This extension provides a common harvesting framework for ckan extensions and adds a CLI and a WUI to CKAN to manage harvesting sources and jobs.
This extension requires CKAN v2.0 or later on both the CKAN it is installed into and the CKANs it harvests. However you are unlikely to encounter a CKAN running a version lower than 2.0.
The harvest extension can use two different backends. You can choose whichever you prefer depending on your needs, but Redis has been found to be more stable and reliable so it is the recommended one:
Redis (recommended): To install it, run:
sudo apt-get update sudo apt-get install redis-server
On your CKAN configuration file, add in the [app:main] section:
ckan.harvest.mq.type = redis
RabbitMQ: To install it, run:
sudo apt-get update sudo apt-get install rabbitmq-server
On your CKAN configuration file, add in the [app:main] section:
ckan.harvest.mq.type = amqp
Activate your CKAN virtual environment, for example:
$ . /usr/lib/ckan/default/bin/activate
Install the ckanext-harvest Python package into your virtual environment:
(pyenv) $ pip install -e git+https://github.com/ckan/ckanext-harvest.git#egg=ckanext-harvest
Install the python modules required by the extension (adjusting the path according to where ckanext-harvest was installed in the previous step):
(pyenv) $ cd /usr/lib/ckan/default/src/ckanext-harvest/ (pyenv) $ pip install -r pip-requirements.txt
Make sure the CKAN configuration ini file contains the harvest main plugin, as well as the harvester for CKAN instances if you need it (included with the extension):
ckan.plugins = harvest ckan_harvester
If you haven't done it yet on the previous step, define the backend that you are using with the
ckan.harvest.mq.typeoption in the [app:main] section (it defaults toamqp):ckan.harvest.mq.type = redis
There are a number of configuration options available for the backends. These don't need to be modified at all if you are using the default Redis or RabbitMQ install (step 1). However you may wish to add them with custom options to the into the CKAN config file the [app:main] section. The list below shows the available options and their default values:
- Redis:
ckan.harvest.mq.hostname(localhost)ckan.harvest.mq.port(6379)ckan.harvest.mq.redis_db(0)
- RabbitMQ:
ckan.harvest.mq.user_id(guest)ckan.harvest.mq.password(guest)ckan.harvest.mq.hostname(localhost)ckan.harvest.mq.port(5672)ckan.harvest.mq.virtual_host(/)
Note: it is safe to use the same backend server (either Redis or RabbitMQ)
for different CKAN instances, as long as they have different site ids. The ckan.site_id
config option (or default) will be used to namespace the relevant things:
- On RabbitMQ it will be used to name the queues used, eg
ckan.harvest.site1.gatherandckan.harvest.site1.fetch. - On Redis, it will namespace the keys used, so only the relevant instance gets them, eg
site1:harvest_job_id,site1:harvest_object__id:804f114a-8f68-4e7c-b124-3eb00f66202f
Run the following command to create the necessary tables in the database (ensuring the pyenv is activated):
(pyenv) $ paster --plugin=ckanext-harvest harvester initdb --config=/etc/ckan/default/production.ini
Finally, restart CKAN to have the changes take affect:
sudo service apache2 restart
After installation, the harvest source listing should be available under /harvest, eg:
http://localhost/harvest
Logging to the database is disabled by default. If you want your ckan harvest logs to be exposed to the CKAN API you need to properly configure the logger with the following configuration parameter:
ckan.harvest.log_scope = 0
- -1 - Do not log in the database - DEFAULT
- 0 - Log everything
- 1 - model, logic.action, logic.validators, harvesters
- 2 - model, logic.action, logic.validators
- 3 - model, logic.action
- 4 - logic.action
- 5 - model
- 6 - plugin
- 7 - harvesters
Setup time frame(in days) for the clean-up mechanism with the following config parameter (in the [app:main] section):
ckan.harvest.log_timeframe = 10
If no value is present the default is 30 days.
Setup log level for the database logger:
ckan.harvest.log_level = info
If no log level is set the default is
debug.
API Usage
You can access CKAN harvest logs via the API:
$ curl {ckan_url}/api/3/action/harvest_log_list
Replace {ckan_url} with the url from your CKAN instance.
