<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2 Final//EN">
<!--Converted with LaTeX2HTML 2002-2-1 (1.71)
original version by: Nikos Drakos, CBLU, University of Leeds
* revised and updated by: Marcus Hennecke, Ross Moore, Herb Swan
* with significant contributions from:
Jens Lippmann, Marek Rouchal, Martin Wilck and others -->
<HTML>
<HEAD>
<TITLE>JFlex User's Manual</TITLE>
<META NAME="description" CONTENT="JFlex User's Manual">
<META NAME="keywords" CONTENT="manual">
<META NAME="resource-type" CONTENT="document">
<META NAME="distribution" CONTENT="global">
<META NAME="Generator" CONTENT="LaTeX2HTML v2002-2-1">
<META HTTP-EQUIV="Content-Style-Type" CONTENT="text/css">
<LINK REL="STYLESHEET" HREF="manual.css">
</HEAD>
<BODY >
<P>
<CENTER>
<A NAME="TOP"></a>
<A HREF="http://www.jflex.de"><IMG SRC="logo.gif" BORDER=0 HEIGHT=223 WIDTH=577></a>
</CENTER>
<P>
<DIV ALIGN="CENTER">
<I><FONT SIZE="+2">The Fast Lexical Analyser Generator</FONT>
<BR></I></DIV>
<P></P>
<DIV ALIGN="CENTER"></DIV>
<P></P>
<DIV ALIGN="CENTER"><I>Copyright ©1998-2009 by <A NAME="tex2html1"
HREF="http://www.doclsf.de">Gerwin Klein</A>
<BR></I></DIV>
<P><P><BR>
<DIV ALIGN="CENTER"><I><FONT SIZE="+4"><I><B>JFlex User's Manual</B></I></FONT>
<BR></I></DIV>
<P><P><BR>
<DIV ALIGN="CENTER"><I>Version 1.4.3, January 31, 2009
</I></DIV>
<P>
<BR>
<H2><A NAME="SECTION00010000000000000000">
Contents</A>
</H2>
<!--Table of Contents-->
<UL>
<LI><A NAME="tex2html79"
HREF="manual.html#SECTION00020000000000000000">Introduction</A>
<UL>
<LI><A NAME="tex2html80"
HREF="manual.html#SECTION00021000000000000000">Design goals</A>
<LI><A NAME="tex2html81"
HREF="manual.html#SECTION00022000000000000000">About this manual</A>
</UL><BR>
<LI><A NAME="tex2html82"
HREF="manual.html#SECTION00030000000000000000">Installing and Running JFlex</A>
<UL>
<LI><A NAME="tex2html83"
HREF="manual.html#SECTION00031000000000000000">Installing JFlex</A>
<LI><A NAME="tex2html84"
HREF="manual.html#SECTION00032000000000000000">Running JFlex</A>
</UL><BR>
<LI><A NAME="tex2html85"
HREF="manual.html#SECTION00040000000000000000">A simple Example: How to work with JFlex</A>
<UL>
<LI><A NAME="tex2html86"
HREF="manual.html#SECTION00041000000000000000">Code to include</A>
<LI><A NAME="tex2html87"
HREF="manual.html#SECTION00042000000000000000">Options and Macros</A>
<LI><A NAME="tex2html88"
HREF="manual.html#SECTION00043000000000000000">Rules and Actions</A>
<LI><A NAME="tex2html89"
HREF="manual.html#SECTION00044000000000000000">How to get it going</A>
</UL><BR>
<LI><A NAME="tex2html90"
HREF="manual.html#SECTION00050000000000000000">Lexical Specifications</A>
<UL>
<LI><A NAME="tex2html91"
HREF="manual.html#SECTION00051000000000000000">User code</A>
<LI><A NAME="tex2html92"
HREF="manual.html#SECTION00052000000000000000">Options and declarations</A>
<LI><A NAME="tex2html93"
HREF="manual.html#SECTION00053000000000000000">Lexical rules</A>
</UL><BR>
<LI><A NAME="tex2html94"
HREF="manual.html#SECTION00060000000000000000">Encodings, Platforms, and Unicode</A>
<UL>
<LI><A NAME="tex2html95"
HREF="manual.html#SECTION00061000000000000000">The Problem</A>
<LI><A NAME="tex2html96"
HREF="manual.html#SECTION00062000000000000000">Scanning text files</A>
<LI><A NAME="tex2html97"
HREF="manual.html#SECTION00063000000000000000">Scanning binaries</A>
</UL><BR>
<LI><A NAME="tex2html98"
HREF="manual.html#SECTION00070000000000000000">A few words on performance</A>
<UL>
<LI><A NAME="tex2html99"
HREF="manual.html#SECTION00071000000000000000">Comparison of JLex and JFlex</A>
<LI><A NAME="tex2html100"
HREF="manual.html#SECTION00072000000000000000">How to write a faster specification</A>
</UL><BR>
<LI><A NAME="tex2html101"
HREF="manual.html#SECTION00080000000000000000">Porting Issues</A>
<UL>
<LI><A NAME="tex2html102"
HREF="manual.html#SECTION00081000000000000000">Porting from JLex</A>
<LI><A NAME="tex2html103"
HREF="manual.html#SECTION00082000000000000000">Porting from lex/flex</A>
</UL><BR>
<LI><A NAME="tex2html104"
HREF="manual.html#SECTION00090000000000000000">Working together</A>
<UL>
<LI><A NAME="tex2html105"
HREF="manual.html#SECTION00091000000000000000">JFlex and CUP</A>
<LI><A NAME="tex2html106"
HREF="manual.html#SECTION00092000000000000000">JFlex and BYacc/J</A>
</UL><BR>
<LI><A NAME="tex2html107"
HREF="manual.html#SECTION000100000000000000000">Bugs and Deficiencies</A>
<UL>
<LI><A NAME="tex2html108"
HREF="manual.html#SECTION000101000000000000000">Deficiencies</A>
<LI><A NAME="tex2html109"
HREF="manual.html#SECTION000102000000000000000">Bugs</A>
</UL><BR>
<LI><A NAME="tex2html110"
HREF="manual.html#SECTION000110000000000000000">Copying and License</A>
<LI><A NAME="tex2html111"
HREF="manual.html#SECTION000120000000000000000">Bibliography</A>
</UL>
<!--End of Table of Contents-->
<H1><A NAME="SECTION00020000000000000000"></A><A NAME="Intro"></A><BR>
Introduction
</H1>
JFlex is a lexical analyser generator for Java<A NAME="tex2html2"
HREF="#foot33"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]" SRC="footnote.png"></SUP></A>written in Java. It is also a rewrite of the very useful tool JLex [<A
HREF="manual.html#JLex">3</A>] which
was developed by Elliot Berk at Princeton University. As Vern Paxson states
for his C/C++ tool flex [<A
HREF="manual.html#flex">11</A>]: they do not share any code though.
<P>
<H2><A NAME="SECTION00021000000000000000">
Design goals</A>
</H2>
The main design goals of JFlex are:
<UL>
<LI><B>Full unicode support</B>
</LI>
<LI><B>Fast generated scanners </B>
</LI>
<LI><B>Fast scanner generation</B>
</LI>
<LI><B>Convenient specification syntax</B>
</LI>
<LI><B>Platform independence</B>
</LI>
<LI><B>JLex compatibility</B>
</LI>
</UL>
<P>
<H2><A NAME="SECTION00022000000000000000">
About this manual</A>
</H2>
This manual gives a brief but complete description of the tool JFlex. It
assumes that you are familiar with the issue of lexical analysis. The references [<A
HREF="manual.html#Aho">1</A>],
[<A
HREF="manual.html#Appel">2</A>], and [<A
HREF="manual.html#Maurer">13</A>] provide a good introduction to this topic.
<P>
The next section of this manual describes <A HREF="#Installing"><I>installation procedures</I></A>
for JFlex. If you never worked with JLex or
just want to compare a JLex and a JFlex scanner specification you
should also read <A HREF="#Example"><I>Working with JFlex - an example</I></A>
(section <A HREF="#Example">3</A>). All options and the complete
specification syntax are presented in
<A HREF="#Specifications"><I>Lexical specifications</I></A> (section <A HREF="#Specifications">4</A>);
<A HREF="#sec:encodings"><I>Encodings, Platforms, and Unicode</I></A> (section <A HREF="#sec:encodings">5</A>)
provides information about scanning text vs. binary files.
If you are interested in performance
considerations and comparing JLex with JFlex speed,
<A HREF="#performance"><I>a few words on performance</I></A> (section <A HREF="#performance">6</A>)
might be just right for you. Those who want to
use their old JLex specifications may want to check out section <A HREF="#Porting">7.1</A>
<A HREF="#Porting"><I>Porting from JLex</I></A> to avoid possible problems
with not portable or non standard JLex behaviour that has been fixed in
JFlex. Section <A HREF="#lexport">7.2</A> talks about porting scanners from the
Unix tools lex and flex. Interfacing JFlex scanners with the LALR
parser generators CUP and BYacc/J is explained in <A HREF="#WorkingTog"><I>working
together</I></A> (section <A HREF="#WorkingTog">8</A>). Section <A HREF="#Bugs">9</A>
<A HREF="#Bugs"><I>Bugs</I></A> gives a list of currently known active bugs.
The manual concludes with notes about
<A HREF="#Copyright"><I>Copying and License</I></A> (section <A HREF="#Copyright">10</A>) and
<A HREF="#References">references</A>.
<P>
<H1><A NAME="SECTION00030000000000000000"></A><A NAME="Installing"></A><BR>
Installing and Running JFlex
</H1>
<P>
<H2><A NAME="SECTION00031000000000000000">
Installing JFlex</A>
</H2>
<P>
<H3><A NAME="SECTION00031100000000000000"></A><A NAME="install:windows"></A><BR>
Windows
</H3>
To install JFlex on Windows 95/98/NT/XP, follow these three steps:
<OL>
<LI>Unzip the file you downloaded into the directory you want JFlex in (using
something like
<A NAME="tex2html3"
HREF="http://www.winzip.com">WinZip</A>).
If you unzipped it to say <code>C:\</code>, the following directory structure
should be generated:
<PRE>
C:\JFlex\
+--bin\ (start scripts)
+--doc\ (FAQ and manual)
+--examples\
+--binary\ (scanning binary files)
+--byaccj\ (calculator example for BYacc/J)
+--cup\ (calculator example for cup)
+--interpreter\ (interpreter example for cup)
+--java\ (Java lexer specification)
+--simple\ (example scanner)
+--simple-maven\ (example with maven)
+--standalone\ (a simple standalone scanner)
+--standalone-maven\ (above with maven)
+--lib\ (the precompiled classes)
+--src\
+--JFlex\ (source code of JFlex)
+--JFlex\gui (source code of JFlex UI classes)
+--java_cup\runtime\ (source code of cup runtime classes)
</PRE>
<P>
</LI>
<LI>Edit the file <B><code>bin\jflex.bat</code></B>
(in the example it's <code>C:\JFlex\bin\jflex.bat</code>)
such that
<P>
<UL>
<LI><B><TT>JAVA_HOME</TT></B> contains the directory where your Java JDK is installed
(for instance <code>C:\java</code>) and
</LI>
<LI><B><TT>JFLEX_HOME</TT></B> the directory that contains JFlex (in the example:
<code>C:\JFlex</code>)
</LI>
</UL>
<P>
</LI>
<LI>Include the <code>bin\</code> directory of JFlex in your path.
(the one that contains the start script, in the example: <code>C:\JFlex\bin</code>).
</LI>
</OL>
<P>
<H3><A NAME="SECTION00031200000000000000">
Unix with tar archive</A>
</H3>
<P>
To install JFlex on a Unix system, follow these two steps:
<UL>
<LI>Decompress the archive into a directory of your choice
with GNU tar, for instance to <TT>/usr/share</TT>:
<P>
<TT>tar -C /usr/share -xvzf jflex-1.4.3.tar.gz</TT>
<P>
(The example is for site wide installation. You need to
be root for that. User installation works exactly the
same way--just choose a directory where you have write
permission)
<P>
</LI>
<LI>Make a symbolic link from somewhere in your binary
path to <TT>bin/jflex</TT>, for instance:
<P>
<TT>ln -s /usr/share/JFlex/bin/jflex /usr/bin/jflex</TT>
<P>
If the Java interpreter is not in your binary path, you
need to supply its location in the script <TT>bin/jflex</TT>.
</LI>
</UL>
<P>
You can verify the integrity of the downloaded file with
the MD5 checksum available on the <A NAME="tex2html4"
HREF="http://www.jflex.de/download.html">JFlex download page</A>.
If you put the checksum file in the same directory
as the archive, you run:
<P>
<code>md5sum --check </code><TT>jflex-1.4.3.tar.gz.md5</TT>
<P>
It should tell you
<P>
<TT>jflex-1.4.3.tar.gz: OK</TT>
<P>
<H3><A NAME="SECTION00031300000000000000">
Linux with RPM</A>
</H3>
<P>
<UL>
<LI>become root
</LI>
<LI>issue
<BR> <TT>rpm -U jflex-1.4.3-0.rpm</TT>
</LI>
</UL>
<P>
You can verify the integrity of the downloaded <TT>rpm</TT> file with
<P>
<code>rpm --checksig </code><TT>jflex-1.4.3-0.rpm</TT>
<P>
<H2><A NAME="SECTION00032000000000000000">
Running JFlex</A>
</H2>
You run JFlex with:
<P>
<TT>jflex <options> <inputfiles></TT>
<P>
It is also possible to skip the start script in <code>bin\</code>
and include the file <code>lib\JFlex.jar</code>
in your <TT>CLASSPATH</TT> environment variable instead.
<P>
Then you run JFlex with:
<P>
<TT>java JFlex.Main <options> <inputfiles></TT>
<P>
The input files and options are in both cases optional. If you don't provide a file name on
the command line, JFlex will pop up a window to ask you for one.
<P>
JFlex knows about the following options:
<P>
<DL>
<DT></DT>
<DD><code>-d <directory></code>
<BR> writes the generated file to the directory <code><directory></code>
<P>
</DD>
<DT></DT>
<DD><code>--skel <file></code>
<BR> uses external skeleton <code><file></code>. This is mainly for JFlex
maintenance and special low level customisations. Use only when you
know what you are doing! JFlex comes with a skeleton file in the
<TT>src</TT> directory that reflects exactly the internal, pre-compiled
skeleton and can be used with the <TT>-skel</TT> option.
<P>
</DD>
<DT></DT>
<DD><code>--nomin</code>
<BR> skip the DFA minimisation step during scanner generation.
<P>
</DD>
<DT></DT>
<DD><code>--jlex</code>
<BR> tries even harder to comply to JLex interpretation of specs.
<P>
</DD>
<DT></DT>
<DD><code>--dot</code>
<BR> generate graphviz dot files for the NFA, DFA and minimised
DFA. This feature is still in alpha status, and not
fully implemented yet.
<P>
</DD>
<DT></DT>
<DD><code>--dump</code>
<BR> display transition tables of NFA, initial DFA, and minimised DFA
<P>
</DD>
<DT></DT>
<DD><code>--verbose</code> or <TT>-v</TT>
<BR> display generation progress messages (enabled by default)
<P>
</DD>
<DT></DT>
<DD><code>--quiet</code> or <TT>-q</TT>
<BR> display error messages only (no chatter about what JFlex is
currently doing)
<P>
</DD>
<DT></DT>
<DD><code>--time</code>
<BR> display time statistics about the code generation process
(not very accurate)
<P>
</DD>
<DT></DT>
<DD><code>--version</code>
<BR> print version number
<P>
</DD>
<DT></DT>
<DD><code>--info</code>
<BR> print system and JDK information (useful if you'd like
to report a problem)
<P>
</DD>
<DT></DT>
<DD><code>--pack</code>
<BR> use the %pack code generation method by default
<P>
</DD>
<DT></DT>
<DD><code>--table</code>
<BR> use the %table code generation method by default
<P>
</DD>
<DT></DT>
<DD><code>--switch</code>
<BR> use the %switch code generation method by default
<P>
</DD>
<DT></DT>
<DD><code>--help</code> or <TT>-h</TT>
<BR> print a help message explaining options and usage of JFlex.
</DD>
</DL>
<P>
<H1><A NAME="SECTION00040000000000000000"></A><A NAME="Example"></A><BR>
A simple Example: How to work with JFlex
</H1>
To demonstrate what a lexical specification with JFlex looks like, this
section presents a part of the specification for the Java language.
