Skip to content

Latest commit

 

History

History
31 lines (25 loc) · 2.22 KB

ROADMAP.md

File metadata and controls

31 lines (25 loc) · 2.22 KB
Raw

Roadmap

Priorities:

  • query subcommand in the CLI
  • expose reference documentation by turning it into HTML with python-inspector and pandoc.
  • fix bug in date handling of start/stop datetimes (see also: "extract date handling functionality")
  • in-depth reference documentation of all major functionality (authenticate, Account, CoreQuery, Report) in docstrings

Can wait:

  • nicer and more powerful CoreQuery#segment interface, perhaps using query#users and query#sessions, with a signature like *segments, scope=None) (where scope would refer to the inner scope, which can be more precise than the segment scope) and then use the or, and and any functions as well as methods on Column objects to fill out the remaining functionality
  • example report-to-html template
  • unit tests for the authentication functionality
  • clean up some of the object initializations (init should be clean, from_api can be dirty)
  • the ability to pluck rows and/or metrics in blueprints (e.g. it often happens that you query for a single metric in aggregate, and in that case the data you want back is query.rows[0].mymetric, that is, a single number, not a list of dictionaries)
  • consider turning sort and filter into an additive interface so that e.g. query.sort('-pageviews').sort('exits') does the right thing
  • caching of queries when using the lazy-loading shortcuts
  • suggestions for similar keys when a KeyError is raised on a column
  • support for the Real Time Reporting API
  • clean up column serialization code
  • nicer filtering interface
  • drop into a Python shell, automatically log in and expose accounts
  • wrap errors into ones that are easier to understand and test against in try/except blocks.

As part of supporting / other modules:

  • extract date handling functionality (ultimately, this is generic and needed in different contexts) -- perhaps make it a part of hrange
  • utilities for humanized output (ratios, cognitively rounded numbers, etc.)
  • verbose mode for blueprints and CLI (fetching query x...)
  • better diagnostics / google-analytics-specific errors rather than generic Python ones