Info file regex, produced by Makeinfo, -*- Text -*- from input file

File: regex,  Node: top,  Next: syntax,  Up: (dir)

"regex" regular expression matching library.


Regular expression matching allows you to test whether a string fits
into a specific syntactic shape.  You can also search a string for a
substring that fits a pattern.

A regular expression describes a set of strings.  The simplest case
is one that describes a particular string; for example, the string
`foo' when regarded as a regular expression matches `foo' and nothing
else.  Nontrivial regular expressions use certain special constructs
so that they can match more than one string.  For example, the
regular expression `foo\|bar' matches either the string `foo' or the
string `bar'; the regular expression `c[ad]*r' matches any of the
strings `cr', `car', `cdr', `caar', `cadddar' and all other such
strings with any number of `a''s and `d''s.

The first step in matching a regular expression is to compile it. 
You must supply the pattern string and also a pattern buffer to hold
the compiled result.  That result contains the pattern in an internal
format that is easier to use in matching.

Having compiled a pattern, you can match it against strings.  You can
match the compiled pattern any number of times against different

* Menu:

* syntax::	Syntax of regular expressions
* directives::	Meaning of characters as regex string directives.
* emacs::	Additional character directives available
		  only for use within Emacs.
* programming:: Using the regex library from C programs
* unix::	Unix-compatible entry-points to regex library

File: regex,  Node: syntax,  Next: directives,  Prev: top,  Up: top

Syntax of Regular Expressions

Regular expressions have a syntax in which a few characters are
special constructs and the rest are "ordinary".  An ordinary
character is a simple regular expression which matches that character
and nothing else.  The special characters are `$', `^', `.', `*',
`+', `?', `[', `]' and `\'.  Any other character appearing in a
regular expression is ordinary, unless a `\' precedes it.

For example, `f' is not a special character, so it is ordinary, and
therefore `f' is a regular expression that matches the string `f' and
no other string.  (It does *not* match the string `ff'.)  Likewise,
`o' is a regular expression that matches only `o'.

Any two regular expressions A and B can be concatenated.  The result
is a regular expression which matches a string if A matches some
amount of the beginning of that string and B matches the rest of the

As a simple example, we can concatenate the regular expressions `f'
and `o' to get the regular expression `fo', which matches only the
string `fo'.  Still trivial.

Note: for Unix compatibility, special characters are treated as
ordinary ones if they are in contexts where their special meanings
make no sense.  For example, `*foo' treats `*' as ordinary since
there is no preceding expression on which the `*' can act.  It is
poor practice to depend on this behavior; better to quote the special
character anyway, regardless of where is appears.

File: regex,  Node: directives,  Next: emacs,  Prev: syntax,  Up: top

The following are the characters and character sequences which have
special meaning within regular expressions.  Any character not
mentioned here is not special; it stands for exactly itself for the
purposes of searching and matching.  *Note syntax::.

     is a special character that matches anything except a newline. 
     Using concatenation, we can make regular expressions like `a.b'
     which matches any three-character string which begins with `a'
     and ends with `b'.

     is not a construct by itself; it is a suffix, which means the
     preceding regular expression is to be repeated as many times as
     possible.  In `fo*', the `*' applies to the `o', so `fo*'
     matches `f' followed by any number of `o''s.

     The case of zero `o''s is allowed: `fo*' does match `f'.

     `*' always applies to the *smallest* possible preceding
     expression.  Thus, `fo*' has a repeating `o', not a repeating

     The matcher processes a `*' construct by matching, immediately,
     as many repetitions as can be found.  Then it continues with the
     rest of the pattern.  If that fails, backtracking occurs,
     discarding some of the matches of the `*''d construct in case
     that makes it possible to match the rest of the pattern.  For
     example, matching `c[ad]*ar' against the string `caddaar', the
     `[ad]*' first matches `addaa', but this does not allow the next
     `a' in the pattern to match.  So the last of the matches of
     `[ad]' is undone and the following `a' is tried again.  Now it

     `+' is like `*' except that at least one match for the preceding
     pattern is required for `+'.  Thus, `c[ad]+r' does not match
     `cr' but does match anything else that `c[ad]*r' would match.

