CHANGELOG

Changelog
=========

All notable changes to this project will be documented in this file.

The format is based on `Keep a Changelog <http://keepachangelog.com/>`_
and this project adheres to `Semantic Versioning <http://semver.org/>`_

0.5.1 - 2023-03-30
------------------

Added
~~~~~

- Testing infrastructure for functional tests based on xnat4tests
- Added the `insert_date` and `is_history` properties to XNAT objects (Project, Subject, Experiment, Scan) to
  allow access to these metadata fields
- Option to retrieve users per project

Fixed
~~~~~

- Fixed bugs in upload where files would be opened in string instead of bytes mode, leading to encoding errors
- Fixed bugs in upload where return values were omitted in some nested function, leading to breaking responses
- Fixed bug in object creation when secondary id was not supplied, see issue #37

0.5.0 - 2023-02-27
------------------

Changed
~~~~~~~

- Dropped Python 2 support, code can now use more modern features and is easier to maintain

Added
~~~~~

- Support for XNAT search engine usings XNATpy
- Exposes more server information (xnat_uptime, xnat_build_info, plugins list)


Improved
~~~~~~~~

- Reworked upload functions to have more clear behaviour

Fixed
~~~~~

- Creating experiments with an existing label now does not acccidentally move the
  experiment to a different subject (added safety checks)
- Files had the ID/path swapped, making it possible to have non-unique IDs
- ProjectData creation now uses the id parameter correctly


0.4.3 - 2022-10-31
------------------

Added
~~~~~

- Added initial support for XNAT/jupyterhub

Improved
~~~~~~~~

- Fixed some of the plumbing for searching

0.4.2 - 2022-04-07
------------------

Fixed
~~~~~

- Problem with the namespace prefix parsing with XNAT 1.8.4 cause the parser to fail

0.4.1 - 2022-02-22
------------------

Added
~~~~~

- New `connection.upload_file()` method that makes sure the argument is point to a valid file, avoiding
  accidentally attempting to upload a string.

Changed
~~~~~~~

- The `services.import_` method now has two mutually exclusive arguments for supplying data: `path` and
  `data`. For a given `path` a check is done to see if it actually points to a valid file on the filesystem
  for data it is just passed on to `upload`.

Fixed
~~~~~

- Some fuctions depending on `services.import_` were broken because a check was added to see
  if the data pointed to a file, not allowing in-memory data to be uploaded, breaking `import_dir`
  methods.

0.4.0 - 2022-01-27
------------------

Fixed
~~~~~

- Nested SimpleListing failed to recognise correct parent, breaking `scan.parameters.add_param`
- Fixes in xpath and xsi_type resolving

Added
~~~~~

- xnatpy command-line interface, now xnatpy comes with a number of CLI options
  to use directly from the shell
- A `scan.fields` shortcut to `scan.parameters.add_param` for convenience
- Objects with custom variables now have the `obj.custom_variables` property which gives access to the custom
  variable defined in the interface, including some type casting and value checking when updating them.

0.3.28 - 2021-10-28
-------------------

Fixed
~~~~~

- Additional parameters were not set properly during object creation
- It was impossible to login using tokens because the auth provides could not be found, added workaround

Changed
~~~~~~~

- Updated upload_dir and import_dir to default use a SpooledTemporaryFile and improved the zip and tar functionality

0.3.27 - 2021-09-29
-------------------

Added
~~~~~

- Options to create a session with re-using a JSESSION ID instead of logging in and a BaseXNATSession that does
  not destroy the session on disconnect
- Catalog refresh method for resources

Changed
~~~~~~~

- Allow import functions to take project, subject and experiment as xnatpy objects as well as strings

Fixed
~~~~~

- Failure to connect with redirection when url has a trailing ``/``
- Added an special auth for requests to avoid it automatically reading the netrc and using
  basic auth, ruining the efficiency from cookie based sessions.
- Deleting an object should also remove it from the connection and listing caches

0.3.26 - 2021-04-16
-------------------

Improved
~~~~~~~~

