getbib

#!/bin/bash
# Usage: getbib [AUXFILE_NAME] [-o|--output OUTPUTFILE_NAME]
# getbib exports a bib file containing only the cited terms according to the current aux
# file. $BAKDIR contains recent backups.

AUXFILE=""
OUTPUTFILE=""
TIME=$(date +%Y.%m.%d-%H.%M.%S)

# Log files
GETBIBLOG="getbib.log"
CHECKBIBLOG=checkbib.log

# Directory to contain the backups.
BAKDIR=./backup/


if [[ $# -gt 3 ]]; then
    printf "Usage: getbib [AUXFILE_NAME] [-o|--output OUTPUTFILE_NAME]\n"
    exit 1
fi

# Parse the arguments
while [[ -n "$1" ]]; do
    case "$1" in
        -o|--output)
            if [[ $# -gt 1 ]]; then
                OUTPUTFILE="$2"
                shift
            fi
            ;;
        *)
            if [[ -z "$AUXFILE" ]]; then
                AUXFILE="$1"
            fi
            ;;
    esac
    shift
done

# Get the name of the aux file with full path.
# If the input is empty, then use the latest aux file under the current directory.
if [[ $# -eq 0 ]] ||  [[ -z "$1" ]]; then
    AUXFILE=$(find ./ -maxdepth 1 -type f -name "*.aux" -print0 | xargs -r -0 ls -1 -t | head -1)
fi
AUXFILE=$(readlink -f "$AUXFILE")
AUXNAME=${AUXFILE##*/}
# AUXNAME is the name of the aux file without path or extension.
AUXNAME=${AUXNAME%.*}
if [[ -z "$AUXNAME" ]]; then
    printf "\nNo aux file found in the current directory.\n"
    exit 2
elif [[ ! -f "$AUXFILE" ]]; then
    printf "\n%s does not exist.\n" "$AUXFILE"
    exit 2
fi

# If no output file is specified, set OUTPUTFILE to $AUXNAME.bib.
if [[ -z "$OUTPUTFILE" ]]; then
    OUTPUTFILE="$AUXNAME".bib
fi
# Get the full path of $OUTPUTFILE. Note that we do not resolve the link if $OUTPUTFILE exists
# and is a symlink.
OUTPUTFILE="$(cd "$(dirname -- "$OUTPUTFILE")" && pwd)/$(basename -- "$OUTPUTFILE")"


# Temporary files.
# We work on OUTPUTBIB and copy it to OUTPUTFILE in the very end. This is necessary, because
# OUTPUTFILE may be identical to $AUXNAME.bib; we do not want to destroy $AUXNAME.bib until the very
# end, because the script may be interrupted in the middle.
OUTPUTBIB="outputbib-$TIME.bib"
BIBMACROS="bibmacros-$TIME"
BIBENTRIES="bibentries-$TIME"


# Export the cited entries.
printf "\nExporting bib entries ..."
rm -f "$BIBENTRIES" "$BIBMACROS"
touch "$BIBENTRIES" "$BIBMACROS"

BIBDATAFILES=$(grep bibdata "$AUXFILE" | sed 's/.*\\bibdata{//' | sed 's/}.*$//' | sed 's/,/\ /' | tr '\n' ' ')
CITEDBIBS=$(grep bibcite "$AUXFILE" | sed 's/.*\\bibcite{//' | sed 's/}.*$//' | sort -u)
#BIBDATAFILES=`grep bibdata "$AUXFILE" | sed 's/.*\\\\bibdata{//' | sed 's/}.*$//' | sed 's/,/\ /'`
#CITEDBIBS=`grep bibcite "$AUXFILE" | sed 's/.*\\\\bibcite{//' | sed 's/}.*$//' | sort -u`

CITEDBIBS_REDUCED="$CITEDBIBS"
for FILE in $BIBDATAFILES
do
    if [[ ! -f $FILE && -f $FILE.bib ]]; then
        FILE=$FILE.bib
    fi
    CITEDBIBS="$CITEDBIBS_REDUCED"
    for BIB in $CITEDBIBS
    do
        if grep -q "@.*{\s*$BIB\s*," "$FILE" ; then
            CITEDBIBS_REDUCED=$(echo "$CITEDBIBS_REDUCED" | sed "/^\s*$BIB\s*$/d")
            BIBENTRY=$(sed -n "/@.*{\s*$BIB/,/^\s*}/{p;/^\s*}/q}" "$FILE")
            echo "" >> "$BIBENTRIES"
            echo "$BIBENTRY" >> "$BIBENTRIES"
            BIBMACRO=$(echo "$BIBENTRY" | grep "journal\s*=\|journaltitle\s*=\|series\s*=" \
                | sed 's/.*journal\s*=\s*//' \
                | sed 's/.*journaltitle\s*=\s*//' \
                | sed 's/.*series\s*=\s*//' \
                | sed s/,.*$// | sed 's/{.*}//')
                if [[ -n $BIBMACRO ]]; then
                    echo "" >> "$BIBMACROS"
                    grep "@string{\s*$BIBMACRO" "$FILE" | sed 's/.*@string/@string/' >> "$BIBMACROS"
                fi
        fi
    done
done