Allowed parameters are:
- level (filter log records by level)
- limit (used for pagination)
- offset (used for pagination)
e.g. Fetch all logs with log level INFO:
$ curl {ckan_url}/api/3/action/harvest_log_list?level=info
- {
"help":"http://127.0.0.1:5000/api/3/action/help_show?name=harvest_log_list",
"success":true,
"result": [{"content":"Sent job aa987717-2316-4e47-b0f2-cbddfb4c4dfc to the gather queue","level":"INFO","created":"2016-06-03 10:59:40.961657"}, {"content":"Sent job aa987717-2316-4e47-b0f2-cbddfb4c4dfc to the gather queue","level":"INFO","created":"2016-06-03 10:59:40.951548"}]
}
The following operations can be run from the command line as described underneath:
harvester initdb
- Creates the necessary tables in the database
harvester source {name} {url} {type} [{title}] [{active}] [{owner_org}] [{frequency}] [{config}]
- create new harvest source
harvester source {source-id/name}
- shows a harvest source
harvester rmsource {source-id/name}
- remove (deactivate) a harvester source, whilst leaving any related
datasets, jobs and objects
harvester clearsource {source-id/name}
- clears all datasets, jobs and objects related to a harvest source,
but keeps the source itself
harvester sources [all]
- lists harvest sources
If 'all' is defined, it also shows the Inactive sources
harvester job {source-id/name}
- create new harvest job
harvester jobs
- lists harvest jobs
harvester job_abort {source-id/name}
- marks a job as "Aborted" so that the source can be restarted afresh.
It ensures that the job's harvest objects status are also marked
finished. You should ensure that neither the job nor its objects are
currently in the gather/fetch queues.
harvester run
- starts any harvest jobs that have been created by putting them onto
the gather queue. Also checks running jobs - if finished it
changes their status to Finished.
harvester run_test {source-id/name}
- runs a harvest - for testing only.
This does all the stages of the harvest (creates job, gather, fetch,
import) without involving the web UI or the queue backends. This is
useful for testing a harvester without having to fire up
gather/fetch_consumer processes, as is done in production.
harvester gather_consumer
- starts the consumer for the gathering queue
harvester fetch_consumer
- starts the consumer for the fetching queue
harvester purge_queues
- removes all jobs from fetch and gather queue
WARNING: if using Redis, this command purges all data in the current
Redis database
harvester clean_harvest_log
- Clean-up mechanism for the harvest log table.
You can configure the time frame through the configuration
parameter 'ckan.harvest.log_timeframe'. The default time frame is 30 days
harvester [-j] [-o] [--segments={segments}] import [{source-id}]
- perform the import stage with the last fetched objects, for a certain
source or a single harvest object. Please note that no objects will
be fetched from the remote server. It will only affect the objects
already present in the database.
To import a particular harvest source, specify its id as an argument.
To import a particular harvest object use the -o option.
To import a particular package use the -p option.
You will need to specify the -j flag in cases where the datasets are
not yet created (e.g. first harvest, or all previous harvests have
failed)
The --segments flag allows to define a string containing hex digits that represent which of
the 16 harvest object segments to import. e.g. 15af will run segments 1,5,a,f
harvester job-all
- create new harvest jobs for all active sources.
harvester reindex
- reindexes the harvest source datasets
The commands should be run with the pyenv activated and refer to your CKAN configuration file:
(pyenv) $ paster --plugin=ckanext-harvest harvester sources --config=/etc/ckan/default/production.ini
Starting from CKAN 2.0, harvest sources behave exactly the same as datasets (they are actually internally implemented as a dataset type). That means they can be searched and faceted, and that the same authorization rules can be applied to them. The default authorization settings are based on organizations.
Have a look at the Authorization documentation on CKAN core to see how to configure your instance depending on your needs.
The plugin includes a harvester for remote CKAN instances. To use it, you need to add the ckan_harvester plugin to your options file:
ckan.plugins = harvest ckan_harvester
After adding it, a 'CKAN' option should appear in the 'New harvest source' form.
The CKAN harvesters support a number of configuration options to control their behaviour. Those need to be defined as a JSON object in the configuration form field. The currently supported configuration options are:
- api_version: You can force the harvester to use either version 1 or 2 of the CKAN API. Default is 2.
