The harvest_source_delete logic function proxies to package delete,
which will delete the harvest source dataset. The harvest plugin then
hooks to the after_delete extension point in order to inactivate the
actual HarvestSource object and abort any pending jobs.
Also added the Delete button to the harvest source form.
We want to take into account objects with errors that where created or
updated anyway (eg bbox errors), so we bascially query for the number of
objects that have object errors.
Also add the number of gather errors to this count.
Basically handle the 'owner_org' field in form_to_db and db_to_form.
Added 'owner_org', 'frequency' (has default) and 'config' to surplus
keys in check_data_dict.
Also remove schema tweaks to let package_show call the appropiate schema
function.
When the run command flags a job as finished, it will query the most
recent harvest object for this job and use its import_finished value as
the job finishing time.
It will return the counts for the 20 most common errors for that
particular job. These will available when calling harvest_job_show.
Also refactor the harvest source status object to just call
harvest_job_dictize on the 'last_job' key, as it has all the
interesting fields anyway.
The authorization functions have been refactored to take into account
both the new organizaton based authorization on CKAN core and the
harvest source datasets.
Basically at the source level, authorization checks are forwarded to the
relevant package auth function (package_create, package_update, etc.)
wich will check for organizations membership, sysadmin, etc.
Also we only use functions available on the plugins toolkit whenever
possible.
Until now, harvest jobs were set to Finished just after sending all
objects to the fetch stage. Now every time the run command is run, jobs
are set to Running, and all previous Running jobs are checked to see if
all harvest objects have a state of Complete or Error. Only then the job
is flagged as Finished.
The status dict is added automatically to harvest source packages.
Note that the actual queries still need to be updated as they proabably
won't scale.
amercader
John Martin
kindly