\n
+ ,,,,.,,.,...
+ BOOKDARTS |''| |'''''''
+ | \ /
+ | \/ And sop
+ | What do we
+ | and so dog
+::::::::::::::::::::::::::::::::::::::::::::::::::
+\n
+
+EOF
INDEX=0
while [ $(( $INDEX < $DOCUMENTS_COUNT )) = 1 ]; do
- get_document_attribute "$(nth_element $INDEX)" title
+ printf "Publishing document $INDEX \n"
+ ITEM=$(nth_element $INDEX)
+ echo "$ITEM" > $INDEX.temp
+ TITLE=$(printf "$ITEM" | get_document_attribute title)
+ AUTHOR=$(printf "$ITEM" | get_document_attribute author)
+ DOCUMENT_FILENAME=$(printf "$TITLE - $AUTHOR" \
+ | sed -E 's/[^a-zA-Z0-9?=&\-\ ]//g' \
+ | sed -E 's/ /_/g' \
+ | tr '[:upper:]' '[:lower:]')
+ DOCUMENT_PUBLISHING_FILE="$MARKDOWN_DOCUMENTS_DIR/$DOCUMENT_FILENAME.txt"
+ ENTRIES_SORTED=$(printf "$ITEM" \
+ | get_document_attribute entries \
+ | jq 'sort_by(.page)')
+ ENTRIES_COUNT=$(printf "$ENTRIES_SORTED" \
+ | jq 'length')
+ # Create the file that'll contain all our highlights for
+ # this document.
+ printf "$ITEM" | create_entries_file $DOCUMENT_PUBLISHING_FILE
+ CURR_ENTRY=0
+ # Populate the file with highlights.
+ if [ $ENTRIES_COUNT > 0 ]; then
+ while [ $(( $CURR_ENTRY < $ENTRIES_COUNT )) = 1 ]; do
+ log "Publishing entry $CURR_ENTRY..."
+ printf "$ITEM" | get_document_attribute entry $CURR_ENTRY \
+ | publish_entry $DOCUMENT_PUBLISHING_FILE
+ CURR_ENTRY=$(( $CURR_ENTRY + 1 ))
+ done
+ # Construct gph menu item for this document's highlights file.
+ log "Constructing gph menu for document $INDEX..."
+ create_gph_menu_item "1" "$TITLE" \
+ "$DOCUMENT_PUBLISHING_FILE" \
+ >> $GPH_MENU_FILE
+ fi
+ # Prepare for next document.
INDEX=$(( $INDEX + 1 ))
done
DIR diff --git a/export-from-device.sh b/export-from-device.sh
@@ -4,20 +4,24 @@
BOOKDARTS_DIR=$PWD
DEVICE_DIR=/media/kobo
-HIGHLIGHTS_DIR=/.adds/koreader/clipboard
+HIGHLIGHTS_DIR=.adds/koreader/clipboard
HIGHLIGHTS_FILENAME_REGEX=".*-all-books.json"
-HIGHLIGHTS_FILE=$(ls -a "$DEVICE_DIR/$HIGHLIGHTS_DIR" | sort | grep -E "$HIGHLIGHTS_FILENAME_REGEX" | head -n 1)
+HIGHLIGHTS_FILE=$(ls -a "$DEVICE_DIR/$HIGHLIGHTS_DIR" \
+ | sort \
+ | grep -E "$HIGHLIGHTS_FILENAME_REGEX" \
+ | head -n 1)
if [ -z "$HIGHLIGHTS_FILE" ]; then
printf "Could not find highlights file. Exiting..."
exit 1
fi
-printf "Found highlights file: %s" "$HIGHLIGHTS_DIR/$HIGHLIGHTS_FILE\n"
+HIGHLIGHTS_FILE="$DEVICE_DIR/$HIGHLIGHTS_DIR/$HIGHLIGHTS_FILE"
+printf "Found highlights file: %s\n" "$HIGHLIGHTS_FILE"
-if [ ! -r "$HIGHLIGHTS_DIR/HIGHLIGHTS_FILE" ]; then
- printf "Could not read from file. Exiting..."
+if [ ! -r "$HIGHLIGHTS_FILE" ]; then
+ printf "Could not read from file. Exiting...\n"
exit 1
fi
-cp "$HIGHLIGHTS_DIR/$HIGHLIGHTS_FILE" "$BOOKDARTS_DIR"
+cp "$HIGHLIGHTS_FILE" "$BOOKDARTS_DIR/highlights-from-device.json"