path: root/usr.bin/lex/flex.1

.\"	$OpenBSD: flex.1,v 1.30 2010/10/18 14:42:16 jmc Exp $
.\"
.\" Copyright (c) 1990 The Regents of the University of California.
.\" All rights reserved.
.\"
.\" This code is derived from software contributed to Berkeley by
.\" Vern Paxson.
.\"
.\" The United States Government has rights in this work pursuant
.\" to contract no. DE-AC03-76SF00098 between the United States
.\" Department of Energy and the University of California.
.\"
.\" Redistribution and use in source and binary forms, with or without
.\" modification, are permitted provided that the following conditions
.\" are met:
.\"
.\" 1. Redistributions of source code must retain the above copyright
.\"    notice, this list of conditions and the following disclaimer.
.\" 2. Redistributions in binary form must reproduce the above copyright
.\"    notice, this list of conditions and the following disclaimer in the
.\"    documentation and/or other materials provided with the distribution.
.\"
.\" Neither the name of the University nor the names of its contributors
.\" may be used to endorse or promote products derived from this software
.\" without specific prior written permission.
.\"
.\" THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR
.\" IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED
.\" WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
.\" PURPOSE.
.\"
.Dd $Mdocdate: October 18 2010 $
.Dt FLEX 1
.Os
.Sh NAME
.Nm flex
.Nd fast lexical analyzer generator
.Sh SYNOPSIS
.Nm
.Bk -words
.Op Fl 78BbcdFfhIiLlnpsTtVvw+?
.Op Fl C Ns Op Cm aeFfmr
.Op Fl Fl help
.Op Fl Fl version
.Op Fl o Ns Ar output
.Op Fl P Ns Ar prefix
.Op Fl S Ns Ar skeleton
.Op Ar
.Ek
.Sh DESCRIPTION
.Nm
is a tool for generating
.Em scanners :
programs which recognize lexical patterns in text.
.Nm
reads the given input files, or its standard input if no file names are given,
for a description of a scanner to generate.
The description is in the form of pairs of regular expressions and C code,
called
.Em rules .
.Nm
generates as output a C source file,
.Pa lex.yy.c ,
which defines a routine
.Fn yylex .
This file is compiled and linked with the
.Fl lfl
library to produce an executable.
When the executable is run, it analyzes its input for occurrences
of the regular expressions.
Whenever it finds one, it executes the corresponding C code.
.Pp
The manual includes both tutorial and reference sections:
.Bl -ohang
.It Sy Some Simple Examples
.It Sy Format of the Input File
.It Sy Patterns
The extended regular expressions used by
.Nm .
.It Sy How the Input is Matched
The rules for determining what has been matched.
.It Sy Actions
How to specify what to do when a pattern is matched.
.It Sy The Generated Scanner
Details regarding the scanner that
.Nm
produces;
how to control the input source.
.It Sy Start Conditions
Introducing context into scanners, and managing
.Qq mini-scanners .
.It Sy Multiple Input Buffers
How to manipulate multiple input sources;
how to scan from strings instead of files.
.It Sy End-of-File Rules
Special rules for matching the end of the input.
.It Sy Miscellaneous Macros
A summary of macros available to the actions.
.It Sy Values Available to the User
A summary of values available to the actions.
.It Sy Interfacing with Yacc
Connecting flex scanners together with
.Xr yacc 1
parsers.
.It Sy Options
.Nm
command-line options, and the
.Dq %option
directive.
.It Sy Performance Considerations
How to make scanners go as fast as possible.
.It Sy Generating C++ Scanners
The
.Pq experimental
facility for generating C++ scanner classes.
.It Sy Incompatibilities with Lex and POSIX
How
.Nm
differs from AT&T lex and the
.Tn POSIX
lex standard.
.It Sy Files
Files used by
.Nm .
.It Sy Diagnostics
Those error messages produced by
.Nm
.Pq or scanners it generates
whose meanings might not be apparent.
.It Sy See Also
Other documentation, related tools.
.It Sy Authors
Includes contact information.
.It Sy Bugs
Known problems with
.Nm .
.El
.Sh SOME SIMPLE EXAMPLES
First some simple examples to get the flavor of how one uses
.Nm .
The following
.Nm
input specifies a scanner which whenever it encounters the string
.Qq username
will replace it with the user's login name:
.Bd -literal -offset indent
%%
username    printf("%s", getlogin());
.Ed
.Pp
By default, any text not matched by a
.Nm
scanner is copied to the output, so the net effect of this scanner is
to copy its input file to its output with each occurrence of
.Qq username
expanded.
In this input, there is just one rule.
.Qq username
is the
.Em pattern
and the
.Qq printf
is the
.Em action .
The
.Qq %%
marks the beginning of the rules.
.Pp
Here's another simple example:
.Bd -literal -offset indent
%{
int num_lines = 0, num_chars = 0;
%}

%%
\en      ++num_lines; ++num_chars;
\&.       ++num_chars;

%%
main()
{
	yylex();
	printf("# of lines = %d, # of chars = %d\en",
            num_lines, num_chars);
}
.Ed
.Pp
This scanner counts the number of characters and the number
of lines in its input
(it produces no output other than the final report on the counts).
The first line declares two globals,
.Qq num_lines
and
.Qq num_chars ,
which are accessible both inside
.Fn yylex
and in the
.Fn main
routine declared after the second
.Qq %% .
There are two rules, one which matches a newline
.Pq \&"\en\&"
and increments both the line count and the character count,
and one which matches any character other than a newline
(indicated by the
.Qq \&.
regular expression).
.Pp
A somewhat more complicated example:
.Bd -literal -offset indent
/* scanner for a toy Pascal-like language */

%{
/* need this for the call to atof() below */
#include <math.h>
%}

DIGIT    [0-9]
ID       [a-z][a-z0-9]*

%%

{DIGIT}+ {
        printf("An integer: %s (%d)\en", yytext,
            atoi(yytext));
}

{DIGIT}+"."{DIGIT}* {
        printf("A float: %s (%g)\en", yytext,
            atof(yytext));
}

if|then|begin|end|procedure|function {
        printf("A keyword: %s\en", yytext);
}

{ID}    printf("An identifier: %s\en", yytext);

"+"|"-"|"*"|"/"   printf("An operator: %s\en", yytext);

"{"[^}\en]*"}"     /* eat up one-line comments */

[ \et\en]+          /* eat up whitespace */

\&.       printf("Unrecognized character: %s\en", yytext);

%%

main(int argc, char *argv[])
{
        ++argv; --argc;  /* skip over program name */
        if (argc > 0)
                yyin = fopen(argv[0], "r");
        else
                yyin = stdin;

