We want to download several layers from geoserver every day in order to create snapshots that can be used as geo-temporal data later on.

• support for GML files only
• create shapshots once per day or less frequently
• retry download if failed (but only up to some retry limit, e.g max 3 times)
• use content from previous day if failed to download (but keep track of it)
• the data can potentially be several GB of XML
• store the snapshots efficiently (compressed, only diffs ...)
• use cron to trigger the download every N minutes
• when triggered, only a single GML file should be downloaded at a time
• keep low profile - avoid parallel downloads, limit transfer rate
• every file is downloaded only once per day

• we need a URL of the geoserver and a directory DIR where the layers will be downloaded

script download_layer.sh URL DIR:

• picks the oldest *.gml file from DIR
• downloads the corresponding layer from URL using WFS as GML (XML)
• formats the XML files using xmllint --format
• removes fid XML attributes (because they are always newly generated by the geoserver)
• for every *.gml file a corresponding *.meta file will be generated which contains some accounting information about the download

duplicity:

• a useful tool for incremental backups (see usage examples below)

duplicity -vi --allow-source-mismatch --no-encryption path/to/src/dir file:https://path/to/snapshot/dir

• -vi = verbosity level is "info"
• --allow-source-mismatch allows that the names of source dirs can be changed

duplicity colletion-status file:https://path/to/my/snapshot/dir

duplicity restore --no-encryption --time 2016-06-30T11:00:00 file:https://path/to/snapshot/dir path/to/output/dir

Assuming we want to compare differences between directories dir1 and dir2 and that we want to ignore files matching a pattern *.meta:

diff -x '*.meta' dir1 dir2 | diffstats

Output should look like this:

include/net/bluetooth/l2cap.h |    6 ++++++
 net/bluetooth/l2cap.c         |   18 +++++++++---------
 2 files changed, 15 insertions(+), 9 deletions(-)

Assuming you are in some directory which contains files and the prefix is "PREFIX" (This is just a quick and dirty method, there is certainly a better way to do so)

find . | while read F; do mv $F ${F#./PREFIX}; done