Version 3.2

• Small bug fixes to the NFN extractor and text reducers.
• Bump dependency versions to latest releases

Version 3.1.0

Changes

New features

• Extractors and reducer for TESS
• New text extractor for transcription projects
  • Clusters using OPTICS
• NfN extractor added
• Effect userify added
• Place API behind a Gunicorn server
• Add running reducers for TESS

Package Updates

• Update Numpy to 1.17.0
• Update Sci-kit Learn to 0.21.3
• Update Pandas to 0.25.0
• Update to python 3.7

Other updates

• poly-line-text-reducer updated to track user_id, gold_standard, and extract_index (this matches the new OPTICS reducer)
  • These changes are needed for the new text-editor app that is being built

Version 3

Major changes

Command line argument update

The workflow auto-config no long accepts -l or -c as inputs. These were for picking the language and setting the location of the workflow-contents csv file. The workflow-contents file is no longer available from a project's data export page. The information contained in this file is now included in the workflow data export.

Since this data has moved the workflow auto-config script has been updated to always create the task label look-up-table based on the strings column of the workflow data export. It will always be in the primary language of the project.

New API route

A running_reducers route has been added for use with Caesar. These reducers take in only the latest extract and a store and return a new reduction and an updated store. These are useful for "user reductions" rather than "subject reductions" since one user could have a very large number of extracts where a subject will not.

relevant_reductions and user_id can be passed into reducers

This is being tested out for the TESS project and these options can be turned out with keywords passed to the reducer_wrapper decorator.

PRs that are included in this release:

#121: Update workflow auto-config for new workflow data dump format
#118: lock number version
#114: treat shortcut task as a separate extractor/reducer type
#103: Add custom extractors and reducers for TESS

Fix vertial text aggregation

The slope for near vertical text was not being calculated correctly, this release fixes the edge case.

Add a citable DOI to the repository

This release does not change any code but does add a DOI to the repository to enable citation of the code.

Add verbose flag to command line

All warning messages for the command line scripts have been placed behind an optional -vv or -verbose flag.

Fix windows GUI

The GUI script for windows will now run without crashing. The original bug is detailed in #100 and #101.

This fix works by making a copy of the panoptes_aggregation_gui.exe script without the .exe extension.

Add icons for GUI

Bug fix

• Include GUI icons in the package resources

GUI and unified command line scripts

Changes

• All command line scripts are unified under a single command panoptes_aggregation
  • config_workflow_panoptes is now panoptes_aggregation config
  • extract_panoptes_csv is now panoptes_aggregation extract
  • reduce_panoptes_csv is now panoptes_aggregation reduce
• All scripts have a new flag to specify the output directory --dir for the resulting files
• Documentation has been updated for installing the package with anaconda builds of python
• Scripting example updated
  • added link to example Penguin Watch data files
  • minor typos fixed in examples
  • command help text updated to match the new grouping being used
• setup.py will prevent pip from installing on python version below 3
• GUI added
  • the GUI can be installed with the command pip install -U panoptes_aggregation[gui] (this is to keep the size of the docker container down)
  • the GUI can be launched with the command panoptes_aggregation_gui
  • example documentation added for using the GUI

Bug fixes

• Fixed bug in the code that calculates the distance between angles

Slider task

This version adds support of the slider task. The reducer for this task provides the mean, median, and var for the extracted slider values.