Feature request: client-side row filtering (like less "&")

[lvbuwei]

I'd like to propose a small feature: client-side row filtering.

Motivation

When inspecting wide tables (e.g. \d+ output, CSV, or query results),
it is often useful to temporarily hide every row that does not match a
substring, to focus on the rows of interest. The less pager offers
exactly this with its & command.

Proposed behavior

• Pressing & prompts for a substring (same prompt UX as the / search).
• Only data rows that contain the substring stay visible; non-matching
  rows are hidden. The header, footer and corner windows stay intact.
• Cursor navigation (j / k / g / G) moves between matching rows.
• An empty pattern (press & then Enter) clears the filter and restores
  all rows.
• Matching honors the global ignore_case option.

Implementation

I have a working implementation ready. It:

• adds a cmd_Filter command bound to & plus a "Filter rows" menu item,
• stores the pattern in ScrDesc (so it survives re-layout),
• makes window_fill skip non-matching data rows, deriving the row
  number and cursor position from the real line number so rownum/cursor
  stay in sync.

Known limitations:

• Case handling follows ignore_case only (no smart-case
  ignore_lower_case heuristic like search).
• When no row matches, the data area is shown empty with a
  "Pattern not found" notice; clear it with an empty pattern.

A patch is attached / a pull request is available if you are
interested. Would you be open to accepting such a feature?

[okbob]

I am not against this feature, but I think it needs more thinking about cursor, and first row (on the screen) positioning. Your simple implementation can produce some unwanted visual artefacts. Probably it needs an alternative total_rows, and switching when filter is active. Everything should to work although the cursors can be invisible (deactivated).