set-char-mask(2)

NAME

set-char-mask - Set character word mask

SYNOPSIS

n set-char-mask "flags" ["value"]

DESCRIPTION

set-char-mask returns or modifies the setting of MicroEmacs internal character tables. The argument n defines the action to be taken, as follows:-

-1


Removes characters from the given set.

0


Returns characters in the given set in $result(5).

1


Adds characters to the given set.

The first argument "flags" determines the required character set as follows:-

M


Character set Map. Internally MicroEmacs uses a standard character set, or code page, for each language so that only one set of spelling rules and dictionaries are required, however the current display character set may be different so a mapping is required to enable MicroEmacs to convert between the two. By convention MicroEmacs uses the most appropriate Windows code page for its internal character set, e.g. Windows CP-1252 is the internal character set used for American, British, French, German, Spanish, Portuguese and Italian while CP-1253 is used for Greek. Windows code pages are used in preference to ISO-8859 character sets because they tend to have more characters available so have slightly better language coverage.

The "value" for the M flag must be a string containing pairs of characters, an internal character set character followed by its display character set equivalent. For example, if the current internal character set is CP-1252 and the display character set is CP-437 (a common DOS code page), then 'e' acute must be mapped from 0xe9 in CP-1252 to 0x82 in CP-437 so "value" should contain "\xe9\x82" as well as many other mappings. All characters in the display character set should be mapped if possible rather than just the letters most commonly used by the current language as this creates the best support for any text entered.

Some display character sets may not have all the characters available in the internal character set, for instance DOS code page CP-437 does not have an upper-case 'E' grave. In this case an ordinary 'E' should be used as a sensible replacement, i.e. "`EE" (where `E is an 'E' grave) as this is the best that can be done given the limitations of the current display character set.

This flag cannot be incrementally altered, any calls to alter this set leads to the resetting of all the character tables so the character mapping must be performed first and in a single call. No other set may be altered in the same call.

c


Unicode character mapping. This flag is used to define a mapping between the upper half of the current display character set (the lower half is assumed to be ASCII) and the Unicode character set. This flag cannot be incrementally altered, any call to set this mapping first resets the mapping table to 0's so the mapping must be performed in a single call. No other set may be altered in the same call. The value must be a continuous string of up to 128 4-digit hexadecimal numbers, each value being either 0 for no mapping or the U+#### Unicode character value. The values must be given in ascending order with the first being for character 0x80 etc.

The mapping is used by MicroEmacs on Windows and UNIX XTerm systems to better map the system clipboard text to the current display character set. For example XTerm typically uses an A-F code page which does not support a euro currency symbol, however CP-1252 based fonts can be installed and used correctly by MicroEmacs allowing support for the euro and many other characters. This mapping then allows MicroEmacs to correctly handle these characters when copied between different applications. The mapping table is also used by the &uni(4) functions, expand-iso-accents(3) and change-buffer-charset(3) commands.

d


Is Displayable. Characters in this set can be directly displayed to the screen (as a single character) when occurring in a buffer. When a character not in this set is to be displayed it is performed using more than one character. Characters in the range 1-31 are displayed as "^?" where ? is the ASCII character plus 64, (e.g. 0x01 -> 65, i.e. "^A") otherwise the character is displayed in the form "\xhh" where hh is the hex form of the ASCII value. One notable exception is the tab character (0x09), by default this character is not displayable, instead it is displayed as a sequence of one or more spaces up to the next tab stop.

p


Is Pokable. Similar to d, characters in this set can be poked to the screen when using screen-poke(2). When found in a binary file the character is displayed in the right hand column. Unlike d, any character outside this set will be displayed as a single period '.', indicating that it cannot be displayed.

P


Is Printable. Similar to d, characters in this set may be printed as a single character when using print-buffer(2) or print-region(2). Any character not in this set is printed in a similar fashion to d.

I


Internal character set map list. This set cannot be altered using this flag, character mappings must be set up using flag M. The order of the characters in the returned $result string is the same as the order for flag D.

D


Display character set map list. This set cannot be altered using this flag, character mappings must be set up using flag M. The order of the characters in $result when returned is the same as the order for flag I.

a


Is Alphabetic letter. Characters in this set are alphabetical characters, used by many MicroEmacs commands such as forward-word(2). When setting, the "value" must specify pairs of internal character set chars, an upper-case character followed by its lower-case equivalent. This enables commands such as lower-case-word(2) to operate correctly regardless of the font and language being used. Some character sets may not have all the characters available for rendering, for instance PC DOS Code page 437 does not have an upper-case 'e' grave. In this case an ordinary 'E' should be used as a sensible replacement, i.e. "E`e" (where `e is an 'e' grave). However, this will lead to all upper-case 'E's to map to a lower-case 'e' grave in a case changing operation, this may be corrected by adding a further mapping of 'E' to 'e' to over-ride the 'e' grave mapping, i.e. "E`eEe". This technique does fail when changing the case more than once, when all lower case 'e' graves will be lost.

Note that the returned character list will pair all lower-case characters with their upper-case equivalent letters first.

l


Is Lower case letter. This set cannot be altered using this flag, alterations to the alphabetic set must be performed using flag a. Characters in this set are all the lower-case letters, typically the characters 'a' to 'z'. The order may not be the same as returned by flag u.

u


Is Upper case letter. This set cannot be altered using this flag, alterations to the alphabetic set must be performed using flag a. Characters in this set are all the upper-case letters, typically the characters 'A' to 'Z'. The order may not be the same as returned by

h


Is Hex-decimal Digit. The set is rarely used as it is invariably the digits '0' to '9' and the letters 'a' to 'f' in upper and lower case. It is often used in the setting of $buffer-mask(5).

A


Is Alpha-numeric. This set cannot be altered using this flag, alterations to the alphabetic set must be performed using flag a. Characters in this set are either alphabetic characters or the digits 0-9.

s


Is Spell extended word character. The characters in this set are recognized by the spell checker as characters which may be considered part of a word, for example the period '.'s in e.g. or the hyphen '-' in hyphenated-words. Typically this set contains the characters ''', '-' and '.'.

1, 2, 3 & 4


Is in Word. These user definable sets are used to add characters to a buffer's word character set, affecting the operation of commands like forward-word(2). Many different file types operate better with a different word character set, e.g. it is preferable to include the '_' character when editing C files. See variable $buffer-mask(5).

k


Keyboard character map. This flag is used to configure a mapping from a foreign non-latin keyboard (such as Russian) to Latin characters, setting the mapping correctly can greatly improve usability in MicroEmacs as changing keyboards is required less frequently. The mapping is used by character input routines which first check the main key typed (e.g. the Cyrillic character) and if not bound or invalid MicroEmacs uses this keyboard mapping to translate the character into a Latin character. This allows the user to execute binding such as "C-x k" or responding to "[y/n]" prompts without having to change keyboards.

As with flag M, this cannot be incrementally altered, any call to set this mapping first resets the mapping table so the mapping must be performed in a single call. No other set may be altered in the same call. When setting, the "value" must supply pairs of characters, the keyboard character followed by the character to map it to, typically an ASCII character.

Unless stated otherwise, multiple flags may be specified at the same time returning a combined character set or setting multiple properties for the given "value" characters.

EXAMPLE

For many UNIX XTerm fonts the best characters to use for $box-chars(5) (used in drawing osd(2) dialogs) lie in the range 0x0B to 0x19. For example the vertical bar is '\x19', the top left hand corner is '\x0D' etc. These characters are by default set to be not displayable or pokable which renders them useless. They can be made displayable and pokable as follows:-

set-char-mask "dp" "\x19\x0D\x0C\x0E\x0B\x18\x15\x0F\x16\x17\x12" 

MicroEmacs variables have either '$', '#', '%', ':' or a '.' character prepended to their name, they may also contain a '-' character in the body of their name. It is preferable for these characters to be part of the variable 'word' so commands like forward-kill-word(2) can work correctly. This may be achieved by adding these characters to user set 2 and setting the buffer-mask variable to include set 2, as follows:

set-char-mask "2" "$#%:.-" 

define-macro fhook-emf 
    set-variable $buffer-mask "luh2" 
    . 
    . 
!emacro 

For the examples below only the following subset of characters will be used:-

Character               Win CP-1252   Cmd CP-850     DOS CP-437 

Capital A (A)           A             A              A 
Capital A grave (`A)    \xC0          \xB7           No equivalent 
Capital A acute ('A)    \xC1          \x90           No equivalent 
Small a (a)             a             a              a 
Small A grave (`a)      \xE0          \x85           \x85 
Small A acute ('a)      \xE1          \xA0           \xA0 

As the spell checker for French will operates in Windows CP-1252, the character font mapping (flag M) must be correctly setup for spell checking to operate correctly. When CP-1252 is also used as the display character set the mapping is the empty string as the internal and display character set are fully in-sync, but for both Windows Console CP-850 and DOS code page CP-437 the mappings should be set as follows:-

; CP-850 mapping setup 
set-char-mask "M" "\xC0\xB7\xC1\x90\xE0\x85\xE1\xA0" 
; CP-437 mapping setup 
set-char-mask "M" "\xC0A\xC1AAA\xE0\x85\xE1\xA0" 

As all the characters in CP-1252 have equivalents in CP-850, the mapping for Windows console is a simple 1-to-1 lossless character list. However the missing capital A's in CP-437 causes problems, for the command change-buffer-charset(3) it is preferable for a mapping of `A to be given, otherwise the document being converted may become corrupted and unreadable. Therefore a mapping of `A to A is given to alleviate this problem, similarly 'A is also mapped to A leading to loss of information.

This leads to a further problem with the conversion of CP-437 back to CP-1252, if the mapping the 'A's was left as just "\xC0A\xC1A" the last mapping ('A to A) would also be the back conversion for A, i.e. ALL A's would be converted back to 'A's. To solve this problem, a further seemingly pointless mapping of A to A is given to correct the back conversion.

While ISO-8859-1 (Latin 1) supports a very similar set of characters to CP-1252, it lacks some accented 'S', 'Y' & 'Z' characters must be mapped to their plain letter equivalents.

For languages which use accented characters, the alphabetic character set must be extended to include these characters for letter based commands like forward-word(2) and upper-case-word(2) to operate correctly. However, the letter set should be fully extended for each code page regardless of the language being used as an 'a' acute should always be considered a letter even though it is unlikely to occur. The addition of extra letters must achieve two goals, firstly to define whether a character is a letter, enabling commands like forward-word to work correctly. The second is to provide an upper case to lower case character mapping, enabling commands like upper-case-word to work correctly. This is achieved with a single call to set-char-mask using the a flag as follows:-

set-char-mask "a" "\xC0\xE0\xC1\xE1" 

Note that this flag always expects an internal character set based string, this allows the same map character list to be used regardless of the display character set being used, i.e. the above line can be used for CP-1252, CP-850, CP-437 & ISO-8859-1 code pages. But it does mean that the internal to display character set mapping (flag M) must already have been provided.

Similar mapping problems are encountered with the a flag as with flag M above. The problem is not immediately obvious because the mapping is always given in internal character set which will support the widest set of characters, but when CP-437 is used the mapping string of "A\x85A\xA0" must be used. As can be seen, A is mapped last to 'a so an upper to lower character operation will convert a A to 'a. A similar solution is used, a further mapping of A to a is given to correct the default case mapping for both A and a, i.e. the following line should always be used instead:-

set-char-mask "a" "\xC0\xE0\xC1\xE1Aa" 

SEE ALSO

forward-word(2), upper-case-word(2), change-buffer-charset(3), &uni(4), $buffer-mask(5), screen-poke(2), spell(2), $buffer-tab-width(5).

(c) Copyright JASSPA 2025
Last Modified: 2024/05/12
Generated On: 2025/09/29