This manual documents version 1.35 of Bison, updated 25 February 2002.
Tutorial sections:
Reference sections:
yyparse.
The Concepts of Bison
Examples
Reverse Polish Notation Calculator
Grammar Rules for rpcalc
Location Tracking Calculator: ltcalc
Multi-Function Calculator: mfcalc
Bison Grammar Files
Outline of a Bison Grammar
Defining Language Semantics
Bison Declarations
Parser C-Language Interface
yyparse and what it returns.
yylex
which reads tokens.
yyerror.
The Lexical Analyzer Function yylex
yyparse calls yylex.
yylex must return the semantic value
of the token it has read.
yylex must return the text position
(line number, etc.) of the token, if the
actions want that.
The Bison Parser Algorithm
Operator Precedence
Handling Context Dependencies
Invoking Bison
Copying This Manual
Bison is a general-purpose parser generator that converts a grammar description for an LALR(1) context-free grammar into a C program to parse that grammar. Once you are proficient with Bison, you may use it to develop a wide range of language parsers, from those used in simple desk calculators to complex programming languages.
Bison is upward compatible with Yacc: all properly-written Yacc grammars ought to work with Bison with no change. Anyone familiar with Yacc should be able to use Bison with little trouble. You need to be fluent in C programming in order to use Bison or to understand this manual.
We begin with tutorial chapters that explain the basic concepts of using Bison and show three explained examples, each building on the last. If you don't know Bison or Yacc, start by reading these chapters. Reference chapters follow which describe specific aspects of Bison in detail.
Bison was written primarily by Robert Corbett; Richard Stallman made it Yacc-compatible. Wilfred Hansen of Carnegie Mellon University added multi-character string literals and other features.
This edition corresponds to version 1.35 of Bison.
As of Bison version 1.24, we have changed the distribution terms for
yyparse to permit using Bison's output in nonfree programs.
Formerly, Bison parsers could be used only in programs that were free
software.
The other GNU programming tools, such as the GNU C compiler, have never had such a requirement. They could always be used for nonfree software. The reason Bison was different was not due to a special policy decision; it resulted from applying the usual General Public License to all of the Bison source code.
The output of the Bison utility--the Bison parser file--contains a
verbatim copy of a sizable piece of Bison, which is the code for the
yyparse function. (The actions from your grammar are inserted
into this function at one point, but the rest of the function is not
changed.) When we applied the GPL terms to the code for yyparse,
the effect was to restrict the use of Bison output to free software.
We didn't change the terms because of sympathy for people who want to make software proprietary. Software should be free. But we concluded that limiting Bison's use to free software was doing little to encourage people to make other software free. So we decided to make the practical conditions for using Bison match the practical conditions for using the other GNU tools.
Copyright © 1989, 1991 Free Software Foundation, Inc. 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too.
When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things.
To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it.
For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights.
We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software.
Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations.
Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all.
The precise terms and conditions for copying, distribution and modification follow.
Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does.
You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee.
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program.
In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License.
The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable.
If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code.
If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances.
It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice.
This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License.
Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation.
If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms.
To do so, attach the following notices to the program. It is safest
to attach them to the start of each source file to most effectively
convey the exclusion of warranty; and each file should have at least
the "copyright" line and a pointer to where the full notice is found.
one line to give the program's name and a brief idea of what it does. Copyright (C) yyyy name of author This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
Also add information on how to contact you by electronic and paper mail.
If the program is interactive, make it output a short notice like this
when it starts in an interactive mode:
Gnomovision version 69, Copyright (C) 19yy name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details.
The hypothetical commands show w and show c should show
the appropriate parts of the General Public License. Of course, the
commands you use may be called something other than show w and
show c; they could even be mouse-clicks or menu items--whatever
suits your program.
You should also get your employer (if you work as a programmer) or your
school, if any, to sign a "copyright disclaimer" for the program, if
necessary. Here is a sample; alter the names:
Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. signature of Ty Coon, 1 April 1989 Ty Coon, President of Vice
This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License.
This chapter introduces many of the basic concepts without which the details of Bison will not make sense. If you do not already know how to use Bison or Yacc, we suggest you start by reading this chapter carefully.
In order for Bison to parse a language, it must be described by a context-free grammar. This means that you specify one or more syntactic groupings and give rules for constructing them from their parts. For example, in the C language, one kind of grouping is called an `expression'. One rule for making an expression might be, "An expression can be made of a minus sign and another expression". Another would be, "An expression can be an integer". As you can see, rules are often recursive, but there must be at least one rule which leads out of the recursion.
The most common formal system for presenting such rules for humans to read is Backus-Naur Form or "BNF", which was developed in order to specify the language Algol 60. Any grammar expressed in BNF is a context-free grammar. The input to Bison is essentially machine-readable BNF.
Not all context-free languages can be handled by Bison, only those that are LALR(1). In brief, this means that it must be possible to tell how to parse any portion of an input string with just a single token of look-ahead. Strictly speaking, that is a description of an LR(1) grammar, and LALR(1) involves additional restrictions that are hard to explain simply; but it is rare in actual practice to find an LR(1) grammar that fails to be LALR(1). See Mysterious Reduce/Reduce Conflicts, for more information on this.
In the formal grammatical rules for a language, each kind of syntactic unit or grouping is named by a symbol. Those which are built by grouping smaller constructs according to grammatical rules are called nonterminal symbols; those which can't be subdivided are called terminal symbols or token types. We call a piece of input corresponding to a single terminal symbol a token, and a piece corresponding to a single nonterminal symbol a grouping.
