GitLab

• change the strategy:
  • In the current implementation reports are based on a query from the database.
  • It would be more general if reports are based on a DataFrame.
  • The DataFrame can be filled with the content of a database table including foreign fields or by a dedicated algorithm mixing, for example, different events.
• In the model register the source producing the DataFrame, a dictionary linking a name to a function.
• The report configuration should contains 6 sections
  • Select the source
  • Define selection criteria / query. It is based on a string written in natural language and on the DataFrame.query method.
  • Transform the source using pandas chaining. It is a string interpreted using python eval, a general approach which should allow all kind of postprocessing.
  • View defining the DataFrame columns which are shown in the Ext JS Grid, the grouping and the summary.
  • Summary to customize summary row of the grid: per group, total, etc
  • Permission to define roles with can used this report. Add a JSON field containing the list of role allowed to see the report
  • ...
• In order to implement this approach:
  • Remove dedicated selector.
  • Migrate general function from lhcbfrance.modules.ana: active_period, coverage, db2df, full_name, get_people,is_end, is_start, normalize_history_data to the module report_tools.
  • Create the functions returning a DataFrame with the content of the history table
  • Create the functions / method in order to transforme a DataFrame into an Ext JS store, grid
  • Modify the database schema to handle additional configuration parameters fo report
  • Adapt the tools List, Metric1D and Metric2D.
  • Adapt controllers.
  • Update the documentation.
  • ...