- Switched to cookie based login instead of basic auth with every requests (should be more performant)
- Close underlying requests session when closing a connection
- Added CI/CD infrastructure for testing and releasing
- Improved import_dicom_inbox

Fixed
~~~~~

- InvestigatorData had nonsensical string conversion
- Option to flatten directories when downloading an entire resource
- Bug with resource labels doing unwanted quotation on creation
- Added timeouts and caught them on initial requests (before the xnat session was established)


0.3.25 - 2020-09-24
-------------------

Added
~~~~~

- Added ``import_dicom_inbox`` function to import dicoms from server filesystem
  via a path
- Support of querying information for FileData via the containing resource file
  listing, giving access to file_content, file_format, digest, etc

Fixed
~~~~~

- Fixed issue when creating objects with a space in the label, the object would
  be created but the uri with the label (and spaces) would not work, making the
  return object useless. Now it updates the URI to use the generated ID if
  possible avoiding the issue.

0.3.24 - 2020-06-19
-------------------

Fixed
~~~~~

- Stupid import bug in previous release

0.3.23 - 2020-06-19
-------------------

Improved
~~~~~~~~

- Connections now check if they are still valid and will raise an appropriate
  error if they are closed. This avoid strange errors when trying to use a
  connection after ``.disconnect`` has been called.
- Avoid warning about collections.abc use on Python 3.7+

0.3.22 - 2020-02-26
-------------------

Added
~~~~~

- The ``services.import_dir`` function to import a local directory into XNAT.
  The function will automatically zip the contents in a temporary file (or
  memory if desired) before upload.

Changed
~~~~~~~

- Updated documentation and setup to refer to new gitlab repository instead
  of bitbucket, finalizing the move
- Add a default timeout of 300 seconds (5 minutes) to all requests to avoid
  stale connections blocking the entire code (without error, just infinite
  hangs). Can be changed when connecting or in the XNATSession object
  afterwards. Note that the 300 seconds is not the total duration of the
  request allowed, but means that after there is nothing received 300 seconds
  the connection is considered timed-out, a download that takes 10 minutes
  would still be possible.

Improved
~~~~~~~~

- The ``import_`` function is more resilient. On windows programs like 7zip could
  mess with the mimetypes and cause zip files get a different estimated
  content_type. If a non-recognized content type is found, xnatpy will try to
  re-estimate it based just on the extension.
- Added supper to change the ``import-handler`` in the ``import_`` function

0.3.21 - 2019-10-09
-------------------

Improved
~~~~~~~~

- Removed annoying warning message when resource was accessed

0.3.20 - 2019-10-09
-------------------

Fixed
~~~~~

- now possible to create/retrieve data for resources properly. XNAT would
  only return the catalog when queried and not the object information. Created
  a work around that catches the case in get_json and retrieves the json data
  from the listing instead and presents it in a format consistent with the
  REST API so other code still functions.

0.3.19 - 2019-09-20
-------------------

Added
~~~~~

- xnatpy changes the user-agent header to give detailed information about
  the xnatpy client running.
- option to have some progress logging/printing added to the download_dir
  methods for projects and subjects
- add method to find specific sessions in the prearchive

Improved
~~~~~~~~

- Avoid use of get_source when generating xnatpy module code, which should help
  when compiling xnatpy to a binary (py2exe etc)

0.3.18 - 2019-06-06
-------------------

Improved
~~~~~~~~

- Added ``force`` flag to ``ScanData.read_dicom``
- Added ``open`` to ``PrearchiveFile`` (same as for FileData)
- Added ``read_dicom`` to ``PrearchiveScan`` (same as for ScanData)
- Documentation improved, added code reference and changelog into docs

Fixed
~~~~~