        yylex();
}
.Ed
.Pp
This is the beginnings of a simple scanner for a language like Pascal.
It identifies different types of
.Em tokens
and reports on what it has seen.
.Pp
The details of this example will be explained in the following sections.
.Sh FORMAT OF THE INPUT FILE
The
.Nm
input file consists of three sections, separated by a line with just
.Qq %%
in it:
.Bd -unfilled -offset indent
definitions
%%
rules
%%
user code
.Ed
.Pp
The
.Em definitions
section contains declarations of simple
.Em name
definitions to simplify the scanner specification, and declarations of
.Em start conditions ,
which are explained in a later section.
.Pp
Name definitions have the form:
.Pp
.D1 name definition
.Pp
The
.Qq name
is a word beginning with a letter or an underscore
.Pq Sq _
followed by zero or more letters, digits,
.Sq _ ,
or
.Sq -
.Pq dash .
The definition is taken to begin at the first non-whitespace character
following the name and continuing to the end of the line.
The definition can subsequently be referred to using
.Qq {name} ,
which will expand to
.Qq (definition) .
For example:
.Bd -literal -offset indent
DIGIT    [0-9]
ID       [a-z][a-z0-9]*
.Ed
.Pp
This defines
.Qq DIGIT
to be a regular expression which matches a single digit, and
.Qq ID
to be a regular expression which matches a letter
followed by zero-or-more letters-or-digits.
A subsequent reference to
.Pp
.Dl {DIGIT}+"."{DIGIT}*
.Pp
is identical to
.Pp
.Dl ([0-9])+"."([0-9])*
.Pp
and matches one-or-more digits followed by a
.Sq .\&
followed by zero-or-more digits.
.Pp
The
.Em rules
section of the
.Nm
input contains a series of rules of the form:
.Pp
.D1 pattern	action
.Pp
The pattern must be unindented and the action must begin
on the same line.
.Pp
See below for a further description of patterns and actions.
.Pp
Finally, the user code section is simply copied to
.Pa lex.yy.c
verbatim.
It is used for companion routines which call or are called by the scanner.
The presence of this section is optional;
if it is missing, the second
.Qq %%
in the input file may be skipped too.
.Pp
In the definitions and rules sections, any indented text or text enclosed in
.Sq %{
and
.Sq %}
is copied verbatim to the output
.Pq with the %{}'s removed .
The %{}'s must appear unindented on lines by themselves.
.Pp
In the rules section,
any indented or %{} text appearing before the first rule may be used to
declare variables which are local to the scanning routine and
.Pq after the declarations
code which is to be executed whenever the scanning routine is entered.
Other indented or %{} text in the rule section is still copied to the output,
but its meaning is not well-defined and it may well cause compile-time
errors (this feature is present for
.Tn POSIX
compliance; see below for other such features).
.Pp
In the definitions section
.Pq but not in the rules section ,
an unindented comment
(i.e., a line beginning with
.Qq /* )
is also copied verbatim to the output up to the next
.Qq */ .
.Sh PATTERNS
The patterns in the input are written using an extended set of regular
expressions.
These are:
.Bl -tag -width "XXXXXXXX"
.It x
Match the character
.Sq x .
.It .\&
Any character
.Pq byte
except newline.
.It [xyz]
A
.Qq character class ;
in this case, the pattern matches either an
.Sq x ,
a
.Sq y ,
or a
.Sq z .
.It [abj-oZ]
A
.Qq character class
with a range in it; matches an
.Sq a ,
a
.Sq b ,
any letter from
.Sq j
through
.Sq o ,
or a
.Sq Z .
.It [^A-Z]
A
.Qq negated character class ,
i.e., any character but those in the class.
In this case, any character EXCEPT an uppercase letter.
.It [^A-Z\en]
Any character EXCEPT an uppercase letter or a newline.
.It r*
Zero or more r's, where
.Sq r
is any regular expression.
.It r+
One or more r's.
.It r?
Zero or one r's (that is,
.Qq an optional r ) .
.It r{2,5}
Anywhere from two to five r's.
.It r{2,}
Two or more r's.
.It r{4}
Exactly 4 r's.
.It {name}
The expansion of the
.Qq name
definition
.Pq see above .
.It \&"[xyz]\e\&"foo\&"
The literal string: [xyz]"foo.
.It \eX
If
.Sq X
is an
.Sq a ,
.Sq b ,
.Sq f ,
.Sq n ,
.Sq r ,
.Sq t ,
or
.Sq v ,
then the ANSI-C interpretation of
.Sq \eX .
Otherwise, a literal
.Sq X
(used to escape operators such as
.Sq * ) .
.It \e0
A NUL character
.Pq ASCII code 0 .
.It \e123
The character with octal value 123.
.It \ex2a
The character with hexadecimal value 2a.
.It (r)
Match an
.Sq r ;
parentheses are used to override precedence
.Pq see below .
.It rs
The regular expression
.Sq r
followed by the regular expression
.Sq s ;
called
.Qq concatenation .
.It r|s
Either an
.Sq r
or an
.Sq s .
.It r/s
An
.Sq r ,
but only if it is followed by an
.Sq s .
The text matched by
.Sq s
is included when determining whether this rule is the
.Qq longest match ,
but is then returned to the input before the action is executed.
So the action only sees the text matched by
.Sq r .
This type of pattern is called
.Qq trailing context .
(There are some combinations of r/s that
.Nm
cannot match correctly; see notes in the
.Sx BUGS
section below regarding
.Qq dangerous trailing context . )
.It ^r
An
.Sq r ,
but only at the beginning of a line
(i.e., just starting to scan, or right after a newline has been scanned).
.It r$
An
.Sq r ,
but only at the end of a line
.Pq i.e., just before a newline .
Equivalent to
.Qq r/\en .
.Pp
Note that
.Nm flex Ns 's
notion of
.Qq newline
is exactly whatever the C compiler used to compile
.Nm
interprets
.Sq \en
as.
.\" In particular, on some DOS systems you must either filter out \er's in the
.\" input yourself, or explicitly use r/\er\en for
.\" .Qq r$ .
.It <s>r
An
.Sq r ,
but only in start condition
.Sq s
.Pq see below for discussion of start conditions .
.It <s1,s2,s3>r
The same, but in any of start conditions s1, s2, or s3.
.It <*>r
An
.Sq r
in any start condition, even an exclusive one.
.It <<EOF>>
An end-of-file.
.It <s1,s2><<EOF>>
An end-of-file when in start condition s1 or s2.
.El
.Pp
Note that inside of a character class, all regular expression operators
lose their special meaning except escape
.Pq Sq \e
and the character class operators,
.Sq - ,
.Sq ]\& ,
and, at the beginning of the class,
.Sq ^ .
.Pp
The regular expressions listed above are grouped according to
precedence, from highest precedence at the top to lowest at the bottom.
Those grouped together have equal precedence.
For example,
.Pp
.D1 foo|bar*
.Pp
is the same as
.Pp
.D1 (foo)|(ba(r*))
.Pp
since the
.Sq *
operator has higher precedence than concatenation,
and concatenation higher than alternation
.Pq Sq |\& .
This pattern therefore matches
.Em either
the string
.Qq foo
.Em or
the string
.Qq ba
followed by zero-or-more r's.
To match
.Qq foo
or zero-or-more "bar"'s,
use:
.Pp
.D1 foo|(bar)*
.Pp
and to match zero-or-more "foo"'s-or-"bar"'s:
.Pp
.D1 (foo|bar)*
.Pp
In addition to characters and ranges of characters, character classes
can also contain character class
.Em expressions .
These are expressions enclosed inside
.Sq [:
and
.Sq :]
delimiters (which themselves must appear between the
.Sq \&[
and
.Sq ]\&
of the
character class; other elements may occur inside the character class, too).
The valid expressions are:
.Bd -unfilled -offset indent
[:alnum:] [:alpha:] [:blank:]
[:cntrl:] [:digit:] [:graph:]
[:lower:] [:print:] [:punct:]
[:space:] [:upper:] [:xdigit:]
.Ed
.Pp
These expressions all designate a set of characters equivalent to
the corresponding standard C
.Fn isXXX
function.
For example, [:alnum:] designates those characters for which
.Xr isalnum 3
returns true \- i.e., any alphabetic or numeric.
Some systems don't provide
.Xr isblank 3 ,
so
.Nm
defines [:blank:] as a blank or a tab.
.Pp
For example, the following character classes are all equivalent:
.Bd -unfilled -offset indent
[[:alnum:]]
[[:alpha:][:digit:]]
[[:alpha:]0-9]
[a-zA-Z0-9]
.Ed
.Pp
If the scanner is case-insensitive (the
.Fl i
flag), then [:upper:] and [:lower:] are equivalent to [:alpha:].
.Pp
Some notes on patterns:
.Bl -dash
.It
A negated character class such as the example
.Qq [^A-Z]
above will match a newline unless "\en"
.Pq or an equivalent escape sequence
is one of the characters explicitly present in the negated character class
(e.g.,
.Qq [^A-Z\en] ) .
This is unlike how many other regular expression tools treat negated character
classes, but unfortunately the inconsistency is historically entrenched.
Matching newlines means that a pattern like
.Qq [^"]*
can match the entire input unless there's another quote in the input.
.It
A rule can have at most one instance of trailing context
(the
.Sq /
operator or the
.Sq $
operator).
The start condition,
.Sq ^ ,
and
.Qq <<EOF>>
patterns can only occur at the beginning of a pattern, and, as well as with
.Sq /
and
.Sq $ ,
cannot be grouped inside parentheses.
A
.Sq ^
which does not occur at the beginning of a rule or a
.Sq $
which does not occur at the end of a rule loses its special properties
and is treated as a normal character.
.It
The following are illegal:
.Bd -unfilled -offset indent
foo/bar$
<sc1>foo<sc2>bar
.Ed
.Pp
Note that the first of these, can be written
.Qq foo/bar\en .
.It
The following will result in
.Sq $
or
.Sq ^
being treated as a normal character:
.Bd -unfilled -offset indent
foo|(bar$)
foo|^bar
.Ed
.Pp
If what's wanted is a
.Qq foo
or a bar-followed-by-a-newline, the following could be used
(the special
.Sq |\&
action is explained below):
.Bd -unfilled -offset indent
foo      |
bar$     /* action goes here */
.Ed
.Pp
A similar trick will work for matching a foo or a
bar-at-the-beginning-of-a-line.
.El
.Sh HOW THE INPUT IS MATCHED
When the generated scanner is run,
it analyzes its input looking for strings which match any of its patterns.
If it finds more than one match,
it takes the one matching the most text
(for trailing context rules, this includes the length of the trailing part,
even though it will then be returned to the input).
If it finds two or more matches of the same length,
the rule listed first in the
.Nm
input file is chosen.
.Pp
Once the match is determined, the text corresponding to the match
(called the
.Em token )
is made available in the global character pointer
.Fa yytext ,
and its length in the global integer
.Fa yyleng .
The
.Em action
corresponding to the matched pattern is then executed
.Pq a more detailed description of actions follows ,
and then the remaining input is scanned for another match.
.Pp
If no match is found, then the default rule is executed:
the next character in the input is considered matched and
copied to the standard output.
Thus, the simplest legal
.Nm
input is:
.Pp
.D1 %%
.Pp
which generates a scanner that simply copies its input
.Pq one character at a time
to its output.
.Pp
Note that
.Fa yytext
can be defined in two different ways:
either as a character pointer or as a character array.
Which definition
.Nm
uses can be controlled by including one of the special directives
.Dq %pointer
or
.Dq %array
in the first
.Pq definitions
section of flex input.
The default is
.Dq %pointer ,
unless the
.Fl l
lex compatibility option is used, in which case
.Fa yytext
will be an array.
The advantage of using
.Dq %pointer
is substantially faster scanning and no buffer overflow when matching
very large tokens
.Pq unless not enough dynamic memory is available .
The disadvantage is that actions are restricted in how they can modify
.Fa yytext
.Pq see the next section ,
and calls to the
.Fn unput
function destroy the present contents of
.Fa yytext ,
which can be a considerable porting headache when moving between different
.Nm lex
versions.
.Pp
The advantage of
.Dq %array
is that
.Fa yytext
can be modified as much as wanted, and calls to
.Fn unput
do not destroy
.Fa yytext
.Pq see below .
Furthermore, existing
.Nm lex
programs sometimes access
.Fa yytext
externally using declarations of the form:
.Pp
.D1 extern char yytext[];
.Pp
This definition is erroneous when used with
.Dq %pointer ,
but correct for
.Dq %array .
.Pp
.Dq %array
defines
.Fa yytext
to be an array of
.Dv YYLMAX
characters, which defaults to a fairly large value.
The size can be changed by simply #define'ing
.Dv YYLMAX
to a different value in the first section of
.Nm
input.
As mentioned above, with
.Dq %pointer
yytext grows dynamically to accommodate large tokens.
While this means a
.Dq %pointer
scanner can accommodate very large tokens
.Pq such as matching entire blocks of comments ,
bear in mind that each time the scanner must resize
.Fa yytext
it also must rescan the entire token from the beginning, so matching such
tokens can prove slow.
.Fa yytext
presently does not dynamically grow if a call to
.Fn unput
results in too much text being pushed back; instead, a run-time error results.
.Pp
Also note that
.Dq %array
cannot be used with C++ scanner classes
.Pq the c++ option; see below .
.Sh ACTIONS
Each pattern in a rule has a corresponding action,
which can be any arbitrary C statement.
The pattern ends at the first non-escaped whitespace character;
the remainder of the line is its action.
If the action is empty,
then when the pattern is matched the input token is simply discarded.
For example, here is the specification for a program
which deletes all occurrences of
.Qq zap me
from its input:
.Bd -literal -offset indent
%%
"zap me"
.Ed
.Pp
(It will copy all other characters in the input to the output since
they will be matched by the default rule.)
.Pp
Here is a program which compresses multiple blanks and tabs down to
a single blank, and throws away whitespace found at the end of a line:
.Bd -literal -offset indent
%%
[ \et]+        putchar(' ');
[ \et]+$       /* ignore this token */
.Ed
.Pp
If the action contains a
.Sq { ,
then the action spans till the balancing
.Sq }
is found, and the action may cross multiple lines.
.Nm
knows about C strings and comments and won't be fooled by braces found
within them, but also allows actions to begin with
.Sq %{
and will consider the action to be all the text up to the next
.Sq %}
.Pq regardless of ordinary braces inside the action .
.Pp
An action consisting solely of a vertical bar
.Pq Sq |\&
means
.Qq same as the action for the next rule .
See below for an illustration.
.Pp
Actions can include arbitrary C code,
including return statements to return a value to whatever routine called
.Fn yylex .
Each time
.Fn yylex
is called, it continues processing tokens from where it last left off
until it either reaches the end of the file or executes a return.
.Pp
Actions are free to modify
.Fa yytext
except for lengthening it
(adding characters to its end \- these will overwrite later characters in the
input stream).
This, however, does not apply when using
.Dq %array
.Pq see above ;
in that case,
.Fa yytext
may be freely modified in any way.
.Pp
Actions are free to modify
.Fa yyleng
except they should not do so if the action also includes use of
.Fn yymore
.Pq see below .
.Pp
There are a number of special directives which can be included within
an action:
.Bl -tag -width Ds
.It ECHO
Copies
.Fa yytext
to the scanner's output.
.It BEGIN
Followed by the name of a start condition, places the scanner in the
corresponding start condition
.Pq see below .
.It REJECT
Directs the scanner to proceed on to the
.Qq second best
rule which matched the input
.Pq or a prefix of the input .
The rule is chosen as described above in
.Sx HOW THE INPUT IS MATCHED ,
and
.Fa yytext
and
.Fa yyleng
set up appropriately.
It may either be one which matched as much text
as the originally chosen rule but came later in the
.Nm
input file, or one which matched less text.
For example, the following will both count the
words in the input and call the routine
.Fn special
whenever
.Qq frob
is seen:
.Bd -literal -offset indent
int word_count = 0;
%%

frob        special(); REJECT;
[^ \et\en]+   ++word_count;
.Ed
.Pp
Without the
.Em REJECT ,
any "frob"'s in the input would not be counted as words,
since the scanner normally executes only one action per token.
Multiple
.Em REJECT Ns 's
are allowed,
each one finding the next best choice to the currently active rule.
For example, when the following scanner scans the token
.Qq abcd ,
it will write
.Qq abcdabcaba
to the output:
.Bd -literal -offset indent
%%
a        |
ab       |
abc      |
abcd     ECHO; REJECT;
\&.|\en     /* eat up any unmatched character */
.Ed
.Pp
(The first three rules share the fourth's action since they use
the special
.Sq |\&
action.)
.Em REJECT
is a particularly expensive feature in terms of scanner performance;
if it is used in any of the scanner's actions it will slow down
all of the scanner's matching.
Furthermore,
.Em REJECT
cannot be used with the
.Fl Cf
or
.Fl CF
options
.Pq see below .
.Pp
Note also that unlike the other special actions,
.Em REJECT
is a
.Em branch ;
code immediately following it in the action will not be executed.
.It yymore()
Tells the scanner that the next time it matches a rule, the corresponding
token should be appended onto the current value of
.Fa yytext
rather than replacing it.
For example, given the input
.Qq mega-kludge
the following will write
.Qq mega-mega-kludge
to the output:
.Bd -literal -offset indent
%%
mega-    ECHO; yymore();
kludge   ECHO;
.Ed
.Pp
First
.Qq mega-
is matched and echoed to the output.
Then
.Qq kludge
is matched, but the previous
.Qq mega-
is still hanging around at the beginning of
.Fa yytext
so the
.Em ECHO
for the
.Qq kludge
rule will actually write
.Qq mega-kludge .
.Pp
Two notes regarding use of
.Fn yymore :
First,
.Fn yymore
depends on the value of
.Fa yyleng
correctly reflecting the size of the current token, so
.Fa yyleng
must not be modified when using
.Fn yymore .
Second, the presence of
.Fn yymore
in the scanner's action entails a minor performance penalty in the
scanner's matching speed.
.It yyless(n)
Returns all but the first
.Ar n
characters of the current token back to the input stream, where they
will be rescanned when the scanner looks for the next match.
.Fa yytext
and
.Fa yyleng
are adjusted appropriately (e.g.,
.Fa yyleng
will now be equal to
.Ar n ) .
For example, on the input
.Qq foobar
the following will write out
.Qq foobarbar :
.Bd -literal -offset indent
%%
foobar    ECHO; yyless(3);
[a-z]+    ECHO;
.Ed
.Pp
An argument of 0 to
.Fa yyless
will cause the entire current input string to be scanned again.
Unless how the scanner will subsequently process its input has been changed
(using
.Em BEGIN ,
for example),
this will result in an endless loop.
.Pp
Note that
.Fa yyless
is a macro and can only be used in the
.Nm
input file, not from other source files.
.It unput(c)
Puts the character
.Ar c
back into the input stream.
It will be the next character scanned.
The following action will take the current token and cause it
to be rescanned enclosed in parentheses.
.Bd -literal -offset indent
{
        int i;
        char *yycopy;

        /* Copy yytext because unput() trashes yytext */
        if ((yycopy = strdup(yytext)) == NULL)
                err(1, NULL);
        unput(')');
        for (i = yyleng - 1; i >= 0; --i)
                unput(yycopy[i]);
        unput('(');
        free(yycopy);
}
.Ed
.Pp
Note that since each
.Fn unput
puts the given character back at the beginning of the input stream,
pushing back strings must be done back-to-front.
.Pp
An important potential problem when using
.Fn unput
is that if using
.Dq %pointer
.Pq the default ,
a call to
.Fn unput
destroys the contents of
.Fa yytext ,
starting with its rightmost character and devouring one character to
the left with each call.
If the value of
.Fa yytext
should be preserved after a call to
.Fn unput
.Pq as in the above example ,
it must either first be copied elsewhere, or the scanner must be built using
.Dq %array
instead (see
.Sx HOW THE INPUT IS MATCHED ) .
.Pp
Finally, note that EOF cannot be put back
to attempt to mark the input stream with an end-of-file.
.It input()
Reads the next character from the input stream.
For example, the following is one way to eat up C comments:
.Bd -literal -offset indent
%%
"/*" {
        int c;

        for (;;) {
                while ((c = input()) != '*' && c != EOF)
                        ; /* eat up text of comment */

                if (c == '*') {
                        while ((c = input()) == '*')
                                ;
                        if (c == '/')
                                break; /* found the end */
                }

                if (c == EOF) {
                        errx(1, "EOF in comment");
                        break;
                }
        }
}
.Ed
.Pp
(Note that if the scanner is compiled using C++, then
.Fn input
is instead referred to as
.Fn yyinput ,
in order to avoid a name clash with the C++ stream by the name of input.)
.It YY_FLUSH_BUFFER
Flushes the scanner's internal buffer
so that the next time the scanner attempts to match a token,
it will first refill the buffer using
.Dv YY_INPUT
(see
.Sx THE GENERATED SCANNER ,
below).
This action is a special case of the more general
.Fn yy_flush_buffer
function, described below in the section
.Sx MULTIPLE INPUT BUFFERS .
.It yyterminate()
Can be used in lieu of a return statement in an action.
It terminates the scanner and returns a 0 to the scanner's caller, indicating
.Qq all done .
By default,
.Fn yyterminate
is also called when an end-of-file is encountered.
It is a macro and may be redefined.
.El
.Sh THE GENERATED SCANNER
The output of
.Nm
is the file
.Pa lex.yy.c ,
which contains the scanning routine
.Fn yylex ,
a number of tables used by it for matching tokens,
and a number of auxiliary routines and macros.
By default,
.Fn yylex
is declared as follows:
.Bd -unfilled -offset indent
int yylex()
{
    ... various definitions and the actions in here ...
}
.Ed
.Pp
(If the environment supports function prototypes, then it will
be "int yylex(void)".)
This definition may be changed by defining the
.Dv YY_DECL
macro.
For example:
.Bd -literal -offset indent
#define YY_DECL float lexscan(a, b) float a, b;
.Ed
.Pp
would give the scanning routine the name
.Em lexscan ,
returning a float, and taking two floats as arguments.
Note that if arguments are given to the scanning routine using a
K&R-style/non-prototyped function declaration,
the definition must be terminated with a semi-colon
.Pq Sq ;\& .
.Pp
Whenever
.Fn yylex
is called, it scans tokens from the global input file
.Pa yyin
.Pq which defaults to stdin .
It continues until it either reaches an end-of-file
.Pq at which point it returns the value 0
or one of its actions executes a
.Em return
statement.
.Pp
If the scanner reaches an end-of-file, subsequent calls are undefined
unless either
.Em yyin
is pointed at a new input file
.Pq in which case scanning continues from that file ,
or
.Fn yyrestart
is called.
.Fn yyrestart
takes one argument, a
.Fa FILE *
pointer (which can be nil, if
.Dv YY_INPUT
has been set up to scan from a source other than
.Em yyin ) ,
and initializes
.Em yyin
for scanning from that file.
Essentially there is no difference between just assigning
.Em yyin
to a new input file or using
.Fn yyrestart
to do so; the latter is available for compatibility with previous versions of
.Nm ,
and because it can be used to switch input files in the middle of scanning.
It can also be used to throw away the current input buffer,
by calling it with an argument of
.Em yyin ;
but better is to use
.Dv YY_FLUSH_BUFFER
.Pq see above .
Note that
.Fn yyrestart
does not reset the start condition to
.Em INITIAL
(see
.Sx START CONDITIONS ,
below).
.Pp
If
.Fn yylex
stops scanning due to executing a
.Em return
statement in one of the actions, the scanner may then be called again and it
will resume scanning where it left off.
.Pp
By default
.Pq and for purposes of efficiency ,
the scanner uses block-reads rather than simple
.Xr getc 3
calls to read characters from
.Em yyin .
The nature of how it gets its input can be controlled by defining the
.Dv YY_INPUT
macro.
.Dv YY_INPUT Ns 's
calling sequence is
.Qq YY_INPUT(buf,result,max_size) .
Its action is to place up to
.Dv max_size
characters in the character array
.Em buf
and return in the integer variable
.Em result
either the number of characters read or the constant
.Dv YY_NULL
(0 on
.Ux
systems)
to indicate
.Dv EOF .
The default
.Dv YY_INPUT
reads from the global file-pointer
.Qq yyin .
.Pp
A sample definition of
.Dv YY_INPUT
.Pq in the definitions section of the input file :
.Bd -unfilled -offset indent
%{
#define YY_INPUT(buf,result,max_size) \e
{ \e
        int c = getchar(); \e
        result = (c == EOF) ? YY_NULL : (buf[0] = c, 1); \e
}
%}
.Ed
.Pp
This definition will change the input processing to occur
one character at a time.
.Pp
When the scanner receives an end-of-file indication from
.Dv YY_INPUT ,
it then checks the
.Fn yywrap
function.
If
.Fn yywrap
returns false
.Pq zero ,
then it is assumed that the function has gone ahead and set up
.Em yyin
to point to another input file, and scanning continues.
If it returns true
.Pq non-zero ,
then the scanner terminates, returning 0 to its caller.
Note that in either case, the start condition remains unchanged;
it does not revert to
.Em INITIAL .
.Pp
If you do not supply your own version of
.Fn yywrap ,
then you must either use
.Dq %option noyywrap
(in which case the scanner behaves as though
.Fn yywrap
returned 1), or you must link with
.Fl lfl
to obtain the default version of the routine, which always returns 1.
.Pp
Three routines are available for scanning from in-memory buffers rather
than files:
.Fn yy_scan_string ,
.Fn yy_scan_bytes ,
and
.Fn yy_scan_buffer .
See the discussion of them below in the section
.Sx MULTIPLE INPUT BUFFERS .
.Pp
The scanner writes its
.Em ECHO
output to the
.Em yyout
global
.Pq default, stdout ,
which may be redefined by the user simply by assigning it to some other
.Va FILE
pointer.
.Sh START CONDITIONS
.Nm
provides a mechanism for conditionally activating rules.
Any rule whose pattern is prefixed with
.Qq Aq sc
will only be active when the scanner is in the start condition named
.Qq sc .
For example,
.Bd -literal -offset indent
<STRING>[^"]* { /* eat up the string body ... */
        ...
}
.Ed
.Pp
will be active only when the scanner is in the
.Qq STRING
start condition, and
.Bd -literal -offset indent
<INITIAL,STRING,QUOTE>\e. { /* handle an escape ... */
        ...
}
.Ed
.Pp
will be active only when the current start condition is either
.Qq INITIAL ,
.Qq STRING ,
or
.Qq QUOTE .
.Pp
Start conditions are declared in the definitions
.Pq first
section of the input using unindented lines beginning with either
.Sq %s
or
.Sq %x
followed by a list of names.
The former declares
.Em inclusive
start conditions, the latter
.Em exclusive
start conditions.
A start condition is activated using the
.Em BEGIN
action.
Until the next
.Em BEGIN
action is executed, rules with the given start condition will be active and
rules with other start conditions will be inactive.
If the start condition is inclusive,
then rules with no start conditions at all will also be active.
If it is exclusive,
then only rules qualified with the start condition will be active.
A set of rules contingent on the same exclusive start condition
describe a scanner which is independent of any of the other rules in the
.Nm
input.
Because of this, exclusive start conditions make it easy to specify
.Qq mini-scanners
which scan portions of the input that are syntactically different
from the rest
.Pq e.g., comments .
.Pp
If the distinction between inclusive and exclusive start conditions
is still a little vague, here's a simple example illustrating the
connection between the two.
The set of rules:
.Bd -literal -offset indent
%s example
%%

<example>foo   do_something();

bar            something_else();
.Ed
.Pp
is equivalent to
.Bd -literal -offset indent
%x example
%%

<example>foo   do_something();

<INITIAL,example>bar    something_else();
.Ed
.Pp
Without the
.Aq INITIAL,example
qualifier, the
.Dq bar
pattern in the second example wouldn't be active
.Pq i.e., couldn't match
when in start condition
.Dq example .
If we just used
.Aq example
to qualify
.Dq bar ,
though, then it would only be active in
.Dq example
and not in
.Em INITIAL ,
while in the first example it's active in both,
because in the first example the
.Dq example
start condition is an inclusive
.Pq Sq %s
start condition.
.Pp
Also note that the special start-condition specifier
.Sq Aq *
matches every start condition.
Thus, the above example could also have been written:
.Bd -literal -offset indent
%x example
%%

<example>foo   do_something();

<*>bar         something_else();
.Ed
.Pp
The default rule (to
.Em ECHO
any unmatched character) remains active in start conditions.
It is equivalent to:
.Bd -literal -offset indent
<*>.|\en     ECHO;
.Ed
.Pp
.Dq BEGIN(0)
returns to the original state where only the rules with
no start conditions are active.
This state can also be referred to as the start-condition
.Em INITIAL ,
so
.Dq BEGIN(INITIAL)
is equivalent to
.Dq BEGIN(0) .
(The parentheses around the start condition name are not required but
are considered good style.)
.Pp
.Em BEGIN
actions can also be given as indented code at the beginning
of the rules section.
For example, the following will cause the scanner to enter the
.Qq SPECIAL
start condition whenever
.Fn yylex
is called and the global variable
.Fa enter_special
is true:
.Bd -literal -offset indent
int enter_special;

%x SPECIAL
%%
        if (enter_special)
                BEGIN(SPECIAL);

<SPECIAL>blahblahblah
\&...more rules follow...
.Ed
.Pp
To illustrate the uses of start conditions,
here is a scanner which provides two different interpretations
of a string like
.Qq 123.456 .
By default it will treat it as three tokens: the integer
.Qq 123 ,
a dot
.Pq Sq .\& ,
and the integer
.Qq 456 .
But if the string is preceded earlier in the line by the string
.Qq expect-floats
it will treat it as a single token, the floating-point number 123.456:
.Bd -literal -offset indent
%{
#include <math.h>
%}
%s expect

%%
expect-floats        BEGIN(expect);

<expect>[0-9]+"."[0-9]+ {
        printf("found a float, = %f\en",
            atof(yytext));
}
<expect>\en {
        /*
         * That's the end of the line, so
         * we need another "expect-number"
         * before we'll recognize any more
         * numbers.
         */
        BEGIN(INITIAL);
}

[0-9]+ {
        printf("found an integer, = %d\en",
            atoi(yytext));
}

"."     printf("found a dot\en");
.Ed
.Pp
Here is a scanner which recognizes
.Pq and discards
C comments while maintaining a count of the current input line:
.Bd -literal -offset indent
%x comment
%%
int line_num = 1;

"/*"                    BEGIN(comment);

<comment>[^*\en]*        /* eat anything that's not a '*' */
<comment>"*"+[^*/\en]*   /* eat up '*'s not followed by '/'s */
<comment>\en             ++line_num;
<comment>"*"+"/"        BEGIN(INITIAL);
.Ed
.Pp
This scanner goes to a bit of trouble to match as much
text as possible with each rule.
In general, when attempting to write a high-speed scanner
try to match as much as possible in each rule, as it's a big win.
.Pp
Note that start-condition names are really integer values and
can be stored as such.
Thus, the above could be extended in the following fashion:
.Bd -literal -offset indent
%x comment foo
%%
int line_num = 1;
int comment_caller;

"/*" {
        comment_caller = INITIAL;
        BEGIN(comment);
}

\&...

<foo>"/*" {
        comment_caller = foo;
        BEGIN(comment);
}

<comment>[^*\en]*        /* eat anything that's not a '*' */
<comment>"*"+[^*/\en]*   /* eat up '*'s not followed by '/'s */
<comment>\en             ++line_num;
<comment>"*"+"/"        BEGIN(comment_caller);
.Ed
.Pp
Furthermore, the current start condition can be accessed by using
the integer-valued
.Dv YY_START
macro.
For example, the above assignments to
.Em comment_caller
could instead be written
.Pp
.Dl comment_caller = YY_START;
.Pp
Flex provides
.Dv YYSTATE
as an alias for
.Dv YY_START
(since that is what's used by AT&T
.Nm lex ) .
.Pp
Note that start conditions do not have their own name-space;
%s's and %x's declare names in the same fashion as #define's.
.Pp
Finally, here's an example of how to match C-style quoted strings using
exclusive start conditions, including expanded escape sequences
(but not including checking for a string that's too long):
.Bd -literal -offset indent
%x str

%%
#define MAX_STR_CONST 1024
char string_buf[MAX_STR_CONST];
char *string_buf_ptr;

\e"      string_buf_ptr = string_buf; BEGIN(str);

<str>\e" { /* saw closing quote - all done */
        BEGIN(INITIAL);
        *string_buf_ptr = '\e0';
        /*
         * return string constant token type and
         * value to parser
         */
}

<str>\en {
        /* error - unterminated string constant */
        /* generate error message */
}

<str>\e\e[0-7]{1,3} {
        /* octal escape sequence */
        int result;

        (void) sscanf(yytext + 1, "%o", &result);

        if (result > 0xff) {
                /* error, constant is out-of-bounds */
	} else
	        *string_buf_ptr++ = result;
}

<str>\e\e[0-9]+ {
        /*
         * generate error - bad escape sequence; something
         * like '\e48' or '\e0777777'
         */
}

<str>\e\en  *string_buf_ptr++ = '\en';
<str>\e\et  *string_buf_ptr++ = '\et';
<str>\e\er  *string_buf_ptr++ = '\er';
<str>\e\eb  *string_buf_ptr++ = '\eb';
<str>\e\ef  *string_buf_ptr++ = '\ef';

<str>\e\e(.|\en)  *string_buf_ptr++ = yytext[1];

<str>[^\e\e\en\e"]+ {
        char *yptr = yytext;

        while (*yptr)
                *string_buf_ptr++ = *yptr++;
}
.Ed
.Pp
Often, such as in some of the examples above,
a whole bunch of rules are all preceded by the same start condition(s).
.Nm
makes this a little easier and cleaner by introducing a notion of
start condition
.Em scope .
A start condition scope is begun with:
.Pp
.Dl <SCs>{
.Pp
where
.Dq SCs
is a list of one or more start conditions.
Inside the start condition scope, every rule automatically has the prefix
.Aq SCs
applied to it, until a
.Sq }
which matches the initial
.Sq { .
So, for example,
.Bd -literal -offset indent
<ESC>{
    "\e\en"   return '\en';
    "\e\er"   return '\er';
    "\e\ef"   return '\ef';
    "\e\e0"   return '\e0';
}
.Ed
.Pp
is equivalent to:
.Bd -literal -offset indent
<ESC>"\e\en"  return '\en';
<ESC>"\e\er"  return '\er';
<ESC>"\e\ef"  return '\ef';
<ESC>"\e\e0"  return '\e0';
.Ed
.Pp
Start condition scopes may be nested.
.Pp
Three routines are available for manipulating stacks of start conditions:
.Bl -tag -width Ds
.It void yy_push_state(int new_state)
Pushes the current start condition onto the top of the start condition
stack and switches to
.Fa new_state
as though
.Dq BEGIN new_state
had been used
.Pq recall that start condition names are also integers .
.It void yy_pop_state()
Pops the top of the stack and switches to it via
.Em BEGIN .
.It int yy_top_state()
Returns the top of the stack without altering the stack's contents.
.El
.Pp
The start condition stack grows dynamically and so has no built-in
size limitation.
If memory is exhausted, program execution aborts.
.Pp
To use start condition stacks, scanners must include a
.Dq %option stack
directive (see
.Sx OPTIONS
below).
.Sh MULTIPLE INPUT BUFFERS
Some scanners
(such as those which support
.Qq include
files)
require reading from several input streams.
As
.Nm
scanners do a large amount of buffering, one cannot control
where the next input will be read from by simply writing a
.Dv YY_INPUT
which is sensitive to the scanning context.
.Dv YY_INPUT
is only called when the scanner reaches the end of its buffer, which
may be a long time after scanning a statement such as an
.Qq include
which requires switching the input source.
.Pp
To negotiate these sorts of problems,
.Nm
provides a mechanism for creating and switching between multiple
input buffers.
An input buffer is created by using:
.Pp
.D1 YY_BUFFER_STATE yy_create_buffer(FILE *file, int size)
.Pp
which takes a
.Fa FILE
pointer and a
.Fa size
and creates a buffer associated with the given file and large enough to hold
.Fa size
characters (when in doubt, use
.Dv YY_BUF_SIZE
for the size).
It returns a
.Dv YY_BUFFER_STATE
handle, which may then be passed to other routines
.Pq see below .
The
.Dv YY_BUFFER_STATE
type is a pointer to an opaque
.Dq struct yy_buffer_state
structure, so
.Dv YY_BUFFER_STATE
variables may be safely initialized to
.Dq ((YY_BUFFER_STATE) 0)
if desired, and the opaque structure can also be referred to in order to
correctly declare input buffers in source files other than that of scanners.
Note that the
.Fa FILE
pointer in the call to
.Fn yy_create_buffer
is only used as the value of
.Fa yyin
seen by
.Dv YY_INPUT ;
if
.Dv YY_INPUT
is redefined so that it no longer uses
.Fa yyin ,
then a nil
.Fa FILE
pointer can safely be passed to
.Fn yy_create_buffer .
To select a particular buffer to scan:
.Pp
.D1 void yy_switch_to_buffer(YY_BUFFER_STATE new_buffer)
.Pp
It switches the scanner's input buffer so subsequent tokens will
come from
.Fa new_buffer .
Note that
.Fn yy_switch_to_buffer
may be used by
.Fn yywrap
to set things up for continued scanning,
instead of opening a new file and pointing
.Fa yyin
at it.
Note also that switching input sources via either
.Fn yy_switch_to_buffer
or
.Fn yywrap
does not change the start condition.
.Pp
.D1 void yy_delete_buffer(YY_BUFFER_STATE buffer)
.Pp
is used to reclaim the storage associated with a buffer.
.Pf ( Fa buffer
can be nil, in which case the routine does nothing.)
To clear the current contents of a buffer:
.Pp
.D1 void yy_flush_buffer(YY_BUFFER_STATE buffer)
.Pp
This function discards the buffer's contents,
so the next time the scanner attempts to match a token from the buffer,
it will first fill the buffer anew using
.Dv YY_INPUT .
.Pp
.Fn yy_new_buffer
is an alias for
.Fn yy_create_buffer ,
provided for compatibility with the C++ use of
.Em new
and
.Em delete
for creating and destroying dynamic objects.
.Pp
Finally, the
.Dv YY_CURRENT_BUFFER
macro returns a
.Dv YY_BUFFER_STATE
handle to the current buffer.
.Pp
Here is an example of using these features for writing a scanner
which expands include files (the
.Aq Aq EOF
feature is discussed below):
.Bd -literal -offset indent
/*
 * the "incl" state is used for picking up the name
 * of an include file
 */
%x incl

%{
#define MAX_INCLUDE_DEPTH 10
YY_BUFFER_STATE include_stack[MAX_INCLUDE_DEPTH];
int include_stack_ptr = 0;
%}

%%
include             BEGIN(incl);

[a-z]+              ECHO;
[^a-z\en]*\en?        ECHO;

<incl>[ \et]*        /* eat the whitespace */
<incl>[^ \et\en]+ {   /* got the include file name */
        if (include_stack_ptr >= MAX_INCLUDE_DEPTH)
                errx(1, "Includes nested too deeply");

        include_stack[include_stack_ptr++] =
            YY_CURRENT_BUFFER;

        yyin = fopen(yytext, "r");

        if (yyin == NULL)
                err(1, NULL);

        yy_switch_to_buffer(
            yy_create_buffer(yyin, YY_BUF_SIZE));

        BEGIN(INITIAL);
}

<<EOF>> {
        if (--include_stack_ptr < 0)
                yyterminate();
        else {
                yy_delete_buffer(YY_CURRENT_BUFFER);
                yy_switch_to_buffer(
                    include_stack[include_stack_ptr]);
       }
}
.Ed
.Pp
Three routines are available for setting up input buffers for
scanning in-memory strings instead of files.
All of them create a new input buffer for scanning the string,
and return a corresponding
.Dv YY_BUFFER_STATE
handle (which should be deleted afterwards using
.Fn yy_delete_buffer ) .
They also switch to the new buffer using
.Fn yy_switch_to_buffer ,
so the next call to
.Fn yylex
will start scanning the string.
.Bl -tag -width Ds
.It yy_scan_string(const char *str)
Scans a NUL-terminated string.
.It yy_scan_bytes(const char *bytes, int len)
Scans
.Fa len
bytes
.Pq including possibly NUL's
starting at location
.Fa bytes .
.El
.Pp
Note that both of these functions create and scan a copy
of the string or bytes.
(This may be desirable, since
.Fn yylex
modifies the contents of the buffer it is scanning.)
The copy can be avoided by using:
.Bl -tag -width Ds
.It yy_scan_buffer(char *base, yy_size_t size)
Which scans the buffer starting at
.Fa base ,
consisting of
.Fa size
bytes, the last two bytes of which must be
.Dv YY_END_OF_BUFFER_CHAR
.Pq ASCII NUL .
These last two bytes are not scanned; thus, scanning consists of
base[0] through base[size-2], inclusive.
.Pp
If
.Fa base
is not set up in this manner
(i.e., forget the final two
.Dv YY_END_OF_BUFFER_CHAR
bytes), then
.Fn yy_scan_buffer
returns a nil pointer instead of creating a new input buffer.
.Pp
The type
.Fa yy_size_t
is an integral type which can be cast to an integer expression
reflecting the size of the buffer.
.El
.Sh END-OF-FILE RULES
The special rule
.Qq Aq Aq EOF
indicates actions which are to be taken when an end-of-file is encountered and
.Fn yywrap
returns non-zero
.Pq i.e., indicates no further files to process .
The action must finish by doing one of four things:
.Bl -dash
.It
Assigning
.Em yyin
to a new input file
(in previous versions of
.Nm ,
after doing the assignment, it was necessary to call the special action
.Dv YY_NEW_FILE ;
this is no longer necessary).
.It
Executing a
.Em return
statement.
.It
Executing the special
.Fn yyterminate
action.
.It
Switching to a new buffer using
.Fn yy_switch_to_buffer
as shown in the example above.
.El
.Pp
.Aq Aq EOF
rules may not be used with other patterns;
they may only be qualified with a list of start conditions.
If an unqualified
.Aq Aq EOF
rule is given, it applies to all start conditions which do not already have
.Aq Aq EOF
actions.
To specify an
.Aq Aq EOF
rule for only the initial start condition, use
.Pp
.Dl <INITIAL><<EOF>>
.Pp
These rules are useful for catching things like unclosed comments.
An example:
.Bd -literal -offset indent
%x quote
%%

\&...other rules for dealing with quotes...

<quote><<EOF>> {
         error("unterminated quote");
         yyterminate();
}
<<EOF>> {
         if (*++filelist)
                 yyin = fopen(*filelist, "r");
         else
                 yyterminate();
}
.Ed
.Sh MISCELLANEOUS MACROS
The macro
.Dv YY_USER_ACTION
can be defined to provide an action
which is always executed prior to the matched rule's action.
For example,
it could be #define'd to call a routine to convert yytext to lower-case.
When
.Dv YY_USER_ACTION
is invoked, the variable
.Fa yy_act
gives the number of the matched rule
.Pq rules are numbered starting with 1 .
For example, to profile how often each rule is matched,
the following would do the trick:
.Pp
.Dl #define YY_USER_ACTION ++ctr[yy_act]
.Pp
where
.Fa ctr
is an array to hold the counts for the different rules.
Note that the macro
.Dv YY_NUM_RULES
gives the total number of rules
(including the default rule, even if
.Fl s
is used),
so a correct declaration for
.Fa ctr
is:
.Pp
.Dl int ctr[YY_NUM_RULES];
.Pp
The macro
.Dv YY_USER_INIT
may be defined to provide an action which is always executed before
the first scan
.Pq and before the scanner's internal initializations are done .
For example, it could be used to call a routine to read
in a data table or open a logging file.
.Pp
The macro
.Dv yy_set_interactive(is_interactive)
can be used to control whether the current buffer is considered
.Em interactive .
An interactive buffer is processed more slowly,
but must be used when the scanner's input source is indeed
interactive to avoid problems due to waiting to fill buffers
(see the discussion of the
.Fl I
flag below).
A non-zero value in the macro invocation marks the buffer as interactive,
a zero value as non-interactive.
Note that use of this macro overrides
.Dq %option always-interactive
or
.Dq %option never-interactive
(see
.Sx OPTIONS
below).
.Fn yy_set_interactive
must be invoked prior to beginning to scan the buffer that is
.Pq or is not
to be considered interactive.
.Pp
The macro
.Dv yy_set_bol(at_bol)
can be used to control whether the current buffer's scanning
context for the next token match is done as though at the
beginning of a line.
A non-zero macro argument makes rules anchored with
.Sq ^
active, while a zero argument makes
.Sq ^
rules inactive.
.Pp
The macro
.Dv YY_AT_BOL
returns true if the next token scanned from the current buffer will have
.Sq ^
rules active, false otherwise.
.Pp
In the generated scanner, the actions are all gathered in one large
switch statement and separated using
.Dv YY_BREAK ,
which may be redefined.
By default, it is simply a
.Qq break ,
to separate each rule's action from the following rules.
Redefining
.Dv YY_BREAK
allows, for example, C++ users to
.Dq #define YY_BREAK
to do nothing
(while being very careful that every rule ends with a
.Qq break
or a
.Qq return ! )
to avoid suffering from unreachable statement warnings where because a rule's
action ends with
.Dq return ,
the
.Dv YY_BREAK
is inaccessible.
.Sh VALUES AVAILABLE TO THE USER
This section summarizes the various values available to the user
in the rule actions.
.Bl -tag -width Ds
.It char *yytext
Holds the text of the current token.
It may be modified but not lengthened
.Pq characters cannot be appended to the end .
.Pp
If the special directive
.Dq %array
appears in the first section of the scanner description, then
.Fa yytext
is instead declared
.Dq char yytext[YYLMAX] ,
where
.Dv YYLMAX
is a macro definition that can be redefined in the first section
to change the default value
.Pq generally 8KB .
Using
.Dq %array
results in somewhat slower scanners, but the value of
.Fa yytext
becomes immune to calls to
.Fn input
and
.Fn unput ,
which potentially destroy its value when
.Fa yytext
is a character pointer.
The opposite of
.Dq %array
is
.Dq %pointer ,
which is the default.
.Pp
.Dq %array
cannot be used when generating C++ scanner classes
(the
.Fl +
flag).
.It int yyleng
Holds the length of the current token.
.It FILE *yyin
Is the file which by default
.Nm
reads from.
It may be redefined, but doing so only makes sense before
scanning begins or after an
.Dv EOF
has been encountered.
Changing it in the midst of scanning will have unexpected results since
.Nm
buffers its input; use
.Fn yyrestart
instead.
Once scanning terminates because an end-of-file
has been seen,
.Fa yyin
can be assigned as the new input file
and the scanner can be called again to continue scanning.
.It void yyrestart(FILE *new_file)
May be called to point
.Fa yyin
at the new input file.
The switch-over to the new file is immediate
.Pq any previously buffered-up input is lost .
Note that calling
.Fn yyrestart
with
.Fa yyin
as an argument thus throws away the current input buffer and continues
scanning the same input file.
.It FILE *yyout
Is the file to which
.Em ECHO
actions are done.
It can be reassigned by the user.
.It YY_CURRENT_BUFFER
Returns a
.Dv YY_BUFFER_STATE
handle to the current buffer.
.It YY_START
Returns an integer value corresponding to the current start condition.
This value can subsequently be used with
.Em BEGIN
to return to that start condition.
.El
.Sh INTERFACING WITH YACC
One of the main uses of
.Nm
is as a companion to the
.Xr yacc 1
parser-generator.
yacc parsers expect to call a routine named
.Fn yylex
to find the next input token.
The routine is supposed to return the type of the next token
as well as putting any associated value in the global
.Fa yylval ,
which is defined externally,
and can be a union or any other complex data structure.
To use
.Nm
with yacc, one specifies the
.Fl d
option to yacc to instruct it to generate the file
.Pa y.tab.h
containing definitions of all the
.Dq %tokens
appearing in the yacc input.
This file is then included in the
.Nm
scanner.
For example, if one of the tokens is
.Qq TOK_NUMBER ,
part of the scanner might look like:
.Bd -literal -offset indent
%{
#include "y.tab.h"
%}

