DOKK / manpages / debian 11 / ncurses-doc / echo_wchar.3ncurses.en
add_wch(3NCURSES) add_wch(3NCURSES)

add_wch, wadd_wch, mvadd_wch, mvwadd_wch, echo_wchar, wecho_wchar - add a complex character and rendition to a curses window, then advance the cursor

#include <curses.h>

int add_wch( const cchar_t *wch );
int wadd_wch( WINDOW *win, const cchar_t *wch );
int mvadd_wch( int y, int x, const cchar_t *wch );
int mvwadd_wch( WINDOW *win, int y, int x, const cchar_t *wch );

int echo_wchar( const cchar_t *wch );
int wecho_wchar( WINDOW *win, const cchar_t *wch );

The add_wch, wadd_wch, mvadd_wch, and mvwadd_wch functions put the complex character wch into the given window at its current position, which is then advanced. These functions perform wrapping and special-character processing as follows:

  • If wch refers to a spacing character, then any previous character at that location is removed. A new character specified by wch is placed at that location with rendition specified by wch. The cursor then advances to the next spacing character on the screen.
  • If wch refers to a non-spacing character, all previous characters at that location are preserved. The non-spacing characters of wch are added to the spacing complex character, and the rendition specified by wch is ignored.
  • If the character part of wch is a tab, newline, backspace or other control character, the window is updated and the cursor moves as if addch were called.

The echo_wchar function is functionally equivalent to a call to add_wch followed by a call to refresh(3X). Similarly, the wecho_wchar is functionally equivalent to a call to wadd_wch followed by a call to wrefresh. The knowledge that only a single character is being output is taken into consideration and, for non-control characters, a considerable performance gain might be seen by using the *echo* functions instead of their equivalents.

Like addch(3X), addch_wch accepts symbols which make it simple to draw lines and other frequently used special characters. These symbols correspond to the same VT100 line-drawing set as addch(3X).

ACS Unicode ASCII acsc Glyph
Name Default Default char Name
WACS_BLOCK 0x25ae # 0 solid square block
WACS_BOARD 0x2592 # h board of squares
WACS_BTEE 0x2534 + v bottom tee
WACS_BULLET 0x00b7 o ~ bullet
WACS_CKBOARD 0x2592 : a checker board (stipple)
WACS_DARROW 0x2193 v . arrow pointing down
WACS_DEGREE 0x00b0 ' f degree symbol
WACS_DIAMOND 0x25c6 + ` diamond
WACS_GEQUAL 0x2265 > > greater-than-or-equal-to
WACS_HLINE 0x2500 - q horizontal line
WACS_LANTERN 0x2603 # i lantern symbol
WACS_LARROW 0x2190 < , arrow pointing left
WACS_LEQUAL 0x2264 < y less-than-or-equal-to
WACS_LLCORNER 0x2514 + m lower left-hand corner
WACS_LRCORNER 0x2518 + j lower right-hand corner
WACS_LTEE 0x2524 + t left tee
WACS_NEQUAL 0x2260 ! | not-equal
WACS_PI 0x03c0 * { greek pi
WACS_PLMINUS 0x00b1 # g plus/minus
WACS_PLUS 0x253c + n plus
WACS_RARROW 0x2192 > + arrow pointing right
WACS_RTEE 0x251c + u right tee
WACS_S1 0x23ba - o scan line 1
WACS_S3 0x23bb - p scan line 3
WACS_S7 0x23bc - r scan line 7
WACS_S9 0x23bd _ s scan line 9
WACS_STERLING 0x00a3 f } pound-sterling symbol
WACS_TTEE 0x252c + w top tee
WACS_UARROW 0x2191 ^ - arrow pointing up
WACS_ULCORNER 0x250c + l upper left-hand corner
WACS_URCORNER 0x2510 + k upper right-hand corner
WACS_VLINE 0x2502 | x vertical line

The wide-character configuration of ncurses also defines symbols for thick lines (acsc “J” to “V”):

ACS Unicode ASCII acsc Glyph
Name Default Default char Name
WACS_T_BTEE 0x253b + V thick tee pointing up
WACS_T_HLINE 0x2501 - Q thick horizontal line
WACS_T_LLCORNER 0x2517 + M thick lower left corner
WACS_T_LRCORNER 0x251b + J thick lower right corner
WACS_T_LTEE 0x252b + T thick tee pointing right
WACS_T_PLUS 0x254b + N thick large plus
WACS_T_RTEE 0x2523 + U thick tee pointing left
WACS_T_TTEE 0x2533 + W thick tee pointing down
WACS_T_ULCORNER 0x250f + L thick upper left corner
WACS_T_URCORNER 0x2513 + K thick upper right corner
WACS_T_VLINE 0x2503 | X thick vertical line

and for double-lines (acsc “A” to “I”):

ACS Unicode ASCII acsc Glyph
Name Default Default char Name
WACS_D_BTEE 0x2569 + H double tee pointing up
WACS_D_HLINE 0x2550 - R double horizontal line
WACS_D_LLCORNER 0x255a + D double lower left corner
WACS_D_LRCORNER 0x255d + A double lower right corner
WACS_D_LTEE 0x2560 + F double tee pointing right
WACS_D_PLUS 0x256c + E double large plus
WACS_D_RTEE 0x2563 + G double tee pointing left
WACS_D_TTEE 0x2566 + I double tee pointing down
WACS_D_ULCORNER 0x2554 + C double upper left corner
WACS_D_URCORNER 0x2557 + B double upper right corner
WACS_D_VLINE 0x2551 | Y double vertical line

Unicode's descriptions for these characters differs slightly from ncurses, by introducing the term “light” (along with less important details). Here are its descriptions for the normal, thick, and double horizontal lines:

  • U+2500 BOX DRAWINGS LIGHT HORIZONTAL
  • U+2501 BOX DRAWINGS HEAVY HORIZONTAL
  • U+2550 BOX DRAWINGS DOUBLE HORIZONTAL

All routines return the integer ERR upon failure and OK on success.

X/Open does not define any error conditions. This implementation returns an error

  • if the window pointer is null or
  • if it is not possible to add a complete character in the window.

The latter may be due to different causes:

  • If scrollok is not enabled, writing a character at the lower right margin succeeds. However, an error is returned because it is not possible to wrap to a new line
  • If an error is detected when converting a multibyte character to a sequence of bytes, or if it is not possible to add all of the resulting bytes in the window, an error is returned.

Functions with a “mv” prefix first perform a cursor movement using wmove, and return an error if the position is outside the window, or if the window pointer is null.

Note that add_wch, mvadd_wch, mvwadd_wch, and echo_wchar may be macros.

All of these functions are described in the XSI Curses standard, Issue 4. The defaults specified for line-drawing characters apply in the POSIX locale.

X/Open Curses makes it clear that the WACS_ symbols should be defined as a pointer to cchar_t data, e.g., in the discussion of border_set. A few implementations are problematic:

  • NetBSD curses defines the symbols as a wchar_t within a cchar_t.
  • HPUX curses equates some of the ACS_ symbols to the analogous WACS_ symbols as if the ACS_ symbols were wide characters. The misdefined symbols are the arrows and other symbols which are not used for line-drawing.

X/Open Curses does not define symbols for thick- or double-lines. SVr4 curses implementations defined their line-drawing symbols in terms of intermediate symbols. This implementation extends those symbols, providing new definitions which are not in the SVr4 implementations.

Not all Unicode-capable terminals provide support for VT100-style alternate character sets (i.e., the acsc capability), with their corresponding line-drawing characters. X/Open Curses did not address the aspect of integrating Unicode with line-drawing characters. Existing implementations of Unix curses (AIX, HPUX, Solaris) use only the acsc character-mapping to provide this feature. As a result, those implementations can only use single-byte line-drawing characters. Ncurses 5.3 (2002) provided a table of Unicode values to solve these problems. NetBSD curses incorporated that table in 2010.

In this implementation, the Unicode values are used instead of the terminal description's acsc mapping as discussed in ncurses(3X) for the environment variable NCURSES_NO_UTF8_ACS. In contrast, for the same cases, the line-drawing characters described in curs_addch(3X) will use only the ASCII default values.

Having Unicode available does not solve all of the problems with line-drawing for curses:

  • The closest Unicode equivalents to the VT100 graphics S1, S3, S7 and S9 frequently are not displayed at the regular intervals which the terminal used.
  • The lantern is a special case. It originated with the AT&T 4410 terminal in the early 1980s. There is no accessible documentation depicting the lantern symbol on the AT&T terminal.
Lacking documentation, most readers assume that a storm lantern was intended. But there are several possibilities, all with problems.
Unicode 6.0 (2010) does provide two lantern symbols: U+1F383 and U+1F3EE. Those were not available in 2002, and are irrelevant since they lie outside the BMP and as a result are not generally available in terminals. They are not storm lanterns, in any case.
Most storm lanterns have a tapering glass chimney (to guard against tipping); some have a wire grid protecting the chimney.
For the tapering appearance, ☃ U+2603 was adequate. In use on a terminal, no one can tell what the image represents. Unicode calls it a snowman.
Others have suggested these alternatives: § U+00A7 (section mark), Θ U+0398 (theta), Φ U+03A6 (phi), δ U+03B4 (delta), ⌧ U+2327 (x in a rectangle), ╬ U+256C (forms double vertical and horizontal), and ☒ U+2612 (ballot box with x).

ncurses(3NCURSES), addch(3NCURSES), attr(3NCURSES), clear(3NCURSES), outopts(3NCURSES), refresh(3NCURSES), putwc(3)