reftoget

#!/bin/bash
# Usage: reftoget [-b|--bib BIB_FILE_NAME] [-d|--dir REFERENCE_DIRECTORY]
# reftoget records the references in BIB_FILE_NAME but not in REFERENCE_DIRECTORY. The result is
# written in a bib file named reftoget.bib.
# If BIB_FILE_NAME is not present, then all the bib files in the current directory will be checked.
# If REFERENCE_DIRECTORY is not present, then we check the default reference directory defined below.

# The directory where this scrip resides
TDIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"
# Some configuration, which should define DEFAULT_REFDIR
# shellcheck source=./tconfig
source "$TDIR/tconfig"

REFTOGET="reftoget.bib"
rm -f "$REFTOGET"

# Parse the arguments
if [[ $# -gt 4 ]]; then
    printf "Usage: reftoget [-b|--bib BIB_FILE_NAME] [-d|--dir REFERENCE_DIRECTORY]\n"
    exit 1
fi

while [[ -n "$1" ]]; do
	case "$1" in
    -b|--bib)
        if [[ $# -gt 1 ]]; then
            BIBFILE="$2"
            shift
        fi
        ;;
    -d|--dir)
        if [[ $# -gt 1 ]]; then
            REFFILE="$2"
            shift
        fi
        ;;
	*)
        printf "Usage: reftoget [-b|--bib BIB_FILE_NAME] [-d|--dir REFERENCE_DIRECTORY]\n"
	esac
	shift
done


# Decide the bib files to check
if ! [[ -z "$BIBFILE" ]]; then
    BIBFILES="$BIBFILE"
else
    BIBFILES=$(ls -t *.bib 2>/dev/null)
fi


# Decide the reference directory
if [[ -z "$REFDIR" ]]; then
    REFDIR="$DEFAULT_REFDIR"
fi

LINEBREAKER="LINEBREAKERSTARTING$(date +%s)LINEBREAKERENDING"
TIME=$(date +%Y.%m.%d-%H.%M.%S)
TMPDIR="$(mktemp -d)"
TMPBIB="$TMPDIR"/tmpbib-"$TIME"
GOTREF=$(ls "$REFDIR" | sed 's/\..*$//' | sed 's/-.*$//' | sort -u)
touch "$REFTOGET"
for FILE in $BIBFILES; do
    cat "$FILE" | awk -v ORS="$LINEBREAKER" 1 | sed "s/$LINEBREAKER\s*@/\n@/g" \
        | sed '/^\s*%/d' | sed '/^\s*@string/d' | sed '/^\s*@STRING/d' | sed '/^\s*@String/d' > "$TMPBIB"
    while IFS= read -r LINE
    do
        BIBENTRYID=$(echo "$LINE" | sed 's/^\s*@[^{]*{\s*//' | sed 's/,.*$//')
        if ! [[ -z "$BIBENTRYID" ]]; then
            sed -i "/^\s*@.*{$BIBENTRYID.*$/d" "$TMPBIB"
        fi
    done < "$REFTOGET"
    cat "$TMPBIB" >> "$REFTOGET"
    for REF in $GOTREF; do
        sed -i "/^\s*@.*{$REF.*$/d" "$REFTOGET"
    done
done

sed -i "s/$LINEBREAKER/\n/g" "$REFTOGET"

rm -f "$TMPBIB"

exit 0