/*
  parqit describe - Describe and explore a view or a Parquet file

*!  VERSION 0.1.37  07sep2026

  User > parqit > Describe and explore data...
  Launch directly with:  db parqit_explore      (menu installed by: parqit menu)

  One list box of operations (Stata's interface guideline for more than
  about twelve choices), each an engine-side query that leaves the data in
  memory unchanged: describe/glimpse, ds, lookfor, codebook, head, list,
  count, misstable [patterns], levelsof, distinct, duplicates report/list.
  Summary statistics live in the companion dialog:  db parqit_stats
*/

VERSION 16.0

INCLUDE _std_wide
DEFINE _dlght 400
INCLUDE header
HELP hlp1, view("help parqit##explore")
RESET res1

SCRIPT PREINIT
BEGIN
	create STRING dlgname
	dlgname.storeDialogClassName
	create DOUBLE pq_populate_error
	main.fi_file.formatfilter					///
		"Parquet (*.parquet)|*.parquet|"			///
		"All files (*.*)|*.*"
END

PROGRAM POSTINIT_PROGRAM
BEGIN
	call main.lb_op.forceselchange
	call program main_context
END

DIALOG main, label("parqit describe - Describe and explore a view or a Parquet file") ///
	tabtitle("Main")
BEGIN
  TEXT tx_op			_lft	_top	250	.,		///
	label("Operation:")
  LISTBOX lb_op			_lft	_ss	250	300,		///
	contents(op_words)						///
	values(op_vals)							///
	onselchangelist(op_change)					///
	label("Operation")

  TEXT tx_file			275	_top	315	.,		///
	label("Parquet file, glob, or Hive directory:")
  FILE fi_file			275	_ss	315	.,		///
	error("Filename")						///
	defext(parquet)							///
	label("Browse...")

  TEXT tx_vars			275	_ls	230	.,		///
	label("Variables (leave empty for all):")
  BUTTON bu_pop			510	@	80	.,		///
	onpush(program main_populate)					///
	tooltip("Populate combobox with the variable names of the current view") ///
	label("Populate")
  COMBOBOX cb_vars		275	_ss	315	.,		///
	append								///
	dropdown							///
	contents(vv_list)						///
	label("Variables")

  TEXT tx_if			275	_ls	315	.,		///
	label("If (expression):")
  EXP ex_if			275	_ss	315	.,		///
	label("Create...")

  TEXT tx_rows			275	_ls	60	.,		///
	label("Rows:") right
  EDIT ed_rows			340	@	60	.,		///
	numonly								///
	default(5)							///
	label("Rows")
  TEXT tx_limit			405	@	60	.,		///
	label("Limit:") right
  EDIT ed_limit			470	@	60	.,		///
	numonly								///
	option(limit)							///
	label("Limit")

  TEXT tx_from			275	_ls	60	.,		///
	label("From:") right
  EDIT ed_from			340	@	60	.,		///
	numonly								///
	label("From")
  TEXT tx_to			405	@	60	.,		///
	label("to:") right
  EDIT ed_to			470	@	60	.,		///
	numonly								///
	label("to")

  CHECKBOX ck_joint		275	_ls	315	.,		///
	option(joint)							///
	label("Also count the distinct joint values (joint)")

  TEXT tx_note			275	_ls	315	.,		///
	label("Engine-side query; the data in memory are unchanged.")
  TEXT tx_context _lft 355 470 ., label("Current view: refresh to check")
  BUTTON bu_context _rj80 @ 80 ., label("Refresh") onpush(program main_populate)
END

LIST vv_list
BEGIN
END

LIST op_words
BEGIN
	Describe the current view
	Describe a Parquet file (footer only)
	Glimpse the view or a Parquet file
	List variable names (ds)
	Search names and labels (lookfor)
	Codebook
	Preview the first rows (head)
	List observations (list)
	Count observations (count)
	Missing values by variable (misstable)
	Missing patterns (misstable patterns)
	Levels of one variable (levelsof)
	Distinct values (distinct)
	Report duplicates (duplicates report)
	List duplicates (duplicates list)
END

LIST op_vals
BEGIN
	descview
	descfile
	glimpse
	ds
	lookfor
	codebook
	head
	list
	count
	misstable
	misspatterns
	levelsof
	distinct
	dupreport
	duplist
END

LIST op_change
BEGIN
	script op_descview
	script op_descfile
	script op_glimpse
	script op_descview
	script op_lookfor
	script op_vars
	script op_head
	script op_list
	script op_count
	script op_vars
	script op_patterns
	script op_levelsof
	script op_distinct
	script op_keys
	script op_keys
END

SCRIPT all_off
BEGIN
	main.tx_file.disable
	main.fi_file.disable
	main.tx_vars.disable
	main.bu_pop.disable
	main.cb_vars.disable
	main.tx_if.disable
	main.ex_if.disable
	main.tx_rows.disable
	main.ed_rows.disable
	main.tx_limit.disable
	main.ed_limit.disable
	main.tx_from.disable
	main.ed_from.disable
	main.tx_to.disable
	main.ed_to.disable
	main.ck_joint.disable
	main.tx_vars.setlabel "Variables (leave empty for all):"
	main.tx_file.setlabel "Parquet file, glob, or Hive directory:"
END