- default_tags: A list of tags that will be added to all harvested datasets. Tags don't need to previously exist. This field takes a list of tag dicts (see example), which allows you to optinally specify a vocabulary.
- default_groups: A list of group IDs or names to which the harvested datasets will be added to. The groups must exist.
- default_extras: A dictionary of key value pairs that will be added to extras
of the harvested datasets. You can use the following replacement strings,
that will be replaced before creating or updating the datasets:
- {dataset_id}
- {harvest_source_id}
- {harvest_source_url} # Will be stripped of trailing forward slashes (/)
- {harvest_source_title}
- {harvest_job_id}
- {harvest_object_id}
- override_extras: Assign default extras even if they already exist in the remote dataset. Default is False (only non existing extras are added).
- user: User who will run the harvesting process. Please note that this user needs to have permission for creating packages, and if default groups were defined, the user must have permission to assign packages to these groups.
- api_key: If the remote CKAN instance has restricted access to the API, you can provide a CKAN API key, which will be sent in any request.
- read_only: Create harvested packages in read-only mode. Only the user who performed the harvest (the one defined in the previous setting or the 'harvest' sysadmin) will be able to edit and administer the packages created from this harvesting source. Logged in users and visitors will be only able to read them.
- force_all: By default, after the first harvesting, the harvester will gather only the modified packages from the remote site since the last harvesting. Setting this property to true will force the harvester to gather all remote packages regardless of the modification date. Default is False.
- remote_groups: By default, remote groups are ignored. Setting this property enables the harvester to import the remote groups. There are two alternatives. Setting it to 'only_local' will just import groups which name/id is already present in the local CKAN. Setting it to 'create' will make an attempt to create the groups by copying the details from the remote CKAN.
- remote_orgs: By default, remote organizations are ignored. Setting this property enables the harvester to import remote organizations. There are two alternatives. Setting it to 'only_local' will just import organizations which id is already present in the local CKAN. Setting it to 'create' will make an attempt to create the organizations by copying the details from the remote CKAN.
- clean_tags: By default, tags are not stripped of accent characters, spaces and capital letters for display. If this option is set to True, accent characters will be replaced by their ascii equivalents, capital letters replaced by lower-case ones, and spaces replaced with dashes. Setting this option to False gives the same effect as leaving it unset.
- organizations_filter_include: This configuration option allows you to specify a list of remote organization names (e.g. "arkansas-gov" is the name for organization http://catalog.data.gov/organization/arkansas-gov ). If this property has a value then only datasets that are in one of these organizations will be harvested. All other datasets will be skipped. Only one of organizations_filter_include or organizations_filter_exclude should be configured.
- organizations_filter_exclude: This configuration option allows you to specify a list of remote organization names (e.g. "arkansas-gov" is the name for organization http://catalog.data.gov/organization/arkansas-gov ). If this property is set then all datasets from the remote source will be harvested unless it belongs to one of the organizations in this option. Only one of organizations_filter_exclude or organizations_filter_include should be configured.
Here is an example of a configuration object (the one that must be entered in the configuration field):
{
"api_version": 1,
"default_tags": [{"name": "geo"}, {"name": "namibia"],
"default_groups": ["science", "spend-data"],
"default_extras": {"encoding":"utf8", "harvest_url": "{harvest_source_url}/dataset/{dataset_id}"},
"override_extras": true,
"organizations_filter_include": [],
"organizations_filter_exclude": ["remote-organization"],
"user":"harverster-user",
"api_key":"<REMOTE_API_KEY>",
"read_only": true,
"remote_groups": "only_local",
"remote_orgs": "create"
}
Extensions can implement the harvester interface to perform harvesting operations. The harvesting process takes place on three stages:
- The gather stage compiles all the resource identifiers that need to be fetched in the next stage (e.g. in a CSW server, it will perform a GetRecords operation).
- The fetch stage gets the contents of the remote objects and stores them in the database (e.g. in a CSW server, it will perform n GetRecordById operations).
- The import stage performs any necessary actions on the fetched resource (generally creating a CKAN package, but it can be anything the extension needs).
