udata

• Automatically loads default settings from plugins (if plugin.settings module exists) #2058
• Fixes some memory leaks on reindexing #2070
• Fixes minor UI bug #2072
• Prevent ExtrasField failure on null value #2074
• Improve ModelField errors handling #2075
• Fix territories home map #2077
• Prevent timeout on udata index in some cases #2079
• Pin werkzeug dependency to 0.14.1 until incompatibilities are fixed #2081
• Prevent client-side error while handling unparseable API response #2076
• Fix the udata job schedule erroneous help message #2083
• Fix upload button on replace resource file #2085
• Ensure harvest items statuses are updated on the right job #2089
• Added Serbian translations #2055

• Replace "An user" by "A user" #2033
• Use "udata" and fix a few other typos in documentation and UI/translation strings #2023
• Add a surrounding block declaration around community section 2039
• Fix broken form validation on admin discussions and issues #2045
• Fix full reindexation by avoiding SlugField.instance deepcopy in no_dereference() querysets #2048
• Ensure deleted user slug is pseudonymized #2049
• Prevent the "Add resource" modal from closing when using the frontend "Add resource" button #2052

• Fix workers: pin redis version for Celery compatibility #2019

• Remove extra attributes on user deletion #1961
• Pin phantomjs to version 2.1.7 #1975
• Protect membership accept route against flood #1984
• Ensure compatibility with IE11 and Firefox ESR #1990
• Lots of fixes on the resource form. Be explicit about uploading a new file #1991
• Centralize selectize handling and style in base-completer and apply some fixes 1992
• Added the missing number input field widget #1993
• Fix the organization private datasets and reuses counters #1994
• Disable autocorrect, spellcheck... on search and completion fields #1995
• Fix harvest preview in edit form not taking configuration (features and filters) #1996
• Ensure organization page react to URL hash changes (including those from right sidebar) #1997
• Updating community resource as admin keeps original owner #1999
• Major form fixes #2000
• Improved admin errors handling: visual feedback on all errors, Sentry-ID header if present, hide organization unauthorized actions #2005
• Expose and import licenses alternate_urls and alternate_titles fields #2006
• Be consistent on search results wording and icons (Stars vs Followers) #2013
• Switch from a "full facet reset" to a "by term reset" approach in search facets #2014
• Ensures all modals have the same buttons styles and orders, same color code... #2012
• Ensure URLs from assets stored on CDN_DOMAINS are considered as valid and that associated error message is properly translated #2017

• Display the owner/organization on harvester view #1921
• Improve harvest validation errors handling #1920
• Make extra TOS text customizable #1922
• Fixes an UnicodeEncodeError occuring when parsing RDF with unicode URLs #1919
• Fix some external assets handling cases #1918
• Harvest items can now match source.id before source.domain — no more duplicates when changing an harvester URL #1923
• Ensure image picker/cropper only allows images #1925
• Make tags min and max length configurable and ensure admin takes its configuration from the backend #1935
• Prevent errors when there is no date available to focus on the calendar #1937

• Allows arguments and keyword arguments in the task @connect decorator #1908
• Allows to restore assets after being deleted (Datasets, Organizations and Reuses) #1901
• Fixes form events not bubbling (and so fixes harvester config not displaying) #1914

New features

• Harvest sources are now filterable through the harvest source create/edit admin form #1812
• Harvest sources can now enable or disable some optionnal backend features #1875
• Static assets are now compatible with long-term caching (ie. their hash is present in the filename) #1826
• Post UIs have been reworked: publication date, publish/unpublish action, save and continue editing, dynamic sidebar, alignments fixes... #1857

Minor changes

• Only display temporal coverage years on cards and search results #1833
• Add publisher's name on dataset template #1847
• Improved upload error handling: deduplicate notifications, localized generic error message, sentry identifier... #1842
• Allows to filter datasets on resource type (needs reindexing) #1848
• Switch the admin sidebar collapse icon from "hamburger"to left and right arrows #1855
• Discussion add card style coherence #1884
• LINKCHECKING_UNCHECKED_TYPES setting to prevent linkchecking on some ressource types #1892
• swagger.json API specifications now pass validation #1898

Breaking changes

• Theme are now responsible for adding their CSS markup on template (no more assumptions on theme.css and admin.css). Most of the time, overriding raw.html and admin.html should be sufficient
• The discussions API posted_by attribute is now an embedded user instead of an user ID to avoid extra API calls #1839

Bugfixes

• Hide the resource.type attribute from JSON-LD output until handled by a dedicated vocabulary/property #1865
• RDFs, CSVs and resource redirect views are now handling CORS properly #1866
• Fix broken sorts on organization's datasets list in admin #1873
• Ensure harvest previewing is done against current form content #1888
• Ensure deleted objects are unindexed #1891
• Fix the dataset resources list layout wrapping #1893
• Fix wrong behavior for weblinks #1894
• Ensure info config command only displays configuration variables #1897

Internal

• Upgrade to Authlib 0.9 #1760 #1827
• Add a Dataset.on_resource_added signal

• Prevent UnicodeError on unicode URL validation error #1844
• Hide save button in "Add resource" modal until form is visible (and prevent error) #1846
• The purge chunks tasks also remove the directory #1845
• Upgrade to latest Fine-Uploader version to benefit from bug fixes #1849
• Prevent front views from downloading swagger.json #1838
• Ensure API docs works without data #1840
• Expose the default spatial granularity in API specs #1841
• Fix missing dataset title on client-side card listing #1834
• Allows to clear the dataset form temporal coverage. #1832
• Ensure that admin notifications are displayed once and with a constant width. #1831
• Fix broken date range picker date parsing (ie. manual keyboard input) #1863
• Normalize uploaded filenames to avoid encoding issues, filesystem incompatibilities... #1852

• Fix client-side temporal coverage rendering #1821
• Prevent word breaking when wrapping discussions messages #1822
• Properly render message content on issues and discussions mails #1823

• Ensure OEmbed compatibility with external CDN #1815
• Fixes some static URL serialization #1815

New features

• Slugs are now redirected on change when changed until old slug are free #1771
• Improve usability of new organization form #1777
• Allows to serve assets on an external CDN domain using CDN_DOMAIN #1804

Breaking changes

None

Bug fixes and minor changes

• Sort dataset update frequencies by ascending frequency #1758
• Skip gov.uk references tests when site is unreachable #1767
• Fix resources reorder (registered extras validation logic) #1796
• Fix checksum display on resource modal #1797
• Use metrics.views on resource card #1778
• Fix dataset collapse on ie11 #1802
• Upgrade i18next (security) #1803

Internals

• Backports some Python 3 forward compatible changes and fixes some bugs #1769:
  • avoid filter and map usage instead of list comprehension
  • explicit encoding handling
  • avoid comparison to None
  • use next() instead of .next() to iterate
  • unhide some implicit casts (in particular search weight)
• Tests are now run against local.test instead of localhost to avoid pytest warnings

• Fix community resource creation and display #1733
• Failsafe JS cache storage: use a custom in-memory storage as fallback when access to sessionStorage is not allowed #1742
• Prevent errors when handling API errors without data/payload #1743
• Improve/fix validation error formatting on harvesting #1745
• Ensure daterange can be parsed from full iso datetime #1748
• API: enforce application/json content-type for forms #1751
• RDF parser can now process european frequencies #1752
• Fix images upload broken by chunked upload #1756

New features

• Typed resources #1398
• Initial data preview implementation #1581 #1632
• Handle some alternate titles and alternate URLs on licenses for improved match on harvesting #1592
• Allow to specify a dataset acronym #1217
• Starts using harvest backend config (validation, API exposition, HarvestFilters...) #1716
• The map widget can now be configured (tiles URL, initial position...) #1672
• New discussions layout #1623
• Dynamic API documentation, Enhancement to Pull #1542 - #1542
• Resource modal overhaul with markdown support #1547

Breaking changes

• Normalize resource.format (migration - ⚠️ need reindexing). #1563
• Enforce a domain whitelist when resource.filetype is file. See RESOURCES_FILE_ALLOWED_DOMAINS settings variable for details and configuration. #1567
• Remove extras from datasets search index (needs reindexation) #1718

Bug fixes and minor changes

• Switch to PyPI.org for package links #1583
• Show resource type in modal (front) #1714
• Adds ETag to internal avatar for efficient caching control #1712
• Fix 404/missing css on front pages #1709
• Fix markdown max image width (front) #1707
• Ensure registered extras types are properly parsed from JSON. Remove the need for custom db.Extra classes #1699
• Fix the temporal coverage facet query string parsing #1676
• Fix search auto-complete hitbox #1687
• Fix Firefox custom error handling, part 2 #1671
• Add resend confirmation email link to login screen #1653
• Audience metrics: use only views #1607
• Add missing spatial granularities translations #1636
• Protocol-relative URLs support #1599

Internals

• Simplify ExtrasField form field signature (no need anymore for the extras parameter) #1698
• Register known extras types #1700

• Fix side menu on mobile #1701
• Fix update frequency field #1702

• Protect Resource.need_check against malformed/string dates #1691
• Fix search auto-complete loading on new page #1693

• Expose Resource.extras as writable in the API #1660
• Fix Firefox custom errors handling #1662

• Prevent linkchecker to pollute timeline as a side-effect. (migration). Warning, the migration will delete all dataset update activities #1643
• Fix OAuth authorization screen failing with unicode SITE_TITLE #1624
• Fix markdown handling of autolinks with angle brackets and factorize (and test) markdown parse_html() #1625
• Fix timeline order #1642
• Fix markdown rendering on IE11 #1645
• Consider bad UUID as 404 in routing #1646
• Add missing email templates #1647
• Polyfill ChildNode.remove() for IE11 #1648
• Improve Raven-js/Sentry error handling #1649
• Prevent regex special characters to break site search #1650

• Fix sendmail regression #1620

• Fix some search parameters validation #1601
• Prevent API tracking errors with unicode #1602
• Prevent a race condition error when uploading file with concurrent chunking #1606
• Disallow resources dict in API #1603
• Test and fix territories routing #1611
• Fix the client-side Raven/Sentry configuration #1612
• Raise a 404 in case of unknown RDF content type #1613
• Ensure current theme is available to macros requiring it in mails #1614
• Fix documentation about NGinx configuration for https #1615
• Remove unwanted commas in default SECURITY_EMAIL_SUBJECT_* parameters #1616

• Prevent OEmbed card to be styled when loaded in bootstrap 4 #1569
• Fix organizations sort by last_modified #1576
• Fix dataset creation form (and any other form) #1584
• Fix an XSS on client-side markdown parsing #1585
• Ensure URLs validation is the same everywhere #1586