Skip to main content

gpx2tcx Shell Script

· 6 min read
Paul Colby
Paul Colby
gpx2tcx.sh screenshot

Following on from my gpx2tcx AWK script and Windows batch file, I now present what will probably the last separate component in the series... my gpx2tcx.sh Bash shell script.

This script is similar in nature to the Windows equivalent presented earlier, however it is considerably more flexible. Unlike the Windows batch file, this Bash script supports a number of command line arguments that allow its operation to be tweaked more easily.

The script's basic help message is as follows:

Usage: gpx2tcx [options] [base_filename[.gpx|hrm]] ... [base_filename[.gpx|hrm]]

Allowed options:
  -a meters  user meters as the starting altitude if no HRM altitude data present
  -f         overwrite destination file if any exists
  -h         show this help text and exit
  -s sport   set TCX activity type to sport; should be one of "Biking", "Running", or "Other";
             if not set, will use "Biking" if cadence data is present, otherwise "Running"
  -v         report this script's version and exit
  -z offset  replace UTZ "Z" timezone suffixes with offset (eg "+11:00")

If called with no arguments, this script will perform essentially the same as the Windows batch file - ie will attempt to convert all GPX files in the current directory that do not yet have corresponding TCX files. However, as you can see, the script allows you to set explicit GPX / HRM files to convert on the command line.

I dare say the -f, -h and -v options are self-explanatory, but I'll take just a second to explain the other two options.

The -s option allows you to specify the sport or TCX activity type. The TCX format includes the concept of an "sport type", which of course, specifies that type of sport the TCX file has recorded data for. According to the TCX schema, valid sport types are: Running, Biking, and Other, so you should probably only use one of those.

As noted in the help message above, if -s not specified, the following default behaviour is used: if a corresponding HRM file is being used, and that HRM file includes cadence information, then the default sport will be Biking, otherwise Running will be used instead. Note, this default behaviour is actually implemented in the AWK script (so exists for the Windows batch file too), this Bash script merely provides an easy way to override that behaviour.

Finally, the -z option may be used to specify the correct timezone to used in place of any UTC timezones found in the GPX / TCX files. This is because although the Polar RCX5 exports data with local timestamps, those timestamps all purport to be UTC instead - which is wrong, obviously. So specifying a timezone is an easy way to have those erroneous UTC timestamps corrected (the Windows batch file has equivalent functionality, but its not quite so easy to use).

So, here's the script:

#!/bin/bash
# gpx2tcx.sh by Paul Colby (https://colby.id.au), no rights reserved ;)
# $Id: gpx2tcx.sh 263 2012-02-11 03:16:18Z paul $

function showUsage {
  SCRIPT_NAME=`basename $0 .sh`
  echo -e "\nUsage: $SCRIPT_NAME [options] [base_filename[.gpx|hrm]] ... [base_filename[.gpx|hrm]]\n"
  echo 'Allowed options:'
  echo '  -a meters  user meters as the starting altitude if no HRM altitude data present'
  echo '  -f         overwrite destination file if any exists'
  echo '  -h         show this help text and exit'
  echo '  -s sport   set TCX activity type to sport; should be one of "Biking", "Running", or "Other";'
  echo '             if not set, will use "Biking" if cadence data is present, otherwise "Running"'
  echo "  -v         report this script's version and exit"
  echo '  -z offset  replace UTZ "Z" timezone suffixes with offset (eg "+11:00")'
  echo
}

function showVersion {
  # Note, Subversion will automatically update the following line.
  SCRIPT_VERSION=( 1 0 0 `echo '$Revision: 263 $' | sed -e 's/.*: *\([0-9]\+\).*$/\1/'` )
  echo "${SCRIPT_VERSION[0]}.${SCRIPT_VERSION[1]}.${SCRIPT_VERSION[2]}.${SCRIPT_VERSION[3]}"
}

function parseCommandLine {
  unset ALTITUDE
  unset FORCE
  unset SPORT
  unset TIMEZONE
  if [ -r ~/".`basename $0 .sh`" ]; then . ~/".`basename $0 .sh`"; fi
  local -a ERRORS
  while getopts ':a:fhs:vz:' OPTION; do
    case $OPTION in
      a) ALTITUDE="$OPTARG"   ;;
      f) FORCE=Y              ;;
      h) showUsage   ; exit 0 ;;
      s) SPORT="$OPTARG"      ;;
      v) showVersion ; exit 0 ;;
      z) TIMEZONE="$OPTARG"   ;;
      :) ERRORS[${#ERRORS[@]}]="Option -$OPTARG requires an argument." ;;
     \?) ERRORS[${#ERRORS[@]}]="Option -$OPTARG not known."            ;;
      *) ERRORS[${#ERRORS[@]}]="Option -$OPTION not implemented yet."  ;;
    esac
  done
  shift $(( $OPTIND - 1))

  if [ ${#ERRORS[@]} -gt 0 ]; then
    for ERROR in "${ERRORS[@]}"; do
      echo "$ERROR" >&2
    done
    showUsage ; exit 1
  fi

  # All positional arguments are [base] filenames.
  unset $FILENAMES
  for POSITIONAL_ARGUMENT; do
    FILENAMES[${#FILENAMES[@]}]=$POSITIONAL_ARGUMENT
  done

  # If no filenames where specified, look for GPX/HRM files in the current directory.
  if [ ${#FILENAMES[@]} -eq 0 ]; then
    local PREVIOUS_IFS="$IFS"
    IFS=$'\n'
    FILENAMES=( `find . -name '*.gpx' | sed -e 's/\.[^.]*$//' | sort | uniq` )
    if [ -z "$FORCE" ]; then
      # Not forcing output, so ignore any input files that have already have corresponding TCX files.
      FILENAMES=( $( for NAME in "${FILENAMES[@]}"; do if [ ! -e "$NAME.tcx" ]; then echo "$NAME"; fi; done ) )
    fi
    IFS="$PREVIOUS_IFS"

    # See if we actually found any files to process.
    if [ ${#FILENAMES[@]} -eq 0 ]; then
      echo "Found no files to process."
      exit 0
    fi
  else
    # Remove any .gpx/.hrm suffixes from the user-supplied list, and uniquify.
    local PREVIOUS_IFS="$IFS"
    IFS=$'\n'
    FILENAMES=( $( for NAME in "${FILENAMES[@]}"; do echo $NAME; done | sed -e 's/\.\(hrm\|gpx\)$//' | sort -u ) )
    IFS="$PREVIOUS_IFS"
  fi
}

function merge {
  if [ -z "$FORCE" ] && [ -e "$1.tcx" ]; then
    echo "Skipping \"$1\" - output file (\"$1.tcx\") already exists." >&2
    return 0
  fi

  echo "Processing $1"
  awk -f "`dirname $0`/`basename $0 sh`awk" -v ALTITUDE="$ALTITUDE" -v HRMFILE="$1.hrm" -v SPORT="$SPORT" "$1.gpx" > "$1.tcx"

  if [ -n "$TIMEZONE" ]; then
    sed -i -re "s/([>\"][0-9]{4}(-[0-9]{2}){2}T([0-9]{2}:){2}[0-9]{2})Z([<\"])/\1$TIMEZONE\4/g" "$1.tcx"
  fi
}

function mergeAll {
  for FILENAME in "${FILENAMES[@]}"; do
    merge "$FILENAME"
  done
}

parseCommandLine "$@"
mergeAll

Finally, here's a quick summary of some relevant links:

Update​

Just updated the script to include support for the ALTITUDE variable just added to the AWK script.

Attachments​