Plugins willing to implement the harvesting interface must provide the following methods:
from ckan.plugins.core import SingletonPlugin, implements
from ckanext.harvest.interfaces import IHarvester
class MyHarvester(SingletonPlugin):
'''
A Test Harvester
'''
implements(IHarvester)
def info(self):
'''
Harvesting implementations must provide this method, which will return
a dictionary containing different descriptors of the harvester. The
returned dictionary should contain:
* name: machine-readable name. This will be the value stored in the
database, and the one used by ckanext-harvest to call the appropiate
harvester.
* title: human-readable name. This will appear in the form's select box
in the WUI.
* description: a small description of what the harvester does. This
will appear on the form as a guidance to the user.
A complete example may be::
{
'name': 'csw',
'title': 'CSW Server',
'description': 'A server that implements OGC's Catalog Service
for the Web (CSW) standard'
}
:returns: A dictionary with the harvester descriptors
'''
def validate_config(self, config):
'''
[optional]
Harvesters can provide this method to validate the configuration
entered in the form. It should return a single string, which will be
stored in the database. Exceptions raised will be shown in the form's
error messages.
:param harvest_object_id: Config string coming from the form
:returns: A string with the validated configuration options
'''
def get_original_url(self, harvest_object_id):
'''
[optional]
This optional but very recommended method allows harvesters to return
the URL to the original remote document, given a Harvest Object id.
Note that getting the harvest object you have access to its guid as
well as the object source, which has the URL.
This URL will be used on error reports to help publishers link to the
original document that has the errors. If this method is not provided
or no URL is returned, only a link to the local copy of the remote
document will be shown.
Examples:
* For a CKAN record: http://{ckan-instance}/api/rest/{guid}
* For a WAF record: http://{waf-root}/{file-name}
* For a CSW record: http://{csw-server}/?Request=GetElementById&Id={guid}&...
:param harvest_object_id: HarvestObject id
:returns: A string with the URL to the original document
'''
def gather_stage(self, harvest_job):
'''
The gather stage will receive a HarvestJob object and will be
responsible for:
- gathering all the necessary objects to fetch on a later.
stage (e.g. for a CSW server, perform a GetRecords request)
- creating the necessary HarvestObjects in the database, specifying
the guid and a reference to its job. The HarvestObjects need a
reference date with the last modified date for the resource, this
may need to be set in a different stage depending on the type of
source.
- creating and storing any suitable HarvestGatherErrors that may
occur.
- returning a list with all the ids of the created HarvestObjects.
- to abort the harvest, create a HarvestGatherError and raise an
exception. Any created HarvestObjects will be deleted.
:param harvest_job: HarvestJob object
:returns: A list of HarvestObject ids
'''
def fetch_stage(self, harvest_object):
'''
The fetch stage will receive a HarvestObject object and will be
responsible for:
- getting the contents of the remote object (e.g. for a CSW server,
perform a GetRecordById request).
- saving the content in the provided HarvestObject.
- creating and storing any suitable HarvestObjectErrors that may
occur.
- returning True if everything is ok (ie the object should now be
imported), "unchanged" if the object didn't need harvesting after
all (ie no error, but don't continue to import stage) or False if
there were errors.
:param harvest_object: HarvestObject object
:returns: True if successful, 'unchanged' if nothing to import after
all, False if not successful
'''
def import_stage(self, harvest_object):
'''
The import stage will receive a HarvestObject object and will be
responsible for:
- performing any necessary action with the fetched object (e.g.
create, update or delete a CKAN package).
Note: if this stage creates or updates a package, a reference
to the package should be added to the HarvestObject.
- setting the HarvestObject.package (if there is one)
- setting the HarvestObject.current for this harvest:
- True if successfully created/updated
- False if successfully deleted
- setting HarvestObject.current to False for previous harvest
objects of this harvest source if the action was successful.
- creating and storing any suitable HarvestObjectErrors that may
occur.
- creating the HarvestObject - Package relation (if necessary)
- returning True if the action was done, "unchanged" if the object
didn't need harvesting after all or False if there were errors.
NB You can run this stage repeatedly using 'paster harvest import'.
:param harvest_object: HarvestObject object
:returns: True if the action was done, "unchanged" if the object didn't
need harvesting after all or False if there were errors.