if ! grep -q '[^[:space:]]' "$BIBENTRIES" ; then
    printf "\n%s is not an aux file containing citations or it refers to invalid bib files. Abort.\n" "$AUXFILE"
    rm "$BIBENTRIES" "$BIBMACROS"
    exit 3
fi

{
    printf "%% bib file generated according to %s.aux at %s.\n" "$AUXNAME" "$TIME"
    printf "%%\n"
    printf "%% The recommended format of the bib entry label is:\n"
    printf "%%\n"
    printf "%%     <AuthorLastnames_YearLetter>\n"
    printf "%%\n"
    printf "%%  For example, the key for an article by Yuan may be:\n"
    printf "%%\n"
    printf "%%     Yuan_2000a\n%%\n"
    printf "%% It is advised to use the abbreviations of names of serials in Math. Rev.\n"
    printf "%% See https://mathscinet.ams.org/msnhtml/serials.pdf .\n"
    printf "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n\n\n"
} > "$OUTPUTBIB"

if grep -q '[^[:space:]]' "$BIBMACROS" ; then
    {
        printf "%% String macros for abbreviations of names of serials.\n"
        printf "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%MACRO_BEGIN%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n"

        sed -r '/^\s*$/d' "$BIBMACROS" | sort -u
        printf "%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%MACRO_END%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%\n\n"
    } >> "$OUTPUTBIB"
fi
rm "$BIBMACROS"

cat "$BIBENTRIES" >> "$OUTPUTBIB"
rm "$BIBENTRIES"


# Make a backup of $OUTPUTFILE under $BAKDIR if $OUTPUTFILE exists.
FULLBIBNAME=${OUTPUTFILE//\//%}
BAKBIBFILE=$FULLBIBNAME-$TIME
BIBDIFFERENT=0
echo "" > $GETBIBLOG
if [[ -f "$OUTPUTFILE" ]]; then
    printf "******** Difference between the new bib and the original one ********\n" >> $GETBIBLOG
    # Remove the comments before comparing.
    diff <(grep -v "^\s*%" "$OUTPUTBIB") <(grep -v "^\s*%" "$OUTPUTFILE") >> $GETBIBLOG
    if [[ $? -eq 1 ]]; then
        BIBDIFFERENT=1
        # $BAKDIR/$FULLBIBNAME-latest_backup is always the latest backup.
        mkdir -p $BAKDIR
        cp "$OUTPUTFILE" "$BAKDIR/$FULLBIBNAME-latest_backup"
        mv "$OUTPUTFILE" "$BAKDIR/$BAKBIBFILE"
        # Keep at most 50 latest files under $BAKDIR.
        (cd $BAKDIR && ls -tp | grep -v '/$' | grep "$FULLBIBNAME" | tail -n +51 | xargs -I {} rm -- {})
    fi
fi

# Write the output to $OUTPUTFILE
mv "$OUTPUTBIB" "$OUTPUTFILE"


# Check $OUTPUTFILE.
# First, decide whether this is a BibLaTeX file or a BibTeX file.
if grep -q "journaltitle\s*=\s*\|date\s*=\s*\|location\s*=\s*\|type\s*=\s*" "$OUTPUTFILE" ; then
    # This is likely a BibLaTeX file.
    checkbib -l "$OUTPUTFILE"
else
    # This is likely a BibTeX file.
    checkbib "$OUTPUTFILE"
fi

echo "*************************** checkbib.log ****************************" | cat - "$CHECKBIBLOG" > "tmp-$TIME" && mv "tmp-$TIME" "$CHECKBIBLOG"
cat $GETBIBLOG  >> $CHECKBIBLOG
mv $CHECKBIBLOG $GETBIBLOG

echo "Done. See $GETBIBLOG for $CHECKBIBLOG and the difference between the new bib and the original one."

if [[ $BIBDIFFERENT -eq 1 ]]; then
    printf "\nA copy of the original bib file is saved at\n%s" "$BAKDIR/$BAKBIBFILE"
    printf "\nA copy of the original bib file is saved at\n%s" "$BAKDIR/$BAKBIBFILE" >> $GETBIBLOG
fi

echo ""

exit 0