     `?' is like `*' except that it allows either zero or one match
     for the preceding pattern.  Thus, `c[ad]?r' matches `cr' or
     `car' or `cdr', and nothing else.

`[ ... ]'
     `[' begins a "character set", which is terminated by a `]'.  In
     the simplest case, the characters between the two form the set. 
     Thus, `[ad]' matches either `a' or `d', and `[ad]*' matches any
     string of `a''s and `d''s (including the empty string), from
     which it follows that `c[ad]*r' matches `car', etc.

     Character ranges can also be included in a character set, by
     writing two characters with a `-' between them.  Thus, `[a-z]'
     matches any lower-case letter.  Ranges may be intermixed freely
     with individual characters, as in `[a-z$%.]', which matches any
     lower case letter or `$', `%' or period.

     Note that the usual special characters are not special any more
     inside a character set.  A completely different set of special
     characters exists inside character sets: `]', `-' and `^'.

     To include a `]' in a character set, you must make it the first
     character.  For example, `[]a]' matches `]' or `a'.  To include
     a `-', you must use it in a context where it cannot possibly
     indicate a range: that is, as the first character, or
     immediately after a range.

`[^ ... ]'
     `[^' begins a "complement character set", which matches any
     character except the ones specified.  Thus, `[^a-z0-9A-Z]'
     matches all characters *except* letters and digits.

     `^' is not special in a character set unless it is the first
     character.  The character following the `^' is treated as if it
     were first (it may be a `-' or a `]').

     is a special character that matches the empty string -- but only
     if at the beginning of a line in the text being matched. 
     Otherwise it fails to match anything.  Thus, `^foo' matches a
     `foo' which occurs at the beginning of a line.

     is similar to `^' but matches only at the end of a line.  Thus,
     `xx*$' matches a string of one or more `x''s at the end of a line.

     has two functions: it quotes the above special characters
     (including `\'), and it introduces additional special constructs.

     Because `\' quotes special characters, `\$' is a regular
     expression which matches only `$', and `\[' is a regular
     expression which matches only `[', and so on.

     For the most part, `\' followed by any character matches only
     that character.  However, there are several exceptions:
     characters which, when preceded by `\', are special constructs. 
     Such characters are always ordinary when encountered on their own.

     No new special characters will ever be defined.  All extensions
     to the regular expression syntax are made by defining new
     two-character constructs that begin with `\'.

     specifies an alternative.  Two regular expressions A and B with
     `\|' in between form an expression that matches anything that
     either A or B will match.

     Thus, `foo\|bar' matches either `foo' or `bar' but no other

     `\|' applies to the largest possible surrounding expressions. 
     Only a surrounding `\( ... \)' grouping can limit the grouping
     power of `\|'.

     Full backtracking capability exists when multiple `\|''s are used.

`\( ... \)'
     is a grouping construct that serves three purposes:

       1. To enclose a set of `\|' alternatives for other operations.
          Thus, `\(foo\|bar\)x' matches either `foox' or `barx'.

       2. To enclose a complicated expression for the postfix `*' to
          operate on.  Thus, `ba\(na\)*' matches `bananana', etc.,
          with any (zero or more) number of `na''s.

       3. To mark a matched substring for future reference.

     This last application is not a consequence of the idea of a
     parenthetical grouping; it is a separate feature which happens
     to be assigned as a second meaning to the same `\( ... \)'
     construct because there is no conflict in practice between the
     two meanings.  Here is an explanation of this feature:

     After the end of a `\( ... \)' construct, the matcher remembers
     the beginning and end of the text matched by that construct. 
     Then, later on in the regular expression, you can use `\'
     followed by DIGIT to mean "match the same text matched the
     DIGIT'th time by the `\( ... \)' construct."  The `\( ... \)'
     constructs are numbered in order of commencement in the regexp.