The example does not describe the whole lexical structure of Java programs,
but only a small and simplified part of it (some keywords, some operators,
comments and only two kinds of literals). It also shows how to interface
with the LALR parser generator CUP [<A
HREF="manual.html#CUP">8</A>] and therefore
uses a class <TT>sym</TT> (generated by CUP), where integer constants for
the terminal tokens of the CUP grammar are declared. JFlex comes with a
directory <TT>examples</TT>, where you can find a small standalone scanner
that doesn't need other tools like CUP to give you a running example.
The "<TT>examples</TT>" directory also contains a <EM>complete</EM> JFlex
specification of the lexical structure of Java programs together with the
CUP parser specification for Java by
<A NAME="tex2html5"
HREF="mailto:cananian@alumni.princeton.edu">C. Scott Ananian</A>, obtained
from the CUP [<A
HREF="manual.html#CUP">8</A>] web site (it was modified to interface with the JFlex scanner).
Both specifications adhere to the Java Language Specification [<A
HREF="manual.html#LangSpec">7</A>].
<P>
<FONT SIZE="-1"><A NAME="CodeTop"></A></FONT><PRE>
/* JFlex example: part of Java language lexer specification */
import java_cup.runtime.*;
/**
* This class is a simple example lexer.
*/
%%
</PRE><FONT SIZE="-1">
<A NAME="CodeOptions"></A></FONT><PRE>
%class Lexer
%unicode
%cup
%line
%column
</PRE><FONT SIZE="-1">
<A NAME="CodeScannerCode"></A></FONT><PRE>
%{
StringBuffer string = new StringBuffer();
private Symbol symbol(int type) {
return new Symbol(type, yyline, yycolumn);
}
private Symbol symbol(int type, Object value) {
return new Symbol(type, yyline, yycolumn, value);
}
%}
</PRE><FONT SIZE="-1">
<A NAME="CodeMacros"></A></FONT><PRE>
LineTerminator = \r|\n|\r\n
InputCharacter = [^\r\n]
WhiteSpace = {LineTerminator} | [ \t\f]
/* comments */
Comment = {TraditionalComment} | {EndOfLineComment} | {DocumentationComment}
TraditionalComment = "/*" [^*] ~"*/" | "/*" "*"+ "/"
EndOfLineComment = "//" {InputCharacter}* {LineTerminator}
DocumentationComment = "/**" {CommentContent} "*"+ "/"
CommentContent = ( [^*] | \*+ [^/*] )*
Identifier = [:jletter:] [:jletterdigit:]*
DecIntegerLiteral = 0 | [1-9][0-9]*
</PRE><FONT SIZE="-1">
<A NAME="CodeStateDecl"></A></FONT><PRE>
%state STRING
%%
</PRE><FONT SIZE="-1">
<A NAME="CodeRulesYYINITIAL"></A></FONT><PRE>
/* keywords */
<YYINITIAL> "abstract" { return symbol(sym.ABSTRACT); }
<YYINITIAL> "boolean" { return symbol(sym.BOOLEAN); }
<YYINITIAL> "break" { return symbol(sym.BREAK); }
</PRE><FONT SIZE="-1">
<A NAME="CodeRulesBunch"></A></FONT><PRE>
<YYINITIAL> {
/* identifiers */
{Identifier} { return symbol(sym.IDENTIFIER); }
/* literals */
{DecIntegerLiteral} { return symbol(sym.INTEGER_LITERAL); }
\" { string.setLength(0); yybegin(STRING); }
/* operators */
"=" { return symbol(sym.EQ); }
"==" { return symbol(sym.EQEQ); }
"+" { return symbol(sym.PLUS); }
/* comments */
{Comment} { /* ignore */ }
/* whitespace */
{WhiteSpace} { /* ignore */ }
}
</PRE><FONT SIZE="-1">
<A NAME="CodeRulesYYtext"></A></FONT><PRE>
<STRING> {
\" { yybegin(YYINITIAL);
return symbol(sym.STRING_LITERAL,
string.toString()); }
[^\n\r\"\\]+ { string.append( yytext() ); }
\\t { string.append('\t'); }
\\n { string.append('\n'); }
\\r { string.append('\r'); }
\\\" { string.append('\"'); }
\\ { string.append('\\'); }
}
</PRE><FONT SIZE="-1">
<A NAME="CodeRulesAllStates"></A></FONT><PRE>
/* error fallback */
.|\n { throw new Error("Illegal character <"+
yytext()+">"); }
</PRE>
<P>
From this specification JFlex generates a <TT>.java</TT> file with one
class that contains code for the scanner. The class will have a
constructor taking a <TT>java.io.Reader</TT> from which the input is
read. The class will also have a function <TT>yylex()</TT> that runs the
scanner and that can be used to get the next token from the input (in this
example the function actually has the name <TT>next_token()</TT> because
the specification uses the <TT>%cup</TT> switch).
<P>
As with JLex, the specification consists of three parts, divided by <TT>%%</TT>:
<UL>
<LI><A HREF="#ExampleUserCode">usercode</A>,
</LI>
<LI><A HREF="#ExampleOptions">options and declarations</A> and
</LI>
<LI><A HREF="#ExampleLexRules">lexical rules</A>.
</LI>
</UL>
<P>
<H2><A NAME="SECTION00041000000000000000"></A><A NAME="ExampleUserCode"></A><BR>
Code to include
</H2>
Let's take a look at the first section, ``user code'': The text up to the
first line starting with <TT>%%</TT> is copied verbatim to the top
of the generated lexer class (before the actual class declaration).
Beside <TT>package</TT> and <TT>import</TT> statements there is usually not much
to do here. If the code ends with a javadoc class comment, the generated class
will get this comment, if not, JFlex will generate one automatically.
<P>
<H2><A NAME="SECTION00042000000000000000"></A><A NAME="ExampleOptions"></A><BR>
Options and Macros
</H2>
The second section ``options and declarations'' is more interesting. It consists
of a set of options, code that is included inside the generated scanner
class, lexical states and macro declarations. Each JFlex option must begin
a line of the specification and starts with a <TT>%</TT>. In our example
the following options are used:
<P>
<UL>
<LI><TT><A HREF="#CodeOptions">%class Lexer</A></TT> tells JFlex to give the
generated class the name ``Lexer'' and to write the code to a file ``<TT>Lexer.java</TT>''.
<P>
</LI>
<LI><TT><A HREF="#CodeOptions">%unicode</A></TT> defines the set of characters the scanner will
work on. For scanning text files, <TT>%unicode</TT> should always be used. See also
section <A HREF="#sec:encodings">5</A> for more information on character sets, encodings, and
scanning text vs. binary files.
<P>
</LI>
<LI><TT><A HREF="#CodeOptions">%cup</A></TT> switches to CUP compatibility
mode to interface with a CUP generated parser.
<P>
</LI>
<LI><TT><A HREF="#CodeOptions">%line</A></TT> switches line counting on (the
current line number can be accessed via the variable <TT>yyline</TT>)
<P>
</LI>
<LI><TT><A HREF="#CodeOptions">%column</A></TT> switches column counting on
(current column is accessed via <TT>yycolumn</TT>)
<P>
</LI>
</UL>
<A NAME="ExampleScannerCode"></A>
<P>
The code included in <TT><A HREF="#CodeScannerCode">%{...%}</A></TT>
is copied verbatim into the generated lexer class source.
Here you can declare member variables and functions that are used
inside scanner actions. In our example we declare a <TT>StringBuffer</TT> ``<TT>string</TT>''
in which we will store parts of string literals and two helper functions
``<TT>symbol</TT>'' that create <TT>java_cup.runtime.Symbol</TT> objects
with position information of the current token (see section <A HREF="#CUPWork">8.1</A>
<A HREF="#CUPWork"><I>JFlex and CUP</I></A>
for how to interface with the parser generator CUP). As JFlex options, both
<code>%{</code> and <code>\%}</code> must begin a line.
<A NAME="ExampleMacros"></A>
<P>
The specification continues with macro declarations. Macros are
abbreviations for regular expressions, used to make lexical specifications
easier to read and understand. A macro declaration
consists of a macro identifier followed by <TT>=</TT>, then followed by
the regular expression it represents. This regular expression may
itself contain macro usages. Although this allows a grammar like specification
style, macros are still just abbreviations and not non terminals - they
cannot be recursive or mutually recursive. Cycles in macro definitions
are detected and reported at generation time by JFlex.
<P>
Here some of the example macros in more detail:
<UL>
<LI><TT><A HREF="#CodeMacros">LineTerminator</A></TT> stands for the regular
expression that matches an ASCII CR, an ASCII LF or an CR followed by LF.
<P>
</LI>
<LI><TT><A HREF="#CodeMacros">InputCharacter</A></TT> stands for all characters
that are not a CR or LF.
<P>
</LI>
<LI><TT><A HREF="#CodeMacros">TraditionalComment</A></TT> is the expression
that matches the string <TT>"/*"</TT> followed by a character that
is not a <TT>*</TT>, followed by anything that does not contain, but
ends in <TT>"/*"</TT>. As this would not match comments like
<TT>/****/</TT>, we add <TT>"/*"</TT> followed by an arbitrary
number (at least one) of <TT>"*"</TT> followed by the closing
<TT>"/"</TT>. This is not the only, but one of the simpler
expressions matching non-nesting Java comments. It is tempting to
just write something like the expression <TT>"/*" .* "*/"</TT>, but
this would match more than we want. It would for instance match the
whole of <TT>/* */ x = 0; /* */</TT>, instead of two comments and
four real tokens. See DocumentationComment and CommentContent for an
alternative.
<P>
</LI>
<LI><TT><A HREF="#CodeMacros">CommentContent</A></TT> matches zero or more
occurrences of any character except a <TT>*</TT> or any number of
<TT>*</TT> followed by a character that is not a <TT>/</TT>
<P>
</LI>
<LI><TT><A HREF="#CodeMacros">Identifier</A></TT> matches each string that
starts with a character of class <TT>jletter</TT> followed by zero or more characters
of class <TT>jletterdigit</TT>. <TT>jletter</TT> and <TT>jletterdigit</TT>
are predefined character classes. <TT>jletter</TT> includes all characters for which
the Java function <TT>Character.isJavaIdentifierStart</TT> returns <TT>true</TT> and
<TT>jletterdigit</TT> all characters for that <TT>Character.isJavaIdentifierPart</TT>
returns <TT>true</TT>.
</LI>
</UL>
<A NAME="ExampleStateDecl"></A>
<P>
The last part of the second section in our
lexical specification is a lexical state declaration:
<TT><A HREF="#CodeStateDecl">%state STRING</A></TT>
declares a lexical state <TT>STRING</TT> that can be
used in the ``lexical rules'' part of the specification. A state declaration
is a line starting with <TT>%state</TT> followed by a space or comma
separated list of state identifiers. There can be more than one line starting
with <TT>%state</TT>.
<P>
<H2><A NAME="SECTION00043000000000000000"></A><A NAME="ExampleLexRules"></A><BR>
Rules and Actions
</H2>
The "lexical rules" section of a JFlex specification contains regular expressions
and actions (Java code) that are executed when the scanner matches the
associated regular expression. As the scanner reads its input, it keeps
track of all regular expressions and activates the action of the expression
that has the longest match. Our specification above for instance would with input
"<TT>breaker</TT>" match the regular expression for <TT><A HREF="#CodeMacros">Identifier</A></TT>
and not the keyword "<TT><A HREF="#CodeRulesYYINITIAL">break</A></TT>"
followed by the Identifier "<TT>er</TT>", because rule <code>{Identifier}</code>
matches more of this input at once (i.e. it matches all of it)
than any other rule in the specification. If two regular expressions both
have the longest match for a certain input, the scanner chooses the action
of the expression that appears first in the specification. In that way, we
get for input "<TT>break</TT>" the keyword "<TT>break</TT>" and not an
Identifier "<TT>break</TT>".
<P>
Additional to regular expression matches, one can use lexical states to
refine a specification. A lexical state acts like a start condition.
If the scanner is in lexical state <TT>STRING</TT>, only expressions that
are preceded by the start condition <TT><STRING></TT> can be matched.
A start condition of a regular expression can contain more than one lexical
state. It is then matched when the lexer is in any of these lexical states.
The lexical state <TT>YYINITIAL</TT> is predefined and is also the state
in which the lexer begins scanning. If a regular expression has no start
conditions it is matched in <EM>all</EM> lexical states.
<A NAME="ExampleRulesStateBunch"></A>
<P>
Since you often have a bunch of expressions with the same start conditions,
JFlex allows the same abbreviation as the Unix tool <TT>flex</TT>:
<PRE>
<STRING> {
expr1 { action1 }
expr2 { action2 }
}
</PRE>
means that both <TT>expr1</TT> and <TT>expr2</TT> have start condition <TT><STRING></TT>.
<A NAME="ExampleRulesYYINITIAL"></A>
<P>
The first three rules in our example demonstrate the syntax of a regular
expression preceded by the start condition <TT><YYINITIAL></TT>.
<P>
<TT><A HREF="#CodeRulesYYINITIAL"><YYINITIAL> "abstract"</A><code> {</code> return symbol(sym.ABSTRACT); <code>}</code></TT>
<P>
matches the input "<TT>abstract</TT>" only if the scanner is in its
start state "<TT>YYINITIAL</TT>". When the string "<TT>abstract</TT>" is
matched, the scanner function returns the CUP symbol <TT>sym.ABSTRACT</TT>.
If an action does not return a value, the scanning process is resumed immediately
after executing the action.
<A NAME="ExampleRulesBunch"></A>
<P>
The rules enclosed in
<P>
<TT><A HREF="#CodeRulesBunch"><YYINITIAL> {
<BR> ...
<BR>}</A></TT>
<P>
demonstrate the abbreviated syntax and are also only matched in state <TT>YYINITIAL</TT>.
<A NAME="ExampleRulesYYbegin"></A>
<P>
Of these rules, one may be of special interest:
<P>
<code>\" { </code> <TT><A HREF="#CodeRulesBunch">string.setLength(0); yybegin(STRING);</A></TT><code> }</code>
<P>
If the scanner matches a double quote in state <TT>YYINITIAL</TT> we
have recognised the start of a string literal. Therefore we clear our <TT>StringBuffer</TT>
that will hold the content of this string literal and tell the scanner
with <TT>yybegin(STRING)</TT> to switch into the lexical state <TT>STRING</TT>.
Because we do not yet return a value to the parser, our scanner proceeds
immediately.
<A NAME="ExampleRulesYYtext"></A>
<P>
In lexical state <TT>STRING</TT> another
rule demonstrates how to refer to the input that has been matched:
<P>
<code>[^\n\r\"]+ { </code> <TT><A HREF="#CodeRulesYYtext">string.append( yytext() );</A></TT><code> }</code>
<P>
The expression <code>[^\n\r\"]+</code> matches
all characters in the input up to the next backslash (indicating an
escape sequence such as <code>\n</code>), double quote (indicating the end
of the string), or line terminator (which must not occur in a string literal).
The matched region of the input is referred to with <TT><A HREF="#CodeRulesYYtext">yytext()</A></TT>
and appended to the content of the string literal parsed so far.
<A NAME="ExampleRuleLast"></A>
<P>
The last lexical rule in the example specification
is used as an error fallback. It matches any character in any state that
has not been matched by another rule. It doesn't conflict with any other
rule because it has the least priority (because it's the last rule) and
because it matches only one character (so it can't have longest match
precedence over any other rule).
<P>
<H2><A NAME="SECTION00044000000000000000">
How to get it going</A>
</H2>
<UL>
<LI>Install JFlex (see section <A HREF="#Installing">2</A> <A HREF="#Installing"><I>Installing JFlex</I></A>)
<P>
</LI>
<LI>If you have written your specification file (or chosen one from the <TT>examples</TT>
directory), save it (say under the name <TT>java-lang.flex</TT>).
<P>
</LI>
<LI>Run JFlex with
<P>
<TT>jflex java-lang.flex</TT>
<P>
</LI>
<LI>JFlex should then report some progress messages about generating the scanner
and write the generated code to the directory of your specification file.
<P>
</LI>
<LI>Compile the generated <TT>.java</TT> file and your own classes. (If you
use CUP, generate your parser classes first)
<P>
</LI>
<LI>That's it.
</LI>
</UL>
<P>
<H1><A NAME="SECTION00050000000000000000"></A><A NAME="Specifications"></A><BR>
Lexical Specifications
</H1>
As shown above, a lexical specification file for JFlex consists of three
parts divided by a single line starting with <TT>%%</TT>:
<P>
<TT><A HREF="#SpecUsercode">UserCode</A></TT>
<BR><TT>%%</TT>
<BR><TT><A HREF="#SpecOptions">Options and declarations</A></TT>
<BR><TT>%%</TT>
<BR><TT><A HREF="#LexRules">Lexical rules</A></TT>
<P>
In all parts of the specification comments of the form
<TT>/* comment text */</TT> and the Java style end of line comments starting with <TT>//</TT>
are permitted. JFlex comments do nest - so the number of <TT>/*</TT> and <TT>*/</TT>
should be balanced.