SCRIPT vars_on
BEGIN
	main.tx_vars.enable
	main.bu_pop.enable
	main.cb_vars.enable
END

SCRIPT op_descview
BEGIN
	script all_off
END

SCRIPT op_descfile
BEGIN
	script all_off
	main.tx_file.enable
	main.fi_file.enable
END

SCRIPT op_glimpse
BEGIN
	script all_off
	main.tx_file.enable
	main.fi_file.enable
	main.tx_file.setlabel "Parquet file, glob, or Hive directory (leave empty for the current view):"
END

SCRIPT op_lookfor
BEGIN
	script all_off
	script vars_on
	main.tx_vars.setlabel "Words to search for in variable names and labels:"
END

SCRIPT op_vars
BEGIN
	script all_off
	script vars_on
END

SCRIPT op_patterns
BEGIN
	script all_off
	script vars_on
	main.tx_vars.setlabel "Variables (maximum 14; empty selects all):"
END

SCRIPT op_head
BEGIN
	script all_off
	main.tx_rows.enable
	main.ed_rows.enable
END

SCRIPT op_list
BEGIN
	script all_off
	script vars_on
	main.tx_if.enable
	main.ex_if.enable
	main.tx_from.enable
	main.ed_from.enable
	main.tx_to.enable
	main.ed_to.enable
END

SCRIPT op_count
BEGIN
	script all_off
	main.tx_if.enable
	main.ex_if.enable
END

SCRIPT op_levelsof
BEGIN
	script all_off
	script vars_on
	main.tx_vars.setlabel "Variable:"
	main.tx_limit.enable
	main.ed_limit.enable
END

SCRIPT op_distinct
BEGIN
	script all_off
	script vars_on
	main.ck_joint.enable
END

SCRIPT op_keys
BEGIN
	script all_off
	script vars_on
	main.tx_vars.setlabel "Key variables:"
	main.tx_limit.enable
	main.ed_limit.enable
END

PROGRAM main_populate
BEGIN
	call program main_context
	call pq_populate_error.setvalue 0
	put "parqit _dlgvars "
	put dlgname
	put " vv_list"
	stata hidden immediate
	if pq_populate_error {
		stopbox stop "Unable to get the variable names of the current view." ///
			"Open a view first (User > parqit > Read data...)."
	}
	call main.cb_vars.repopulate
END

PROGRAM main_context
BEGIN
	put "parqit _dlgcontext "
	put dlgname
	stata hidden queue
	clear
END

PROGRAM command
BEGIN
	if main.lb_op.equals("descview") {
		put "parqit describe"
	}
	if main.lb_op.equals("descfile") {
		require main.fi_file
		put "parqit describe "
		put /smartquote main.fi_file
	}
	if main.lb_op.equals("glimpse") {
		put "parqit glimpse"
		if main.fi_file {
			put " "
			put /smartquote main.fi_file
		}
	}
	if main.lb_op.equals("ds") {
		put "parqit ds"
	}
	if main.lb_op.equals("lookfor") {
		require main.cb_vars
		put "parqit lookfor " main.cb_vars
	}
	if main.lb_op.equals("codebook") {
		put "parqit codebook"
		if main.cb_vars {
			put " " main.cb_vars
		}
	}
	if main.lb_op.equals("head") {
		require main.ed_rows
		put "parqit head " main.ed_rows
	}
	if main.lb_op.equals("list") {
		put "parqit list"
		if main.cb_vars {
			put " " main.cb_vars
		}
		if main.ex_if {
			put " if " main.ex_if
		}
		if main.ed_from | main.ed_to {
			require main.ed_from
			require main.ed_to
			put " in " main.ed_from "/" main.ed_to
		}
	}
	if main.lb_op.equals("count") {
		put "parqit count"
		if main.ex_if {
			put " if " main.ex_if
		}
	}
	if main.lb_op.equals("misstable") {
		put "parqit misstable"
		if main.cb_vars {
			put " " main.cb_vars
		}
	}
	if main.lb_op.equals("misspatterns") {
		put "parqit misstable patterns"
		if main.cb_vars {
			put " " main.cb_vars
		}
	}
	if main.lb_op.equals("levelsof") {
		require main.cb_vars
		put "parqit levelsof " main.cb_vars
		beginoptions
			optionarg main.ed_limit
		endoptions
	}
	if main.lb_op.equals("distinct") {
		put "parqit distinct"
		if main.cb_vars {
			put " " main.cb_vars
		}
		beginoptions
			option main.ck_joint
		endoptions
	}
	if main.lb_op.equals("dupreport") {
		require main.cb_vars
		put "parqit duplicates report " main.cb_vars
		beginoptions
			optionarg main.ed_limit
		endoptions
	}
	if main.lb_op.equals("duplist") {
		require main.cb_vars
		put "parqit duplicates list " main.cb_vars
		beginoptions
			optionarg main.ed_limit
		endoptions
	}
END