     The strings matching the first nine `\( ... \)' constructs
     appearing in a regular expression are assigned numbers 1 through
     9 in order of their beginnings.  `\1' through `\9' may be used
     to refer to the text matched by the corresponding `\( ... \)'

     For example, `\(.*\)\1' matches any string that is composed of
     two identical halves.  The `\(.*\)' matches the first half,
     which may be anything, but the `\1' that follows must match the
     same exact text.

     matches the empty string, but only if it is at the beginning or
     end of a word.  Thus, `\bfoo\b' matches any occurrence of `foo'
     as a separate word.  `\bball\(s\|\)\b' matches `ball' or `balls'
     as a separate word.

     matches the empty string, provided it is *not* at the beginning
     or end of a word.

     matches the empty string, but only if it is at the beginning of
     a word.

     matches the empty string, but only if it is at the end of a word.

     matches any word-constituent character.

     matches any character that is not a word-constituent.

There are a number of additional `\' regexp directives available for
use within Emacs only.

(*note emacs::.).

File: regex,  Node: emacs,  Next: programming,  Prev: directives,  Up: top

Constructs Available in Emacs Only

     matches the empty string, but only if it is at the beginning of
     the buffer.

     matches the empty string, but only if it is at the end of the

     matches any character whose syntax is CODE.  CODE is a letter
     which represents a syntax code: thus, `w' for word constituent,
     `-' for whitespace, `(' for open-parenthesis, etc.  See the
     documentation for the Emacs function `modify-syntax-entry' for
     further details.

     Thus, `\s(' matches any character with open-parenthesis syntax.

     matches any character whose syntax is not CODE.

File: regex,  Node: programming,  Next: compiling,  Prev: emacs,  Up: top

Programming using the `regex' library

The subnodes accessible from this menu give information on entry
points and data structures which C programs need to interface to the
`regex' library.

* Menu:

* compiling::	How to compile regular expressions
* matching::	Matching compiled regular expressions
* searching::	Searching for compiled regular expressions
* translation::	Translating characters into other characters
		  (for both compilation and matching)
* registers::	determining what was matched
* split::	matching data which is split into two pieces
* unix::	Unix-compatible entry-points to regex library

File: regex,  Node: compiling,  Next: matching,  Prev: programming,  Up: programming

Compiling a Regular Expression

To compile a regular expression, you must supply a pattern buffer. 
This is a structure defined, in the include file `regex.h', as

     struct re_pattern_buffer
         char *buffer   /* Space holding the compiled pattern commands. */
         int allocated  /* Size of space that  buffer  points to */
         int used       /* Length of portion of buffer actually occupied */
         char *fastmap; /* Pointer to fastmap, if any, or zero if none. */
                        /* re_search uses the fastmap, if there is one,
                           to skip quickly over totally implausible
                           characters */
         char *translate;
                        /* Translate table to apply to characters before
                           comparing, or zero for no translation.
                           The translation is applied to a pattern when
                           it is compiled and to data when it is matched. */
         char fastmap_accurate;
                        /* Set to zero when a new pattern is stored,
                           set to one when the fastmap is updated from it. */

Before compiling a pattern, you must initialize the `buffer' field to
point to a block of memory obtained with `malloc', and the
`allocated' field to the size of that block, in bytes.  The pattern
compiler will replace this block with a larger one if necessary.

You must also initialize the `translate' field to point to the
translate table that you will use when you match the compiled
pattern, or to zero if you will use no translate table when you
match.  *Note translation::.

Then call `re_compile_pattern' to compile a regular expression into
the buffer:

     re_compile_pattern (REGEX, REGEX_SIZE, BUF)

REGEX is the address of the regular expression (`char *'), REGEX_SIZE
is its length (`int'), BUF is the address of the buffer (`struct
re_pattern_buffer *').