- Flag extension_types=False now also working for 1.7 servers

0.3.17 - 2019-04-04
-------------------

Added
~~~~~

- Can open FileData object with ``file.open()`` to get a file-like object
  that can be used similar to a local file.
- Added ``read_dicom()`` to ScanData to read the dicom header/file with pydicom
- ``external_uri`` method to get a full external uri of an XNAT object

0.3.16 - 2019-03-28
-------------------

Fixed
~~~~~

- Support for changing subject and experiment labels
- Support creation of ScanData types with the id and type set on creation
- Fix a bug in scanning extension types where there are new-lines in the xs:schema tag

0.3.14 - 2019-02-22
-------------------

Added
~~~~~

- Check which user is logged in and expose that in ``connection.logged_in_user``
- Check the cookies to set the appropriate heartbeat interval for the server.
- Allow getting the session expiration information with ``connection.session_expiration_time``

Changed
~~~~~~~

- Refactored some code in the model building, which is optional if you only want
  to use xnatpy for a convenience layer about requests. Giving ``no_build_model=True``
  to the connect function will disable the scraping of the server xml structure and
  not create all classes, but will log in and keep alive a connection. Only the simple
  connection ``get``, ``head``, ``put``, ``post``, ``delete``, ``download``, ``upload``
  methods are really safe to use in that case.
- XNAT objects (including subject and experiments) will use the listing to get their
  label and xsitype to avoid the need to get each experiment when creating a listing.
  This makes listings way more efficient.

Fixed
~~~~~

- Bug with auth when xnat was not running in the server root, but rather in a subdirectory

0.3.13 - 2019-01-07
-------------------

Fixed
~~~~~

- Import problem in Python 2 which broke xnatpy

0.3.12 - 2019-01-03
-------------------

Added
~~~~~

- Adds fields argument to the dicom_dump method to filter on dicom tags server side.
- Adds dicom_dump method to prearchive scan.
- Allow deleting variables by useing del object.variable, this works in most
  cases but seem to fail server-side on restriction such as gender (it does
  not match any valid options)

Changed
~~~~~~~

- Beter computation for the uri's of resources

Fixed
~~~~~

- Fixed xml deprecation warning due to the use of .getchildren()

0.3.11 - 2018-11-12
-------------------

Fixed
~~~~~

- Functions with an async parameter had them renamed to asynchronous as async
  is a keyword as of Python 3.7
- Fix a bug in the XSD parsing when an XSD contains a schema-level simpleType
- Bug in upload_dir with python3 when using a method based on a temporary file

Added
~~~~~

- Resource upload methods can now forward kwargs to the ``connection.upload``
  method.
- Resource constructor also optionally takes a ``data_dir`` and ``upload_method``
  arguments for uploading data immediately after creation.


0.3.10 - 2018-08-31
-------------------

Added
~~~~~

- The experiment.create_resource and scan.create_resource now take two extra
  arguments: data_dir and method, which allow the uploading of the content of
  a directory as the content of the newly created resource. The method is the
  method for resource.upload_dir method
- Command line callable scripts that copies an entire project to another xnat.
  See ``xnat_cp_project --help``

Changed
~~~~~~~

- Removed wrong default argument for create_assessor (invalid assessor type)
- The lower level get/put/post/delete methods now can process full uris as well
  as paths as long as the uri start matches the server uri (e.g. instead of
  using /data/projects you can also give https://serveruri.com/data/projects.

Fixed
~~~~~

- Fixed a small bug where an incorrect error message was giving when not giving
  a value for the secondary label during object creation.

0.3.9 - 2018-07-02
------------------

Fixed
~~~~~

- xnatpy had issues with shared subjects and sessions as the REST API would
  return the original object (with sharing information in it). Now xnatpy
  check the requested uri and makes sure the information of the correct project
  is used. Now shared objects can be used properly in xnatpy.

Added
~~~~~

- resource.upload now takes an ``extract`` parameter indicating data should be
  extracted into files after upload
- resource.upload_dir to upload an entire directory to a resource, the directory
  will be added into the resources so that e.g. directory/a.txt becomes resource/a.txt
- redirection detections, if the server has moved and is being redirected (e.g. using 
  a 302 or 301 response), xnatpy will detect that and use the new url instead.

