Render subject in oai_dc metadata regardless if the study.keyword has
a value (Fixes #31).

Added
^^^^^

* Support for deleted records. (Implements `#3`_)

  * The deletedRecords value (in response to Identify-verb) is configurable via
    ``--oai-pmh-deleted-records`` and defaults to value `transient`.

* Configuration option to limit the list size of each metadataformat separately.
  (Implements `#2`_)
* New metadataformat OAI Datacite with prefix `oai_datacite`.
* Plugin architecture for developing arbitrary metadataformats & OAI-sets.
* Support insecure (unverified SSL certificate chain) https connections in list_records.py
  via configuration flag ``--insecure``. Default is to reject insecure connections.
* Generate setName values for ``language`` and ``data_kind`` sets which do not have one.
  (Implements `#24`_)
* Command line entrypoint `kuha_list_records` outputs deleted record count and
  supports HTTP Basic Authentication. (Implements `#25`_)

Changed
^^^^^^^

* Require Kuha Common 1.0.0.
* Plenty of API changes to facilitate the developing of metadataformats as plugins:

  * Delete configure.py. Serve-endpoint configuration is now declared in
    serve.py. Modules declare configurations via their respective add_cli_args
    functions.
  * Add controller.py. The controller is responsible for gathering
    the OAI responses from metadataformats.
  * Add http_api.py. This module declares the HTTP interface and dispatches
    requests to the configured controller. The module contains minimum
    amount of non-server imports and calls to controller.
  * Add metadataformats-package, which replaces the oai/metadata_formats.py
    and oai/records.py modules.
  * serve.py loads and configures metadataformats using entrypoint-group
    ``kuha.oai.metadataformats``. No metadataformats are directly imported.
  * oai/__init__.py now contains add_cli_args & configure methods to
    configure package classes. It defines a Protocol-class which
    wraps all OAI-PMH protocol objects.

* Update dependencies in requirements.txt.

  * Tornado 6.1.0
  * ConfigArgParse 1.5.3
  * six 1.16.0

Removed
^^^^^^^

* Drop support for Python versions below Python 3.8.

Fixed
^^^^^

* Repeat EAD3 daterange element when a record has multiple date
  ranges. (Fixes `#20`_)
* Metadata rendered by prefix `oai_ddi25` should place
  distrbtr-element before distDate-element in distStmt.
  (Fixes `#21`_)

Document upgrade procedure

Fix #29 at Bitbucket

Render distrbtr before distDate in oai_ddi25.xml-template. Fixes #21 at Bitbucket.

Render distrbtr before distDate in oai_ddi25.xml-template. Add test to
make sure the element order is correct. Write 0.14.1 changelog entry.

Fixes #21 at Bitbucket.

Fix changelog entry for 0.14.0 regarding a XPATH to
/codeBook/stdyDscr/citation/holdings/@URI

Fix XPATH in 0.14.0 changelog entry.

The correct XPATH is /codeBook/stdyDscr/citation/holdings/@URI

No version bump since this is a documentation issue.

Fix issue #23 at Bitbucket.

Ensure future compatibility with CESSDA Data Catalogue

  * Require kuha_common 0.15.1
  * Include elements `/codeBook/docDscr/citation/titlStmt/titl` and
    `/codeBook/stdyDscr/holdings@URI` to DDI 2.5 serializations.
  * Use `study.document_titles` to populate
    `/codeBook/docDscr/citation/titlStmt/titl` and `study.study_uris` to
    populate `/codeBook/stdyDscr/holdings@URI`.

Drop testing with Python interpreters below py38.

* Dev / CI maintenance

Drop tests with Python interpreters below Py38.

Bump VERSION to 0.14.0

* DDI-C contains Study.document_title and Study.study_uri attributes

Include elements `/codeBook/docDscr/citation/titlStmt/titl` and
`/codeBook/stdyDscr/holdings@URI` to DDI 2.5 serializations. Use
`study.document_titles` to populate
`/codeBook/docDscr/citation/titlStmt/titl` and `study.study_uris` to
populate `/codeBook/stdyDscr/holdings@URI`.

* Require kuha_common 0.15.1

Fix #22 at BitBucket.

Changes to EAD3 mapping:

  /ead/archdesc/dsc/c01/c02/did/daoset/dao/@daotype value is now
  "unknown" instead of "derived"

  study.principal_investigators are now divided into corpname &
  persname elements. If a principal investigator has an affiliated
  organization the value is placed into persname and the
  organization is placed into corpname. If a principal investigator
  has no affiliated organization, it's value is expected to be an
  organization and the value is placed into corpname.

Lock pip version to 20.3.4 in install script, which is the latest
pip that supports Python 3.5. The install script should be
compatible with Ubuntu 16.04, which defaults to Python 3.5. The
latest pip does not support Python 3.5 and therefore cannot be
upgraded into.

Use ''python-latest'' in Jenkinsfile. It is guaranteed to point to
the latest python on the CI server. Note that this is a CI specific
configuration and is not portable to some arbitrary Jenkins
instance.

Add py39 to tox environments and use it in Jenkinsfile.

Upgrade to latest genshi 0.7.5 in requirements.txt. Python 3.9
introduced changes to ast module that were incompatible with
previous genshi version 0.7.3 used.

Add six to requirements.txt, since genshi 0.7.5 requires it.

CHANGES.rst: Add todays date to 0.13.0

The default pip (8.1.1) does not support upgrade-strategy, but we'll
need it to upgrade dependencies. Lock the pip version to 20.3.4 when
using the install script.

Add six to requirements.txt, since genshi 0.7.5 requires it.

Update documentation regarding versions of Python & Ubuntu

/ead/archdesc/dsc/c01/c02/did/daoset/dao/@daotype value is now
"unknown" instead of "derived"

study.principal_investigators are now divided into corpname & persname
elements. If a principal investigator has an affiliated organization
the value is placed into persname and the organization is placed into
corpname. If a principal investigator has no affiliated organization,
it's value is expected to be an organization and the value is placed
into corpname.

Remove 'pip --upgrade pip' from install script. Since we are
supporting Ubuntu 16.04, upgrading pip will not work since pip no
longer supports py35, which is the default in Ubuntu 16.04.

Use 'python-latest' in Jenkinsfile.

Add py39 to tox environments and use it in Jenkinsfile.

Add 'unreleased' changelog entry.

Bump version.

Upgrade genshi to 0.7.5. Python 3.9 and genshi 0.7.3 don't work
together. Upgrade genshi to 0.7.5 to try to fix the error.

* Fix unhandled AttributeError when requesting a list response with
  an unsupported set-parameter. Respond with OAI error code
  "noRecordsMatch". (Fixes #19)

Server now returns an OAI response with an error code "noRecordsMatch"
when requesting a list response with an unsupported OAI-set. The fix
treats the return value None of
oai.records.get_record_query_field_by_setspec() as an unsupported
set. If the returned value is None, the
oai.records.get_query_filter_for_set() should also return None. Then
the handler will handle the request correctly.

Add unittests against the handler to confirm the fix.

Bump version to 0.12.1

CHANGES.rst: Add today's date to 0.12.1 entry.

* Support Python 3.8
* Require kuha_common 0.14.0

Support for Python 3.8 requires changes to test code and latest
kuha_common. No new functionality, bug fixes or tests are introduced
here.

Require kuha_common 0.14.0 in requirements.txt and setup.py. Release
versionlock of genshi in requirements-dev.txt.

Update all copyright headers to 2020.

Add py38 environment to tox.ini and use it in Jenkinsfile.

Write changelog entry for 0.12.0.

Bump version to 0.12.0

* Check for query result item's set.record_field_setspec using
item.get() since the item may not have such key and we wish to avoid
KeyError. (Fixes #18)
* Remove check for the record_field_setname since it is acceptable to
have and OAI set without a setName.

Use dict.get for checking item's setspec value since an item may not
have the required key. This way such record won't result in a KeyError
but will be discarded from ListSets response.

Add test against such condition using study_group.

Write changelog entry for 0.11.1.

Bump version to 0.11.1

* Render relpubl elements in DDI 2.5. templates using
Study.related_publications
* Require kuha_common 0.13.0

Require kuha_common 0.13.0, which brings in support for
Study.related_publications. Use Study.related_publications to render
relpubl elements in DDI 2.5. templates.

Bump version.

Write changelog entry for 0.11.0

Write entry to CHANGES.rst about release 0.11.0. Bump VERSION to 0.11.0.

DDI templates, ddi_c.xml and oai_ddi25.xml now use
Study.related_publications to serialize relpubl elements and
contents. Add Study.related_publications to
DDIMetadataFormat.record_fields. Add OAIRecord.iter_relpubls() to
iterate over related_publications by distinct description and lang. This
way the templates can serialize DDI so that a relpubl element may have
multiple citation child elements if the relpubls share same
description and lang.

Add tests to make sure the serialization is correct and queries to
document store include Study.related_publications field.

* Project source code management changed to git. Does not contain code changes to
  OAI-PMH Repo Handler.
* INSTALL.rst now instructs to use Git instead of Mercurial.
* requirements.txt: kuha_common 0.12.0

Change to git: Add .gitignore for Python-project. Jenkinsfile
uses git to dig up commit authors. Update requirements.txt to
use git instead of hg. INSTALL.rst instructs to use git instead
of hg.

Prepare 0.10.0: Bump version. Write changelog entry.

Bump version. Write changelog entry.

Add .gitignore for Python-project. Jenkinsfile uses git to dig up
commit authors. Update requirements.txt to use git instead of hg.
INSTALL.rst instructs to use git instead of hg.

Merge: stable <- fix false positive test in tests/unit/test_list_records.py::TestCliRuns::test_reads_optional_args

Fix failing false positive: tests/unit/test_list_records.py::TestCliRuns::test_reads_optional_args. Tests need to take into account that when unpacking a dict the order of keys is not consistent.