%%

[0-9]+        yylval = atoi(yytext); return TOK_NUMBER;
.Ed
.Sh OPTIONS
.Nm
has the following options:
.Bl -tag -width Ds
.It Fl 7
Instructs
.Nm
to generate a 7-bit scanner, i.e., one which can only recognize 7-bit
characters in its input.
The advantage of using
.Fl 7
is that the scanner's tables can be up to half the size of those generated
using the
.Fl 8
option
.Pq see below .
The disadvantage is that such scanners often hang
or crash if their input contains an 8-bit character.
.Pp
Note, however, that unless generating a scanner using the
.Fl Cf
or
.Fl CF
table compression options, use of
.Fl 7
will save only a small amount of table space,
and make the scanner considerably less portable.
.Nm flex Ns 's
default behavior is to generate an 8-bit scanner unless
.Fl Cf
or
.Fl CF
is specified, in which case
.Nm
defaults to generating 7-bit scanners unless it was
configured to generate 8-bit scanners
(as will often be the case with non-USA sites).
It is possible tell whether
.Nm
generated a 7-bit or an 8-bit scanner by inspecting the flag summary in the
.Fl v
output as described below.
.Pp
Note that if
.Fl Cfe
or
.Fl CFe
are used
(the table compression options, but also using equivalence classes as
discussed below),
.Nm
still defaults to generating an 8-bit scanner,
since usually with these compression options full 8-bit tables
are not much more expensive than 7-bit tables.
.It Fl 8
Instructs
.Nm
to generate an 8-bit scanner, i.e., one which can recognize 8-bit
characters.
This flag is only needed for scanners generated using
.Fl Cf
or
.Fl CF ,
as otherwise
.Nm
defaults to generating an 8-bit scanner anyway.
.Pp
See the discussion of
.Fl 7
above for
.Nm flex Ns 's
default behavior and the tradeoffs between 7-bit and 8-bit scanners.
.It Fl B
Instructs
.Nm
to generate a
.Em batch
scanner, the opposite of
.Em interactive
scanners generated by
.Fl I
.Pq see below .
In general,
.Fl B
is used when the scanner will never be used interactively,
and you want to squeeze a little more performance out of it.
If the aim is instead to squeeze out a lot more performance,
use the
.Fl Cf
or
.Fl CF
options
.Pq discussed below ,
which turn on
.Fl B
automatically anyway.
.It Fl b
Generate backing-up information to
.Pa lex.backup .
This is a list of scanner states which require backing up
and the input characters on which they do so.
By adding rules one can remove backing-up states.
If all backing-up states are eliminated and
.Fl Cf
or
.Fl CF
is used, the generated scanner will run faster (see the
.Fl p
flag).
Only users who wish to squeeze every last cycle out of their
scanners need worry about this option.
(See the section on
.Sx PERFORMANCE CONSIDERATIONS
below.)
.It Fl C Ns Op Cm aeFfmr
Controls the degree of table compression and, more generally, trade-offs
between small scanners and fast scanners.
.Bl -tag -width Ds
.It Fl Ca
Instructs
.Nm
to trade off larger tables in the generated scanner for faster performance
because the elements of the tables are better aligned for memory access
and computation.
On some
.Tn RISC
architectures, fetching and manipulating longwords is more efficient
than with smaller-sized units such as shortwords.
This option can double the size of the tables used by the scanner.
.It Fl Ce
Directs
.Nm
to construct
.Em equivalence classes ,
i.e., sets of characters which have identical lexical properties
(for example, if the only appearance of digits in the
.Nm
input is in the character class
.Qq [0-9]
then the digits
.Sq 0 ,
.Sq 1 ,
.Sq ... ,
.Sq 9
will all be put in the same equivalence class).
Equivalence classes usually give dramatic reductions in the final
table/object file sizes
.Pq typically a factor of 2\-5
and are pretty cheap performance-wise
.Pq one array look-up per character scanned .
.It Fl CF
Specifies that the alternate fast scanner representation
(described below under the
.Fl F
option)
should be used.
This option cannot be used with
.Fl + .
.It Fl Cf
Specifies that the
.Em full
scanner tables should be generated \-
.Nm
should not compress the tables by taking advantage of
similar transition functions for different states.
.It Fl \&Cm
Directs
.Nm
to construct
.Em meta-equivalence classes ,
which are sets of equivalence classes
(or characters, if equivalence classes are not being used)
that are commonly used together.
Meta-equivalence classes are often a big win when using compressed tables,
but they have a moderate performance impact
(one or two
.Qq if
tests and one array look-up per character scanned).
.It Fl Cr
Causes the generated scanner to
.Em bypass
use of the standard I/O library
.Pq stdio
for input.
Instead of calling
.Xr fread 3
or
.Xr getc 3 ,
the scanner will use the
.Xr read 2
system call,
resulting in a performance gain which varies from system to system,
but in general is probably negligible unless
.Fl Cf
or
.Fl CF
are being used.
Using
.Fl Cr
can cause strange behavior if, for example, reading from
.Fa yyin
using stdio prior to calling the scanner
(because the scanner will miss whatever text previous reads left
in the stdio input buffer).
.Pp
.Fl Cr
has no effect if
.Dv YY_INPUT
is defined
(see
.Sx THE GENERATED SCANNER
above).
.El
.Pp
A lone
.Fl C
specifies that the scanner tables should be compressed but neither
equivalence classes nor meta-equivalence classes should be used.
.Pp
The options
.Fl Cf
or
.Fl CF
and
.Fl \&Cm
do not make sense together \- there is no opportunity for meta-equivalence
classes if the table is not being compressed.
Otherwise the options may be freely mixed, and are cumulative.
.Pp
The default setting is
.Fl Cem
which specifies that
.Nm
should generate equivalence classes and meta-equivalence classes.
This setting provides the highest degree of table compression.
It is possible to trade off faster-executing scanners at the cost of
larger tables with the following generally being true:
.Bd -unfilled -offset indent
slowest & smallest
      -Cem
      -Cm
      -Ce
      -C
      -C{f,F}e
      -C{f,F}
      -C{f,F}a
fastest & largest
.Ed
.Pp
Note that scanners with the smallest tables are usually generated and
compiled the quickest,
so during development the default is usually best,
maximal compression.
.Pp
.Fl Cfe
is often a good compromise between speed and size for production scanners.
.It Fl c
A do-nothing, deprecated option included for
.Tn POSIX
compliance.
.It Fl d
Makes the generated scanner run in debug mode.
Whenever a pattern is recognized and the global
.Fa yy_flex_debug
is non-zero
.Pq which is the default ,
the scanner will write to stderr a line of the form:
.Pp
.D1 --accepting rule at line 53 ("the matched text")
.Pp
The line number refers to the location of the rule in the file
defining the scanner
(i.e., the file that was fed to
.Nm ) .
Messages are also generated when the scanner backs up,
accepts the default rule,
reaches the end of its input buffer
(or encounters a NUL;
at this point, the two look the same as far as the scanner's concerned),
or reaches an end-of-file.
.It Fl F
Specifies that the fast scanner table representation should be used
.Pq and stdio bypassed .
This representation is about as fast as the full table representation
.Pq Fl f ,
and for some sets of patterns will be considerably smaller
.Pq and for others, larger .
In general, if the pattern set contains both
.Qq keywords
and a catch-all,
.Qq identifier
rule, such as in the set:
.Bd -unfilled -offset indent
"case"    return TOK_CASE;
"switch"  return TOK_SWITCH;
\&...
"default" return TOK_DEFAULT;
[a-z]+    return TOK_ID;
.Ed
.Pp
then it's better to use the full table representation.
If only the
.Qq identifier
rule is present and a hash table or some such is used to detect the keywords,
it's better to use
.Fl F .
.Pp
This option is equivalent to
.Fl CFr
.Pq see above .
It cannot be used with
.Fl + .
.It Fl f
Specifies
.Em fast scanner .
No table compression is done and stdio is bypassed.
The result is large but fast.
This option is equivalent to
.Fl Cfr
.Pq see above .
.It Fl h
Generates a help summary of
.Nm flex Ns 's
options to stdout and then exits.
.Fl ?\&
and
.Fl Fl help
are synonyms for
.Fl h .
.It Fl I
Instructs
.Nm
to generate an
.Em interactive
scanner.
An interactive scanner is one that only looks ahead to decide
what token has been matched if it absolutely must.
It turns out that always looking one extra character ahead,
even if the scanner has already seen enough text
to disambiguate the current token, is a bit faster than
only looking ahead when necessary.
But scanners that always look ahead give dreadful interactive performance;
for example, when a user types a newline,
it is not recognized as a newline token until they enter
.Em another
token, which often means typing in another whole line.
.Pp
.Nm
scanners default to
.Em interactive
unless
.Fl Cf
or
.Fl CF
table-compression options are specified
.Pq see above .
That's because if high-performance is most important,
one of these options should be used,
so if they weren't,
.Nm
assumes it is preferable to trade off a bit of run-time performance for
intuitive interactive behavior.
Note also that
.Fl I
cannot be used in conjunction with
.Fl Cf
or
.Fl CF .
Thus, this option is not really needed; it is on by default for all those
cases in which it is allowed.
.Pp
A scanner can be forced to not be interactive by using
.Fl B
.Pq see above .
.It Fl i
Instructs
.Nm
to generate a case-insensitive scanner.
The case of letters given in the
.Nm
input patterns will be ignored,
and tokens in the input will be matched regardless of case.
The matched text given in
.Fa yytext
will have the preserved case
.Pq i.e., it will not be folded .
.It Fl L
Instructs
.Nm
not to generate
.Dq #line
directives.
Without this option,
.Nm
peppers the generated scanner with #line directives so error messages
in the actions will be correctly located with respect to either the original
.Nm
input file
(if the errors are due to code in the input file),
or
.Pa lex.yy.c
(if the errors are
.Nm flex Ns 's
fault \- these sorts of errors should be reported to the email address
given below).
.It Fl l
Turns on maximum compatibility with the original AT&T
.Nm lex
implementation.
Note that this does not mean full compatibility.
Use of this option costs a considerable amount of performance,
and it cannot be used with the
.Fl + , f , F , Cf ,
or
.Fl CF
options.
For details on the compatibilities it provides, see the section
.Sx INCOMPATIBILITIES WITH LEX AND POSIX
below.
This option also results in the name
.Dv YY_FLEX_LEX_COMPAT
being #define'd in the generated scanner.
.It Fl n
Another do-nothing, deprecated option included only for
.Tn POSIX
compliance.
.It Fl o Ns Ar output
Directs
.Nm
to write the scanner to the file
.Ar output
instead of
.Pa lex.yy.c .
If
.Fl o
is combined with the
.Fl t
option, then the scanner is written to stdout but its
.Dq #line
directives
(see the
.Fl L
option above)
refer to the file
.Ar output .
.It Fl P Ns Ar prefix
Changes the default
.Qq yy
prefix used by
.Nm
for all globally visible variable and function names to instead be
.Ar prefix .
For example,
.Fl P Ns Ar foo
changes the name of
.Fa yytext
to
.Fa footext .
It also changes the name of the default output file from
.Pa lex.yy.c
to
.Pa lex.foo.c .
Here are all of the names affected:
.Bd -unfilled -offset indent
yy_create_buffer
yy_delete_buffer
yy_flex_debug
yy_init_buffer
yy_flush_buffer
yy_load_buffer_state
yy_switch_to_buffer
yyin
yyleng
yylex
yylineno
yyout
yyrestart
yytext
yywrap
.Ed
.Pp
(If using a C++ scanner, then only
.Fa yywrap
and
.Fa yyFlexLexer
are affected.)
Within the scanner itself, it is still possible to refer to the global variables
and functions using either version of their name; but externally, they
have the modified name.
.Pp
This option allows multiple
.Nm
programs to be easily linked together into the same executable.
Note, though, that using this option also renames
.Fn yywrap ,
so now either an
.Pq appropriately named
version of the routine for the scanner must be supplied, or
.Dq %option noyywrap
must be used, as linking with
.Fl lfl
no longer provides one by default.
.It Fl p
Generates a performance report to stderr.
The report consists of comments regarding features of the
.Nm
input file which will cause a serious loss of performance in the resulting
scanner.
If the flag is specified twice,
comments regarding features that lead to minor performance losses
will also be reported>
.Pp
Note that the use of
.Em REJECT ,
.Dq %option yylineno ,
and variable trailing context
(see the
.Sx BUGS
section below)
entails a substantial performance penalty; use of
.Fn yymore ,
the
.Sq ^
operator, and the
.Fl I
flag entail minor performance penalties.
.It Fl S Ns Ar skeleton
Overrides the default skeleton file from which
.Nm
constructs its scanners.
This option is needed only for
.Nm
maintenance or development.
.It Fl s
Causes the default rule
.Pq that unmatched scanner input is echoed to stdout
to be suppressed.
If the scanner encounters input that does not
match any of its rules, it aborts with an error.
This option is useful for finding holes in a scanner's rule set.
.It Fl T
Makes
.Nm
run in
.Em trace
mode.
It will generate a lot of messages to stderr concerning
the form of the input and the resultant non-deterministic and deterministic
finite automata.
This option is mostly for use in maintaining
.Nm .
.It Fl t
Instructs
.Nm
to write the scanner it generates to standard output instead of
.Pa lex.yy.c .
.It Fl V
Prints the version number to stdout and exits.
.Fl Fl version
is a synonym for
.Fl V .
.It Fl v
Specifies that
.Nm
should write to stderr
a summary of statistics regarding the scanner it generates.
Most of the statistics are meaningless to the casual
.Nm
user, but the first line identifies the version of
.Nm
(same as reported by
.Fl V ) ,
and the next line the flags used when generating the scanner,
including those that are on by default.
.It Fl w
Suppresses warning messages.
.It Fl +
Specifies that
.Nm
should generate a C++ scanner class.
See the section on
.Sx GENERATING C++ SCANNERS
below for details.
.El
.Pp
.Nm
also provides a mechanism for controlling options within the
scanner specification itself, rather than from the
.Nm
command-line.
This is done by including
.Dq %option
directives in the first section of the scanner specification.
Multiple options can be specified with a single
.Dq %option
directive, and multiple directives in the first section of the
.Nm
input file.
.Pp
Most options are given simply as names, optionally preceded by the word
.Qq no
.Pq with no intervening whitespace
to negate their meaning.
A number are equivalent to
.Nm
flags or their negation:
.Bd -unfilled -offset indent
7bit            -7 option
8bit            -8 option
align           -Ca option
backup          -b option
batch           -B option
c++             -+ option

caseful or
case-sensitive  opposite of -i (default)

case-insensitive or
caseless        -i option

debug           -d option
default         opposite of -s option
ecs             -Ce option
fast            -F option
full            -f option
interactive     -I option
lex-compat      -l option
meta-ecs        -Cm option
perf-report     -p option
read            -Cr option
stdout          -t option
verbose         -v option
warn            opposite of -w option
                (use "%option nowarn" for -w)

array           equivalent to "%array"
pointer         equivalent to "%pointer" (default)
.Ed
.Pp
Some %option's provide features otherwise not available:
.Bl -tag -width Ds
.It always-interactive
Instructs
.Nm
to generate a scanner which always considers its input
.Qq interactive .
Normally, on each new input file the scanner calls
.Fn isatty
in an attempt to determine whether the scanner's input source is interactive
and thus should be read a character at a time.
When this option is used, however, no such call is made.
.It main
Directs
.Nm
to provide a default
.Fn main
program for the scanner, which simply calls
.Fn yylex .
This option implies
.Dq noyywrap
.Pq see below .
.It never-interactive
Instructs
.Nm
to generate a scanner which never considers its input
.Qq interactive
(again, no call made to
.Fn isatty ) .
This is the opposite of
.Dq always-interactive .
.It stack
Enables the use of start condition stacks
(see
.Sx START CONDITIONS
above).
.It stdinit
If set (i.e.,
.Dq %option stdinit ) ,
initializes
.Fa yyin
and
.Fa yyout
to stdin and stdout, instead of the default of
.Dq nil .
Some existing
.Nm lex
programs depend on this behavior, even though it is not compliant with ANSI C,
which does not require stdin and stdout to be compile-time constant.
.It yylineno
Directs
.Nm
to generate a scanner that maintains the number of the current line
read from its input in the global variable
.Fa yylineno .
This option is implied by
.Dq %option lex-compat .
.It yywrap
If unset (i.e.,
.Dq %option noyywrap ) ,
makes the scanner not call
.Fn yywrap
upon an end-of-file, but simply assume that there are no more files to scan
(until the user points
.Fa yyin
at a new file and calls
.Fn yylex
again).
.El
.Pp
.Nm
scans rule actions to determine whether the
.Em REJECT
or
.Fn yymore
features are being used.
The
.Dq reject
and
.Dq yymore
options are available to override its decision as to whether to use the
options, either by setting them (e.g.,
.Dq %option reject )
to indicate the feature is indeed used,
or unsetting them to indicate it actually is not used
(e.g.,
.Dq %option noyymore ) .
.Pp
Three options take string-delimited values, offset with
.Sq = :
.Pp
.D1 %option outfile="ABC"
.Pp
is equivalent to
.Fl o Ns Ar ABC ,
and
.Pp
.D1 %option prefix="XYZ"
.Pp
is equivalent to
.Fl P Ns Ar XYZ .
Finally,
.Pp
.D1 %option yyclass="foo"
.Pp
only applies when generating a C++ scanner
.Pf ( Fl +
option).
It informs
.Nm
that
.Dq foo
has been derived as a subclass of yyFlexLexer, so
.Nm
will place actions in the member function
.Dq foo::yylex()
instead of
.Dq yyFlexLexer::yylex() .
It also generates a
.Dq yyFlexLexer::yylex()
member function that emits a run-time error (by invoking
.Dq yyFlexLexer::LexerError() )
if called.
See
.Sx GENERATING C++ SCANNERS ,
below, for additional information.
.Pp
A number of options are available for
.Xr lint 1
purists who want to suppress the appearance of unneeded routines
in the generated scanner.
Each of the following, if unset
(e.g.,
.Dq %option nounput ) ,
results in the corresponding routine not appearing in the generated scanner:
.Bd -unfilled -offset indent
input, unput
yy_push_state, yy_pop_state, yy_top_state
yy_scan_buffer, yy_scan_bytes, yy_scan_string
.Ed
.Pp
(though
.Fn yy_push_state
and friends won't appear anyway unless
.Dq %option stack
is being used).
.Sh PERFORMANCE CONSIDERATIONS
The main design goal of
.Nm
is that it generate high-performance scanners.
It has been optimized for dealing well with large sets of rules.
Aside from the effects on scanner speed of the table compression
.Fl C
options outlined above,
there are a number of options/actions which degrade performance.
These are, from most expensive to least:
.Bd -unfilled -offset indent
REJECT
%option yylineno
arbitrary trailing context

pattern sets that require backing up
%array
%option interactive
%option always-interactive

\&'^' beginning-of-line operator
yymore()
.Ed
.Pp
with the first three all being quite expensive
and the last two being quite cheap.
Note also that
.Fn unput
is implemented as a routine call that potentially does quite a bit of work,
while
.Fn yyless
is a quite-cheap macro; so if just putting back some excess text,
use
.Fn yyless .
.Pp
.Em REJECT
should be avoided at all costs when performance is important.
It is a particularly expensive option.
.Pp
Getting rid of backing up is messy and often may be an enormous
amount of work for a complicated scanner.
In principal, one begins by using the
.Fl b
flag to generate a
.Pa lex.backup
file.
For example, on the input
.Bd -literal -offset indent
%%
foo        return TOK_KEYWORD;
foobar     return TOK_KEYWORD;
.Ed
.Pp
the file looks like:
.Bd -literal -offset indent
State #6 is non-accepting -
 associated rule line numbers:
       2       3
 out-transitions: [ o ]
 jam-transitions: EOF [ \e001-n  p-\e177 ]

State #8 is non-accepting -
 associated rule line numbers:
       3
 out-transitions: [ a ]
 jam-transitions: EOF [ \e001-`  b-\e177 ]

State #9 is non-accepting -
 associated rule line numbers:
       3
 out-transitions: [ r ]
 jam-transitions: EOF [ \e001-q  s-\e177 ]

Compressed tables always back up.
.Ed
.Pp
The first few lines tell us that there's a scanner state in
which it can make a transition on an
.Sq o
but not on any other character,
and that in that state the currently scanned text does not match any rule.
The state occurs when trying to match the rules found
at lines 2 and 3 in the input file.
If the scanner is in that state and then reads something other than an
.Sq o ,
it will have to back up to find a rule which is matched.
With a bit of headscratching one can see that this must be the
state it's in when it has seen
.Sq fo .
When this has happened, if anything other than another
.Sq o
is seen, the scanner will have to back up to simply match the
.Sq f
.Pq by the default rule .
.Pp
The comment regarding State #8 indicates there's a problem when
.Qq foob
has been scanned.
Indeed, on any character other than an
.Sq a ,
the scanner will have to back up to accept
.Qq foo .
Similarly, the comment for State #9 concerns when
.Qq fooba
has been scanned and an
.Sq r
does not follow.
.Pp
The final comment reminds us that there's no point going to
all the trouble of removing backing up from the rules unless we're using
.Fl Cf
or
.Fl CF ,
since there's no performance gain doing so with compressed scanners.
.Pp
The way to remove the backing up is to add
.Qq error
rules:
.Bd -literal -offset indent
%%
foo    return TOK_KEYWORD;
foobar return TOK_KEYWORD;

fooba  |
foob   |
fo {
        /* false alarm, not really a keyword */
        return TOK_ID;
}
.Ed
.Pp
Eliminating backing up among a list of keywords can also be done using a
.Qq catch-all
rule:
.Bd -literal -offset indent
%%
foo    return TOK_KEYWORD;
foobar return TOK_KEYWORD;

[a-z]+ return TOK_ID;
.Ed
.Pp
This is usually the best solution when appropriate.
.Pp
Backing up messages tend to cascade.
With a complicated set of rules it's not uncommon to get hundreds of messages.
If one can decipher them, though,
it often only takes a dozen or so rules to eliminate the backing up
(though it's easy to make a mistake and have an error rule accidentally match
a valid token; a possible future
.Nm
feature will be to automatically add rules to eliminate backing up).
.Pp
It's important to keep in mind that the benefits of eliminating
backing up are gained only if
.Em every
instance of backing up is eliminated.
Leaving just one gains nothing.
.Pp
.Em Variable
trailing context
(where both the leading and trailing parts do not have a fixed length)
entails almost the same performance loss as
.Em REJECT
.Pq i.e., substantial .
So when possible a rule like:
.Bd -literal -offset indent
%%
mouse|rat/(cat|dog)   run();
.Ed
.Pp
is better written:
.Bd -literal -offset indent
%%
mouse/cat|dog         run();
rat/cat|dog           run();
.Ed
.Pp
or as
.Bd -literal -offset indent
%%
mouse|rat/cat         run();
mouse|rat/dog         run();
.Ed
.Pp
Note that here the special
.Sq |\&
action does not provide any savings, and can even make things worse (see
.Sx BUGS
below).
.Pp
Another area where the user can increase a scanner's performance
.Pq and one that's easier to implement
arises from the fact that the longer the tokens matched,
the faster the scanner will run.
This is because with long tokens the processing of most input
characters takes place in the
.Pq short
inner scanning loop, and does not often have to go through the additional work
of setting up the scanning environment (e.g.,
.Fa yytext )
for the action.
Recall the scanner for C comments:
.Bd -literal -offset indent
%x comment
%%
int line_num = 1;

"/*"                    BEGIN(comment);

<comment>[^*\en]*
<comment>"*"+[^*/\en]*
<comment>\en             ++line_num;
<comment>"*"+"/"        BEGIN(INITIAL);
.Ed
.Pp
This could be sped up by writing it as:
.Bd -literal -offset indent
%x comment
%%
int line_num = 1;

"/*"                    BEGIN(comment);

<comment>[^*\en]*
<comment>[^*\en]*\en      ++line_num;
<comment>"*"+[^*/\en]*
<comment>"*"+[^*/\en]*\en ++line_num;
<comment>"*"+"/"        BEGIN(INITIAL);
.Ed
.Pp
Now instead of each newline requiring the processing of another action,
recognizing the newlines is
.Qq distributed
over the other rules to keep the matched text as long as possible.
Note that adding rules does
.Em not
slow down the scanner!
The speed of the scanner is independent of the number of rules or
(modulo the considerations given at the beginning of this section)
how complicated the rules are with regard to operators such as
.Sq *
and
.Sq |\& .
.Pp
A final example in speeding up a scanner:
scan through a file containing identifiers and keywords, one per line
and with no other extraneous characters, and recognize all the keywords.
A natural first approach is:
.Bd -literal -offset indent
%%
asm      |
auto     |
break    |
\&... etc ...
volatile |
while    /* it's a keyword */

\&.|\en     /* it's not a keyword */
.Ed
.Pp
To eliminate the back-tracking, introduce a catch-all rule:
.Bd -literal -offset indent
%%
asm      |
auto     |
break    |
\&... etc ...
volatile |
while    /* it's a keyword */

[a-z]+   |
\&.|\en     /* it's not a keyword */
.Ed
.Pp
Now, if it's guaranteed that there's exactly one word per line,
then we can reduce the total number of matches by a half by
merging in the recognition of newlines with that of the other tokens:
.Bd -literal -offset indent
%%
asm\en      |
auto\en     |
break\en    |
\&... etc ...
volatile\en |
while\en    /* it's a keyword */

[a-z]+\en   |
\&.|\en       /* it's not a keyword */
.Ed
.Pp
One has to be careful here,
as we have now reintroduced backing up into the scanner.
In particular, while we know that there will never be any characters
in the input stream other than letters or newlines,
.Nm
can't figure this out, and it will plan for possibly needing to back up
when it has scanned a token like
.Qq auto
and then the next character is something other than a newline or a letter.
Previously it would then just match the
.Qq auto
rule and be done, but now it has no
.Qq auto
rule, only an
.Qq auto\en
rule.
To eliminate the possibility of backing up,
we could either duplicate all rules but without final newlines, or,
since we never expect to encounter such an input and therefore don't
how it's classified, we can introduce one more catch-all rule,
this one which doesn't include a newline:
.Bd -literal -offset indent
%%
asm\en      |
auto\en     |
break\en    |
\&... etc ...
volatile\en |
while\en    /* it's a keyword */

[a-z]+\en   |
[a-z]+     |
\&.|\en       /* it's not a keyword */
.Ed
.Pp
Compiled with
.Fl Cf ,
this is about as fast as one can get a
.Nm
scanner to go for this particular problem.
.Pp
A final note:
.Nm
is slow when matching NUL's,
particularly when a token contains multiple NUL's.
It's best to write rules which match short
amounts of text if it's anticipated that the text will often include NUL's.
.Pp
Another final note regarding performance: as mentioned above in the section
.Sx HOW THE INPUT IS MATCHED ,
dynamically resizing
.Fa yytext
to accommodate huge tokens is a slow process because it presently requires that
the
.Pq huge
token be rescanned from the beginning.
Thus if performance is vital, it is better to attempt to match
.Qq large
quantities of text but not
.Qq huge
quantities, where the cutoff between the two is at about 8K characters/token.
.Sh GENERATING C++ SCANNERS
.Nm
provides two different ways to generate scanners for use with C++.
The first way is to simply compile a scanner generated by
.Nm
using a C++ compiler instead of a C compiler.
This should not generate any compilation errors
(please report any found to the email address given in the
.Sx AUTHORS
section below).
C++ code can then be used in rule actions instead of C code.
Note that the default input source for scanners remains
.Fa yyin ,
and default echoing is still done to
.Fa yyout .
Both of these remain
.Fa FILE *
variables and not C++ streams.
.Pp
.Nm
can also be used to generate a C++ scanner class, using the
.Fl +
option (or, equivalently,
.Dq %option c++ ) ,
which is automatically specified if the name of the flex executable ends in a
.Sq + ,
such as
.Nm flex++ .
When using this option,
.Nm
defaults to generating the scanner to the file
.Pa lex.yy.cc
instead of
.Pa lex.yy.c .
The generated scanner includes the header file
.Aq Pa g++/FlexLexer.h ,
which defines the interface to two C++ classes.
.Pp
The first class,
.Em FlexLexer ,
provides an abstract base class defining the general scanner class interface.
It provides the following member functions:
.Bl -tag -width Ds
.It const char* YYText()
Returns the text of the most recently matched token, the equivalent of
.Fa yytext .
.It int YYLeng()
Returns the length of the most recently matched token, the equivalent of
.Fa yyleng .
.It int lineno() const
Returns the current input line number
(see
.Dq %option yylineno ) ,
or 1 if
.Dq %option yylineno
was not used.
.It void set_debug(int flag)
Sets the debugging flag for the scanner, equivalent to assigning to
.Fa yy_flex_debug
(see the
.Sx OPTIONS
section above).
Note that the scanner must be built using
.Dq %option debug
to include debugging information in it.
.It int debug() const
Returns the current setting of the debugging flag.
.El
.Pp
Also provided are member functions equivalent to
.Fn yy_switch_to_buffer ,
.Fn yy_create_buffer
(though the first argument is an
.Fa std::istream*
object pointer and not a
.Fa FILE* ) ,
.Fn yy_flush_buffer ,
.Fn yy_delete_buffer ,
and
.Fn yyrestart
(again, the first argument is an
.Fa std::istream*
object pointer).
.Pp
The second class defined in
.Aq Pa g++/FlexLexer.h
is
.Fa yyFlexLexer ,
which is derived from
.Fa FlexLexer .
It defines the following additional member functions:
.Bl -tag -width Ds
.It "yyFlexLexer(std::istream* arg_yyin = 0, std::ostream* arg_yyout = 0)"
Constructs a
.Fa yyFlexLexer
object using the given streams for input and output.
If not specified, the streams default to
.Fa cin
and
.Fa cout ,
respectively.
.It virtual int yylex()
Performs the same role as
.Fn yylex
does for ordinary flex scanners: it scans the input stream, consuming
tokens, until a rule's action returns a value.
If subclass
.Sq S
is derived from
.Fa yyFlexLexer ,
in order to access the member functions and variables of
.Sq S
inside
.Fn yylex ,
use
.Dq %option yyclass="S"
to inform
.Nm
that the
.Sq S
subclass will be used instead of
.Fa yyFlexLexer .
In this case, rather than generating
.Dq yyFlexLexer::yylex() ,
.Nm
generates
.Dq S::yylex()
(and also generates a dummy
.Dq yyFlexLexer::yylex()
that calls
.Dq yyFlexLexer::LexerError()
if called).
.It "virtual void switch_streams(std::istream* new_in = 0, std::ostream* new_out = 0)"
Reassigns
.Fa yyin
to
.Fa new_in
.Pq if non-nil
and
.Fa yyout
to
.Fa new_out
.Pq ditto ,
deleting the previous input buffer if
.Fa yyin
is reassigned.
.It int yylex(std::istream* new_in, std::ostream* new_out = 0)
First switches the input streams via
.Dq switch_streams(new_in, new_out)
and then returns the value of
.Fn yylex .
.El
.Pp
In addition,
.Fa yyFlexLexer
defines the following protected virtual functions which can be redefined
in derived classes to tailor the scanner:
.Bl -tag -width Ds
.It virtual int LexerInput(char* buf, int max_size)
Reads up to
.Fa max_size
characters into
.Fa buf
and returns the number of characters read.
To indicate end-of-input, return 0 characters.
Note that
.Qq interactive
scanners (see the
.Fl B
and
.Fl I
flags) define the macro
.Dv YY_INTERACTIVE .
If
.Fn LexerInput
has been redefined, and it's necessary to take different actions depending on
whether or not the scanner might be scanning an interactive input source,
it's possible to test for the presence of this name via
.Dq #ifdef .
.It virtual void LexerOutput(const char* buf, int size)
Writes out
.Fa size
characters from the buffer
.Fa buf ,
which, while NUL-terminated, may also contain
.Qq internal
NUL's if the scanner's rules can match text with NUL's in them.
.It virtual void LexerError(const char* msg)
Reports a fatal error message.
The default version of this function writes the message to the stream
.Fa cerr
and exits.
.El
.Pp
Note that a
.Fa yyFlexLexer
object contains its entire scanning state.
Thus such objects can be used to create reentrant scanners.
Multiple instances of the same
.Fa yyFlexLexer
class can be instantiated, and multiple C++ scanner classes can be combined
in the same program using the
.Fl P
option discussed above.
.Pp
Finally, note that the
.Dq %array
feature is not available to C++ scanner classes;
.Dq %pointer
must be used
.Pq the default .
.Pp
Here is an example of a simple C++ scanner:
.Bd -literal -offset indent
// An example of using the flex C++ scanner class.

%{
#include <errno.h>
int mylineno = 0;
%}

string  \e"[^\en"]+\e"

ws      [ \et]+

alpha   [A-Za-z]
dig     [0-9]
name    ({alpha}|{dig}|\e$)({alpha}|{dig}|[_.\e-/$])*
num1    [-+]?{dig}+\e.?([eE][-+]?{dig}+)?
num2    [-+]?{dig}*\e.{dig}+([eE][-+]?{dig}+)?
number  {num1}|{num2}

%%

{ws}    /* skip blanks and tabs */

"/*" {
        int c;

        while ((c = yyinput()) != 0) {
                if(c == '\en')
                    ++mylineno;
                else if(c == '*') {
                    if ((c = yyinput()) == '/')
                        break;
                    else
                        unput(c);
                }
        }
}

{number}  cout << "number " << YYText() << '\en';

\en        mylineno++;

{name}    cout << "name " << YYText() << '\en';

{string}  cout << "string " << YYText() << '\en';

%%

int main(int /* argc */, char** /* argv */)
{
	FlexLexer* lexer = new yyFlexLexer;
	while(lexer->yylex() != 0)
	    ;
	return 0;
}
.Ed
.Pp
To create multiple
.Pq different
lexer classes, use the
.Fl P
flag
(or the
.Dq prefix=
option)
to rename each
.Fa yyFlexLexer
to some other
.Fa xxFlexLexer .
.Aq Pa g++/FlexLexer.h
can then be included in other sources once per lexer class, first renaming
.Fa yyFlexLexer
as follows:
.Bd -literal -offset indent
#undef yyFlexLexer
#define yyFlexLexer xxFlexLexer
#include <g++/FlexLexer.h>

#undef yyFlexLexer
#define yyFlexLexer zzFlexLexer
#include <g++/FlexLexer.h>
.Ed
.Pp
If, for example,
.Dq %option prefix="xx"
is used for one scanner and
.Dq %option prefix="zz"
is used for the other.
.Pp
.Sy IMPORTANT :
the present form of the scanning class is experimental
and may change considerably between major releases.
.Sh INCOMPATIBILITIES WITH LEX AND POSIX
.Nm
is a rewrite of the
.At
.Nm lex
tool
(the two implementations do not share any code, though),
with some extensions and incompatibilities, both of which are of concern
to those who wish to write scanners acceptable to either implementation.
.Nm
is fully compliant with the
.Tn POSIX
.Nm lex
specification, except that when using
.Dq %pointer
.Pq the default ,
a call to
.Fn unput
destroys the contents of
.Fa yytext ,
which is counter to the
.Tn POSIX
specification.
.Pp
In this section we discuss all of the known areas of incompatibility between
.Nm ,
AT&T
.Nm lex ,
and the
.Tn POSIX
specification.
.Pp
.Nm flex Ns 's
.Fl l
option turns on maximum compatibility with the original AT&T
.Nm lex
implementation, at the cost of a major loss in the generated scanner's
performance.
We note below which incompatibilities can be overcome using the
.Fl l
option.
.Pp
.Nm
is fully compatible with
.Nm lex
with the following exceptions:
.Bl -dash
.It
The undocumented
.Nm lex
scanner internal variable
.Fa yylineno
is not supported unless
.Fl l
or
.Dq %option yylineno
is used.
.Pp
.Fa yylineno
should be maintained on a per-buffer basis, rather than a per-scanner
.Pq single global variable
basis.
.Pp
.Fa yylineno
is not part of the
.Tn POSIX
specification.
.It
The
.Fn input
routine is not redefinable, though it may be called to read characters
following whatever has been matched by a rule.
If
.Fn input
encounters an end-of-file, the normal
.Fn yywrap
processing is done.
A
.Dq real
end-of-file is returned by
.Fn input
as
.Dv EOF .
.Pp
Input is instead controlled by defining the
.Dv YY_INPUT
macro.
.Pp
The
.Nm
restriction that
.Fn input
cannot be redefined is in accordance with the
.Tn POSIX
specification, which simply does not specify any way of controlling the
scanner's input other than by making an initial assignment to
.Fa yyin .
.It
The
.Fn unput
routine is not redefinable.
This restriction is in accordance with
.Tn POSIX .
.It
.Nm
scanners are not as reentrant as
.Nm lex
scanners.
In particular, if a scanner is interactive and
an interrupt handler long-jumps out of the scanner,
and the scanner is subsequently called again,
the following error message may be displayed:
.Pp
.D1 fatal flex scanner internal error--end of buffer missed
.Pp
To reenter the scanner, first use
.Pp
.Dl yyrestart(yyin);
.Pp
Note that this call will throw away any buffered input;
usually this isn't a problem with an interactive scanner.
.Pp
Also note that flex C++ scanner classes are reentrant,
so if using C++ is an option , they should be used instead.
See
.Sx GENERATING C++ SCANNERS
above for details.
.It
.Fn output
is not supported.
Output from the
.Em ECHO
macro is done to the file-pointer
.Fa yyout
.Pq default stdout .
.Pp
.Fn output
is not part of the
.Tn POSIX
specification.
.It
.Nm lex
does not support exclusive start conditions
.Pq %x ,
though they are in the
.Tn POSIX
specification.
.It
When definitions are expanded,
.Nm
encloses them in parentheses.
With
.Nm lex ,
the following:
.Bd -literal -offset indent
NAME    [A-Z][A-Z0-9]*
%%
foo{NAME}?      printf("Found it\en");
%%
.Ed
.Pp
will not match the string
.Qq foo
because when the macro is expanded the rule is equivalent to
.Qq foo[A-Z][A-Z0-9]*?
and the precedence is such that the
.Sq ?\&
is associated with
.Qq [A-Z0-9]* .
With
.Nm ,
the rule will be expanded to
.Qq foo([A-Z][A-Z0-9]*)?
and so the string
.Qq foo
will match.
.Pp
Note that if the definition begins with
.Sq ^
or ends with
.Sq $
then it is not expanded with parentheses, to allow these operators to appear in
definitions without losing their special meanings.
But the
.Sq Aq s ,
.Sq / ,
and
.Aq Aq EOF
operators cannot be used in a
.Nm
definition.
.Pp
Using
.Fl l
results in the
.Nm lex
behavior of no parentheses around the definition.
.Pp
The
.Tn POSIX
specification is that the definition be enclosed in parentheses.
.It
Some implementations of
.Nm lex
allow a rule's action to begin on a separate line,
if the rule's pattern has trailing whitespace:
.Bd -literal -offset indent
%%
foo|bar<space here>
  { foobar_action(); }
.Ed
.Pp
.Nm
does not support this feature.
.It
The
.Nm lex
.Sq %r
.Pq generate a Ratfor scanner
option is not supported.
It is not part of the
.Tn POSIX
specification.
.It
After a call to
.Fn unput ,
.Fa yytext
is undefined until the next token is matched,
unless the scanner was built using
.Dq %array .
This is not the case with
.Nm lex
or the
.Tn POSIX
specification.
The
.Fl l
option does away with this incompatibility.
.It
The precedence of the
.Sq {}
.Pq numeric range
operator is different.
.Nm lex
interprets
.Qq abc{1,3}
as match one, two, or three occurrences of
.Sq abc ,
whereas
.Nm
interprets it as match
.Sq ab
followed by one, two, or three occurrences of
.Sq c .
The latter is in agreement with the
.Tn POSIX
specification.
.It
The precedence of the
.Sq ^
operator is different.
.Nm lex
interprets
.Qq ^foo|bar
as match either
.Sq foo
at the beginning of a line, or
.Sq bar
anywhere, whereas
.Nm
interprets it as match either
.Sq foo
or
.Sq bar
if they come at the beginning of a line.
The latter is in agreement with the
.Tn POSIX
specification.
.It
The special table-size declarations such as
.Sq %a
supported by
.Nm lex
are not required by
.Nm
scanners;
.Nm
ignores them.
.It
The name
.Dv FLEX_SCANNER
is #define'd so scanners may be written for use with either
.Nm
or
.Nm lex .
Scanners also include
.Dv YY_FLEX_MAJOR_VERSION
and
.Dv YY_FLEX_MINOR_VERSION
indicating which version of
.Nm
generated the scanner
(for example, for the 2.5 release, these defines would be 2 and 5,
respectively).
.El
.Pp
The following
.Nm
features are not included in
.Nm lex
or the
.Tn POSIX
specification:
.Bd -unfilled -offset indent
C++ scanners
%option
start condition scopes
start condition stacks
interactive/non-interactive scanners
yy_scan_string() and friends
yyterminate()
yy_set_interactive()
yy_set_bol()
YY_AT_BOL()
<<EOF>>
<*>
YY_DECL
YY_START
YY_USER_ACTION
YY_USER_INIT
#line directives
%{}'s around actions
multiple actions on a line
.Ed
.Pp
plus almost all of the
.Nm
flags.
The last feature in the list refers to the fact that with
.Nm
Multiple actions ican be placed on the same line,
separated with semi-colons, while with
.Nm lex ,
the following
.Pp
.Dl foo    handle_foo(); ++num_foos_seen;
.Pp
is
.Pq rather surprisingly
truncated to
.Pp
.Dl foo    handle_foo();
.Pp
.Nm
does not truncate the action.
Actions that are not enclosed in braces
are simply terminated at the end of the line.
.Sh FILES
.Bl -tag -width "<g++/FlexLexer.h>"
.It flex.skl
Skeleton scanner.
This file is only used when building flex, not when
.Nm
executes.
.It lex.backup
Backing-up information for the
.Fl b
flag (called
.Pa lex.bck
on some systems).
.It lex.yy.c
Generated scanner
(called
.Pa lexyy.c
on some systems).
.It lex.yy.cc
Generated C++ scanner class, when using
.Fl + .
.It Aq g++/FlexLexer.h
Header file defining the C++ scanner base class,
.Fa FlexLexer ,
and its derived class,
.Fa yyFlexLexer .
.It /usr/lib/libl.*
.Nm
libraries.
The
.Pa /usr/lib/libfl.*\&
libraries are links to these.
Scanners must be linked using either
.Fl \&ll
or
.Fl lfl .
.El
.Sh EXIT STATUS
.Ex -std flex
.Sh DIAGNOSTICS
.Bl -diag
.It warning, rule cannot be matched
Indicates that the given rule cannot be matched because it follows other rules
that will always match the same text as it.
For example, in the following
.Dq foo
cannot be matched because it comes after an identifier
.Qq catch-all
rule:
.Bd -literal -offset indent
[a-z]+    got_identifier();
foo       got_foo();
.Ed
.Pp
Using
.Em REJECT
in a scanner suppresses this warning.
.It "warning, \-s option given but default rule can be matched"
Means that it is possible
.Pq perhaps only in a particular start condition
that the default rule
.Pq match any single character
is the only one that will match a particular input.
Since
.Fl s
was given, presumably this is not intended.
.It reject_used_but_not_detected undefined
.It yymore_used_but_not_detected undefined
These errors can occur at compile time.
They indicate that the scanner uses
.Em REJECT
or
.Fn yymore
but that
.Nm
failed to notice the fact, meaning that
.Nm
scanned the first two sections looking for occurrences of these actions
and failed to find any, but somehow they snuck in
.Pq via an #include file, for example .
Use
.Dq %option reject
or
.Dq %option yymore
to indicate to
.Nm
that these features are really needed.
.It flex scanner jammed
A scanner compiled with
.Fl s
has encountered an input string which wasn't matched by any of its rules.
This error can also occur due to internal problems.
.It token too large, exceeds YYLMAX
The scanner uses
.Dq %array
and one of its rules matched a string longer than the
.Dv YYLMAX
constant
.Pq 8K bytes by default .
The value can be increased by #define'ing
.Dv YYLMAX
in the definitions section of
.Nm
input.
.It "scanner requires \-8 flag to use the character 'x'"
The scanner specification includes recognizing the 8-bit character
.Sq x
and the
.Fl 8
flag was not specified, and defaulted to 7-bit because the
.Fl Cf
or
.Fl CF
table compression options were used.
See the discussion of the
.Fl 7
flag for details.
.It flex scanner push-back overflow
unput() was used to push back so much text that the scanner's buffer
could not hold both the pushed-back text and the current token in
.Fa yytext .
Ideally the scanner should dynamically resize the buffer in this case,
but at present it does not.
.It "input buffer overflow, can't enlarge buffer because scanner uses REJECT"
The scanner was working on matching an extremely large token and needed
to expand the input buffer.
This doesn't work with scanners that use
.Em REJECT .
.It "fatal flex scanner internal error--end of buffer missed"
This can occur in an scanner which is reentered after a long-jump
has jumped out
.Pq or over
the scanner's activation frame.
Before reentering the scanner, use:
.Pp
.Dl yyrestart(yyin);
.Pp
or, as noted above, switch to using the C++ scanner class.
.It "too many start conditions in <> construct!"
More start conditions than exist were listed in a <> construct
(so at least one of them must have been listed twice).
.El
.Sh SEE ALSO
.Xr awk 1 ,
.Xr sed 1 ,
.Xr yacc 1
.Rs
.%A John Levine
.%A Tony Mason
.%A Doug Brown
.%B Lex & Yacc
.%I O'Reilly and Associates
.%N 2nd edition
.Re
.Rs
.%A Alfred Aho
.%A Ravi Sethi
.%A Jeffrey Ullman
.%B Compilers: Principles, Techniques and Tools
.%I Addison-Wesley
.%D 1986
.%O "Describes the pattern-matching techniques used by flex (deterministic finite automata)"
.Re
.Sh STANDARDS
The
.Nm lex
utility is compliant with the
.St -p1003.1-2008
specification,
though its presence is optional.
.Pp
The flags
.Op Fl 78BbCcdFfhIiLloPpSsTVw+? ,
.Op Fl -help ,
and
.Op Fl -version
are extensions to that specification.
.Sh AUTHORS
Vern Paxson, with the help of many ideas and much inspiration from
Van Jacobson.
Original version by Jef Poskanzer.
The fast table representation is a partial implementation of a design done by
Van Jacobson.
The implementation was done by Kevin Gong and Vern Paxson.
.Pp
Thanks to the many
.Nm
beta-testers, feedbackers, and contributors, especially Francois Pinard,
Casey Leedom,
Robert Abramovitz,
Stan Adermann, Terry Allen, David Barker-Plummer, John Basrai,
Neal Becker, Nelson H.F. Beebe, benson@odi.com,
Karl Berry, Peter A. Bigot, Simon Blanchard,
Keith Bostic, Frederic Brehm, Ian Brockbank, Kin Cho, Nick Christopher,
Brian Clapper, J.T. Conklin,
Jason Coughlin, Bill Cox, Nick Cropper, Dave Curtis, Scott David
Daniels, Chris G. Demetriou, Theo de Raadt,
Mike Donahue, Chuck Doucette, Tom Epperly, Leo Eskin,
Chris Faylor, Chris Flatters, Jon Forrest, Jeffrey Friedl,
Joe Gayda, Kaveh R. Ghazi, Wolfgang Glunz,
Eric Goldman, Christopher M. Gould, Ulrich Grepel, Peer Griebel,
Jan Hajic, Charles Hemphill, NORO Hideo,
Jarkko Hietaniemi, Scott Hofmann,
Jeff Honig, Dana Hudes, Eric Hughes, John Interrante,
Ceriel Jacobs, Michal Jaegermann, Sakari Jalovaara, Jeffrey R. Jones,
Henry Juengst, Klaus Kaempf, Jonathan I. Kamens, Terrence O Kane,
Amir Katz, ken@ken.hilco.com, Kevin B. Kenny,
Steve Kirsch, Winfried Koenig, Marq Kole, Ronald Lamprecht,
Greg Lee, Rohan Lenard, Craig Leres, John Levine, Steve Liddle,
David Loffredo, Mike Long,
Mohamed el Lozy, Brian Madsen, Malte, Joe Marshall,
Bengt Martensson, Chris Metcalf,
Luke Mewburn, Jim Meyering, R. Alexander Milowski, Erik Naggum,
G.T. Nicol, Landon Noll, James Nordby, Marc Nozell,
Richard Ohnemus, Karsten Pahnke,
Sven Panne, Roland Pesch, Walter Pelissero, Gaumond Pierre,
Esmond Pitt, Jef Poskanzer, Joe Rahmeh, Jarmo Raiha,
Frederic Raimbault, Pat Rankin, Rick Richardson,
Kevin Rodgers, Kai Uwe Rommel, Jim Roskind, Alberto Santini,
Andreas Scherer, Darrell Schiebel, Raf Schietekat,
Doug Schmidt, Philippe Schnoebelen, Andreas Schwab,
Larry Schwimmer, Alex Siegel, Eckehard Stolz, Jan-Erik Strvmquist,
Mike Stump, Paul Stuart, Dave Tallman, Ian Lance Taylor,
Chris Thewalt, Richard M. Timoney, Jodi Tsai,
Paul Tuinenga, Gary Weik, Frank Whaley, Gerhard Wilhelms, Kent Williams,
Ken Yap, Ron Zellar, Nathan Zelle, David Zuhn,
and those whose names have slipped my marginal mail-archiving skills
but whose contributions are appreciated all the
same.
.Pp
Thanks to Keith Bostic, Jon Forrest, Noah Friedman,
John Gilmore, Craig Leres, John Levine, Bob Mulcahy, G.T.
Nicol, Francois Pinard, Rich Salz, and Richard Stallman for help with various
distribution headaches.
.Pp
Thanks to Esmond Pitt and Earle Horton for 8-bit character support;
to Benson Margulies and Fred Burke for C++ support;
to Kent Williams and Tom Epperly for C++ class support;
to Ove Ewerlid for support of NUL's;
and to Eric Hughes for support of multiple buffers.
.Pp
This work was primarily done when I was with the Real Time Systems Group
at the Lawrence Berkeley Laboratory in Berkeley, CA.
Many thanks to all there for the support I received.
.Pp
Send comments to
.Aq vern@ee.lbl.gov .
.Sh BUGS
Some trailing context patterns cannot be properly matched and generate
warning messages
.Pq "dangerous trailing context" .
These are patterns where the ending of the first part of the rule
matches the beginning of the second part, such as
.Qq zx*/xy* ,
where the
.Sq x*
matches the
.Sq x
at the beginning of the trailing context.
(Note that the POSIX draft states that the text matched by such patterns
is undefined.)
.Pp
For some trailing context rules, parts which are actually fixed-length are
not recognized as such, leading to the above mentioned performance loss.
In particular, parts using
.Sq |\&
or
.Sq {n}
(such as
.Qq foo{3} )
are always considered variable-length.
.Pp
Combining trailing context with the special
.Sq |\&
action can result in fixed trailing context being turned into
the more expensive variable trailing context.
For example, in the following:
.Bd -literal -offset indent
%%
abc      |
xyz/def
.Ed
.Pp
Use of
.Fn unput
invalidates yytext and yyleng, unless the
.Dq %array
directive
or the
.Fl l
option has been used.
.Pp
Pattern-matching of NUL's is substantially slower than matching other
characters.
.Pp
Dynamic resizing of the input buffer is slow, as it entails rescanning
all the text matched so far by the current
.Pq generally huge
token.
.Pp
Due to both buffering of input and read-ahead,
it is not possible to intermix calls to
.Aq Pa stdio.h
routines, such as, for example,
.Fn getchar ,
with
.Nm
rules and expect it to work.
Call
.Fn input
instead.
.Pp
The total table entries listed by the
.Fl v
flag excludes the number of table entries needed to determine
what rule has been matched.
The number of entries is equal to the number of DFA states
if the scanner does not use
.Em REJECT ,
and somewhat greater than the number of states if it does.
.Pp
.Em REJECT
cannot be used with the
.Fl f
or
.Fl F
options.
.Pp
The
.Nm
internal algorithms need documentation.