`re_compile_pattern' returns zero if it succeeds in compiling the
regular expression.  In that case, `*buf' now contains the results. 
Otherwise, `re_compile_pattern' returns a string which serves as an
error message.

After compiling, if you wish to search for the pattern, you must
initialize the `fastmap' component of the pattern buffer.  *Note

File: regex,  Node: matching,  Next: searching,  Prev: compiling,  Up: programming

Matching a Compiled Pattern

Once a regular expression has been compiled into a pattern buffer,
you can match the pattern buffer against a string with `re_match'.

     re_match (BUF, STRING, SIZE, POS, REGS)

BUF is, once again, the address of the buffer (`struct
re_pattern_buffer *').  STRING is the string to be matched (`char *').
sIZE is the length of that string (`int').  POS is the position
within the string at which to begin matching (`int').  The beginning
of the string is position 0.  REGS is described below.  Normally it
is zero.  *Note registers::.

`re_match' returns `-1' if the pattern does not match; otherwise, it
returns the length of the portion of `string' which was matched.

For example, suppose that BUF points to a buffer containing the
result of compiling `x*', STRING points to `xxxxxy', and SIZE is `6'.
Suppose that POS is `2'.  Then the last three `x''s will be matched,
so `re_match' will return `3'.  If POS is zero, the value will be `5'.
If POS is `5' or `6', the value will be zero, meaning that the null
string was successfully matched.  Note that since `x*' matches the
empty string, it will never entirely fail.

It is up to the caller to avoid passing a value of POS that results
in matching outside the specified string.  POS must not be negative
and must not be greater than SIZE.

File: regex,  Node: searching,  Next: translation,  Prev: matching,  Up: programming

Searching for a Match

Searching means trying successive starting positions for a match
until a match is found.  To search, you supply a compiled pattern
buffer.  Before searching you must initialize the `fastmap' field of
the pattern buffer (see below).


is called like `re_match' except that the POS argument is replaced by
two arguments STARTPOS and RANGE.  `re_search' tests for a match
starting at index STARTPOS, then at `STARTPOS + 1', and so on.  It
tries RANGE consecutive positions before giving up and returning
`-1'.  If a match is found, `re_search' returns the index at which
the match was found.

If RANGE is negative, RE_SEARCH tries starting positions STARTPOS,
`STARTPOS - 1', ... in that order.  `|RANGE|' is the number of tries

It is up to the caller to avoid passing value of STARTPOS and RANGE
that result in matching outside the specified string.  STARTPOS must
be between zero and SIZE, inclusive, and so must `STARTPOS + RANGE -
1' (if RANGE is positive) or `STARTPOS + RANGE + 1' (if RANGE is

If you may be searching over a long distance (that is, trying many
different match starting points) with a compiled pattern, you should
use a "fastmap" in it.  This is a block of 256 bytes, whose address
is placed in the `fastmap' component of the pattern buffer.  The
first time you search for a particular compiled pattern, the fastmap
is set so that `FASTMAP[CH]' is nonzero if the character CH might
possibly start a match for this pattern.  `re_search' checks each
character against the fastmap so that it can skip more quickly over

If you do not want a fastmap, store zero in the `fastmap' component
of the pattern buffer before calling `re_search'.

In either case, you must initialize this component in a pattern
buffer before you can use that buffer in a search; but you can choose
as an initial value either zero or the address of a suitable block of

If you compile a new pattern in an existing pattern buffer, it is not
necessary to reinitialize the `fastmap' component (unless you wish to
override your previous choice).

File: regex,  Node: translation,  Next: registers,  Prev: searching,  Up: programming

Translate Tables

With a translate table, you can apply a transformation to all
characters before they are compared.  For example, a table that maps
lower case letters into upper case (or vice versa) causes differences
in case to be ignored by matching.

A translate table is a block of 256 bytes.  Each character of raw
data is used as an index in the translate table.  The value found
there is used instead of the original character.  Each character in a
regular expression, except for the syntactic constructs, is
translated when the expression is compiled.  Each character of a
string being matched is translated whenever it is compared or tested.

A suitable translate table to ignore differences in case maps all
characters into themselves, except for lower case letters, which are
mapped into the corresponding upper case letters.  It could be
initialized by:

     for (i = 0; i < 0400; i++)
       table[i] = i;
     for (i = 'a'; i <= 'z'; i++)
       table[i] = i - 040;

You specify the use of a translate table by putting its address in
the TRANSLATE component of the compiled pattern buffer.  If this
component is zero, no translation is done.  Since both compilation
and matching use the translate table, you must use the same table
contents for both operations or confusing things will happen.

File: regex,  Node: registers,  Next: split,  Prev: translation,  Up: programming

Registers: or "What Did the `\( ... \)' Groupings Actually Match?"

If you want to find out, after the match, what each of the first nine
`\( ... \)' groupings actually matched, you can pass the REGS
argument to the match or search function.  Pass the address of a
structure of this type:

     struct re_registers
         int start[RE_NREGS];
         int end[RE_NREGS];

  `re_match' and `re_search' will store into this structure the data
you want.  `REGS->start[REG]' will be the index in STRING of the
beginning of the data matched by the REG'th `\( ... \)' grouping, and
`REGS->end[REG]' will be the index of the end of that data (the index
of the first character beyond those matched).  The values in the
start and end arrays at indexes greater than the number of `\( ...
\)' groupings present in the regular expression will be set to the
value -1.  Register numbers start at 1 and run to `RE_NREGS - 1'
(normally `9').  `REGS->start[0]' and `REGS->end[0]' are similar but
describe the extent of the substring matched by the entire pattern.

  Both `struct re_registers' and `RE_NREGS' are defined in `regex.h'.

File: regex,  Node: split,  Next: unix,  Prev: registers,  Up: programming

Matching against Split Data

The functions `re_match_2' and `re_search_2' allow one to match in or
search data which is divided into two strings.

`re_match_2' works like `re_match' except that two data strings and
sizes must be given.

     re_match_2 (BUF, STRING1, SIZE1, STRING2, SIZE2, POS, REGS)

The matcher regards the contents of STRING1 as effectively followed
by the contents of STRING2, and matches the combined string against
the pattern in BUF.

`re_search_2' is likewise similar to `re_search':


The value returned by RE_SEARCH_2 is an index into the combined data
made up of STRING1 and STRING2.  It never exceeds `SIZE1 + SIZE2'. 
The values returned in the REGS structure (if there is one) are
likewise indices in the combined data.

File: regex,  Node: unix,  Prev: split,  Up: programming

Unix-Compatible Entry Points

The standard Berkeley Unix way to compile a regular expression is to
call `re_comp'.  This function takes a single argument, the address
of the regular expression, which is assumed to be terminated by a
null character.

`re_comp' does not ask you to specify a pattern buffer because it has
its own pattern buffer -- just one.  Using `re_comp', one may match
only the most recently compiled regular expression.

The value of `re_comp' is zero for success or else an error message
string, as for `re_compile_pattern'.

Calling `re_comp' with the null string as argument it has no effect;
the contents of the buffer remain unchanged.

The standard Berkeley Unix way to match the last regular expression
compiled is to call `re_exec'.  This takes a single argument, the
address of the string to be matched.  This string is assumed to be
terminated by a null character.  Matching is tried starting at each
position in the string.  `re_exec' returns `1' for success or `0' for
failure.  One cannot find out how long a substring was matched, nor
what the `\( ... \)' groupings matched.

Tag Table:
Node: top85
Node: syntax1706
Node: directives3241
Node: emacs10891
Node: programming11653
Node: compiling12381
Node: matching14827
Node: searching16269
Node: translation18534
Node: registers19952
Node: split21245
Node: unix22183

End Tag Table