'''
See the CKAN harvester for an example of how to implement the harvesting interface:
ckanext-harvest/ckanext/harvest/harvesters/ckanharvester.py
Here you can also find other examples of custom harvesters:
- https://github.com/ckan/ckanext-dcat/tree/master/ckanext/dcat/harvesters
- https://github.com/ckan/ckanext-spatial/tree/master/ckanext/spatial/harvesters
There are two ways to run a harvest:
1. ``harvester run_test`` for the command-line, suitable for testing 2. ``harvester run`` used by the Web UI and scheduled runs
You can run a harvester simply using the run_test command. This is handy
for running a harvest with one command in the console and see all the output
in-line. It runs the gather, fetch and import stages all in the same process.
This is useful for developing a harvester because you can insert break-points in your harvester, and rerun a harvest without having to restart the gather_consumer and fetch_consumer processes each time. In addition, because it doesn't use the queue backends it doesn't interfere with harvests of other sources that may be going on in the background.
However running this way, if gather_stage, fetch_stage or import_stage raise an
exception, they are not caught, whereas with harvester run they are handled
slightly differently as they are called by queue.py. So when testing this
aspect its best to use harvester run.
When a harvest job is started by a user in the Web UI, or by a scheduled
harvest, the harvest is started by the harvester run command. This is the
normal method in production systems and scales well.
In this case, the harvesting extension uses two different queues: one that handles the gathering and another one that handles the fetching and importing. To start the consumers run the following command (make sure you have your python environment activated):
(pyenv) $ paster --plugin=ckanext-harvest harvester gather_consumer --config=/etc/ckan/default/production.ini
On another terminal, run the following command:
(pyenv) $ paster --plugin=ckanext-harvest harvester fetch_consumer --config=/etc/ckan/default/production.ini
Finally, on a third console, run the following command to start any pending harvesting jobs:
(pyenv) $ paster --plugin=ckanext-harvest harvester run --config=/etc/ckan/default/production.ini
The run command not only starts any pending harvesting jobs, but also
flags those that are finished, allowing new jobs to be created on that particular
source and refreshing the source statistics. That means that you will need to run
this command before being able to create a new job on a source that was being
harvested. (On a production site you will typically have a cron job that runs the
command regularly, see next section).
Occasionally you can find a harvesting job is in a "limbo state" where the job
has run with errors but the harvester run command will not mark it as
finished, and therefore you cannot run another job. This is due to particular
harvester not handling errors correctly e.g. during development. In this
circumstance, ensure that the gather & fetch consumers are running and have
nothing more to consume, and then run this abort command with the name or id of
the harvest source:
(pyenv) $ paster --plugin=ckanext-harvest harvester job_abort {source-id/name} --config=/etc/ckan/default/production.ini
The previous approach works fine during development or debugging, but it is not recommended for production servers. There are several possible ways of setting up the harvesters, which will depend on your particular infrastructure and needs. The bottom line is that the gather and fetch process should be kept running somehow and then the run command should be run periodically to start any pending jobs.
The following approach is the one generally used on CKAN deployments, and it will probably suit most of the users. It uses Supervisor, a tool to monitor processes, and a cron job to run the harvest jobs, and it assumes that you have already installed and configured the harvesting extension (See Installation if not).
Note: It is recommended to run the harvest process from a non-root user (generally the one you are running CKAN with). Replace the user ckan in the following steps with the one you are using.
Install Supervisor:
sudo apt-get update sudo apt-get install supervisor
You can check if it is running with this command:
ps aux | grep supervisord
You should see a line similar to this one:
root 9224 0.0 0.3 56420 12204 ? Ss 15:52 0:00 /usr/bin/python /usr/bin/supervisord
Supervisor needs to have programs added to its configuration, which will describe the tasks that need to be monitored. This configuration files are stored in
/etc/supervisor/conf.d.Create a file named
/etc/supervisor/conf.d/ckan_harvesting.conf, and copy the following contents:; =============================== ; ckan harvester ; =============================== [program:ckan_gather_consumer] command=/usr/lib/ckan/default/bin/paster --plugin=ckanext-harvest harvester gather_consumer --config=/etc/ckan/default/production.ini ; user that owns virtual environment. user=ckan numprocs=1 stdout_logfile=/var/log/ckan/std/gather_consumer.log stderr_logfile=/var/log/ckan/std/gather_consumer.log autostart=true autorestart=true startsecs=10 [program:ckan_fetch_consumer] command=/usr/lib/ckan/default/bin/paster --plugin=ckanext-harvest harvester fetch_consumer --config=/etc/ckan/default/production.ini ; user that owns virtual environment. user=ckan numprocs=1 stdout_logfile=/var/log/ckan/std/fetch_consumer.log stderr_logfile=/var/log/ckan/std/fetch_consumer.log autostart=true autorestart=true startsecs=10
There are a number of things that you will need to replace with your specific installation settings (the example above shows paths from a ckan instance installed via Debian packages):
- command: The absolute path to the paster command located in the python virtual environment and the absolute path to the config ini file.
