Paul Evans > Term-TermKey-0.15 > Term::TermKey

Download:
Term-TermKey-0.15.tar.gz

Dependencies

Annotate this POD

CPAN RT

Open  1
View/Report Bugs
Module Version: 0.15   Source   Latest Release: Term-TermKey-0.16

NAME ^

Term::TermKey - perl wrapper around libtermkey

SYNOPSIS ^

 use Term::TermKey;

 my $tk = Term::TermKey->new( \*STDIN );

 print "Press any key\n";

 $tk->waitkey( my $key );

 print "You pressed: " . $tk->format_key( $key, 0 );

DESCRIPTION ^

This module provides a light perl wrapper around the libtermkey library. This library attempts to provide an abstract way to read keypress events in terminal-based programs by providing structures that describe keys, rather than simply returning raw bytes as read from the TTY device.

This version of Term::TermKey requires libtermkey version at least 0.16.

Multi-byte keys, ambiguous keys, and waittime

Some keypresses generate multiple bytes from the terminal. There is also the ambiguity between multi-byte CSI or SS3 sequences, and the Escape key itself. The waittime timer is used to distinguish them.

When some bytes arrive that could be the start of possibly multiple different keypress events, the library will attempt to wait for more bytes to arrive that would finish it. If no more bytes arrive after this time, then the bytes will be reported as events as they stand, even if this results in interpreting a partially-complete Escape sequence as a literal Escape key followed by some normal letters or other symbols.

Similarly, if the start of an incomplete UTF-8 sequence arrives when the library is in UTF-8 mode, this will be reported as the UTF-8 replacement character (U+FFFD) if it is incomplete after this time.

CONSTRUCTOR ^

$tk = Term::TermKey->new( $fh, $flags )

Construct a new Term::TermKey object that wraps the given term handle. $fh should be either an IO handle reference, an integer referring to a plain POSIX file descriptor, of undef. $flags is optional, but if given should contain the flags to pass to libtermkey's constructor. Assumes a default of 0 if not supplied. See the FLAG_* constants.

$tk = Term::TermKey->new_abstract( $termtype, $flags )

Construct a new abstract Term::TermKey object not associated with a filehandle. Input may be fed to it using the push_bytes() method rather than waitkey() or advisereadable(). The name of the termtype should be given in the $termtype string.

METHODS ^

$success = $tk->start

$sucess = $tk->stop

Start or stop IO interactions from the instance. Starting will send the terminal initialisation sequence and set up termios(5) settings, stopping will send the terminal shutdown sequence and restore termios(5) back to the initial values. After construction, a Term::TermKey instance is already started, but these methods may be used to suspend and resume, for example, on receipt of a SIGTSTP signal requesting that the application background itself.

Returns false if it fails; $! will contain an error code.

$started = $tk->is_started

Returns true if the instance has been started, or false if it is stopped.

$flags = $tk->get_flags

$tk->set_flags( $newflags )

Accessor and mutator for the flags. One of the FLAG_UTF8 or FLAG_RAW flags will be set, even if neither was present in the constructor, as in this case the library will attempt to detect if the current locale is UTF-8 aware or not.

$canonflags = $tk->get_canonflags

$tk->set_canonflags( $newcanonflags )

Accessor and mutator for the canonicalisation flags.

$msec = $tk->get_waittime

$tk->set_waittime( $msec )

Accessor and mutator for the maximum wait time in miliseconds. The underlying libtermkey library will have specified a default value when the object was constructed.

$bytes = $tk->get_buffer_remaining

Accessor returning the number of bytes of buffer space remaining in the buffer; the space in which push_bytes can write.

$bytes = $tk->get_buffer_size

$tk->set_buffer_size( $size )

Accessor and mutator to for the total buffer size to store pending bytes. If the underlying termkey_set_buffer_size(3) call fails, the set_buffer_size method will throw an exception.

$res = $tk->getkey( $key )

Attempt to retrieve a single keypress event from the buffer, and put it in $key. If successful, will return RES_KEY to indicate that the $key structure now contains a new keypress event. If $key is an undefined lvalue (such as a new scalar variable) it will be initialised to contain a new key structure.

If nothing is in the buffer it will return RES_NONE. If the buffer contains a partial keypress event which does not yet contain all the bytes required, it will return RES_AGAIN (see above section about multibyte events). If no events are ready and the input stream is now closed, will return RES_EOF.

This method will not block, nor will it perform any IO on the underlying file descriptor. For a normal blocking read, see waitkey().

$res = $tk->getkey_force( $key )

Similar to getkey(), but will not return RES_AGAIN if a partial match was found. Instead, it will force an interpretation of the bytes, even if this means interpreting the start of an <Esc>-prefixed multibyte sequence as a literal Escape key followed by normal letters. If $key is an undefined lvalue (such as a new scalar variable) it will be initialised to contain a new key structure.

This method will not block, nor will it perform any IO on the underlying file descriptor. For a normal blocking read, see waitkey().

$res = $tk->waitkey( $key )

Attempt to retrieve a single keypress event from the buffer, or block until one is available. If successful, will return RES_KEY to indicate that the $key structure now contains a new keypress event. If an IO error occurs it will return RES_ERROR, and if the input stream is now closed it will return RES_EOF.

If $key is an undefined lvalue (such as a new scalar variable) it will be initialised to contain a new key structure.

$res = $tk->advisereadable

Inform the underlying library that new input may be available on the underlying file descriptor and so it should call read() to obtain it. Will return RES_AGAIN if it read at least one more byte, RES_NONE if no more input was found, or RES_ERROR if an IO error occurs.

Normally this method would only be used in programs that want to use Term::TermKey asynchronously; see the EXAMPLES section. This method gracefully handles an EAGAIN error from the underlying read() syscall.

$len = $tk->push_bytes( $bytes )

Feed more bytes into the input buffer. This is primarily useful for feeding input into filehandle-less instances, constructed by passing undef or -1 as the filehandle to the constructor. After calling this method, these bytes will be available to read as keypresses by the getkey method.

$str = $tk->get_keyname( $sym )

Returns the name of a key sym, such as returned by Term::TermKey::Key->sym().

$sym = $tk->keyname2sym( $keyname )

Look up the sym for a named key. The result of this method call can be compared directly against the value returned by Term::TermKey::Key->sym(). Because this method has to perform a linear search of key names, it is best called rarely, perhaps during program initialisation, and the result stored for easier comparisons during runtime.

( $cmd, @args ) = $tk->interpret_unknown_csi( $key )

If $key contains an unknown CSI event then its command and arguments are returned in a list. $cmd will be a string of 1 to 3 characters long, containing the initial and intermediate characters if present, followed by the main command character. @args will contain the numerical arguments, where missing arguments are replaced by -1. If $key does not contain an unknown CSI event then an empty list is returned.

Note that this method needs to be called immediately after getkey or waitkey, or at least, before calling either of those methods again. The actual CSI sequence is retained in the libtermkey buffer, and only retrieved by this method call. Calling getkey or waitkey again may overwrite that buffer.

$str = $tk->format_key( $key, $format )

Return a string representation of the keypress event in $key, following the flags given. See the descriptions of the flags, below, for more detail.

This may be useful for matching keypress events against keybindings stored in a hash. See EXAMPLES section for more detail.

$key = $tk->parse_key( $str, $format )

Return a keypress event by parsing the string representation in $str, following the flags given. This method is an inverse of format_key.

This may be useful for parsing entries from a configuration file or similar.

$key = $tk->parse_key_at_pos( $str, $format )

Return a keypress event by parsing the string representation in a region of $str, following the flags given.

Where parse_key will start at the beginning of the string and requires the entire input to be consumed, this method will start at the current pos() position in $str (or at the beginning of the string if none is yet set), and after a successful parse, will update it to the end of the matched section. This position does not have to be at the end of the string. $str must therefore be a real scalar variable, and not a string literal.

This may be useful for incremental parsing of configuration or other data, out of a larger string.

$cmp = $tk->keycmp( $key1, $key2 )

Compares the two given keypress events, returning a number less than, equal to, or greater than zero, depending on the ordering. Keys are ordered first by type (unicode, keysym, function, mouse), then by value within that type, then finally by modifier bits.

This may be useful in sort expressions:

 my @sorted_keys = sort { $tk->keycmp( $a, $b ) } @keys;

KEY OBJECTS ^

The Term::TermKey::Key subclass is used to store a single keypress event. Objects in this class cannot be changed by perl code. getkey(), getkey_force() or waitkey() will overwrite the contents of the structure with a new value.

Keys cannot be constructed, but getkey(), getkey_force() or waitkey() will place a new key structure in the $key variable if it is undefined when they are called. parse_key() and parse_key_at_pos() will return new keys.

$key->type

The type of event. One of TYPE_UNICODE, TYPE_FUNCTION, TYPE_KEYSYM, TYPE_MOUSE, TYPE_POSITION, TYPE_MODEREPORT, TYPE_UNKNOWN_CSI.

$key->type_is_unicode

$key->type_is_function

$key->type_is_keysym

$key->type_is_mouse

$key->type_is_position

$key->type_is_modereport

$key->type_is_unknown_csi

Shortcuts which return a boolean.

$key->codepoint

The Unicode codepoint number for TYPE_UNICODE, or 0 otherwise.

$key->number

The function key number for TYPE_FUNCTION, or 0 otherwise.

$key->sym

The key symbol number for TYPE_KEYSYM, or 0 otherwise. This can be passed to Term::TermKey->get_keyname(), or compared to a result earlier obtained from Term::TermKey->keyname2sym().

$key->modifiers

The modifier bitmask. Can be compared against the KEYMOD_* constants.

$key->modifier_shift

$key->modifier_alt

$key->modifier_ctrl

Shortcuts which return a boolean if the appropriate modifier is present.

$key->utf8

A string representation of the given Unicode codepoint. If the underlying termkey library is in UTF-8 mode then this will be a UTF-8 string. If it is in raw mode, then this will be a single raw byte.

$key->mouseev

$key->button

The details of a mouse event for TYPE_MOUSE, or undef for other types of event.

$key->line

$key->col

The details of a mouse or position event, or undef for other types of event.

$key->termkey

Return the underlying Term::TermKey object this key was retrieved from.

$str = $key->format( $format )

Returns a string representation of the keypress event, identically to calling format_key on the underlying Term::TermKey object.

EXPORTED CONSTANTS ^

The following constant names are all derived from the underlying libtermkey library. For more detail see the documentation on the library.

These constants are possible values of $key->type

TYPE_UNICODE

a Unicode codepoint

TYPE_FUNCTION

a numbered function key

TYPE_KEYSYM

a symbolic key

TYPE_MOUSE

a mouse movement or button press or release

TYPE_POSITION

a cursor position report

TYPE_MODEREPORT

an ANSI or DEC mode report

TYPE_UNKNOWN_CSI

an unrecognised CSI sequence

These constants are result values from getkey(), getkey_force(), waitkey() or advisereadable()

RES_NONE

No key event is ready.

RES_KEY

A key event has been provided.

RES_EOF

No key events are ready and the terminal has been closed, so no more will arrive.

RES_AGAIN

No key event is ready yet, but a partial one has been found. This is only returned by getkey(). To obtain the partial result even if it never completes, call getkey_force().

RES_ERROR

Returned by waitkey or advisereadable if an IO error occurs while trying to read another key event.

These constants are key modifier masks for $key->modifiers

KEYMOD_SHIFT
KEYMOD_ALT
KEYMOD_CTRL

Should be obvious ;)

These constants are types of mouse event which may be returned by $key->mouseev or interpret_mouse:

MOUSE_UNKNOWN

The type of mouse event was not recognised

MOUSE_PRESS

The event reports a mouse button being pressed

MOUSE_DRAG

The event reports the mouse being moved while a button is held down

MOUSE_RELEASE

The event reports the mouse buttons being released, or the mouse moved without a button held.

These constants are flags for the constructor, Term::TermKey->new

FLAG_NOINTERPRET

Do not attempt to interpret C0 codes into keysyms (ie. Backspace, Tab, Enter, Escape). Instead report them as plain Ctrl-letter events.

FLAG_CONVERTKP

Convert xterm's alternate keypad symbols into the plain ASCII codes they would represent.

FLAG_RAW

Ignore locale settings; do not attempt to recombine UTF-8 sequences. Instead report only raw values.

FLAG_UTF8

Ignore locale settings; force UTF-8 recombining on.

FLAG_NOTERMIOS

Even if the terminal file descriptor represents a TTY device, do not call the tcsetattr() termios function on it to set in canonical input mode.

FLAG_SPACESYMBOL

Sets the CANON_SPACESYMBOL canonicalisation flag. See below.

FLAG_CTRLC

Disable the SIGINT behaviour of the Ctrl-C key, allowing it to be read as a modified Unicode keypress.

FLAG_EINTR

Disable retry on signal interrupt; instead report it as an error with RES_ERROR and $! set to EINTR. Without this flag, IO operations will be retried if interrupted.

These constants are canonicalisation flags for set_canonflags and get_canonflags

CANON_SPACESYMBOL

With this flag set, the Space key will appear as a TYPE_KEYSYM key event whose symname is "Space". Without this flag, it appears as a normal TYPE_UNICODE character.

CANON_DELBS

With this flag set, the ASCII DEL byte is interpreted as the "Backspace" keysym, rather than "DEL". This flag does not affect the interpretation of ASCII BS, which is always represented as "Backspace".

These constants are flags to format_key

FORMAT_LONGMOD

Print full modifier names e.g. Shift- instead of abbreviating to S-.

FORMAT_CARETCTRL

If the only modifier is Ctrl on a plain character, render it as ^X.

FORMAT_ALTISMETA

Use the name Meta or the letter M instead of Alt or A.

FORMAT_WRAPBRACKET

If the key event is a special key instead of unmodified Unicode, wrap it in <brackets>.

FORMAT_MOUSE_POS

If the event is a mouse event, also include the cursor position; rendered as @ ($col,$line)

FORMAT_VIM

Shortcut to FORMAT_ALTISMETA|FORMAT_WRAPBRACKET; which gives an output close to the format the vim editor uses.

EXAMPLES ^

A simple print-until-Ctrl-C loop

This program just prints every keypress until the user presses Ctrl-C.

 use Term::TermKey qw( FLAG_UTF8 RES_EOF FORMAT_VIM );
 
 my $tk = Term::TermKey->new(\*STDIN);
 
 # ensure perl and libtermkey agree on Unicode handling
 binmode( STDOUT, ":encoding(UTF-8)" ) if $tk->get_flags & FLAG_UTF8;
 
 while( ( my $ret = $tk->waitkey( my $key ) ) != RES_EOF ) {
    print "Got key: ".$tk->format_key( $key, FORMAT_VIM )."\n";
 }

Configuration of custom keypresses

Because format_key() yields a plain string representation of a keypress it can be used as a hash key to look up a "handler" routine for the key.

The following implements a simple line input program, though obviously lacking many features in a true line editor like readline.

 use Term::TermKey qw( FLAG_UTF8 RES_EOF FORMAT_LONGMOD );
 
 my $tk = Term::TermKey->new(\*STDIN);
 
 # ensure perl and libtermkey agree on Unicode handling
 binmode( STDOUT, ":encoding(UTF-8)" ) if $tk->get_flags & FLAG_UTF8;

 my $line = "";

 $| = 1;

 my %key_handlers = (
    "Enter"  => sub { 
       print "\nThe line is: $line\n";
       $line = "";
    },

    "Backspace" => sub {
       return unless length $line;
       substr( $line, -1, 1 ) = "";
       print "\cH \cH"; # erase it
    },

    # other handlers ...
 );
 
 while( ( my $ret = $tk->waitkey( my $key ) ) != RES_EOF ) {
    my $handler = $key_handlers{ $tk->format_key( $key, FORMAT_LONGMOD ) };
    if( $handler ) {
       $handler->( $key );
    }
    elsif( $key->type_is_unicode and !$key->modifiers ) {
       my $char = $key->utf8;

       $line .= $char;
       print $char;
    }
 }

Asynchronous operation

Because the getkey() method performs no IO itself, it can be combined with the advisereadable() method in an asynchronous program.

 use IO::Select;
 use Term::TermKey qw(
    FLAG_UTF8 RES_KEY RES_AGAIN RES_EOF FORMAT_VIM
 );
 
 my $select = IO::Select->new();
 
 my $tk = Term::TermKey->new(\*STDIN);
 $select->add(\*STDIN);
 
 # ensure perl and libtermkey agree on Unicode handling
 binmode( STDOUT, ":encoding(UTF-8)" ) if $tk->get_flags & FLAG_UTF8;
 
 sub on_key
 {
    my ( $tk, $key ) = @_;
 
    print "You pressed " . $tk->format_key( $key, FORMAT_VIM ) . "\n";
 }
 
 my $again = 0;
 
 while(1) {
    my $timeout = $again ? $tk->get_waittime/1000 : undef;
    my @ready = $select->can_read($timeout);
 
    if( !@ready ) {
       my $ret;
       while( ( $ret = $tk->getkey_force( my $key ) ) == RES_KEY ) {
          on_key( $tk, $key );
       }
    }
 
    while( my $fh = shift @ready ) {
       if( $fh == \*STDIN ) {
          $tk->advisereadable;
          my $ret;
          while( ( $ret = $tk->getkey( my $key ) ) == RES_KEY ) {
             on_key( $tk, $key );
          }
 
          $again = ( $ret == RES_AGAIN );
          exit if $ret == RES_EOF;
       }
       # Deal with other filehandles here
    }
 }

There may also be more appropriate modules on CPAN for particular event frameworks; see the SEE ALSO section below.

SEE ALSO ^

AUTHOR ^

Paul Evans <leonerd@leonerd.org.uk>

syntax highlighting: