Add additional selection helpers.

[ericemc3]

These are suggestions from a R practice, with handy functionalities i frequently use:

1. select
   current possibilities such as selecting by name, indices, not, range are excellent. I'd like also: starts_with, contains, and the possibility to test on column types (for instance select all numeric columns)

2. rename
   of course select in Arquero allows to rename, i have noticed that.
   Within R, janitor::clean_names is very useful for tidying and fixing complex column_names that you can find, sometimes, in CSV files (with blanks, accented characters, caps) => DB compliant column names (one single world, lowercase, without accented or special chars...)

[jheer]

Thanks! I've created a PR that adds more selection helpers (matches, startswith, endswith) for parity with dplyr.

As for more advanced renaming support, you can always add (and share!) your own utilities outside of the core Arquero library. For example:

function normalize_name(str) {
  // perform custom column name normalization
}

// create a mapping of original column names to normalized names
const nameMap = {};
table.columnNames.forEach(s => nameMap[name] = normalize_name(s));

// rename the columns
const newTable = table.select(nameMap);

[ericemc3]

Thank you for the suggestion, so here is my share for normalizing table columns names:

function normalize_column_name(str, options = {lower:true}) {
    str = str.replace(/[%#$£()\'\"]/g, '')
             .replace( /[ /,+.*:\-\r\n@]/g, "_" ) 
             .replace(/_+/g, '_') 
             .normalize("NFD").replace(/[\u0300-\u036f]/g, "") //get rid of accents
   
  if (options.lower) 
    str = str.toLowerCase() ;
  
  return str ;
}

function normalize_table_names(tb, options = {lower:true}) {
  return tb.select( tb.columnNames().map(d => ({[d]: normalize_column_name(d, options)})) ) ;
}

[jheer]

Thanks for sharing! @ericemc3, do you mind if I later add this (or a minor revision thereof) to the Arquero Cookbook? If that's OK, let me know how you'd like me to acknowledge your contribution (by name, GitHub id, Observable id, or...?).

[ericemc3]

My pleasure, you can of course reuse this as you want. I published them on https://observablehq.com/@ericmauviere/arquero-table-utils

[ericemc3]

I added a new utility function for renaming columns :

function rename_columns(tb, upd) {
  let colsdef = tb.columnNames().map(d => ({[d]:d}) ).reduce((a,v) => Object.assign(a,v)) ;
  
  for (const k in upd) {
    if (colsdef.hasOwnProperty(k)) 
        colsdef[k] = upd[k] ; 
  }
  
  return tb.select(colsdef) ;
}

rename_columns(tb, {currentname1:'newname1', currentname2:'newname2'})

[jheer]

New selection helpers now staged for v1.2.0.