Ckanext-xloader

================

Fixes:
* populate_full_text_trigger error when doing 'datastore set-permissions' 72
* '%' in column name causes TypeError("'dict' object does not support indexing") 65
* numpy >= 1.16 causes 'RuntimeError: implement_array_function' on CKAN startup 79
* CKAN 2.9 compatibility - fix error about 'resource_revision_table' 81

=================

Features:
* 'ignore_hash' config option added to bypass the hash check which can skip loads 34
* File size is logged 35
* Retries are no done when downloading CSV 39
* xloader_status action now available to GET (not just POST) 46
* ANALYZE run after load, needed for CKAN 2.9 resource preview speed-up 47
* CLI load of multiple datasets now resumes following an error with a resource 57
* Added 'max_excerpt_lines' config to load an excerpt of files larger than max_content_length 63

Fixes:
* Unicode characters in CSV headers caused error 28
* Column headers longer than 63 characters caused error 14
* Floats in CSV headers caused error 49
* Integers in xlsx header caused error 53
* Extraneous commas in combination with non-ascii chars caused error 51
* CSV with no rows caused error 38
* Template compatibility with CKAN 2.9 40
* CLI submitted resources with non-xloader formats 43
* ckanext.xloader.max_content_length wasn't working 62
* KeyError: 'resource_id' when xloading a resource 68

=================

* Fix for exception about 'ckan.datapusher.url' not being set when you check xload status 26

=================

* Trigger any resource_views to be created straight after load, rather than wait for the index.

=================

* Renamed ckanext-xloader
* Added to PyPI
* The user is given access to the data earlier in the job - the column indexing now occurs afterwards, since this is only an optimization for queries and takes much longer than the load itself
* Fixed exception during error-handling for files too long and of non-accepted schemes

===============

* Initial code, named ckanext-shift.