<P>
<H2><A NAME="SECTION00051000000000000000"></A><A NAME="SpecUsercode"></A><BR>
User code
</H2>
The first part contains user code that is copied verbatim into the beginning
of the source file of the generated lexer before the scanner class is declared.
As shown in the example above, this is the place to put <TT>package</TT>
declarations and <TT>import</TT>
statements. It is possible, but not considered as good Java programming
style to put own helper class (such as token classes) in this section.
They should get their own <TT>.java</TT> file instead.
<P>
<H2><A NAME="SECTION00052000000000000000"></A><A NAME="SpecOptions"></A><BR>
Options and declarations
</H2>
The second part of the lexical specification contains <A HREF="#SpecOptDirectives">options</A>
to customise your generated lexer (JFlex directives and Java code to include in
different parts of the lexer), declarations of <A HREF="#StateDecl">lexical states</A> and
<A HREF="#MacroDefs">macro definitions</A> for use in the third section
<A HREF="#LexRules">``Lexical rules''</A> of the lexical specification file.
<A NAME="SpecOptDirectives"></A>
<P>
Each JFlex directive must be situated at the beginning of a line
and starts with the <TT>%</TT> character. Directives that have one or
more parameters are described as follows:
<P>
<TT>%class "classname"</TT>
<P>
means that you start a line with <TT>%class</TT> followed by a space followed
by the name of the class for the generated scanner (the double quotes are
<I>not</I> to be entered, see the <A HREF="#CodeOptions">example specification</A> in
section <A HREF="#CodeOptions">3</A>).
<P>
<H3><A NAME="SECTION00052100000000000000"></A><A NAME="ClassOptions"></A><BR>
Class options and user class code
</H3>
These options regard name, constructor, API, and related parts of the
generated scanner class.
<UL>
<LI><B><TT>%class "classname"</TT></B>
<P>
Tells JFlex to give the generated class the name "<TT>classname</TT>" and to
write the generated code to a file "<TT>classname.java</TT>". If the
<TT>-d <directory></TT> command line option is not used, the code
will be written to the directory where the specification file resides. If
no <TT>%class</TT> directive is present in the specification, the generated
class will get the name "<TT>Yylex</TT>" and will be written to a file
"<TT>Yylex.java</TT>". There should be only one <TT>%class</TT> directive
in a specification.
<P>
</LI>
<LI><B><TT>%implements "interface 1"[, "interface 2", ..]</TT></B>
<P>
Makes the generated class implement the specified interfaces. If more than
one <TT>%implements</TT> directive is present, all the specified interfaces
will be implemented.
<P>
</LI>
<LI><B><TT>%extends "classname"</TT></B>
<P>
Makes the generated class a subclass of the class ``<TT>classname</TT>''.
There should be only one <TT>%extends</TT> directive in a specification.
<P>
</LI>
<LI><B><TT>%public</TT></B>
<P>
Makes the generated class public (the class is only accessible in its
own package by default).
<P>
</LI>
<LI><B><TT>%final</TT></B>
<P>
Makes the generated class final.
<P>
</LI>
<LI><B><TT>%abstract</TT></B>
<P>
Makes the generated class abstract.
<P>
</LI>
<LI><B><TT>%apiprivate</TT></B>
<P>
Makes all generated methods and fields of the class
private. Exceptions are the constructor, user code in the
specification, and, if <code>%cup</code> is present, the method
<TT>next_token</TT>. All occurrences of
<TT>" public "</TT> (one space character before and after <TT>public</TT>)
in the skeleton file are replaced by
<TT>" private "</TT> (even if a user-specified skeleton is used).
Access to the generated class is expected to be mediated by user class
code (see next switch).
<P>
</LI>
<LI><B><code>%{</code></B>
<BR><B><TT>...</TT></B>
<BR><B><code>%}</code></B>
<P>
The code enclosed in <code>%{</code> and <code>%}</code> is copied verbatim
into the generated class. Here you can define your own member variables
and functions in the generated scanner. Like all options, both <code>%{</code>
and <code>%}</code> must start a line in the specification. If more than one
class code directive <code>%{...%}</code> is present, the code is concatenated
in order of appearance in the specification.
<P>
</LI>
<LI><B><code>%init{</code></B>
<BR><B><TT>...</TT></B>
<BR><B><code>%init}</code></B>
<P>
The code enclosed in <code>%init{</code> and <code>%init}</code> is copied
verbatim into the constructor of the generated class. Here, member
variables declared in the <code>%{...%}</code> directive can be initialised.
If more than one initialiser option is present, the code is concatenated
in order of appearance in the specification.
<P>
</LI>
<LI><B><code>%initthrow{</code></B>
<BR><B><TT>"exception1"[, "exception2", ...]</TT></B>
<BR><B><code>%initthrow}</code></B>
<P>
or (on a single line) just
<P>
<B><TT>%initthrow "exception1" [, "exception2", ...]</TT></B>
<P>
Causes the specified exceptions to be declared in the <TT>throws</TT>
clause of the constructor. If more than one <code>%initthrow{</code> <TT>...</TT> <code>%initthrow}</code>
directive is present in the specification, all specified exceptions will
be declared.
<P>
</LI>
<LI><B><TT>%ctorarg "type" "ident"</TT></B>
<P>
Adds the specified argument to the constructors of the generated scanner.
If more than one such directive is present, the arguments are added in order
of occurrence in the specification. Note that this option conflicts with
the <code>%standalone</code> and <code>%debug</code> directives, because there is no
sensible default that can be created automatically for such parameters
in the generated <TT>main</TT> methods. JFlex will warn in this case and
generate an additional default constructor without these parameters and without user init code (which might potentially refer to the parameters).
<P>
</LI>
<LI><B><TT>%scanerror "exception"</TT></B>
<P>
Causes the generated scanner to throw an instance of the specified
exception in case of an internal error (default is
<TT>java.lang.Error</TT>). Note that this exception is only for
internal scanner errors. With usual specifications it should never
occur (i.e. if there is an error fallback rule in the specification
and only the documented scanner API is used).
<P>
</LI>
<LI><B><TT>%buffer "size"</TT></B>
<P>
Set the initial size of the scan buffer to the specified value
(decimal, in bytes). The default value is 16384.
<P>
</LI>
<LI><B><TT>%include "filename"</TT></B>
<P>
Replaces the <TT>%include</TT> verbatim by the specified file. This
feature is still experimental. It works, but error reporting can be
strange if a syntax error occurs on the last token in the included
file.
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052200000000000000"></A><A NAME="ScanningMethod"></A><BR>
Scanning method
</H3>
This section shows how the scanning method can be customised. You can redefine
the name and return type of the method and it is possible to declare
exceptions that may be thrown in one of the actions of the specification.
If no return type is specified, the scanning method will be declared as
returning values of class <TT>Yytoken</TT>.
<UL>
<LI><B><TT>%function "name"</TT></B>
<P>
Causes the scanning method to get the specified name. If no <TT>%function</TT>
directive is present in the specification, the scanning method gets the
name ``<TT>yylex</TT>''. This directive overrides settings of the
<TT><A HREF="#CupMode">%cup</A></TT> switch. Please note that the default name
of the scanning method with the <TT><A HREF="#CupMode">%cup</A></TT> switch is
<TT>next_token</TT>. Overriding this name might lead to the generated scanner
being implicitly declared as <TT>abstract</TT>, because it does not provide
the method <TT>next_token</TT> of the interface <TT>java_cup.runtime.Scanner</TT>.
It is of course possible to provide a dummy implementation of that method
in the class code section if you still want to override the function name.
<P>
</LI>
<LI><B><TT>%integer</TT></B>
<BR><B><TT>%int</TT></B>
<P>
Both cause the scanning method to be declared as of Java type <TT>int</TT>.
Actions in the specification can then return <TT>int</TT> values as tokens.
The default end of file value under this setting is <TT>YYEOF</TT>, which is a <TT>public
static final int</TT> member of the generated class.
<P>
</LI>
<LI><B><TT>%intwrap</TT></B>
<P>
Causes the scanning method to be declared as of the Java wrapper type
<TT>Integer</TT>. Actions in the specification can then return <TT>Integer</TT>
values as tokens. The default end of file value under this setting is <TT>null</TT>.
<P>
</LI>
<LI><B><TT>%type "typename"</TT></B>
<P>
Causes the scanning method to be declared as returning values of the specified type.
Actions in the specification can then return values of <TT>typename</TT>
as tokens. The default end of file value under this setting is <TT>null</TT>.
If <TT>typename</TT> is not a subclass of <TT>java.lang.Object</TT>,
you should specify another end of file value using the
<A HREF="#eofval"><TT>%eofval{</TT> <TT>...</TT> <TT>%eofval}</TT></A>
directive or the <A HREF="#EOFRule"><TT><<EOF>></TT> rule</A>.
The <TT>%type</TT> directive overrides settings of the
<TT><A HREF="#CupMode">%cup</A></TT> switch.
<P>
</LI>
<LI><B><code>%yylexthrow{</code></B>
<BR><B><TT>"exception1"[, "exception2", ... ]</TT></B>
<BR><B><code>%yylexthrow}</code></B>
<P>
or (on a single line) just
<P>
<B><TT>%yylexthrow "exception1" [, "exception2", ...]</TT></B>
<P>
The exceptions listed inside <code>%yylexthrow{</code> <TT>...</TT> <code>%yylexthrow}</code>
will be declared in the throws clause of the scanning method. If there is
more than one <code>%yylexthrow{</code> <TT>...</TT> <code>%yylexthrow}</code> clause in
the specification, all specified exceptions will be declared.
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052300000000000000"></A><A NAME="EOF"></A><BR>
The end of file
</H3>
There is always a default value that the scanning method will return when
the end of file has been reached. You may however define a specific value
to return and a specific piece of code that should be executed when the
end of file is reached.
<P>
The default end of file values depends on the return type of the scanning method:
<UL>
<LI>For <B><TT>%integer</TT></B>, the scanning method will return the value
<B><TT>YYEOF</TT></B>, which is a <TT>public static final int</TT> member
of the generated class.
<P>
</LI>
<LI>For <B><TT>%intwrap</TT></B>,
</LI>
<LI>no specified type at all, or a
</LI>
<LI>user defined type, declared using <B><TT>%type</TT></B>, the value is <B><TT>null</TT></B>.
<P>
</LI>
<LI>In CUP compatibility mode, using <B><TT>%cup</TT></B>, the value is
<P>
<B><TT>new java_cup.runtime.Symbol(sym.EOF)</TT></B>
</LI>
</UL>
<P>
User values and code to be executed at the end of file can be defined using these directives:
<A NAME="eofval"></A><UL>
<LI><B><code>%eofval{</code></B>
<BR><B><TT>...</TT></B>
<BR><B><code>%eofval}</code></B>
<P>
The code included in <code>%eofval{</code> <TT>...</TT> <code>%eofval}</code> will
be copied verbatim into the scanning method and will be executed <EM>each time</EM>
when the end of file is reached (this is possible when
the scanning method is called again after the end of file has been
reached). The code should return the value that indicates the end of
file to the parser. There should be only one <code>%eofval{</code>
<TT>...</TT> <code>%eofval}</code> clause in the specification.
The <code>%eofval{ ... %eofval}</code> directive overrides settings of the
<TT><A HREF="#CupMode">%cup</A></TT> switch and <TT><A HREF="#YaccMode">%byaccj</A></TT> switch.
As of version 1.2 JFlex provides
a more readable way to specify the end of file value using the
<A HREF="#EOFRule"><TT><<EOF>></TT> rule</A> (see also section <A HREF="#EOFRule">4.3.2</A>).
<P>
</LI>
<LI><A NAME="eof"></A> <B><code>%eof{</code></B>
<BR> <B><TT>...</TT></B>
<BR> <B><code>%eof}</code></B>
<P>
The code included in <code>%{eof ... %eof}</code> will be executed
exactly once, when the end of file is reached. The code is included
inside a method <TT>void yy_do_eof()</TT> and should not return any
value (use <code>%eofval{...%eofval}</code> or
<A HREF="#EOFRule"><TT><<EOF>></TT></A> for this purpose). If more than one
end of file code directive is present, the code will be concatenated
in order of appearance in the specification.
<P>
</LI>
<LI><B><code>%eofthrow{</code></B>
<BR> <B><TT>"exception1"[,"exception2", ... ]</TT></B>
<BR> <B><code>%eofthrow}</code></B>
<P>
or (on a single line) just
<P>
<B><TT>%eofthrow "exception1" [, "exception2", ...]</TT></B>
<P>
The exceptions listed inside <code>%eofthrow{...%eofthrow}</code> will
be declared in the throws clause of the method <TT>yy_do_eof()</TT>
(see <A HREF="#eof"><TT>%eof</TT></A> for more on that method).
If there is more than one <code>%eofthrow{...%eofthrow}</code> clause
in the specification, all specified exceptions will be declared.
<P>
<A NAME="eofclose"></A></LI>
<LI><B><TT>%eofclose</TT></B>
<P>
Causes JFlex to close the input stream at the end of file. The code
<TT>yyclose()</TT> is appended to the method <TT>yy_do_eof()</TT>
(together with the code specified in <code>%eof{...%eof}</code>) and
the exception <TT>java.io.IOException</TT> is declared in the throws
clause of this method (together with those of
<code>%eofthrow{...%eofthrow}</code>)
<P>
</LI>
<LI><B><TT>%eofclose false</TT></B>
<P>
Turns the effect of <TT>%eofclose</TT> off again (e.g. in case closing of
input stream is not wanted after <TT>%cup</TT>).
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052400000000000000"></A><A NAME="Standalone"></A><BR>
Standalone scanners
</H3>
<UL>
<LI><B><TT>%debug</TT></B>
<P>
Creates a main function in the generated class that expects the name
of an input file on the command line and then runs the scanner on this
input file by printing information about each returned token to the Java
console until the end of file is reached. The information includes:
line number (if line counting is enabled), column (if column counting is enabled),
the matched text, and the executed action (with line number in the specification).
<P>
</LI>
<LI><B><TT>%standalone</TT></B>
<P>
Creates a main function in the generated class that expects the name
of an input file on the command line and then runs the scanner on this
input file. The values returned by the scanner are ignored, but any unmatched
text is printed to the Java console instead (as the C/C++ tool flex does, if
run as standalone program). To avoid having to use an extra token class, the
scanning method will be declared as having default type <TT>int</TT>, not <TT>YYtoken</TT>
(if there isn't any other type explicitly specified).
This is in most cases irrelevant, but could be useful to know when making
another scanner standalone for some purpose. You should also consider using
the <TT>%debug</TT> directive, if you just want to be able to run the scanner
without a parser attached for testing etc.
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052500000000000000"></A><A NAME="CupMode"></A><BR>
CUP compatibility
</H3>
You may also want to read section <A HREF="#CUPWork">8.1</A> <A HREF="#CUPWork"><I>JFlex and CUP</I></A>
if you are interested in how to interface your generated
scanner with CUP.
<UL>
<LI><B><TT>%cup</TT></B>
<P>
The <TT>%cup</TT> directive enables the CUP compatibility mode and is equivalent
to the following set of directives:
<P>
<PRE>
%implements java_cup.runtime.Scanner
%function next_token
%type java_cup.runtime.Symbol
%eofval{
return new java_cup.runtime.Symbol(<CUPSYM>.EOF);
%eofval}
%eofclose
</PRE>
<P>
The value of <TT><CUPSYM></TT> defaults to <TT>sym</TT> and can be
changed with the <TT>%cupsym</TT> directive. In JLex compatibility
mode (<TT>-jlex</TT> switch on the command line), <TT>%eofclose</TT>
will not be turned on.
<P>
</LI>
<LI><B><TT>%cupsym "classname"</TT></B>
<P>
Customises the name of the CUP generated class/interface
containing the names of terminal tokens. Default is <TT>sym</TT>.
The directive should not be used after <TT>%cup</TT>, but before.
<P>
</LI>
<LI><B><TT>%cupdebug</TT></B>
<P>
Creates a main function in the generated class that expects the name
of an input file on the command line and then runs the scanner on this
input file. Prints line, column, matched text, and CUP symbol name for
each returned token to standard out.
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052600000000000000"></A><A NAME="YaccMode"></A><BR>
BYacc/J compatibility
</H3>
You may also want to read section <A HREF="#YaccWork">8.2</A> <A HREF="#YaccWork"><I>JFlex and BYacc/J</I></A>
if you are interested in how to interface your generated
scanner with Byacc/J.