- user: The unix user you are running CKAN with
- stdout_logfile and stderr_logfile: All output coming from the harvest consumers will be written to this file. Ensure that the necessary permissions are setup.
The rest of the configuration options are pretty self explanatory. Refer to the Supervisor documentation to know more about these and other options available.
Start the supervisor tasks with the following commands:
sudo supervisorctl reread sudo supervisorctl add ckan_gather_consumer sudo supervisorctl add ckan_fetch_consumer sudo supervisorctl start ckan_gather_consumer sudo supervisorctl start ckan_fetch_consumer
To check that the processes are running, you can run:
sudo supervisorctl status ckan_fetch_consumer RUNNING pid 6983, uptime 0:22:06 ckan_gather_consumer RUNNING pid 6968, uptime 0:22:45
Some problems you may encounter when starting the processes:
- ckan_gather_consumer: ERROR (no such process)
Double-check your supervisor configuration file and stop and restart the supervisor daemon:
sudo service supervisor start; sudo service supervisor stop
- ckan_gather_consumer: ERROR (abnormal termination)
Something prevented the command from running properly. Have a look at the log file that you defined in the stdout_logfile section to see what happened. Common errors include:
`socket.error: [Errno 111] Connection refused` RabbitMQ is not running:: sudo service rabbitmq-server start
Once we have the two consumers running and monitored, we just need to create a cron job that will run the run harvester command periodically. To do so, edit the cron table with the following command (it may ask you to choose an editor):
sudo crontab -e -u ckan
Note that we are running this command as the same user we configured the processes to be run with (ckan in our example).
Paste this line into your crontab, again replacing the paths to paster and the ini file with yours:
# m h dom mon dow command */15 * * * * /usr/lib/ckan/default/bin/paster --plugin=ckanext-harvest harvester run --config=/etc/ckan/default/production.ini
This particular example will check for pending jobs every fifteen minutes. You can of course modify this periodicity, this Wikipedia page has a good overview of the crontab syntax.
In order to setup clean-up mechanism for the harvest log one more cron job needs to be scheduled:
sudo crontab -e -u ckan
Paste this line into your crontab, again replacing the paths to paster and the ini file with yours:
# m h dom mon dow command 0 5 * * * /usr/lib/ckan/default/bin/paster --plugin=ckanext-harvest harvester clean_harvest_log --config=/etc/ckan/default/production.ini
This particular example will perform clean-up each day at 05 AM. You can tweak the value according to your needs.
You can run the tests like this:
cd ckanext-harvest nosetests --reset-db --ckan --with-pylons=test-core.ini ckanext/harvest/tests
Here are some common errors and solutions:
(OperationalError) no such table: harvest_object_error u'delete from "harvest_object_error"The database has got into in a bad state. Run the tests again but with the--reset-dbparameter.(ProgrammingError) relation "harvest_object_extra" does not existThe database has got into in a bad state. Run the tests again but without the--reset-dbparameter.(OperationalError) near "SET": syntax errorYou are testing with SQLite as the database, but the CKAN Harvester needs PostgreSQL. Specify test-core.ini instead of test.ini.
- Developer mailing list: [email protected]
- Developer IRC channel: #ckan on irc.freenode.net
- Issue tracker
For contributing to ckanext-spatial or its documentation, follow the same guidelines that apply to CKAN core, described in CONTRIBUTING.
This extension is open and licensed under the GNU Affero General Public License (AGPL) v3.0. Its full text may be found at: