Vim documentation: usr_44

*usr_44.txt*	For Vim version 7.3.  Last change: 2008 Dec 28

		     VIM USER MANUAL - by Bram Moolenaar

			 Your own syntax highlighted


Vim comes with highlighting for a couple of hundred different file types.  If
the file you are editing isn't included, read this chapter to find out how to
get this type of file highlighted.  Also see |:syn-define| in the reference
manual.

|44.1|	Basic syntax commands
|44.2|	Keywords
|44.3|	Matches
|44.4|	Regions
|44.5|	Nested items
|44.6|	Following groups
|44.7|	Other arguments
|44.8|	Clusters
|44.9|	Including another syntax file
|44.10|	Synchronizing
|44.11|	Installing a syntax file
|44.12|	Portable syntax file layout

     Next chapter: |usr_45.txt|  Select your language
 Previous chapter: |usr_43.txt|  Using filetypes
Table of contents: |usr_toc.txt|

==============================================================================

*44.1*	Basic syntax commands

Using an existing syntax file to start with will save you a lot of time.  Try
finding a syntax file in $VIMRUNTIME/syntax for a language that is similar.
These files will also show you the normal layout of a syntax file.  To
understand it, you need to read the following.

Let's start with the basic arguments.  Before we start defining any new
syntax, we need to clear out any old definitions:

	:syntax clear

This isn't required in the final syntax file, but very useful when
experimenting.

There are more simplifications in this chapter.  If you are writing a syntax
file to be used by others, read all the way through the end to find out the
details.


LISTING DEFINED ITEMS

To check which syntax items are currently defined, use this command:

	:syntax

You can use this to check which items have actually been defined.  Quite
useful when you are experimenting with a new syntax file.  It also shows the
colors used for each item, which helps to find out what is what.
   To list the items in a specific syntax group use:

	:syntax list {group-name}

This also can be used to list clusters (explained in |44.8|).  Just include
the @ in the name.


MATCHING CASE

Some languages are not case sensitive, such as Pascal.  Others, such as C, are
case sensitive.  You need to tell which type you have with the following
commands:
	:syntax case match
	:syntax case ignore

The "match" argument means that Vim will match the case of syntax elements.
Therefore, "int" differs from "Int" and "INT".  If the "ignore" argument is
used, the following are equivalent: "Procedure", "PROCEDURE" and "procedure".
   The ":syntax case" commands can appear anywhere in a syntax file and affect
the syntax definitions that follow.  In most cases, you have only one ":syntax
case" command in your syntax file; if you work with an unusual language that
contains both case-sensitive and non-case-sensitive elements, however, you can
scatter the ":syntax case" command throughout the file.

==============================================================================

*44.2*	Keywords

The most basic syntax elements are keywords.  To define a keyword, use the
following form:

	:syntax keyword {group} {keyword} ...

The {group} is the name of a syntax group.  With the ":highlight" command you
can assign colors to a {group}.  The {keyword} argument is an actual keyword.
Here are a few examples:

	:syntax keyword xType int long char
	:syntax keyword xStatement if then else endif

This example uses the group names "xType" and "xStatement".  By convention,
each group name is prefixed by the filetype for the language being defined.
This example defines syntax for the x language (eXample language without an
interesting name).  In a syntax file for "csh" scripts the name "cshType"
would be used.  Thus the prefix is equal to the value of 'filetype'.
   These commands cause the words "int", "long" and "char" to be highlighted
one way and the words "if", "then", "else" and "endif" to be highlighted
another way.  Now you need to connect the x group names to standard Vim
names.  You do this with the following commands:

	:highlight link xType Type
	:highlight link xStatement Statement

This tells Vim to highlight "xType" like "Type" and "xStatement" like
"Statement".  See |group-name| for the standard names.


UNUSUAL KEYWORDS

The characters used in a keyword must be in the 'iskeyword' option.  If you
use another character, the word will never match.  Vim doesn't give a warning
message for this.
   The x language uses the '-' character in keywords.  This is how it's done:

	:setlocal iskeyword+=-
	:syntax keyword xStatement when-not

The ":setlocal" command is used to change 'iskeyword' only for the current
buffer.  Still it does change the behavior of commands like "w" and "*".  If
that is not wanted, don't define a keyword but use a match (explained in the
next section).

The x language allows for abbreviations.  For example, "next" can be
abbreviated to "n", "ne" or "nex".  You can define them by using this command:

	:syntax keyword xStatement n[ext]

This doesn't match "nextone", keywords always match whole words only.

==============================================================================

*44.3*	Matches

Consider defining something a bit more complex.  You want to match ordinary
identifiers.  To do this, you define a match syntax item.  This one matches
any word consisting of only lowercase letters:

	:syntax match xIdentifier /\<\l\+\>/
 
	Note:
	Keywords overrule any other syntax item.  Thus the keywords "if",
	"then", etc., will be keywords, as defined with the ":syntax keyword"
	commands above, even though they also match the pattern for
	xIdentifier.

The part at the end is a pattern, like it's used for searching.  The // is
used to surround the pattern (like how it's done in a ":substitute" command).
You can use any other character, like a plus or a quote.

Now define a match for a comment.  In the x language it is anything from # to
the end of a line:

	:syntax match xComment /#.*/

Since you can use any search pattern, you can highlight very complex things
with a match item.  See |pattern| for help on search patterns.

==============================================================================

*44.4*	Regions

In the example x language, strings are enclosed in double quotation marks (").
To highlight strings you define a region.  You need a region start (double
quote) and a region end (double quote).  The definition is as follows:

	:syntax region xString start=/"/ end=/"/

The "start" and "end" directives define the patterns used to find the start
and end of the region.  But what about strings that look like this?

	"A string with a double quote (\") in it" 

This creates a problem: The double quotation marks in the middle of the string
will end the region.  You need to tell Vim to skip over any escaped double
quotes in the string.  Do this with the skip keyword:

	:syntax region xString start=/"/ skip=/\\"/ end=/"/

The double backslash matches a single backslash, since the backslash is a
special character in search patterns.

When to use a region instead of a match?  The main difference is that a match
item is a single pattern, which must match as a whole.  A region starts as
soon as the "start" pattern matches.  Whether the "end" pattern is found or
not doesn't matter.  Thus when the item depends on the "end" pattern to match,
you cannot use a region.  Otherwise, regions are often simpler to define.  And
it is easier to use nested items, as is explained in the next section.

==============================================================================

*44.5*	Nested items

Take a look at this comment:

	%Get input  TODO: Skip white space 

You want to highlight TODO in big yellow letters, even though it is in a
comment that is highlighted blue.  To let Vim know about this, you define the
following syntax groups:

	:syntax keyword xTodo TODO contained
	:syntax match xComment /%.*/ contains=xTodo

In the first line, the "contained" argument tells Vim that this keyword can
exist only inside another syntax item.  The next line has "contains=xTodo".
This indicates that the xTodo syntax element is inside it.  The result is that
the comment line as a whole is matched with "xComment" and made blue.  The
word TODO inside it is matched by xTodo and highlighted yellow (highlighting
for xTodo was setup for this).


RECURSIVE NESTING

The x language defines code blocks in curly braces.  And a code block may
contain other code blocks.  This can be defined this way:

	:syntax region xBlock start=/{/ end=/}/ contains=xBlock

Suppose you have this text:

	while i < b { 
		if a { 
			b = c; 
		} 
	} 

First a xBlock starts at the { in the first line.  In the second line another
{ is found.  Since we are inside a xBlock item, and it contains itself, a
nested xBlock item will start here.  Thus the "b = c" line is inside the
second level xBlock region.  Then a } is found in the next line, which matches
with the end pattern of the region.  This ends the nested xBlock.  Because the
} is included in the nested region, it is hidden from the first xBlock region.
Then at the last } the first xBlock region ends.


KEEPING THE END

Consider the following two syntax items:

	:syntax region xComment start=/%/ end=/$/ contained
	:syntax region xPreProc start=/#/ end=/$/ contains=xComment

You define a comment as anything from % to the end of the line.  A
preprocessor directive is anything from # to the end of the line.  Because you
can have a comment on a preprocessor line, the preprocessor definition
includes a "contains=xComment" argument.  Now look what happens with this
text:

	#define X = Y  % Comment text 
	int foo = 1; 

What you see is that the second line is also highlighted as xPreProc.  The
preprocessor directive should end at the end of the line.  That is why
you have used "end=/$/".  So what is going wrong?
   The problem is the contained comment.  The comment starts with % and ends
at the end of the line.  After the comment ends, the preprocessor syntax
continues.  This is after the end of the line has been seen, so the next
line is included as well.
   To avoid this problem and to avoid a contained syntax item eating a needed
end of line, use the "keepend" argument.  This takes care of
the double end-of-line matching:

	:syntax region xComment start=/%/ end=/$/ contained
	:syntax region xPreProc start=/#/ end=/$/ contains=xComment keepend


CONTAINING MANY ITEMS

You can use the contains argument to specify that everything can be contained.
For example:

	:syntax region xList start=/\[/ end=/\]/ contains=ALL

All syntax items will be contained in this one.  It also contains itself, but
not at the same position (that would cause an endless loop).
   You can specify that some groups are not contained.  Thus contain all
groups but the ones that are listed:

	:syntax region xList start=/\[/ end=/\]/ contains=ALLBUT,xString

With the "TOP" item you can include all items that don't have a "contained"
argument.  "CONTAINED" is used to only include items with a "contained"
argument.  See |:syn-contains| for the details.

==============================================================================

*44.6*	Following groups

The x language has statements in this form:

	if (condition) then 

You want to highlight the three items differently.  But "(condition)" and
"then" might also appear in other places, where they get different
highlighting.  This is how you can do this:

	:syntax match xIf /if/ nextgroup=xIfCondition skipwhite
	:syntax match xIfCondition /([^)]*)/ contained nextgroup=xThen skipwhite
	:syntax match xThen /then/ contained

The "nextgroup" argument specifies which item can come next.  This is not
required.  If none of the items that are specified are found, nothing happens.
For example, in this text:

	if not (condition) then 

The "if" is matched by xIf.  "not" doesn't match the specified nextgroup
xIfCondition, thus only the "if" is highlighted.

The "skipwhite" argument tells Vim that white space (spaces and tabs) may
appear in between the items.  Similar arguments are "skipnl", which allows a
line break in between the items, and "skipempty", which allows empty lines.
Notice that "skipnl" doesn't skip an empty line, something must match after
the line break.

==============================================================================

*44.7*	Other arguments

MATCHGROUP

When you define a region, the entire region is highlighted according to the
group name specified.  To highlight the text enclosed in parentheses () with
the group xInside, for example, use the following command:

	:syntax region xInside start=/(/ end=/)/

Suppose, that you want to highlight the parentheses differently.  You can do
this with a lot of convoluted region statements, or you can use the
"matchgroup" argument.  This tells Vim to highlight the start and end of a
region with a different highlight group (in this case, the xParen group):

	:syntax region xInside matchgroup=xParen start=/(/ end=/)/

The "matchgroup" argument applies to the start or end match that comes after
it.  In the previous example both start and end are highlighted with xParen.
To highlight the end with xParenEnd:

	:syntax region xInside matchgroup=xParen start=/(/
		\ matchgroup=xParenEnd end=/)/

A side effect of using "matchgroup" is that contained items will not match in
the start or end of the region.  The example for "transparent" uses this.


TRANSPARENT

In a C language file you would like to highlight the () text after a "while"
differently from the () text after a "for".  In both of these there can be
nested () items, which should be highlighted in the same way.  You must make
sure the () highlighting stops at the matching ).  This is one way to do this:

	:syntax region cWhile matchgroup=cWhile start=/while\s*(/ end=/)/
		\ contains=cCondNest
	:syntax region cFor matchgroup=cFor start=/for\s*(/ end=/)/
		\ contains=cCondNest
	:syntax region cCondNest start=/(/ end=/)/ contained transparent

Now you can give cWhile and cFor different highlighting.  The cCondNest item
can appear in either of them, but take over the highlighting of the item it is
contained in.  The "transparent" argument causes this.
   Notice that the "matchgroup" argument has the same group as the item
itself.  Why define it then?  Well, the side effect of using a matchgroup is
that contained items are not found in the match with the start item then.
This avoids that the cCondNest group matches the ( just after the "while" or
"for".  If this would happen, it would span the whole text until the matching
) and the region would continue after it.  Now cCondNest only matches after
the match with the start pattern, thus after the first (.


OFFSETS

Suppose you want to define a region for the text between ( and ) after an
"if".  But you don't want to include the "if" or the ( and ).  You can do this
by specifying offsets for the patterns.  Example:

	:syntax region xCond start=/if\s*(/ms=e+1 end=/)/me=s-1

The offset for the start pattern is "ms=e+1".  "ms" stands for Match Start.
This defines an offset for the start of the match.  Normally the match starts
where the pattern matches.  "e+1" means that the match now starts at the end
of the pattern match, and then one character further.
   The offset for the end pattern is "me=s-1".  "me" stands for Match End.
"s-1" means the start of the pattern match and then one character back.  The
result is that in this text:

	if (foo == bar) 

Only the text "foo == bar" will be highlighted as xCond.

More about offsets here: |:syn-pattern-offset|.


ONELINE

The "oneline" argument indicates that the region does not cross a line
boundary.  For example:

	:syntax region xIfThen start=/if/ end=/then/ oneline

This defines a region that starts at "if" and ends at "then".  But if there is
no "then" after the "if", the region doesn't match.

	Note:
	When using "oneline" the region doesn't start if the end pattern
	doesn't match in the same line.  Without "oneline" Vim does _not_
	check if there is a match for the end pattern.  The region starts even
	when the end pattern doesn't match in the rest of the file.


CONTINUATION LINES AND AVOIDING THEM

Things now become a little more complex.  Let's define a preprocessor line.
This starts with a # in the first column and continues until the end of the
line.  A line that ends with \ makes the next line a continuation line.  The
way you handle this is to allow the syntax item to contain a continuation
pattern:

	:syntax region xPreProc start=/^#/ end=/$/ contains=xLineContinue
	:syntax match xLineContinue "\\$" contained

In this case, although xPreProc normally matches a single line, the group
contained in it (namely xLineContinue) lets it go on for more than one line.
For example, it would match both of these lines:

	#define SPAM  spam spam spam \ 
			bacon and spam 

In this case, this is what you want.  If it is not what you want, you can call
for the region to be on a single line by adding "excludenl" to the contained
pattern.  For example, you want to highlight "end" in xPreProc, but only at
the end of the line.  To avoid making the xPreProc continue on the next line,
like xLineContinue does, use "excludenl" like this:

	:syntax region xPreProc start=/^#/ end=/$/
		\ contains=xLineContinue,xPreProcEnd
	:syntax match xPreProcEnd excludenl /end$/ contained
	:syntax match xLineContinue "\\$" contained

"excludenl" must be placed before the pattern.  Since "xLineContinue" doesn't
have "excludenl", a match with it will extend xPreProc to the next line as
before.

==============================================================================

*44.8*	Clusters

One of the things you will notice as you start to write a syntax file is that
you wind up generating a lot of syntax groups.  Vim enables you to define a
collection of syntax groups called a cluster.
   Suppose you have a language that contains for loops, if statements, while
loops, and functions.  Each of them contains the same syntax elements: numbers
and identifiers.  You define them like this:

	:syntax match xFor /^for.*/ contains=xNumber,xIdent
	:syntax match xIf /^if.*/ contains=xNumber,xIdent
	:syntax match xWhile /^while.*/ contains=xNumber,xIdent

You have to repeat the same "contains=" every time.  If you want to add
another contained item, you have to add it three times.  Syntax clusters
simplify these definitions by enabling you to have one cluster stand for
several syntax groups.
   To define a cluster for the two items that the three groups contain, use
the following command:

	:syntax cluster xState contains=xNumber,xIdent

Clusters are used inside other syntax items just like any syntax group.
Their names start with @.  Thus, you can define the three groups like this:

	:syntax match xFor /^for.*/ contains=@xState
	:syntax match xIf /^if.*/ contains=@xState
	:syntax match xWhile /^while.*/ contains=@xState

You can add new group names to this cluster with the "add" argument:

	:syntax cluster xState add=xString

You can remove syntax groups from this list as well:

	:syntax cluster xState remove=xNumber

==============================================================================

*44.9*	Including another syntax file

The C++ language