#!/bin/bash
#  $HeadURL: https://svn.pasta.freemyip.com/main/miniade/trunk/bin/nop-sh $ $LastChangedRevision: 10133 $

#  Modules
. $(miniade) || { echo "${0##*/}: ERROR: miniade failed (hint: run 'miniade' to see error)" >&2; exit 1; }

#  Configurable stuff

#  Other globals
OSMLINK_PRESERVE_FLAG=false
DESC_SIMPLIFY_FLAG=true

#  INSTRUCTIONS
#
#  1) Go to https://overpass-turbo.eu/
#  2) In the left-hand panel replace all the code there with this:
#
#      <query type="node">
#       <has-kv k="tourism" v="camp_site"/>
#       <bbox-query {{bbox}}/>
#      </query>
#      <print/>
#      <query type="way">
#       <has-kv k="tourism" v="camp_site"/>
#       <bbox-query {{bbox}}/>
#      </query>
#      <union>
#       <item/>
#       <recurse type="down"/>
#      </union>
#      <print/>
#  3) In the right-hand panel click the little-box icon (tooltip is 
#     "manually select bbox")
#  4) Select the desired area.
#  5) Click Run.
#  6) Wait.
#  7) Click Export and export as GPX file.
#  8) Create a new directory.
#  9) Call this script from inside the empty directory.
#
#  To deduplicate several (rectangular and possibly overlapping) files
#  then use:
#
#      md5sum ??/*[0-9]* | { 
#          echo ".width 100"
#          echo "CREATE TABLE files (file, md5sum, primary key (file));"
#          echo "BEGIN TRANSACTION;"
#          while read MD5SUM FILE; do 
#              echo "INSERT INTO files VALUES ('$FILE', '$MD5SUM');"
#          done
#          echo "END TRANSACTION;"
#          echo "SELECT GROUP_CONCAT(file,' ') FROM files GROUP BY MD5SUM HAVING COUNT(*) != 1;"
#      } | sqlite3 -column -noheader | sed -r 's/^.{10}//' | xargs rm 
#
#  To then merge those into one file do:
#
#      cat at/atpro */*[0-9]* at/atepi > all.gpx
#  
#  and then strip the UTF line at the top of the file (GORP won't load it
#  with that present).
#
#  To strip that down to a sensible number that are within 50K of the route:
#
#      #  This should create an arc file (see https://www.gpsbabel.org/htmldoc-development/fmt_arc.html)
#      #  but seems not to order the points correctly, which means the
#      #  arc's "corridor" is much wider than it should be, so don't
#      #  use this:
#      #gpsbabel -i gpx -f all.gpx -o arc -F all.arc
#      #  What works better is just to extract the route points out of the
#      #  route file:
#      sed -rn 's/.*<rtept lat="(.*)" lon="(.*)">/\1\t\2/p' seven-countries-route-gpswandern.gpx > seven-countries-route-gpswandern.arcman
#      #  and then get Gpsbabel to remove waypoints from all.gpx that are
#      #  not close to the arc:
#      gpsbabel -i gpx -f all.gpx -x arc,file=seven-countries-route-gpswandern.arcman,distance=50K -o gpx -F all-arcman50K.gpx

main()
{
    local MY_ARGS
    #  Process options
    special_opts_handler()
    {
        case $1 in
            *)      return 1 ;;
        esac
    }
    miniade_process_options --help-handler=help --special-opts-handler=special_opts_handler MY_ARGS "$@" && set -- "${MY_ARGS[@]}"

    #  Process arguments
    [ $# = 1 ] || miniade_bad_usage
    FILE=$1

    #  Sanity checks and derivations
    [[ $FILE =~ \.gpx$ ]] || miniade_error "$FILE: doesn't have .gpx suffix"
    OUTPUT_DIR=${FILE%.gpx}
    COUNTRY_ABBR=${OUTPUT_DIR##*/}
    [[ $COUNTRY_ABBR =~ ^[a-z][a-z]$ ]] || miniade_error "$COUNTRY_ABBR: doesn't look like a country abbreviation"
    [ -f $FILE ] || miniade_error "$FILE: not accessible"
    [ ! -e $OUTPUT_DIR ] || miniade_error "$OUTPUT_DIR: exists already"
 
    #  Guts
    miniade_info "$FILE: pre-processing file ..."
    TMPFILE=/tmp/$PROGNAME.$$.gpx
    cp $FILE $TMPFILE
    #  csplit breaks on *line-boundaries* not on the specified tag, so
    #  ensure all tags are on separate lines.  
    perl -0777 -pi -e 's/</\n</g' $TMPFILE
    #  But join the names back together (we'll need to do that some time
    #  so we might as well do it now).
    perl -0777 -pi -e 's/\n(?=<\/name>)//g' $TMPFILE
    #  Squeeze blank lines
    sed -i '/^ *$/d' $TMPFILE

    miniade_info "$FILE: splitting into waypoints and tracks ..."
    mkdir $OUTPUT_DIR
    cd $OUTPUT_DIR
    csplit -f $COUNTRY_ABBR -k -b %04d $TMPFILE '/\(<trk>\|<wpt\)/' '{*}' > /dev/null || true

    miniade_info "$FILE: saving prologue and epilogue .."
    FIRSTXXFILE=${COUNTRY_ABBR}0000
    LASTXXFILE=$(ls $COUNTRY_ABBR* | sed -n '$p')
    mv $FIRSTXXFILE ${COUNTRY_ABBR}pro
    perl -0777 -pi -e 's/<\/gpx>\n//' $LASTXXFILE
    echo "</gpx>" > ${COUNTRY_ABBR}epi

    for XXFILE in $COUNTRY_ABBR[0-9]*; do
        grep -q trkseg $XXFILE || continue
        miniade_info "$FILE/$XXFILE: converting track into waypoint file ..."
        DESC_LINES=()
        IN_DESC_LINES_FLAG=false
        HAD_FIRST_TRKPT_FLAG=false
        NAME_LINE=
        OSMLINK_LINE=
        while read LINE; do
            if [[ $LINE =~ ^\<trk\>$ ]]; then
                :
            elif [[ $LINE =~ ^\<name\>(.*)\</name\>$ ]]; then
                NAME_LINE=$LINE
            elif [[ $LINE =~ ^\</desc\>$ ]] && $IN_DESC_LINES_FLAG; then
                IN_DESC_LINES_FLAG=false
            elif { ! $IN_DESC_LINES_FLAG; } && [[ $LINE =~ ^\<desc\>(.+)$ ]]; then
                IN_DESC_LINES_FLAG=true
                DESC_LINES=( "${DESC_LINES[@]}" "${BASH_REMATCH[1]}" )
            elif $IN_DESC_LINES_FLAG; then
                DESC_LINES=( "${DESC_LINES[@]}" "$LINE" )
            elif [[ $LINE =~ ^\<link ]]; then
                OSMLINK_LINE="$LINE"
            elif [[ $LINE =~ ^\<trkseg\>$ ]]; then
                :
            elif [[ $LINE =~ ^\<trkpt\ (lat=\".*\"\ lon=\".*\")/\>$ ]]; then
                if ! $HAD_FIRST_TRKPT_FLAG; then
                    FIRST_TRKPT_LAT_LON=${BASH_REMATCH[1]}
                    HAD_FIRST_TRKPT_FLAG=true
                fi
            elif [[ $LINE =~ ^\</trkseg\>$ ]]; then
                :
            elif [[ $LINE =~ ^\</trk\>$ ]]; then
                :
            else
                miniade_error "$XXFILE: $LINE: can't parse"
            fi
        done < $XXFILE
        $HAD_FIRST_TRKPT_FLAG || miniade_error "$XXFILE: track has no trackpoints"
        [ ${#DESC_LINES[*]} -ge 1 ] || miniade_error "$XXFILE: desc looks short"
        [ "X$NAME_LINE" != X ] || miniade_error "$XXFILE: no name"
        [ "X$OSMLINK_LINE" != X ] || miniade_error "$XXFILE: no link"
        {
           echo "<wpt $FIRST_TRKPT_LAT_LON>"
           echo "$NAME_LINE"
           echo -n "<desc>"
           for DESC_LINE in "${DESC_LINES[@]}"; do
               echo "$DESC_LINE"
           done
           echo "</desc>"
           echo "$OSMLINK_LINE"
           echo "</wpt>"
        } > $XXFILE
    done

    if $DESC_SIMPLIFY_FLAG; then
        miniade_info "simplifying description ..."
        for XXFILE in $COUNTRY_ABBR[0-9]*; do
            if ! [[ $(cat $XXFILE) =~ ^(.*)\<desc\>(.*)\</desc\>(.*)$ ]]; then
                miniade_error "$XXFILE: failed to parse"
            fi
            PRE_DESC=${BASH_REMATCH[1]}
            DESC=${BASH_REMATCH[2]}
            POST_DESC=${BASH_REMATCH[3]}
            FOUND_FLAG=false
            for KEY in website url source ref email; do
                if [[ $DESC =~ $KEY=(http[^$'\n']*).* ]]; then
                    URL=${BASH_REMATCH[1]}
                    FOUND_FLAG=true
                    break
                fi
            done
            if $FOUND_FLAG; then
                perl -0777 -pi -e "s@<desc>.*?</desc>@<desc>$URL</desc>@ms" $XXFILE
            else
                miniade_warning "$XXFILE: URL not found"
                #  Previously I erased the description entirely with this:
                #
                #      perl -0777 -pi -e 's@<desc>.*?</desc>\n@@ms' $XXFILE
                #
                #  But, given that there may be some less useful information there
                #  (e.g. address, telno) then now we just leave it.
            fi
        done
    fi
}

help()
{
    local PROGNAME

    miniade_get_progname PROGNAME
    echo "Usage: $PROGNAME [ <standard-options> ] <country-code> <file>"
    exit 0
}

main "$@"