0.3.8 - 2018-06-04
------------------

Added
~~~~~

- Methods to retrieve the DICOM header dump using the dcmdump service. This
  can be used via services.dicom_dump(uri) or experiment.dicom_dump to get the
  dump of the specific experiment

Fixed
~~~~~

- Strict username checking after login disabled to avoid problems with OIDC
- Fix a bug where token result would contain extra data

0.3.7 - 2018-03-12
------------------

Fixed
~~~~~

- Fixed a bug where the prompt for the password on Windows would not work

Changed
~~~~~~~

- Hide certificate warnings if verify=False, just give a one time warning that
  things might not be safe, but no spam at every single request

0.3.6 - 2018-03-09
------------------

Added
~~~~~

- Support for issuing tokens in the service module

Fixed
~~~~~

- Allow user to login using a token (the username check will catch this an allow it)

Changed
~~~~~~~

- Improved the logging by reducing spam at the INFO/DEBUG levels. The debug parameter
  on connect can now be used to enable the logging of xnatpy internals.
- Give a specific error if the XNAt password is outdated and requires an update.

0.3.5 - 2018-01-02
------------------

Fixed
~~~~~

- There were bugs in the prearchive breaking the entire pre-archive funcationaly

0.3.4 - 2017-11-13
------------------

Fixed
~~~~~

- Files in assessors would have a path prefixed with a / in some cases (which should never happen)

0.3.3 - 2017-10-18
------------------

Changed
~~~~~~~

- Abstracted the progress bar for downloading to allow other progress hooks (e.g. GUI)

Fixed
~~~~~

- Set proper minimal versions for requirements (e.g. six can be too old)
- Bug in upload for Python3
- Bug with getting the file size when there are redirects (issue #8)
- Bug with getting files from a project/subject/experiment/scan directly instead of via resource (issue #5)


0.3.2 - 2017-10-15
------------------

Fixed
~~~~~

- Bug in the the create_object function in the selection of the non-history object
- Bug in the setting of project properties (due to the lack of a parent)

0.3.1 - 2017-09-04
------------------

Changed
~~~~~~~

- FileData now has an id and path, the id is the filename and the path is the
  path relative from the resource. This makes working with subdirectories in
  resources possible.

Fixed
~~~~~

- Bug where history of XNAT was misinterpreted and an old version of an object could be loaded
- Resources could loose track of their ID when the cache was cleared
- Resources did not invalidate cache after uploading files


0.3.0 - 2017-08-17
------------------

Added
~~~~~

- Better support for complex data structures, especially data types that
  include lists in their data.
- Support for extension types, xnatpy automatically searches for all extension
  xsd files and will create Python classes for those as well.
- Listings can be indexed with integers to get their n-th element, the order
  is the order given by XNAT.
- Allow overwriting of files on upload
- Support for listing users via /data/users REST endpoint in the session.users

Changed
~~~~~~~

- xsd schema parsing is completely rewritten, allows more support for complex
  data structures

Fixed
~~~~~

- Support for XNAT 1.7.3
- Fixed a bug where opening a second session would ruin the first one, it
  should now be possible to have multiple sessions open concurently.


0.2.3 - 2017-04-03
------------------

Added
~~~~~

- xnatpy now uses the progressbar2 package to deliver fancy progress bars when downloading
- Attributes in the session that allow users to skip/alter the checking of responses


Changed
~~~~~~~

- Logging now using a logger. You can change the log levels or supply your own logger which xnatpy will use in favour of its own
- xnatpy now gets the version information from 1.7 xnat correctly


Fixed
~~~~~

- A bug in XNAT 1.7 caused the prearchive routes to be wrong, added a work around that fixes the prearchive with xnatpy