<UL>
<LI><B><TT>%byacc</TT></B>
<P>
The <TT>%byacc</TT> directive enables the BYacc/J compatibility mode and is equivalent
to the following set of directives:
<P>
<PRE>
%integer
%eofval{
return 0;
%eofval}
%eofclose
</PRE>
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052700000000000000"></A><A NAME="CodeGeneration"></A><BR>
Code generation
</H3>
The following options define what kind of lexical analyser code JFlex
will produce. <TT>%pack</TT> is the default setting and will be used,
when no code generation method is specified.
<P>
<UL>
<LI><B><TT>%switch</TT></B>
<P>
With <TT>%switch</TT> JFlex will generate a scanner that has
the DFA hard coded into a nested switch statement. This method gives
a good deal of compression in terms of the size of the compiled
<TT>.class</TT> file while still providing very good performance. If your
scanner gets to big though (say more than about 200 states)
performance may vastly degenerate and you should consider using one
of the <TT>%table</TT> or <TT>%pack</TT> directives. If your scanner
gets even bigger (about 300 states), the Java compiler <TT>javac</TT>
could produce corrupted code, that will crash when executed or will
give you an <TT>java.lang.VerifyError</TT> when checked by the virtual
machine. This is due to the size limitation of 64 KB of Java
methods as described in the Java Virtual Machine Specification
[<A
HREF="manual.html#MachineSpec">10</A>]. In this case you will be forced to use the
<TT>%pack</TT> directive, since <TT>%switch</TT>
usually provides more compression of the DFA table than the
<TT>%table</TT> directive.
<P>
</LI>
<LI><B><TT>%table</TT></B>
<P>
The <TT>%table</TT> direction causes JFlex to produce a classical
table driven scanner that encodes its DFA table in an array. In
this mode, JFlex only does a small amount of table compression (see
[<A
HREF="manual.html#ParseTable">6</A>], [<A
HREF="manual.html#SparseTable">12</A>], [<A
HREF="manual.html#Aho">1</A>] and [<A
HREF="manual.html#Maurer">13</A>]
for more details on the matter of table compression) and uses the
same method that JLex did up to version 1.2.1. See section <A HREF="#performance">6</A>
<A HREF="#performance">performance</A> of this manual to compare
these methods. The same reason as above (64 KB size limitation of
methods) causes the same problem, when the scanner gets too big.
This is, because the virtual machine treats static initialisers of
arrays as normal methods. You will in this case again be forced to
use the <TT>%pack</TT> directive to avoid the problem.
<P>
</LI>
<LI><B><TT>%pack</TT></B>
<P>
<TT>%pack</TT> causes JFlex to compress the generated DFA table and to
store it in one or more string literals. JFlex takes care that the
strings are not longer than permitted by the class file format.
The strings have to be unpacked when
the first scanner object is created and initialised.
After unpacking the internal access to the DFA table is exactly the
same as with option <TT>%table</TT> -- the only extra work to be done
at runtime is the unpacking process which is quite fast (not noticeable
in normal cases). It is in time complexity proportional to the
size of the expanded DFA table, and it is static,
i.e. it is done only once for a certain scanner class -- no matter
how often it is instantiated. Again, see section
<A HREF="#performance">6</A> <A HREF="#performance">performance</A>
on the performance of these scanners
With <TT>%pack</TT>, there should be practically no
limitation to the size of the scanner. <TT>%pack</TT> is the default
setting and will be used when no code generation method is specified.
</LI>
</UL>
<P>
<H3><A NAME="SECTION00052800000000000000"></A><A NAME="CharacterSets"></A><BR>
Character sets
</H3>
<UL>
<LI><B><TT>%7bit</TT></B>
<P>
Causes the generated scanner to use an 7 bit input character set (character
codes 0-127). If an input character with a code greater than 127 is
encountered in an input at runtime, the scanner will throw an <TT>ArrayIndexOutofBoundsException</TT>.
Not only because of this, you should consider using the <TT>%unicode</TT> directive.
See also section <A HREF="#sec:encodings">5</A> for information about character encodings. This is the default in JLex compatibility mode.
<P>
</LI>
<LI><B><TT>%full</TT></B>
<BR><B><TT>%8bit</TT></B>
<P>
Both options cause the generated scanner to use an 8 bit input character
set (character codes 0-255). If an input character with a code greater
than 255 is encountered in an input at runtime, the scanner will throw
an <TT>ArrayIndexOutofBoundsException</TT>. Note that even if your platform
uses only one byte per character, the Unicode value of a character may
still be greater than 255. If you are scanning text files, you should
consider using the <TT>%unicode</TT> directive. See also section <A HREF="#sec:encodings">5</A>
for more information about character encodings.
<P>
</LI>
<LI><B><TT>%unicode</TT></B>
<BR><B><TT>%16bit</TT></B>
<P>
Both options cause the generated scanner to use the full 16 bit Unicode input
character set that Java supports natively (character code points 0-65535).
There will be no runtime overflow when using this set of input characters.
<TT>%unicode</TT> does not mean that the scanner will read two bytes at a
time. What is read and what constitutes a character depends on the runtime
platform. See also section <A HREF="#sec:encodings">5</A> for more information about
character encodings. This is the default unless the JLex compatibility mode is
used (command line option <TT>-jlex</TT>).
<P>
<A NAME="caseless"></A></LI>
<LI><B><TT>%caseless</TT></B>
<BR><B><TT>%ignorecase</TT></B>
<P>
This option causes JFlex to handle all characters and strings in the
specification as if they were specified in both uppercase and lowercase form.
This enables an easy way to specify a scanner for a language with case
insensitive keywords. The string "<TT>break</TT>" in a specification is for
instance handled like the expression <TT>([bB][rR][eE][aA][kK])</TT>. The
<TT>%caseless</TT> option does not change the matched text and does not
effect character classes. So <TT>[a]</TT> still only matches the character
<TT>a</TT> and not <TT>A</TT>, too. Which letters are uppercase and which
lowercase letters, is defined by the Unicode standard and determined by JFlex
with the Java methods <TT>Character.toUpperCase</TT> and
<TT>Character.toLowerCase</TT>. In JLex compatibility mode (<TT>-jlex</TT>
switch on the command line), <TT>%caseless</TT> and <TT>%ignorecase</TT>
also affect character classes.
<P>
</LI>
</UL>
<H3><A NAME="SECTION00052900000000000000"></A><A NAME="Counting"></A><BR>
Line, character and column counting
</H3>
<UL>
<LI><B><TT>%char</TT></B>
<P>
Turns character counting on. The <TT>int</TT> member variable <TT>yychar</TT>
contains the number of characters (starting with 0) from the beginning
of input to the beginning of the current token.
<P>
</LI>
<LI><B><TT>%line</TT></B>
<P>
Turns line counting on. The <TT>int</TT> member variable <TT>yyline</TT>
contains the number of lines (starting with 0) from the beginning of input
to the beginning of the current token.
<P>
</LI>
<LI><B><TT>%column</TT></B>
<P>
Turns column counting on. The <TT>int</TT> member variable <TT>yycolumn</TT>
contains the number of characters (starting with 0) from the beginning
of the current line to the beginning of the current token.
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION000521000000000000000"></A><A NAME="Obsolete"></A><BR>
Obsolete JLex options
</H3>
<UL>
<LI><B><TT>%notunix</TT></B>
<P>
This JLex option is obsolete in JFlex but still recognised as valid directive.
It used to switch between Windows and Unix kind of line terminators (<code>\r\n</code>
and <code>\n</code>) for the <TT>$</TT> operator in regular expressions. JFlex
always recognises both styles of platform dependent line terminators.
<P>
</LI>
<LI><B><TT>%yyeof</TT></B>
<P>
This JLex option is obsolete in JFlex but still recognised as valid directive.
In JLex it declares a public member constant <TT>YYEOF</TT>. JFlex declares it in any case.
</LI>
</UL>
<P>
<H3><A NAME="SECTION000521100000000000000"></A><A NAME="StateDecl"></A><BR>
State declarations
</H3>
State declarations have the following from:
<P>
<TT>%s[tate] "state identifier" [, "state identifier", ... ]</TT> for inclusive or
<BR><TT>%x[state] "state identifier" [, "state identifier", ... ]</TT> for exclusive states
<P>
There may be more than one line of state declarations, each starting with
<TT>%state</TT> or <TT>%xstate</TT> (the first character is sufficient,
<TT>%s</TT> and <TT>%x</TT> works, too). State identifiers are letters followed
by a sequence of letters, digits or underscores. State identifiers can be separated
by white-space or comma.
<P>
The sequence
<P>
<TT>%state STATE1</TT>
<BR><TT>%xstate STATE3, XYZ, STATE_10</TT>
<BR><TT>%state ABC STATE5</TT>
<P>
declares the set of identifiers <TT>STATE1, STATE3, XYZ,
STATE_10, ABC, STATE5</TT> as lexical states, <TT>STATE1</TT>, <TT>ABC</TT>, <TT>STATE5</TT>
as inclusive, and <TT>STATE3</TT>, <TT>XYZ</TT>, <TT>STATE_10</TT> as exclusive.
See also section
<A HREF="#HowMatched">4.3.3</A> on the way lexical states influence how the input is
matched.
<P>
<H3><A NAME="SECTION000521200000000000000"></A><A NAME="MacroDefs"></A><BR>
Macro definitions
</H3>
A macro definition has the form
<P>
<TT>macroidentifier = regular expression</TT>
<P>
That means, a macro definition is a macro identifier (letter followed
by a sequence of letters, digits or underscores), that can later be
used to reference the macro, followed by optional white-space, followed
by an "<TT>=</TT>", followed by optional white-space, followed by a
regular expression (see section <A HREF="#LexRules">4.3</A> <A HREF="#LexRules"><I>lexical
rules</I></A> for more information about regular expressions).
<P>
The regular expression on the right hand side must be well formed and
must not contain the <code>^</code>, <TT>/</TT> or <TT>$</TT> operators. <B>Differently
to JLex, macros are not just pieces of text that are expanded by copying</B>
- they are parsed and must be well formed.
<P>
<B>This is a feature.</B> It eliminates some very hard to find bugs in
lexical specifications (such like not having parentheses around more
complicated macros - which is not necessary with JFlex). See section
<A HREF="#Porting">7.1</A> <A HREF="#Porting"><I>Porting from JLex</I></A> for more
details on the problems of JLex style macros.
<P>
Since it is allowed to have macro usages in macro definitions, it is
possible to use a grammar like notation to specify the desired lexical
structure. Macros however remain just abbreviations of the regular expressions
they represent. They are not non terminals of a grammar and cannot be used
recursively in any way. JFlex detects cycles in macro definitions and reports
them at generation time. JFlex also warns you about macros that have been
defined but never used in the ``lexical rules'' section of the specification.
<P>
<H2><A NAME="SECTION00053000000000000000"></A><A NAME="LexRules"></A><BR>
Lexical rules
</H2>
The ``lexical rules'' section of an JFlex specification contains a set of
regular expressions and actions (Java code) that are executed when the
scanner matches the associated regular expression.
<P>
<H3><A NAME="SECTION00053100000000000000"></A><A NAME="Grammar"></A><BR>
Syntax
</H3>
The syntax of the "lexical rules" section is described by the following
BNF grammar (terminal symbols are enclosed in 'quotes'):
<P>
<PRE>
LexicalRules ::= Rule+
Rule ::= [StateList] ['^'] RegExp [LookAhead] Action
| [StateList] '<<EOF>>' Action
| StateGroup
StateGroup ::= StateList '{' Rule+ '}'
StateList ::= '<' Identifier (',' Identifier)* '>'
LookAhead ::= '$' | '/' RegExp
Action ::= '{' JavaCode '}' | '|'
RegExp ::= RegExp '|' RegExp
| RegExp RegExp
| '(' RegExp ')'
| ('!'|'~') RegExp
| RegExp ('*'|'+'|'?')
| RegExp "{" Number ["," Number] "}"
| '[' ['^'] (Character|Character'-'Character)* ']'
| PredefinedClass
| '{' Identifier '}'
| '"' StringCharacter+ '"'
| Character
PredefinedClass ::= '[:jletter:]'
| '[:jletterdigit:]'
| '[:letter:]'
| '[:digit:]'
| '[:uppercase:]'
| '[:lowercase:]'
| '.'
</PRE>
<P>
<A NAME="Terminals"></A>The grammar uses the following terminal symbols:
<UL>
<LI><TT>JavaCode</TT>
<BR> a sequence of <EM><TT>BlockStatements</TT></EM> as described in the Java
Language Specification [<A
HREF="manual.html#LangSpec">7</A>], section 14.2.
<P>
</LI>
<LI><TT>Number</TT>
<BR> a non negative decimal integer.
<P>
</LI>
<LI><TT>Identifier</TT>
<BR> a letter <code>[a-zA-Z]</code> followed by a sequence of zero or more
letters, digits or underscores <code>[a-zA-Z0-9_]</code>
<P>
</LI>
<LI><TT>Character</TT>
<BR> an escape sequence or any unicode character that is not one of these
meta characters:
<code> | ( ) { } [ ] < > \ . * + ? ^ $ / . " ~ !</code>
<P>
</LI>
<LI><TT>StringCharacter</TT>
<BR> an escape sequence or any unicode character that is not one of these
meta characters:
<code> \ "</code>
<P>
</LI>
<LI>An escape sequence
<P>
<UL>
<LI><code>\n</code> <code>\r</code> <code>\t</code> <code>\f</code> <code>\b</code>
</LI>
<LI>a <code>\x</code> followed by two hexadecimal digits <TT>[a-fA-F0-9]</TT> (denoting
a standard ASCII escape sequence),
<P>
</LI>
<LI>a <code>\u</code> followed by four hexadecimal digits <TT>[a-fA-F0-9]</TT>
(denoting an unicode escape sequence),
<P>
</LI>
<LI>a backslash followed by a three digit octal number from 000 to 377 (denoting
a standard ASCII escape sequence), or
<P>
</LI>
<LI>a backslash followed by any other unicode character that stands for this
character.
<P>
</LI>
</UL>
<P>
</LI>
</UL>
<P>
Please note that the <code>\n</code> escape sequence stands for the ASCII
LF character - not for the end of line. If you would like to match the
line terminator, you should use the expression <code>\r|\n|\r\n</code> if you want
the Java conventions, or <code>\r|\n|\r\n|\u2028|\u2029|\u000B|\u000C|\u0085</code>
if you want to be fully Unicode compliant (see also [<A
HREF="manual.html#unicode_rep">5</A>]).
<P>
As of version 1.1 of JFlex the white-space characters <TT>" "</TT>
(space) and <code>"\t"</code> (tab) can be used to improve the readability of
regular expressions. They will be ignored by JFlex. In character
classes and strings however, white-space characters keep standing for
themselves (so the string <TT>" "</TT> still matches exactly one space
character and <code>[ \n]</code> still matches an ASCII LF or a space
character).
<P>
JFlex applies the following standard operator precedences in regular
expression (from highest to lowest):
<P>
<UL>
<LI>unary postfix operators (<code>'*', '+', '?', {n}, {n,m}</code>)
<P>
</LI>
<LI>unary prefix operators (<code>'!', '~'</code>)
<P>
</LI>
<LI>concatenation (<TT>RegExp::= RegExp Regexp</TT>)
<P>
</LI>
<LI>union (<code>RegExp::= RegExp '|' RegExp</code>)
</LI>
</UL>
<P>
So the expression <code>a | abc | !cd*</code> for instance is parsed as
<code>(a|(abc)) | ((!c)(d*))</code>.
<P>
<H3><A NAME="SECTION00053200000000000000"></A><A NAME="Semantics"></A><BR>
Semantics
</H3>
This section gives an informal description of which text is matched by
a regular expression (i.e. an expression described by the <TT>RegExp</TT>
production of the grammar presented <A HREF="#Grammar">above</A>).
<P>
A regular expression that consists solely of
<UL>
<LI>a <TT>Character</TT> matches this character.
<P>
</LI>
<LI>a character class <code>'[' (Character|Character'-'Character)* ']'</code> matches
any character in that class. A <TT>Character</TT> is to be considered an
element of a class, if it is listed in the class or if its code lies within
a listed character range <TT>Character'-'Character</TT>. So <code>[a0-3\n]</code>
for instance matches the characters
<P>
<code>a 0 1 2 3 \n</code>
<P>
If the list of characters is empty (i.e. just <code>[]</code>), the expression
matches nothing at all (the empty set), not even the empty string. This
may be useful in combination with the negation operator <code>'!'</code>.
<P>
</LI>
<LI>a negated character class <code>'[^' (Character|Character'-'Character)* ']'</code>
matches all characters not listed in the class. If the list of characters
is empty (i.e. <code>[^]</code>), the expression matches any character of the
input character set.
<P>
</LI>
<LI>a string <TT>'"' StringCharacter+ '"</TT> <TT>'</TT> matches the exact
text enclosed in double quotes. All meta characters but <code>\</code> and
<TT>"</TT> loose their special meaning inside a string. See also the
<A HREF="#caseless"><TT>%ignorecase</TT></A> switch.
<P>
</LI>
<LI>a macro usage <code>'{' Identifier '}'</code> matches the input that is matched
by the right hand side of the macro with name "<TT>Identifier</TT>".
<P>
<A NAME="predefCharCl"></A></LI>
<LI>a predefined character class matches any of
the characters in that class. There are the following predefined character
classes:
<P>
<TT>.</TT> contains all characters but <code>\n</code>.
<P>
All other predefined character classes are defined in the Unicode
specification or the Java Language Specification and determined by
Java functions of class
<TT>java</TT>.<TT>lang</TT>.<TT>Character</TT>.
<P>
<PRE>
[:jletter:] isJavaIdentifierStart()
[:jletterdigit:] isJavaIdentifierPart()
[:letter:] isLetter()
[:digit:] isDigit()
[:uppercase:] isUpperCase()
[:lowercase:] isLowerCase()
</PRE>
<P>
They are especially useful when working with the unicode character set.
<P>
</LI>
</UL>
<P>
If <TT>a</TT> and <TT>b</TT> are regular expressions, then
<P>
<DL COMPACT>
<DT><TT>a | b</TT></DT>
<DD>(union)
<P>
is the regular expression, that matches
all input that is matched by <TT>a</TT> or by <TT>b</TT>.
<P>
</DD>
<DT><TT>a b</TT></DT>
<DD>(concatenation)
<P>
is the regular expression,
that matches the input matched by <TT>a</TT> followed by the
input matched by <TT>b</TT>.
<P>
</DD>
<DT><TT>a*</TT></DT>
<DD>(Kleene closure)
<P>
matches zero or more repetitions
of the input matched by <TT>a</TT>
<P>
</DD>
<DT><TT>a+</TT></DT>
<DD>(iteration)
<P>
is equivalent to <TT>aa*</TT>
<P>
</DD>
<DT><TT>a?</TT></DT>
<DD>(option)
<P>
matches the empty input or the input matched
by <TT>a</TT>
<P>
</DD>
<DT><TT>!a</TT></DT>
<DD>(negation)
<P>
matches everything but the strings matched by <TT>a</TT>.
Use with care: the construction of <code>!a</code> involves
an additional, possibly exponential NFA to DFA transformation
on the NFA for <TT>a</TT>. Note that
with negation and union you also have (by applying DeMorgan)
intersection and set difference: the intersection of
<TT>a</TT> and <TT>b</TT> is <code>!(!a|!b)</code>, the expression
that matches everything of <TT>a</TT> not matched by <TT>b</TT> is
<code>!(!a|b)</code>
<P>
</DD>
<DT><TT>~a</TT></DT>
<DD>(upto)
<P>
matches everything up to (and including) the first occurrence of a text
matched by <TT>a</TT>. The expression <code>~a</code> is equivalent
to <code>!([^]* a [^]*) a</code>. A traditional C-style comment
is matched by <code>"/*" ~"*/"</code>
<P>
</DD>
<DT><TT>a{n}</TT></DT>
<DD>(repeat)
<P>
is equivalent to <TT>n</TT> times the concatenation of <TT>a</TT>.
So <code>a{4}</code> for instance is equivalent to the expression <TT>a a a a</TT>.
The decimal integer <TT>n</TT> must be positive.
<P>
</DD>
<DT><TT>a{n,m}</TT></DT>
<DD>is equivalent to at least <TT>n</TT> times and at most <TT>m</TT> times the
concatenation of <TT>a</TT>. So <code>a{2,4}</code> for instance is equivalent
to the expression <code>a a a? a?</code>. Both <TT>n</TT> and <TT>m</TT> are non
negative decimal integers and <TT>m</TT> must not be smaller than <TT>n</TT>.
<P>
</DD>
<DT><TT>( a )</TT></DT>
<DD>matches the same input as <TT>a</TT>.
<P>
</DD>
</DL>
<P>
In a lexical rule, a regular expression <TT>r</TT> may be preceded by a
'<code>^</code>' (the beginning of line operator). <TT>r</TT> is then
only matched at the beginning of a line in the input. A line begins
after each occurrence of <code>\r|\n|\r\n|\u2028|\u2029|\u000B|\u000C|\u0085</code>
(see also [<A
HREF="manual.html#unicode_rep">5</A>]) and at the beginning of input.
The preceding line terminator in the input is not consumed and can
be matched by another rule.
<P>
In a lexical rule, a regular expression <TT>r</TT> may be followed by a
look-ahead expression. A look-ahead expression is either a '<TT>$</TT>'
(the end of line operator) or a <code>'/'</code> followed by an arbitrary
regular expression. In both cases the look-ahead is not consumed and
not included in the matched text region, but it <EM>is</EM> considered
while determining which rule has the longest match (see also
<A HREF="#HowMatched">4.3.3</A> <A HREF="#HowMatched"><I>How the input is matched</I></A>).
<P>
In the '<TT>$</TT>' case <TT>r</TT> is only matched at the end of a line in
the input. The end of a line is denoted by the regular expression
<code>\r|\n|\r\n|\u2028|\u2029|\u000B|\u000C|\u0085</code>.
So <code>a$</code> is equivalent to <code>a / \r|\n|\r\n|\u2028|\u2029|\u000B|\u000C|\u0085</code>.This is a bit different to the situation described in [<A
HREF="manual.html#unicode_rep">5</A>]:
since in JFlex <code>$</code> is a true trailing context, the end of file
does <B>not</B> count as end of line.
<P>
<A NAME="trailingContext"></A>For arbitrary look-ahead (also called <EM>trailing context</EM>) the
expression is matched only when followed by input that matches the
trailing context.
<P>
<A NAME="EOFRule"></A>As of version 1.2, JFlex allows lex/flex style <TT>«EOF»</TT> rules in
lexical specifications. A rule
<PRE>
[StateList] <<EOF>> { some action code }
</PRE>
is very similar to the <A HREF="#eofval"><TT>%eofval</TT> directive</A> (section <A HREF="#eofval">4.2.3</A>).
The difference lies in the optional <TT>StateList</TT> that may precede the <TT>«EOF»</TT> rule. The
action code will only be executed when the end of file is read and the
scanner is currently in one of the lexical states listed in <TT>StateList</TT>.
The same <TT>StateGroup</TT> (see section <A HREF="#HowMatched">4.3.3</A>
<A HREF="#HowMatched"><I>How the input is matched</I></A>) and precedence
rules as in the ``normal'' rule case apply
(i.e. if there is more than one <TT>«EOF»</TT>
rule for a certain lexical state, the action of the one appearing
earlier in the specification will be executed). <TT>«EOF»</TT> rules
override settings of the <TT>%cup</TT> and <TT>%byaccj</TT> options and
should not be mixed with the <TT>%eofval</TT> directive.
<P>
An <TT>Action</TT> consists either of a piece of Java code enclosed in
curly braces or is the special <code>|</code> action. The <code>|</code> action is
an abbreviation for the action of the following expression.
<P>
Example:
<PRE>
expression1 |
expression2 |
expression3 { some action }
</PRE>
is equivalent to the expanded form
<PRE>
expression1 { some action }
expression2 { some action }
expression3 { some action }
</PRE>
<P>
They are useful when you work with trailing context expressions. The
expression <TT>a | (c / d) | b</TT> is not syntactically legal, but can
easily be expressed using the <code>|</code> action:
<PRE>
a |
c / d |
b { some action }
</PRE>
<P>
<H3><A NAME="SECTION00053300000000000000"></A><A NAME="HowMatched"></A><BR>
How the input is matched
</H3>
When consuming its input, the scanner determines the regular expression
that matches the longest portion of the input (longest match rule). If
there is more than one regular expression that matches the longest portion
of input (i.e. they all match the same input), the generated scanner chooses
the expression that appears first in the specification. After determining
the active regular expression, the associated action is executed. If there
is no matching regular expression, the scanner terminates the program with
an error message (if the <TT>%standalone</TT> directive has been used, the
scanner prints the unmatched input to <TT>java.lang.System.out</TT> instead
and resumes scanning).
<P>
Lexical states can be used to further restrict the set of regular expressions
that match the current input.
<P>
<UL>
<LI>A regular expression can only be matched when its associated set of lexical
states includes the currently active lexical state of the scanner or if
the set of associated lexical states is empty and the currently active lexical
state is inclusive. Exclusive and inclusive states only differ at this point:
rules with an empty set of associated states.
<P>
</LI>
<LI>The currently active lexical state of the scanner can be changed from within
an action of a regular expression using the method <TT>yybegin()</TT>.
<P>
</LI>
<LI>The scanner starts in the inclusive lexical state
<TT>YYINITIAL</TT>, which is always declared by default.
<P>
</LI>
<LI>The set of lexical states associated with a regular expression is
the <TT>StateList</TT> that precedes the expression. If a rule is
contained in one or more <TT>StateGroups</TT>, then the states of
these are also associated with the rule, i.e. they accumulate over
<TT>StateGroups</TT>.
<P>
Example:
<PRE>
%states A, B
%xstates C
%%
expr1 { yybegin(A); action }
<YYINITIAL, A> expr2 { action }
<A> {
expr3 { action }
<B,C> expr4 { action }
}
</PRE>
The first line declares two (inclusive) lexical states <TT>A</TT> and <TT>B</TT>,
the second line an exclusive lexical state <TT>C</TT>.
The default (inclusive) state <TT>YYINITIAL</TT> is always implicitly there and
doesn't need to be declared. The rule with <TT>expr1</TT> has no
states listed, and is thus matched in all states but the exclusive
ones, i.e. <TT>A</TT>, <TT>B</TT>, and <TT>YYINITIAL</TT>. In its
action, the scanner is switched to state <TT>A</TT>. The second rule
<TT>expr2</TT> can only match when the scanner is in state
<TT>YYINITIAL</TT> or <TT>A</TT>. The rule <TT>expr3</TT> can only be
matched in state <TT>A</TT> and <TT>expr4</TT> in states <TT>A</TT>, <TT>B</TT>,
and <TT>C</TT>.
<P>
</LI>
<LI>Lexical states are declared and used as Java <TT>int</TT> constants in
the generated class under the same name as they are used in the specification.
There is no guarantee that the values of these integer constants are
distinct. They are pointers into the generated DFA table, and if JFlex
recognises two states as lexically equivalent (if they are used with the
exact same set of regular expressions), then the two constants will get
the same value.
<P>
</LI>
</UL>
<P>
<H3><A NAME="SECTION00053400000000000000">
The generated class</A>
</H3>
JFlex generates exactly one file containing one class from the specification
(unless you have declared another class in the first specification section).
<P>
The generated class contains (among other things) the DFA tables, an input buffer,
the lexical states of the specification, a constructor, and the scanning method
with the user supplied actions.
<P>
The name of the class is by default <TT>Yylex</TT>, it is customisable
with the <TT>%class</TT> directive (see also section
<A HREF="#ClassOptions">4.2.1</A>). The input buffer of the lexer is connected with an
input stream over the <TT>java.io.Reader</TT> object which is passed
to the lexer in the generated constructor. If you want to provide your
own constructor for the lexer, you should always call the generated
one in it to initialise the input buffer. The input buffer should not
be accessed directly, but only over the advertised API (see also
section <A HREF="#ScannerMethods">4.3.5</A>). Its internal implementation may change
between releases or skeleton files without notice.
<P>
The main interface to the outside world is the generated scanning
method (default name <TT>yylex</TT>, default return type
<TT>Yytoken</TT>). Most of its aspects are customisable (name, return
type, declared exceptions etc., see also section
<A HREF="#ScanningMethod">4.2.2</A>). If it is called, it will consume input until
one of the expressions in the specification is matched or an error
occurs. If an expression is matched, the corresponding action is
executed. It may return a value of the specified return type (in which
case the scanning method return with this value), or if it doesn't
return a value, the scanner resumes consuming input until the next
expression is matched. If the end of file is reached, the scanner
executes the EOF action, and (also upon each further call to the scanning
method) returns the specified EOF value (see also section <A HREF="#EOF">4.2.3</A>).
<P>
<H3><A NAME="SECTION00053500000000000000"></A><A NAME="ScannerMethods"></A><BR>
Scanner methods and fields accessible in actions (API)
</H3>
Generated methods and member fields in JFlex scanners are prefixed
with <TT>yy</TT> to indicate that they are generated and to avoid name
conflicts with user code copied into the class. Since user code is
part of the same class, JFlex has no language means like the
<TT>private</TT> modifier to indicate which members and methods are
internal and which ones belong to the API. Instead, JFlex follows a
naming convention: everything starting with a <TT>zz</TT> prefix like
<TT>zzStartRead</TT> is to be considered internal and subject to
change without notice between JFlex releases. Methods and members of
the generated class that do not have a <TT>zz</TT> prefix like
<TT>yycharat</TT> belong to the API that the scanner class provides to
users in action code of the specification. They will be remain stable
and supported between JFlex releases as long as possible.
<P>
Currently, the API consists of the following methods and member fields:
<UL>
<LI><TT>String yytext()</TT>
<BR> returns the matched input text region
<P>
</LI>
<LI><TT>int yylength()</TT>
<BR> returns the length of the matched input text region (does not require
a <TT>String</TT> object to be created)
<P>
</LI>
<LI><TT>char yycharat(int pos)</TT>
<BR> returns the character at position <TT>pos</TT> from the matched text.
It is equivalent to <TT>yytext().charAt(pos)</TT>, but faster. <TT> pos</TT> must be a value from <TT>0</TT> to <TT>yylength()-1</TT>.
<P>
</LI>
<LI><TT>void yyclose()</TT>
<BR> closes the input stream. All subsequent calls to the scanning method will
return the end of file value
<P>
</LI>
<LI><TT>void yyreset(java.io.Reader reader)</TT>
<BR> closes the current input stream, and resets the scanner to read from
a new input stream. All internal variables are reset, the old input
stream <EM>cannot</EM> be reused (content of the internal buffer is
discarded and lost). The lexical state is set to <TT>YY_INITIAL</TT>.
<P>
</LI>
<LI><TT>void yypushStream(java.io.Reader reader)</TT>
<BR> Stores the current input stream on a stack, and
reads from a new stream. Lexical state, line,
char, and column counting remain untouched.
The current input stream can be restored with
<TT>yypopstream</TT> (usually in an <TT>«EOF»</TT> action).
<P>
A typical example for this are include files in
style of the C pre-processor. The corresponding
JFlex specification could look somewhat like this:
<PRE>
"#include" {FILE} { yypushStream(new FileReader(getFile(yytext()))); }
..
<<EOF>> { if (yymoreStreams()) yypopStream(); else return EOF; }
</PRE>
<P>
This method is only available in the skeleton file
<TT>skeleton.nested</TT>. You can find it in the
<TT>src</TT> directory of the JFlex distribution.
<P>
</LI>
<LI><TT>void yypopStream()</TT>
<BR> Closes the current input stream and continues to
read from the one on top of the stream stack.
<P>
This method is only available in the skeleton file
<TT>skeleton.nested</TT>. You can find it in the
<TT>src</TT> directory of the JFlex distribution.
<P>
</LI>
<LI><TT>boolean yymoreStreams()</TT>
<BR> Returns true iff there are still streams for <TT>yypopStream</TT>
left to read from on the stream stack.
<P>
This method is only available in the skeleton file
<TT>skeleton.nested</TT>. You can find it in the
<TT>src</TT> directory of the JFlex distribution.
<P>
</LI>
<LI><TT>int yystate()</TT>
<BR> returns the current lexical state of the scanner.
<P>
</LI>
<LI><TT>void yybegin(int lexicalState)</TT>
<BR> enters the lexical state <TT>lexicalState</TT>
<P>
</LI>
<LI><TT>void yypushback(int number)</TT>
<BR> pushes <TT>number</TT> characters of the matched text back into the input stream.
They will be read again in the next call of the scanning method.
The number of characters to be read again must not be greater than the length
of the matched text. The pushed back characters will after the call of
<TT>yypushback</TT> not be included in <TT>yylength</TT> and <TT>yytext()</TT>.
Please note that in Java strings are unchangeable, i.e. an action code like
<PRE>
String matched = yytext();
yypushback(1);
return matched;
</PRE>
will return the whole matched text, while
<PRE>
yypushback(1);
return yytext();
</PRE>
will return the matched text minus the last character.
<P>
</LI>
<LI><TT>int yyline</TT>
<BR> contains the current line of input (starting with 0, only active with
the <TT><A HREF="#Counting">%line</A></TT> directive)
<P>
</LI>
<LI><TT>int yychar</TT>
<BR> contains the current character count in the input (starting with 0,
only active with the <TT><A HREF="#Counting">%char</A></TT> directive)
<P>
</LI>
<LI><TT>int yycolumn</TT>
<BR> contains the current column of the current line (starting with 0, only
active with the <TT><A HREF="#Counting">%column</A></TT> directive)
<P>
</LI>
</UL>
<P>
<H1><A NAME="SECTION00060000000000000000"></A><A NAME="sec:encodings"></A><BR>
Encodings, Platforms, and Unicode
</H1>
<P>
This section tries to shed some light on the issues of Unicode and
encodings, cross platform scanning, and how to deal with binary data.
My thanks go to Stephen Ostermiller for his input on this topic.
<P>
<H2><A NAME="SECTION00061000000000000000"></A><A NAME="sec:howtoencoding"></A><BR>
The Problem
</H2>
<P>
Before we dive straight into details, let's take a look at what the
problem is. The problem is Java's platform independence when you want
to use it. For scanners the interesting part about platform
independence is character encodings and how they are handled.
<P>
If a program reads a file from disk, it gets a stream of bytes. In
earlier times, when the grass was green, and the world was much
simpler, everybody knew that the byte value 65 is, of course, an A.
It was no problem to see which bytes meant which characters (actually
these times never existed, but anyway). The normal Latin alphabet
only has 26 characters, so 7 bits or 128 distinct values should surely
be enough to map them, even if you allow yourself the luxury of upper
and lower case. Nowadays, things are different. The world suddenly
grew much larger, and all kinds of people wanted all kinds of special
characters, just because they use them in their language and writing.
This is were the mess starts. Since the 128 distinct values were
already filled up with other stuff, people began to use all 8 bits of
the byte, and extended the byte/character mappings to fit their need,
and of course everybody did it differently. Some people for instance
may have said ``let's use the value 213 for the German character ä''. Others
may have found that 213 should much rather mean é, because they didn't need
German and wrote French instead. As long as you use your program and
data files only on one platform, this is no problem, as all know what
means what, and everything gets used consistently.
<P>
Now Java comes into play, and wants to run everywhere (once written,
that is) and now there suddenly is a problem: how do I get the same
program to say ä to a certain byte when it runs in Germany and maybe é
when it runs in France? And also the other way around: when I want to
say é on the screen, which byte value should I send to the operating
system?
<P>
Java's solution to this is to use Unicode internally. Unicode aims to
be a superset of all known character sets and is therefore a perfect base
for encoding things that might get used all over the world. To make
things work correctly, you still have to know where you are and how to
map byte values to Unicode characters and vice versa, but the
important thing is, that this mapping is at least possible (you can
map Kanji characters to Unicode, but you cannot map them to ASCII or
iso-latin-1).
<P>
<H2><A NAME="SECTION00062000000000000000"></A><A NAME="sec:howtotext"></A><BR>
Scanning text files
</H2>
<P>
Scanning text files is the standard application for scanners like
JFlex. Therefore it should also be the most convenient one. Most times
it is.
<P>
The following scenario works like a breeze:
You work on a platform X, write your lexer specification there, can
use any obscure Unicode character in it as you like, and compile the
program. Your users work on any platform Y (possibly but not
necessarily something different from X), they write their input files
on Y and they run your program on Y. No problems.
<P>
Java does this as follows:
If you want to read anything in Java that is supposed to contain text,
you use a <TT>FileReader</TT> or some <TT>InputStream</TT> together with
an <TT>InputStreamReader</TT>. <TT>InputStreams</TT> return the raw bytes, the
<TT>InputStreamReader</TT> converts the bytes into Unicode characters with
the platform's default encoding. If a text file is produced on the
same platform, the platform's default encoding should do the mapping
correctly. Since JFlex also uses readers and Unicode internally, this
mechanism also works for the scanner specifications. If you write an
<TT>A</TT> in your text editor and the editor uses the platform's encoding (say <TT>A</TT> is 65),
then Java translates this into the logical Unicode <TT>A</TT> internally.
If a user writes an <TT>A</TT> on a completely different platform (say <TT>A</TT> is 237 there),
then Java also translates this into the logical Unicode <TT>A</TT> internally. Scanning
is performed after that translation and both match.
<P>
Note that because of this mapping from bytes to characters, you should always
use the <TT>%unicode</TT> switch in you lexer specification if you want to scan
text files. <TT>%8bit</TT> may not be enough, even if
you know that your platform only uses one byte per character. The encoding
Cp1252 used on many Windows machines for instance knows 256 characters, but
the character ´ with Cp1252 code <code>\x92</code> has the Unicode value <code>\u2019</code>, which
is larger than 255 and which would make your scanner throw an
<TT>ArrayIndexOutOfBoundsException</TT> if it is encountered.
<P>
So for the usual case you don't have to do anything but use the
<TT>%unicode</TT> switch in your lexer specification.
<P>
Things may break when you produce a text file on platform X and
consume it on a different platform Y. Let's say you have a file
written on a Windows PC using the encoding Cp1252. Then you move
this file to a Linux PC with encoding ISO 8859-1 and there you want
to run your scanner on it. Java now thinks the file is encoded
in ISO 8859-1 (the platform's default encoding) while it really is
encoded in Cp1252. For most characters
Cp1252 and ISO 8859-1 are the same, but for the byte values <code>\x80</code>
to <code>\x9f</code> they disagree: ISO 8859-1 is undefined there. You can fix
the problem by telling Java explicitly which encoding to use. When
constructing the <TT>InputStreamReader</TT>, you can give the encoding
as argument. The line
<DIV ALIGN="CENTER">
<TT>Reader r = new InputStreamReader(input, "Cp1252"); </TT>
</DIV>
will do the trick.
<P>
Of course the encoding to use can also come from the data itself:
for instance, when you scan a HTML page, it may have embedded
information about its character encoding in the headers.
<P>
More information about encodings, which ones are supported, how
they are called, and how to set them may be found in the
official Java documentation in the chapter about
internationalisation.
The link
<A NAME="tex2html6"
HREF="http://java.sun.com/j2se/1.3/docs/guide/intl/"><TT>http://java.sun.com/j2se/1.3/docs/guide/intl/</TT></A>
leads to an online version of this for Sun's JDK 1.3.
<P>
<H2><A NAME="SECTION00063000000000000000"></A><A NAME="sec:howtobinary"></A><BR>
Scanning binaries
</H2>
<P>
Scanning binaries is both easier and more difficult
than scanning text files. It's easier because you want
the raw bytes and not their meaning, i.e. you don't want
any translation.
It's more difficult because it's not so easy to get
``no translation'' when you use Java readers.
<P>
The problem (for binaries) is that JFlex scanners are
designed to work on text. Therefore the interface is
the <TT>Reader</TT> class (there is a constructor
for <TT>InputStream</TT> instances, but it's just there
for convenience and wraps an <TT>InputStreamReader</TT>
around it to get characters, not bytes).
You can still get a binary scanner when you write
your own custom <TT>InputStreamReader</TT> class that
does explicitly no translation, but just copies
byte values to character codes instead. It sounds
quite easy, and actually it is no big deal, but there
are a few little pitfalls on the way. In the scanner
specification you can only enter positive character
codes (for bytes that is <code>\x00</code>
to <code>\xFF</code>). Java's <TT>byte</TT> type on the other hand
is a signed 8 bit integer (-128 to 127), so you have to convert
them properly in your custom <TT>Reader</TT>. Also, you should
take care when you write your lexer spec: if you
use text in there, it gets interpreted by an encoding
first, and what scanner you get as result might depend
on which platform you run JFlex on when you generate
the scanner (this is what you want for text, but for binaries it
gets in the way). If you are not sure, or if the development
platform might change, it's probably best to use character
code escapes in all places, since they don't change their
meaning.
<P>
To illustrate these points, the example in <TT>examples/binary</TT>
contains a very small binary scanner that tries to
detect if a file is a Java <TT>class</TT> file. For that
purpose it looks if the file begins with the magic number <code>\xCAFEBABE</code>.
<P>
<H1><A NAME="SECTION00070000000000000000"></A><A NAME="performance"></A><BR>
A few words on performance
</H1>
This section gives some empirical results about the speed of JFlex generated
scanners in comparison to those generated by JLex,
compares a JFlex scanner with a <A HREF="#PerformanceHandwritten">handwritten</A>
one, and presents some <A HREF="#PerformanceTips">tips</A> on how to make
your specification produce a faster scanner.
<P>
<H2><A NAME="SECTION00071000000000000000"></A><A NAME="PerformanceJLex"></A><BR>
Comparison of JLex and JFlex
</H2>
Scanners generated by the tool JLex are quite fast. It was however
possible to further improve the performance of generated scanners
using JFlex. The following table shows the results that were produced
by the scanner specification of a small toy programming language (the
example from the JLex web site). The scanner was generated using JLex
1.2.6 and JFlex version 1.3.5 with all three different JFlex code
generation methods. Then it was run on a W98 system using Sun's JDK
1.3 with different sample inputs of that toy programming language. All
test runs were made under the same conditions on an otherwise idle
machine.
<P>
The values presented in the table denote the time from the first call
to the scanning method to returning the EOF value and the speedup in
percent. The tests were run both in the mixed (HotSpot) JVM mode and
the pure interpreted mode. The mixed mode JVM brings
about a factor of 10 performance improvement, the difference between
JLex and JFlex only decreases slightly.
<P>
<TABLE CELLPADDING=3 BORDER="1" WIDTH="100%">
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">KB</TD>
<TD ALIGN="CENTER">JVM</TD>
<TD ALIGN="RIGHT">JLex</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%switch</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%table</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%pack</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">325 ms</TD>
<TD ALIGN="RIGHT">261 ms</TD>
<TD ALIGN="RIGHT">24.5 %</TD>
<TD ALIGN="RIGHT">261 ms</TD>
<TD ALIGN="RIGHT">24.5 %</TD>
<TD ALIGN="RIGHT">261 ms</TD>
<TD ALIGN="RIGHT">24.5 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">127 ms</TD>
<TD ALIGN="RIGHT">98 ms</TD>
<TD ALIGN="RIGHT">29.6 %</TD>
<TD ALIGN="RIGHT">94 ms</TD>
<TD ALIGN="RIGHT">35.1 %</TD>
<TD ALIGN="RIGHT">96 ms</TD>
<TD ALIGN="RIGHT">32.3 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">66 ms</TD>
<TD ALIGN="RIGHT">50 ms</TD>
<TD ALIGN="RIGHT">32.0 %</TD>
<TD ALIGN="RIGHT">50 ms</TD>
<TD ALIGN="RIGHT">32.0 %</TD>
<TD ALIGN="RIGHT">48 ms</TD>
<TD ALIGN="RIGHT">37.5 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">4009 ms</TD>
<TD ALIGN="RIGHT">3025 ms</TD>
<TD ALIGN="RIGHT">32.5 %</TD>
<TD ALIGN="RIGHT">3258 ms</TD>
<TD ALIGN="RIGHT">23.1 %</TD>
<TD ALIGN="RIGHT">3231 ms</TD>
<TD ALIGN="RIGHT">24.1 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">1641 ms</TD>
<TD ALIGN="RIGHT">1155 ms</TD>
<TD ALIGN="RIGHT">42.1 %</TD>
<TD ALIGN="RIGHT">1245 ms</TD>
<TD ALIGN="RIGHT">31.8 %</TD>
<TD ALIGN="RIGHT">1234 ms</TD>
<TD ALIGN="RIGHT">33.0 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">817 ms</TD>
<TD ALIGN="RIGHT">573 ms</TD>
<TD ALIGN="RIGHT">42.6 %</TD>
<TD ALIGN="RIGHT">617 ms</TD>
<TD ALIGN="RIGHT">32.4 %</TD>
<TD ALIGN="RIGHT">613 ms</TD>
<TD ALIGN="RIGHT">33.3 %</TD>
</TR>
</TABLE>
<P><BR>
<P>
Since the scanning time of the lexical analyser examined in the table
above includes lexical actions that often need to create new object instances,
another table shows the execution time for the same specification with empty
lexical actions to compare the pure scanning engines.
<P>
<TABLE CELLPADDING=3 BORDER="1" WIDTH="100%">
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">KB</TD>
<TD ALIGN="CENTER">JVM</TD>
<TD ALIGN="RIGHT">JLex</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%switch</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%table</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%pack</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">204 ms</TD>
<TD ALIGN="RIGHT">140 ms</TD>
<TD ALIGN="RIGHT">45.7 %</TD>
<TD ALIGN="RIGHT">138 ms</TD>
<TD ALIGN="RIGHT">47.8 %</TD>
<TD ALIGN="RIGHT">140 ms</TD>
<TD ALIGN="RIGHT">45.7 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">83 ms</TD>
<TD ALIGN="RIGHT">55 ms</TD>
<TD ALIGN="RIGHT">50.9 %</TD>
<TD ALIGN="RIGHT">52 ms</TD>
<TD ALIGN="RIGHT">59.6 %</TD>
<TD ALIGN="RIGHT">52 ms</TD>
<TD ALIGN="RIGHT">59.6 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">41 ms</TD>
<TD ALIGN="RIGHT">28 ms</TD>
<TD ALIGN="RIGHT">46.4 %</TD>
<TD ALIGN="RIGHT">26 ms</TD>
<TD ALIGN="RIGHT">57.7 %</TD>
<TD ALIGN="RIGHT">26 ms</TD>
<TD ALIGN="RIGHT">57.7 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">2983 ms</TD>
<TD ALIGN="RIGHT">2036 ms</TD>
<TD ALIGN="RIGHT">46.5 %</TD>
<TD ALIGN="RIGHT">2230 ms</TD>
<TD ALIGN="RIGHT">33.8 %</TD>
<TD ALIGN="RIGHT">2232 ms</TD>
<TD ALIGN="RIGHT">33.6 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">1260 ms</TD>
<TD ALIGN="RIGHT">793 ms</TD>
<TD ALIGN="RIGHT">58.9 %</TD>
<TD ALIGN="RIGHT">865 ms</TD>
<TD ALIGN="RIGHT">45.7 %</TD>
<TD ALIGN="RIGHT">867 ms</TD>
<TD ALIGN="RIGHT">45.3 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">628 ms</TD>
<TD ALIGN="RIGHT">395 ms</TD>
<TD ALIGN="RIGHT">59.0 %</TD>
<TD ALIGN="RIGHT">432 ms</TD>
<TD ALIGN="RIGHT">45.4 %</TD>
<TD ALIGN="RIGHT">432 ms</TD>
<TD ALIGN="RIGHT">45.4 %</TD>
</TR>
</TABLE>
<P><BR>
<P>
Execution time of single instructions depends on the platform and
the implementation of the Java Virtual Machine the program is executed
on. Therefore the tables above cannot be used as a reference to which
code generation method of JFlex is the right one to choose in general.
The following table was produced by the same lexical specification and
the same input on a Linux system also using Sun's JDK 1.3.
<P>
With actions:
<P>
<TABLE CELLPADDING=3 BORDER="1" WIDTH="100%">
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">KB</TD>
<TD ALIGN="CENTER">JVM</TD>
<TD ALIGN="RIGHT">JLex</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%switch</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%table</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%pack</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">246 ms</TD>
<TD ALIGN="RIGHT">203 ms</TD>
<TD ALIGN="RIGHT">21.2 %</TD>
<TD ALIGN="RIGHT">193 ms</TD>
<TD ALIGN="RIGHT">27.5 %</TD>
<TD ALIGN="RIGHT">190 ms</TD>
<TD ALIGN="RIGHT">29.5 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">99 ms</TD>
<TD ALIGN="RIGHT">76 ms</TD>
<TD ALIGN="RIGHT">30.3 %</TD>
<TD ALIGN="RIGHT">69 ms</TD>
<TD ALIGN="RIGHT">43.5 %</TD>
<TD ALIGN="RIGHT">70 ms</TD>
<TD ALIGN="RIGHT">41.4 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">48 ms</TD>
<TD ALIGN="RIGHT">36 ms</TD>
<TD ALIGN="RIGHT">33.3 %</TD>
<TD ALIGN="RIGHT">34 ms</TD>
<TD ALIGN="RIGHT">41.2 %</TD>
<TD ALIGN="RIGHT">35 ms</TD>
<TD ALIGN="RIGHT">37.1 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">3251 ms</TD>
<TD ALIGN="RIGHT">2247 ms</TD>
<TD ALIGN="RIGHT">44.7 %</TD>
<TD ALIGN="RIGHT">2430 ms</TD>
<TD ALIGN="RIGHT">33.8 %</TD>
<TD ALIGN="RIGHT">2444 ms</TD>
<TD ALIGN="RIGHT">33.0 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">1320 ms</TD>
<TD ALIGN="RIGHT">848 ms</TD>
<TD ALIGN="RIGHT">55.7 %</TD>
<TD ALIGN="RIGHT">958 ms</TD>
<TD ALIGN="RIGHT">37.8 %</TD>
<TD ALIGN="RIGHT">920 ms</TD>
<TD ALIGN="RIGHT">43.5 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">658 ms</TD>
<TD ALIGN="RIGHT">423 ms</TD>
<TD ALIGN="RIGHT">55.6 %</TD>
<TD ALIGN="RIGHT">456 ms</TD>
<TD ALIGN="RIGHT">44.3 %</TD>
<TD ALIGN="RIGHT">452 ms</TD>
<TD ALIGN="RIGHT">45.6 %</TD>
</TR>
</TABLE>
<P><BR>
<P>
Without actions:
<P>
<TABLE CELLPADDING=3 BORDER="1" WIDTH="100%">
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">KB</TD>
<TD ALIGN="CENTER">JVM</TD>
<TD ALIGN="RIGHT">JLex</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%switch</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%table</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
<TD ALIGN="RIGHT"><FONT SIZE="-1"><TT>%pack</TT></FONT></TD>
<TD ALIGN="RIGHT">speedup</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">136 ms</TD>
<TD ALIGN="RIGHT">78 ms</TD>
<TD ALIGN="RIGHT">74.4 %</TD>
<TD ALIGN="RIGHT">76 ms</TD>
<TD ALIGN="RIGHT">78.9 %</TD>
<TD ALIGN="RIGHT">77 ms</TD>
<TD ALIGN="RIGHT">76.6 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">59 ms</TD>
<TD ALIGN="RIGHT">31 ms</TD>
<TD ALIGN="RIGHT">90.3 %</TD>
<TD ALIGN="RIGHT">48 ms</TD>
<TD ALIGN="RIGHT">22.9 %</TD>
<TD ALIGN="RIGHT">32 ms</TD>
<TD ALIGN="RIGHT">84.4 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">28 ms</TD>
<TD ALIGN="RIGHT">15 ms</TD>
<TD ALIGN="RIGHT">86.7 %</TD>
<TD ALIGN="RIGHT">15 ms</TD>
<TD ALIGN="RIGHT">86.7 %</TD>
<TD ALIGN="RIGHT">15 ms</TD>
<TD ALIGN="RIGHT">86.7 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">1992 ms</TD>
<TD ALIGN="RIGHT">1047 ms</TD>
<TD ALIGN="RIGHT">90.3 %</TD>
<TD ALIGN="RIGHT">1246 ms</TD>
<TD ALIGN="RIGHT">59.9 %</TD>
<TD ALIGN="RIGHT">1215 ms</TD>
<TD ALIGN="RIGHT">64.0 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">187</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">859 ms</TD>
<TD ALIGN="RIGHT">408 ms</TD>
<TD ALIGN="RIGHT">110.5 %</TD>
<TD ALIGN="RIGHT">479 ms</TD>
<TD ALIGN="RIGHT">79.3 %</TD>
<TD ALIGN="RIGHT">487 ms</TD>
<TD ALIGN="RIGHT">76.4 %</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">93</TD>
<TD ALIGN="CENTER">interpr.</TD>
<TD ALIGN="RIGHT">435 ms</TD>
<TD ALIGN="RIGHT">200 ms</TD>
<TD ALIGN="RIGHT">117.5 %</TD>
<TD ALIGN="RIGHT">237 ms</TD>
<TD ALIGN="RIGHT">83.5 %</TD>
<TD ALIGN="RIGHT">242 ms</TD>
<TD ALIGN="RIGHT">79.8 %</TD>
</TR>
</TABLE>
<P><BR>
<P>
Although all JFlex scanners were faster than those generated by JLex,
slight differences between JFlex code generation methods show up when compared
to the run on the W98 system.
<A NAME="PerformanceHandwritten"></A>
<P>
The following table compares a hand-written scanner for the Java language
obtained from the web site of CUP with the JFlex generated scanner for Java
that comes with JFlex in the <TT>examples</TT> directory. They were tested
on different <TT>.java</TT> files on a Linux machine with Sun's JDK 1.3.
<P>
<TABLE CELLPADDING=3 BORDER="1" WIDTH="100%">
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">lines</TD>
<TD ALIGN="RIGHT">KB</TD>
<TD ALIGN="CENTER">JVM</TD>
<TD ALIGN="RIGHT">hand-written scanner</TD>
<TD ALIGN="CENTER" COLSPAN=2>JFlex generated scanner</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">19050</TD>
<TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">824 ms</TD>
<TD ALIGN="RIGHT">248 ms</TD>
<TD ALIGN="RIGHT">235 % faster</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">6350</TD>
<TD ALIGN="RIGHT">165</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">272 ms</TD>
<TD ALIGN="RIGHT">84 ms</TD>
<TD ALIGN="RIGHT">232 % faster</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">1270</TD>
<TD ALIGN="RIGHT">33</TD>
<TD ALIGN="CENTER">hotspot</TD>
<TD ALIGN="RIGHT">53 ms</TD>
<TD ALIGN="RIGHT">18 ms</TD>
<TD ALIGN="RIGHT">194 % faster</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">19050</TD>
<TD ALIGN="RIGHT">496</TD>
<TD ALIGN="CENTER">interpreted</TD>
<TD ALIGN="RIGHT">5.83 s</TD>
<TD ALIGN="RIGHT">3.85 s</TD>
<TD ALIGN="RIGHT">51 % faster</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">6350</TD>
<TD ALIGN="RIGHT">165</TD>
<TD ALIGN="CENTER">interpreted</TD>
<TD ALIGN="RIGHT">1.95 s</TD>
<TD ALIGN="RIGHT">1.29 s</TD>
<TD ALIGN="RIGHT">51 % faster</TD>
</TR>
<TR><TD ALIGN="LEFT"> </TD><TD ALIGN="RIGHT">1270</TD>
<TD ALIGN="RIGHT">33</TD>
<TD ALIGN="CENTER">interpreted</TD>
<TD ALIGN="RIGHT">0.38 s</TD>
<TD ALIGN="RIGHT">0.25 s</TD>
<TD ALIGN="RIGHT">52 % faster</TD>
</TR>
</TABLE>
<P><BR>
<P>
Although JDK 1.3 seems to speed up the hand-written scanner if compared
to JDK 1.1 or 1.2 more than the generated one, the generated scanner is
still up to 3.3 times as fast as the hand-written one. One example of
a hand-written scanner that is
considerably slower than the equivalent generated one is surely no
proof for all generated scanners being faster than hand-written. It is
clearly impossible to prove something like that, since you could
always write the generated scanner by hand. From a software
engineering point of view however, there is no excuse for writing a
scanner by hand since this task takes more time, is more difficult and
therefore more error prone than writing a compact, readable and easy
to change lexical specification. (I'd like to add, that I do <EM>not</EM>
think, that the hand-written scanner from the CUP web site used here in
the test is stupid or badly written or anything like that. I actually
think, Scott did a great job with it)
<P>
<H2><A NAME="SECTION00072000000000000000"></A><A NAME="PerformanceTips"></A><BR>
How to write a faster specification
</H2>
Although JFlex generated scanners show good performance without
special optimisations, there are some heuristics that can make a
lexical specification produce an even faster scanner. Those are
(roughly in order of performance gain):
<P>
<UL>
<LI>Avoid rules that require backtracking
<P>
From the C/C++ flex [<A
HREF="manual.html#flex">11</A>] man page: <EM>``Getting rid
of backtracking is messy and often may be an enormous amount of work for
a complicated scanner.''</EM> Backtracking is introduced by the longest match
rule and occurs for instance on this set of expressions:
<P>
<TT> "averylongkeyword"</TT>
<BR><TT> .</TT>
<P>
With input <TT>"averylongjoke"</TT> the scanner has to read all characters
up to <TT>'j' </TT>to decide that rule <TT>.</TT> should be matched. All
characters of <TT>"verylong"</TT> have to be read again for the next
matching process. Backtracking can be avoided in general by adding
error rules that match those error conditions
<P>
<code> "av"|"ave"|"avery"|"averyl"|..</code>
<P>
While this is impractical in most scanners, there is still the
possibility to add a ``catch all'' rule for a lengthy list of keywords
<PRE>
"keyword1" { return symbol(KEYWORD1); }
..
"keywordn" { return symbol(KEYWORDn); }
[a-z]+ { error("not a keyword"); }
</PRE>
Most programming language scanners already have a rule like this for
some kind of variable length identifiers.
<P>
</LI>
<LI>Avoid line and column counting
<P>
It costs multiple additional comparisons per input character and the
matched text has to be re-scanned for counting. In most scanners it
is possible to do the line counting in the specification by
incrementing <TT>yyline</TT> each time a line terminator has been
matched. Column counting could also be included in actions. This
will be faster, but can in some cases become quite messy.
<P>
</LI>
<LI>Avoid look-ahead expressions and the end of line operator '$'
<P>
In the best case, the trailing context will first have to be read and
then (because it is not to be consumed) re-read again. The cases of
fixed-length look-ahead and fixed-length base expressions are handled efficiently
by matching the concatenation and then pushing back the required amount
of characters. This extends to the case of a disjunction of fixed-length
look-ahead expressions such as <code>r1 / \r|\n|\r\n</code>. All other cases
<code>r1 / r2</code> are handled by first scanning the concatenation of
<code>r1</code> and <code>r2</code>, and then finding the correct end of <code>r1</code>.
The end of <code>r1</code> is found by scanning forwards in the match again,
marking all possible <code>r1</code> terminations, and then scanning the reverse
of <code>r2</code> backwards from the end until a start of <code>r2</code> intersects
with an end of <code>r1</code>. This algorithm is linear in the size of the input
(not quadratic or worse as backtracking is), but about a factor of 2 slower
than normal scanning. It also consumes memory proportional to the size
of the matched input for <code>r1 r2</code>.
<P>
</LI>
<LI>Avoid the beginning of line operator '<code>^</code>'
<P>
It costs multiple additional comparisons per match. In some
cases one extra look-ahead character is needed (when the last character read is
<code>\r</code> the scanner has to read one character ahead to check if
the next one is an <code>\n</code> or not).
<P>
</LI>
<LI>Match as much text as possible in a rule.
<P>
One rule is matched in the innermost loop of the scanner. After
each action some overhead for setting up the internal state of the
scanner is necessary.
</LI>
</UL>
<P>
Note that writing more rules in a specification does not make the generated
scanner slower (except when you have to switch to another code generation
method because of the larger size).
<P>
The two main rules of optimisation apply also for lexical specifications:
<OL>
<LI><B>don't do it</B>
</LI>
<LI><B>(for experts only) don't do it yet</B>
</LI>
</OL>
<P>
Some of the performance tips above contradict a readable and compact
specification style. When in doubt or when requirements are not or not
yet fixed: don't use them -- the specification can always be optimised
in a later state of the development process.
<P>
<H1><A NAME="SECTION00080000000000000000">
Porting Issues</A>
</H1>
<P>
<H2><A NAME="SECTION00081000000000000000"></A><A NAME="Porting"></A><BR>
Porting from JLex
</H2>
JFlex was designed to read old JLex specifications unchanged and to
generate a scanner which behaves exactly the same as the one generated
by JLex with the only difference of being faster.
<P>
This works as expected on all well formed JLex specifications.
<P>
Since the statement above is somewhat absolute, let's take a look at
what ``well formed'' means here. A JLex specification is well formed, when
it
<UL>
<LI>generates a working scanner with JLex
<P>
</LI>
<LI>doesn't contain the unescaped characters <TT>!</TT> and <TT>~</TT>
<P>
They are operators in JFlex while JLex treats them as normal
input characters. You can easily port such a JLex specification
to JFlex by replacing every <TT>!</TT> with <code>\!</code> and every
<code>~</code> with <code>\~</code> in all regular expressions.
<P>
</LI>
<LI>has only complete regular expressions surrounded by parentheses in
macro definitions
<P>
This may sound a bit harsh, but could otherwise be a major problem
- it can also help you find some disgusting bugs in your
specification that didn't show up in the first place. In JLex, a
right hand side of a macro is just a piece of text, that is copied
to the point where the macro is used. With this, some weird kind of
stuff like
<PRE>
macro1 = ("hello"
macro2 = {macro1})*
</PRE>
was possible (with <TT>macro2</TT> expanding to <code>("hello")*</code>). This
is not allowed in JFlex and you will have to transform such
definitions. There are however some more subtle kinds of errors that
can be introduced by JLex macros. Let's consider a definition like
<code>macro = a|b</code> and a usage like <code>{macro}*</code>.
This expands in JLex to <code>a|b*</code> and not to the probably intended
<code>(a|b)*</code>.
<P>
JFlex uses always the second form of expansion, since this is the natural
form of thinking about abbreviations for regular expressions.
<P>
Most specifications shouldn't suffer from this problem, because
macros often only contain (harmless) character classes like
<TT>alpha = [a-zA-Z]</TT> and more dangerous definitions like
<P>
<code> ident = {alpha}({alpha}|{digit})*</code>
<P>
are only used to write rules like
<P>
<code> {ident} { .. action .. }</code>
<P>
and not more complex expressions like
<P>
<code> {ident}* { .. action .. }</code>
<P>
where the kind of error presented above would show up.
</LI>
</UL>
<P>
<H2><A NAME="SECTION00082000000000000000"></A><A NAME="lexport"></A><BR>
Porting from lex/flex
</H2>
This section tries to give an overview of activities and possible
problems when porting a lexical specification from the C/C++ tools lex
and flex [<A
HREF="manual.html#flex">11</A>] available on most Unix systems to JFlex.
<P>
Most of the C/C++ specific features are naturally not present in JFlex,
but most ``clean'' lex/flex lexical specifications can be ported to
JFlex without very much work.
<P>
This section is by far not complete and is based mainly on a survey of
the flex man page and very little personal experience. If you do
engage in any porting activity from lex/flex to JFlex and encounter
problems, have better solutions for points presented here or have just
some tips you would like to share, please do <A NAME="tex2html7"
HREF="mailto:lsf@jflex.de">contact me</A>. I will
incorporate your experiences in this manual (with all due credit to you,
of course).
<P>
<H3><A NAME="SECTION00082100000000000000">
Basic structure</A>
</H3>
A lexical specification for flex has the following basic structure:
<PRE>
definitions
%%
rules
%%
user code
</PRE>
<P>
The <TT>user code</TT> section usually contains some C code that is used
in actions of the <TT>rules</TT> part of the specification. For JFlex most
of this code will have to be included in the class code <code>%{..%}</code>
directive in the <TT>options</TT> <TT>and declarations</TT> section (after
translating the C code to Java, of course).
<P>
<H3><A NAME="SECTION00082200000000000000">
Macros and Regular Expression Syntax</A>
</H3>
The <TT>definitions</TT> section of a flex specification is quite similar
to the <TT>options and declarations</TT> part of JFlex specs.
<P>
Macro definitions in flex have the form:
<PRE>
<identifier> <expression>
</PRE>
To port them to JFlex macros, just insert a <TT>=</TT> between <TT><identifier></TT>
and <TT><expression></TT>.
<P>
The syntax and semantics of regular expressions in flex are pretty much the
same as in JFlex. A little attention is needed for some escape sequences
present in flex (such as <code>\a</code>) that are not supported in JFlex. These
escape sequences should be transformed into their octal or hexadecimal
equivalent.
<P>
Another point are predefined character classes. Flex offers the ones directly
supported by C, JFlex offers the ones supported by Java. These classes will
sometimes have to be listed manually (if there is need for this feature, it
may be implemented in a future JFlex version).
<P>
<H3><A NAME="SECTION00082300000000000000">
Lexical Rules</A>
</H3>
Since flex is mostly Unix based, the '<code>^</code>' (beginning of line) and
'<code>$</code>' (end of line) operators, consider the <code>\n</code> character as only line terminator. This should usually cause not much problems, but you
should be prepared for occurrences of <code>\r</code> or <code>\r\n</code> or one of
the characters <code>\u2028</code>, <code>\u2029</code>, <code>\u000B</code>, <code>\u000C</code>,
or <code>\u0085</code>. They are considered to be line terminators in Unicode and
therefore may not be consumed when
<code>^</code> or <code>$</code> is present in a rule.
<P>
<H1><A NAME="SECTION00090000000000000000"></A><A NAME="WorkingTog"></A><BR>
Working together
</H1>
<P>
<H2><A NAME="SECTION00091000000000000000"></A><A NAME="CUPWork"></A><BR>
JFlex and CUP
</H2>
One of the main design goals of JFlex was to make interfacing with the free
Java parser generator CUP [<A
HREF="manual.html#CUP">8</A>] as easy as possibly.
This has been done by giving
the <TT><A HREF="#CupMode">%cup</A></TT> directive a special meaning. An
interface however always has two sides. This section concentrates on the
CUP side of the story.
<P>
<H3><A NAME="SECTION00091100000000000000">
CUP version 0.10j and above</A>
</H3>
Since CUP version 0.10j, this has been simplified greatly by the new
CUP scanner interface <TT>java_cup.runtime.Scanner</TT>. JFlex lexers now implement
this interface automatically when then <TT><A HREF="#CupMode">%cup</A></TT>
switch is used. There are no special <TT>parser code</TT>, <TT>init
code</TT> or <TT>scan with</TT> options any more that you have to provide
in your CUP parser specification. You can just concentrate on your grammar.
<P>
If your generated lexer has the class name <TT>Scanner</TT>, the parser
is started from the a main program like this:
<P>
<PRE>
...
try {
parser p = new parser(new Scanner(new FileReader(fileName)));
Object result = p.parse().value;
}
catch (Exception e) {
...
</PRE>
<P>
<H3><A NAME="SECTION00091200000000000000">
Custom symbol interface</A>
</H3>
If you have used the <TT>-symbol</TT> command line switch of CUP to change
the name of the generated symbol interface, you have to tell JFlex about
this change of interface so that correct end-of-file code is generated.
You can do so either by using an <code>%eofval{</code> directive or by using
and <TT>«EOF»</TT> rule.
<P>
If your new symbol interface is called <TT>mysym</TT> for example, the
corresponding code in the jflex specification would be either
<P>
<PRE>
%eofval{
return mysym.EOF;
%eofval}
</PRE>
<P>
in the macro/directives section of the spec, or it would be
<P>
<PRE>
<<EOF>> { return mysym.EOF; }
</PRE>
<P>
in the rules section of your spec.
<P>
<H3><A NAME="SECTION00091300000000000000">
Using existing JFlex/CUP specifications with CUP 0.10j</A>
</H3>
If you already have an existing specification and you would like to upgrade
both JFlex and CUP to their newest version, you will probably have to adjust
your specification.
<P>
The main difference between the <TT><A HREF="#CupMode">%cup</A></TT> switch in
JFlex 1.2.1 and lower, and the current JFlex version is, that JFlex scanners
now automatically implement the <TT>java_cup.runtime.Scanner</TT> interface.
This means, that the scanning function now changes its name from <TT>yylex()</TT>
to <TT>next_token()</TT>.
<P>
The main difference from older CUP versions to 0.10j is, that CUP now
has a default constructor that accepts a <TT>java_cup.runtime.Scanner</TT>
as argument and that uses this scanner as
default (so no <TT>scan with</TT> code is necessary any more).
<P>
If you have an existing CUP specification, it will probably look somewhat like this:
<PRE>
parser code {:
Lexer lexer;
public parser (java.io.Reader input) {
lexer = new Lexer(input);
}
:};
scan with {: return lexer.yylex(); :};
</PRE>
<P>
To upgrade to CUP 0.10j, you could change it to look like this:
<PRE>
parser code {:
public parser (java.io.Reader input) {
super(new Lexer(input));
}
:};
</PRE>
<P>
If you do not mind to change the method that is calling the parser,
you could remove the constructor entirely (and if there is nothing else
in it, the whole <TT>parser code</TT> section as well, of course). The calling
main procedure would then construct the parser as shown in the section above.
<P>
The JFlex specification does not need to be changed.
<P>
<H3><A NAME="SECTION00091400000000000000">
Using older versions of CUP</A>
</H3>
For people, who like or have to use older versions of CUP, the following section
explains ``the old way''. Please note, that the standard name of the scanning
function with the <TT><A HREF="#CupMode">%cup</A></TT> switch is not
<TT>yylex()</TT>, but <TT>next_token()</TT>.
<P>
If you have a scanner specification that begins like this:
<P>
<PRE>
package PACKAGE;
import java_cup.runtime.*; /* this is convenience, but not necessary */
%%
%class Lexer
%cup
..
</PRE>
<P>
then it matches a CUP specification starting like
<P>
<PRE>
package PACKAGE;
parser code {:
Lexer lexer;
public parser (java.io.Reader input) {
lexer = new Lexer(input);
}
:};
scan with {: return lexer.next_token(); :};
..
</PRE>
<P>
This assumes that the generated parser will get the name <TT>parser</TT>.
If it doesn't, you have to adjust the constructor name.
<P>
The parser can then be started in a main routine like this:
<P>
<PRE>
..
try {
parser p = new parser(new FileReader(fileName));
Object result = p.parse().value;
}
catch (Exception e) {
..
</PRE>
<P>
If you want the parser specification to be independent of the name of the generated
scanner, you can instead write an interface <TT>Lexer</TT>
<P>
<PRE>
public interface Lexer {
public java_cup.runtime.Symbol next_token() throws java.io.IOException;
}
</PRE>
<P>
change the parser code to:
<P>
<PRE>
package PACKAGE;
parser code {:
Lexer lexer;
public parser (Lexer lexer) {
this.lexer = lexer;
}
:};
scan with {: return lexer.next_token(); :};
..
</PRE>
<P>
tell JFlex about the lexer
interface using the <TT>%implements</TT>
directive:
<P>
<PRE>
..
%class Scanner /* not Lexer now since that is our interface! */
%implements Lexer
%cup
..
</PRE>
<P>
and finally change the main routine to look like
<P>
<PRE>
...
try {
parser p = new parser(new Scanner(new FileReader(fileName)));
Object result = p.parse().value;
}
catch (Exception e) {
...
</PRE>
<P>
If you want to improve the error messages that CUP generated parsers
produce, you can also override the methods <TT>report_error</TT> and <TT>report_fatal_error</TT>
in the ``parser code'' section of the CUP specification. The new methods
could for instance use <TT>yyline</TT> and <TT>yycolumn</TT> (stored in
the <TT>left</TT> and <TT>right</TT> members of class <TT>java_cup.runtime.Symbol</TT>)
to report error positions more conveniently for the user. The lexer and
parser for the Java language in the <TT>examples/java</TT> directory of the
JFlex distribution use this style of error reporting. These specifications
also demonstrate the techniques above in action.
<P>
<H2><A NAME="SECTION00092000000000000000"></A><A NAME="YaccWork"></A><BR>
JFlex and BYacc/J
</H2>
<P>
JFlex has built-in support for the Java extension
<A NAME="tex2html8"
HREF="http://byaccj.sourceforge.net/">BYacc/J</A>
[<A
HREF="manual.html#BYaccJ">9</A>] by Bob Jamison
to the classical Berkeley Yacc parser generator.
This section describes how to interface BYacc/J with JFlex. It
builds on many helpful suggestions and comments from Larry Bell.
<P>
Since Yacc's architecture is a bit different from CUP's, the
interface setup also works in a slightly different manner.
BYacc/J expects a function <TT>int yylex()</TT> in the parser
class that returns each next token. Semantic values are expected
in a field <TT>yylval</TT> of type <TT>parserval</TT> where ``<TT>parser</TT>''
is the name of the generated parser class.
<P>
For a small calculator example, one could use a set up like the
following on the JFlex side:
<P>
<PRE>
%%
%byaccj
%{
/* store a reference to the parser object */
private parser yyparser;
/* constructor taking an additional parser object */
public Yylex(java.io.Reader r, parser yyparser) {
this(r);
this.yyparser = yyparser;
}
%}
NUM = [0-9]+ ("." [0-9]+)?
NL = \n | \r | \r\n
%%
/* operators */
"+" |
..
"(" |
")" { return (int) yycharat(0); }
/* newline */
{NL} { return parser.NL; }
/* float */
{NUM} { yyparser.yylval = new parserval(Double.parseDouble(yytext()));
return parser.NUM; }
</PRE>
<P>
The lexer expects a reference to the parser in its constructor.
Since Yacc allows direct use of terminal characters like <TT>'+'</TT>
in its specifications, we just return the character code for
single char matches (e.g. the operators in the example). Symbolic
token names are stored as <TT>public static int</TT> constants in
the generated parser class. They are used as in the <TT>NL</TT> token
above. Finally, for some tokens, a semantic value may have to be
communicated to the parser. The <TT>NUM</TT> rule demonstrates that
bit.
<P>
A matching BYacc/J parser specification could look like this:
<PRE>
%{
import java.io.*;
%}
%token NL /* newline */
%token <dval> NUM /* a number */
%type <dval> exp
%left '-' '+'
..
%right '^' /* exponentiation */
%%
..
exp: NUM { $$ = $1; }
| exp '+' exp { $$ = $1 + $3; }
..
| exp '^' exp { $$ = Math.pow($1, $3); }
| '(' exp ')' { $$ = $2; }
;
%%
/* a reference to the lexer object */
private Yylex lexer;
/* interface to the lexer */
private int yylex () {
int yyl_return = -1;
try {
yyl_return = lexer.yylex();
}
catch (IOException e) {
System.err.println("IO error :"+e);
}
return yyl_return;
}
/* error reporting */
public void yyerror (String error) {
System.err.println ("Error: " + error);
}
/* lexer is created in the constructor */
public parser(Reader r) {
lexer = new Yylex(r, this);
}
/* that's how you use the parser */
public static void main(String args[]) throws IOException {
parser yyparser = new parser(new FileReader(args[0]));
yyparser.yyparse();
}
</PRE>
<P>
Here, the customised part is mostly in the user code section:
We create the lexer in the constructor of the parser and store
a reference to it for later use in the parser's <TT>int yylex()</TT>
method. This <TT>yylex</TT> in the parser only calls <TT>int yylex()</TT>
of the generated lexer and passes the result on. If something goes
wrong, it returns -1 to indicate an error.
<P>
Runnable versions of the specifications above
are located in the <TT>examples/byaccj</TT> directory of the JFlex
distribution.
<P>
<H1><A NAME="SECTION000100000000000000000"></A><A NAME="Bugs"></A><BR>
Bugs and Deficiencies
</H1>
<P>
<H2><A NAME="SECTION000101000000000000000">
Deficiencies</A>
</H2>
Unicode matching is not fully conforming to the relevant current Unicode report. Instead, the Unicode support in JFlex is the one native to Java. That means, only 16 bit code points are supported and most Unicode character classes are not directly supported (although they can be custom-defined in macros). The Java 5 development version of JFlex contains better support for Unicode, as will the next major release.
<P>
<H2><A NAME="SECTION000102000000000000000">
Bugs</A>
</H2>
As of January 31, 2009, no bugs have been reported for JFlex version 1.4.3. All
bugs reported for earlier versions have been fixed.
<P>
If you find new problems, please use the bugs section of the
<A NAME="tex2html9"
HREF="http://www.jflex.de/">JFlex web site</A>
to report them.
<P>
<H1><A NAME="SECTION000110000000000000000"></A><A NAME="Copyright"></A><BR>
Copying and License
</H1>
JFlex is free software, published under the terms of the
<A NAME="tex2html10"
HREF="http://www.fsf.org/copyleft/gpl.html">GNU General Public License</A>.
<P>
There is absolutely NO WARRANTY for JFlex, its code and its documentation.
<P>
The code generated by JFlex inherits the copyright of the specification it
was produced from. If it was your specification, you may use the generated
code without restriction.
<P>
See the file <A NAME="tex2html11"
HREF="COPYRIGHT"><TT>COPYRIGHT</TT></A>
for more information.
<P>
<H2><A NAME="SECTION000120000000000000000"></A><A NAME="References"></A><BR>
Bibliography
</H2><DL COMPACT><DD>
<P>
<P></P><DT><A NAME="Aho">1</A>
<DD>
A. Aho, R. Sethi, J. Ullman, <EM>Compilers: Principles, Techniques, and Tools</EM>, 1986
<P>
<P></P><DT><A NAME="Appel">2</A>
<DD>
A. W. Appel, <EM>Modern Compiler Implementation in Java: basic techniques</EM>, 1997
<P>
<P></P><DT><A NAME="JLex">3</A>
<DD>
E. Berk, <EM>JLex: A lexical analyser generator for Java</EM>,
<BR> <A NAME="tex2html12"
HREF="http://www.cs.princeton.edu/~appel/modern/java/JLex/"><TT>http://www.cs.princeton.edu/~appel/modern/java/JLex/</TT></A>
<P>
<P></P><DT><A NAME="fast">4</A>
<DD>
K. Brouwer, W. Gellerich,E. Ploedereder,
<EM>Myths and Facts about the Efficient Implementation of Finite Automata and Lexical Analysis</EM>,
in: Proceedings of the 7th International Conference on Compiler Construction (CC '98), 1998
<P>
<P></P><DT><A NAME="unicode_rep">5</A>
<DD>
M. Davis, <EM>Unicode Regular Expression Guidelines</EM>, Unicode Technical Report #18, 2000
<BR> <A NAME="tex2html13"
HREF="http://www.unicode.org/unicode/reports/tr18/tr18-5.1.html"><TT>http://www.unicode.org/unicode/reports/tr18/tr18-5.1.html</TT></A>
<P>
<P></P><DT><A NAME="ParseTable">6</A>
<DD>
P. Dencker, K. Dürre, J. Henft, <EM>Optimization of Parser Tables for portable Compilers</EM>,
in: ACM Transactions on Programming Languages and Systems 6(4), 1984
<P>
<P></P><DT><A NAME="LangSpec">7</A>
<DD>
J. Gosling, B. Joy, G. Steele, <EM>The Java Language Specifcation</EM>, 1996,
<BR> <A NAME="tex2html14"
HREF="http://java.sun.com/docs/books/jls/"><TT>http://java.sun.com/docs/books/jls/</TT></A>
<P>
<P></P><DT><A NAME="CUP">8</A>
<DD>
S. E. Hudson, <EM>CUP LALR Parser Generator for Java</EM>,
<BR> <A NAME="tex2html15"
HREF="http://www.cs.princeton.edu/~appel/modern/java/CUP/"><TT>http://www.cs.princeton.edu/~appel/modern/java/CUP/</TT></A>
<P>
<P></P><DT><A NAME="BYaccJ">9</A>
<DD>
B. Jamison, <EM>BYacc/J</EM>,
<BR> <A NAME="tex2html16"
HREF="http://byaccj.sourceforge.net"><TT>http://byaccj.sourceforge.net/</TT></A>
<P>
<P></P><DT><A NAME="MachineSpec">10</A>
<DD>
T. Lindholm, F. Yellin, <EM>The Java Virtual Machine Specification</EM>, 1996,
<BR> <A NAME="tex2html17"
HREF="http://java.sun.com/docs/books/vmspec/"><TT>http://java.sun.com/docs/books/vmspec/</TT></A>
<P>
<P></P><DT><A NAME="flex">11</A>
<DD>
V. Paxson, <EM>flex - The fast lexical analyzer generator</EM>, 1995
<P>
<P></P><DT><A NAME="SparseTable">12</A>
<DD>
R. E. Tarjan, A. Yao, <EM>Storing a Sparse Table</EM>, in: Communications of the ACM 22(11), 1979
<P>
<P></P><DT><A NAME="Maurer">13</A>
<DD>
R. Wilhelm, D. Maurer, <EM>Übersetzerbau</EM>, Berlin 1997<SUP>2</SUP>
<tex2html_verbatim_mark>mathend000#
<P>
</DL>
<P>
<BR><HR><H4>Footnotes</H4>
<DL>
<DT><A NAME="foot33">... Java</A><A
HREF="manual.html#tex2html2"><SUP><IMG ALIGN="BOTTOM" BORDER="1" ALT="[*]" SRC="footnote.png"></SUP></A></DT>
<DD>Java is a trademark of
Sun Microsystems, Inc., and refers to Sun's Java programming language.
JFlex is not sponsored by or affiliated with Sun Microsystems, Inc.
</DD>
</DL><BR><HR>
<ADDRESS>
Sat 31 Jan 2009 23:43:28 EST, <a href="http://www.doclsf.de">Gerwin Klein</a>
</ADDRESS>
</BODY>
</HTML>
distrib
>
Mageia
>
5
>
x86_64
>
media
>
core-release
>
by-pkgid
>
603cfbd64b904dd3386baed96dee8684
>
files
>
10
