Using the Terminal UI

← Contents

The terminal UI (TUI) runs in a terminal, so you can browse tables on a remote server over SSH. It offers the same features as the GUI, including histograms; of the scatter plot, it shows the 2D density.

• Starting
• The screen
• Moving and selecting
• The sidebar: cell contents and statistics
• Histograms
• 2D density
• Group By
• Sorting
• Filtering
• Searching
• Going to a row, cell or column
• Hiding and freezing columns
• Column types and formats
• Copying
• Sheets
• Running SQL
• Exporting and converting
• Editing cells
• Following a growing file
• Other settings
• Key reference

Starting

tableskim data.csv        # without a graphical display
tableskim -t data.csv     # force the terminal UI
tableskim-tui.exe data.csv   # Windows

The terminal UI starts automatically when there is no graphical display (see Getting Started). It shows exactly one file, given on the command line. Press q (or Ctrl+C) to quit.

The screen

• The header shows the column names (or letters) and the sort and filter marks. The current cell is highlighted. A mark before the name tells what the column seems to hold: # numbers, @ dates, ? booleans (see Column types and formats).
• The row numbers are on the left.
• The status bar at the bottom shows the file name, the current cell (R1C1), messages, the number of rows and columns, and the active options (header, auto-revert, the filter, the search). When there is no message, it lists the keys used most (/ search f filter o sort …, and in edit mode Enter edit u undo ^S save …), as many as fit.
• Files with several sheets show sheet tabs at the top.

Press ? (or F1) at any time for the list of keys.

Moving and selecting

The mouse also works: click a cell to move the cursor, double-click to show the cell in the sidebar, use the wheel to scroll, and click a sheet tab to switch sheets.

The sidebar: cell contents and statistics

• Enter (or a double click) shows the whole value of the current cell in the sidebar, which is useful for long text.
• S shows the statistics of the current column: rows, empty cells, distinct values, minimum and maximum (and the sum, mean and standard deviation for numbers), and the most frequent values. Press S again to go back to the cell.
• s shows or hides the sidebar.

Histograms

Charts and Group By shows, with examples, how to read the charts and the Group By results.

B shows a histogram of the current column in place of the sheet. The title line summarizes what was counted, and each line is one bar with its range, count and percentage.

As in the GUI, numbers are counted into about 30 bins of round widths, dates into bins of whole days, months, years and so on, and text columns show their most frequent values. After Enter, the histogram is computed again for the rows left, so you can narrow down step by step; press Esc to see the filtered sheet.

2D density

M shows how two columns relate, in place of the sheet: the first and last of the selected columns on the horizontal (X) and vertical (Y) axes, or else columns holding numbers or dates chosen as in the GUI: Y is the cursor's column or the nearest such one, X the nearest such one to its left. The plot is a grid of cells, each coloured by the number of rows whose X and Y values fall in it, from dark purple (few) to yellow (many); empty cells stay blank. The grid is sized to the terminal. The densest cell is selected first, and the line at the bottom shows its ranges and count. The title line also shows Pearson's correlation coefficient r, R² and the least-squares line of the rows plotted, as in the GUI.

Y must hold numbers or dates, as for the GUI's scatter plot. The colours need a terminal with 256 or more colours.

When X holds text, each of its 30 most frequent values (and (other) for the rest) gets a line of its own, showing a box plot of its Y values: the whiskers ├──┤, the box ███ from the first to the third quartile, the median ┃ and outliers ∘, on a common scale. The line at the bottom shows the numbers of the selected category.

Group By

A counts the rows of the current sheet (sorted and filtered as shown) by the values of some columns and shows the result in a new sheet named Group By, as the GUI's Group By does. A dialog lists the columns: the selected ones (or the cursor's) are checked to group by.

The functions are count (non-empty values), distinct, sum, mean, min, max and std dev; each one makes a column for each aggregated column, after the group columns and count, the rows of the group. The status bar shows the progress, and Esc cancels. The sheet stays when the file is reloaded; grouping again replaces it.

Sorting

o sorts by the current column: ascending, then descending, then back to the file order.

Filtering

Chart selections are kept as Additional narrowing, ANDed with the editable conditions even in OR mode. In the filter editor, Ctrl+X removes only this additional narrowing; the existing clear-filter command (F on the sheet or chart) removes both. Further chart selections narrow the remaining rows again.

Quick filters

On a cell, = keeps only the rows with the same value in that column, and ! keeps only the rows with a different value. F clears the filter.

The filter dialog

f opens the filter dialog, which edits the same kind of filter as the GUI Filter pane.

The status bar shows how many rows match and the filter in use:

The operators and how values are compared are described in Sorting, Filtering, Searching and Statistics.

Searching

/ opens the search prompt at the bottom. Type the text and press Enter. Ctrl+R in the prompt toggles regular expression search. n / N move to the next / previous hit; the status bar shows the query and the hit number (/watanabe 1/20). An empty search clears the search.

Going to a row, cell or column

Ctrl+G asks for a row (120), a cell (B12 or R12C2) or a column (its letter or name), as in the GUI.

Hiding and freezing columns

• x hides the current column (or the selected columns); X shows all columns again.
• | freezes the columns up to the cursor, so that they stay in place when scrolling sideways; press | again to unfreeze.

Column types and formats

The mark before a column name is judged from the first 1,000 rows read, as in the GUI: # numbers, @ dates and date-times, ? booleans; text and mixed columns have none.

% changes how the current column (or the selected columns) shows its values. The dialog also shows the column's type.

↑ / ↓ move between the lines, Delete resets the format, Enter applies it and Esc closes the dialog. Only the display changes: sorting, filtering, copying and exporting use the values as they are.

Copying

y or c copies the selection (or the current cell) as tab-separated text.

Copying uses the OSC 52 escape sequence, so it also works over SSH: the text goes to the clipboard of the computer where your terminal runs. The terminal must allow OSC 52; inside tmux, enable set-clipboard:

set -g set-clipboard on

Sheets

Tab / Shift+Tab (or ] / [) switch to the next / previous sheet. You can also click a tab.

Running SQL

For DuckDB, SQLite, Parquet and JSON files, : opens the SQL prompt.

• Enter runs the query; a query result is shown as the sheet SQL Result.
• Up / Down recall SQL run before.
• Tab completes table names, column names and keywords.
• Esc cancels.

See Running SQL for details.

Exporting and converting

• w exports the current sheet as shown (sorted, filtered, without hidden columns). A prompt suggests a file name; the extension (.csv, .tsv, .xlsx or .parquet) selects the format.
• P / D convert a text file to Parquet / DuckDB next to the original file and open the result.

Progress is shown in the status bar; Esc cancels (instead of clearing the selection). Until it has finished, the file is not reopened: r, H, # and E ask you to wait, and Auto Revert waits. If the target file exists, TableSkim asks Overwrite? (y/n). See Exporting and Converting.

Editing cells

Cells of CSV / TSV and Excel files can be changed in edit mode only: I turns it on and off, and the status bar shows EDIT. Then Enter, i, F2 or a double click edits the cell in the status line, Delete clears the selected cells, pasting into the terminal fills the selected cells with tab-separated text, u / Ctrl+R undo and redo, and Ctrl+S saves the file. Edited cells are green and the file name starts with * until saved. Before unsaved changes would be lost (q, r, H, #, E, or leaving edit mode), TableSkim asks: y saves, n discards, Esc cancels. See Editing Cells.

Following a growing file

T follows the end of the file, like tail -f: the file is checked every second (with or without Auto Revert), and the cursor moves to the last row whenever rows are added. The status bar shows follow. For CSV / TSV files that are not compressed (in UTF-8 or Shift_JIS), only the appended lines are read, and a line being written is shown once it ends. Other files, and files that were rewritten rather than appended to, are reloaded. Press T again to stop following.

Other settings

When the rows of a large CSV / TSV file take a lot of memory, the status bar asks Continue reading? (y/n), like the GUI (see Supported File Formats).

Log messages are written only with --log <file>, because they would otherwise be drawn over the screen.

Key reference

See Keyboard Shortcuts for the full list.