We can use the C language as an example of what symbols, terminal and nonterminal, mean. The tokens of C are identifiers, constants (numeric and string), and the various keywords, arithmetic operators and punctuation marks. So the terminal symbols of a grammar for C include `identifier', `number', `string', plus one symbol for each keyword, operator or punctuation mark: `if', `return', `const', `static', `int', `char', `plus-sign', `open-brace', `close-brace', `comma' and many more. (These tokens can be subdivided into characters, but that is a matter of lexicography, not grammar.)
Here is a simple C function subdivided into tokens:
int /* keyword `int' */
square (int x) /* identifier, open-paren, identifier, identifier, close-paren */
{ /* open-brace */
return x * x; /* keyword `return', identifier, asterisk, identifier, semicolon */
} /* close-brace */
The syntactic groupings of C include the expression, the statement, the
declaration, and the function definition. These are represented in the
grammar of C by nonterminal symbols `expression', `statement',
`declaration' and `function definition'. The full grammar uses dozens of
additional language constructs, each with its own nonterminal symbol, in
order to express the meanings of these four. The example above is a
function definition; it contains one declaration, and one statement. In
the statement, each x is an expression and so is x * x.
Each nonterminal symbol must have grammatical rules showing how it is made
out of simpler constructs. For example, one kind of C statement is the
return statement; this would be described with a grammar rule which
reads informally as follows:
A `statement' can be made of a `return' keyword, an `expression' and a `semicolon'.
There would be many other rules for `statement', one for each kind of statement in C.
One nonterminal symbol must be distinguished as the special one which defines a complete utterance in the language. It is called the start symbol. In a compiler, this means a complete input program. In the C language, the nonterminal symbol `sequence of definitions and declarations' plays this role.
For example, 1 + 2 is a valid C expression--a valid part of a C
program--but it is not valid as an entire C program. In the
context-free grammar of C, this follows from the fact that `expression' is
not the start symbol.
The Bison parser reads a sequence of tokens as its input, and groups the tokens using the grammar rules. If the input is valid, the end result is that the entire token sequence reduces to a single grouping whose symbol is the grammar's start symbol. If we use a grammar for C, the entire input must be a `sequence of definitions and declarations'. If not, the parser reports a syntax error.
A formal grammar is a mathematical construct. To define the language for Bison, you must write a file expressing the grammar in Bison syntax: a Bison grammar file. See Bison Grammar Files.
A nonterminal symbol in the formal grammar is represented in Bison input
as an identifier, like an identifier in C. By convention, it should be
in lower case, such as expr, stmt or declaration.
The Bison representation for a terminal symbol is also called a token
type. Token types as well can be represented as C-like identifiers. By
convention, these identifiers should be upper case to distinguish them from
nonterminals: for example, INTEGER, IDENTIFIER, IF or
RETURN. A terminal symbol that stands for a particular keyword in
the language should be named after that keyword converted to upper case.
The terminal symbol error is reserved for error recovery.
See Symbols.
A terminal symbol can also be represented as a character literal, just like a C character constant. You should do this whenever a token is just a single character (parenthesis, plus-sign, etc.): use that same character in a literal as the terminal symbol for that token.
A third way to represent a terminal symbol is with a C string constant containing several characters. See Symbols, for more information.
The grammar rules also have an expression in Bison syntax. For example,
here is the Bison rule for a C return statement. The semicolon in
quotes is a literal character token, representing part of the C syntax for
the statement; the naked semicolon, and the colon, are Bison punctuation
used in every rule.
stmt: RETURN expr ';'
;
A formal grammar selects tokens only by their classifications: for example,
if a rule mentions the terminal symbol `integer constant', it means that
any integer constant is grammatically valid in that position. The
precise value of the constant is irrelevant to how to parse the input: if
x+4 is grammatical then x+1 or x+3989 is equally
grammatical.
But the precise value is very important for what the input means once it is parsed. A compiler is useless if it fails to distinguish between 4, 1 and 3989 as constants in the program! Therefore, each token in a Bison grammar has both a token type and a semantic value. See Defining Language Semantics, for details.
The token type is a terminal symbol defined in the grammar, such as
INTEGER, IDENTIFIER or ','. It tells everything
you need to know to decide where the token may validly appear and how to
group it with other tokens. The grammar rules know nothing about tokens
except their types.
The semantic value has all the rest of the information about the
meaning of the token, such as the value of an integer, or the name of an
identifier. (A token such as ',' which is just punctuation doesn't
need to have any semantic value.)
For example, an input token might be classified as token type
INTEGER and have the semantic value 4. Another input token might
have the same token type INTEGER but value 3989. When a grammar
rule says that INTEGER is allowed, either of these tokens is
acceptable because each is an INTEGER. When the parser accepts the
token, it keeps track of the token's semantic value.
Each grouping can also have a semantic value as well as its nonterminal symbol. For example, in a calculator, an expression typically has a semantic value that is a number. In a compiler for a programming language, an expression typically has a semantic value that is a tree structure describing the meaning of the expression.
In order to be useful, a program must do more than parse input; it must also produce some output based on the input. In a Bison grammar, a grammar rule can have an action made up of C statements. Each time the parser recognizes a match for that rule, the action is executed. See Actions.
Most of the time, the purpose of an action is to compute the semantic value of the whole construct from the semantic values of its parts. For example, suppose we have a rule which says an expression can be the sum of two expressions. When the parser recognizes such a sum, each of the subexpressions has a semantic value which describes how it was built up. The action for this rule should create a similar sort of value for the newly recognized larger expression.
For example, here is a rule that says an expression can be the sum of
two subexpressions:
expr: expr '+' expr { $$ = $1 + $3; }
;
The action says how to produce the semantic value of the sum expression from the values of the two subexpressions.
Many applications, like interpreters or compilers, have to produce verbose and useful error messages. To achieve this, one must be able to keep track of the textual position, or location, of each syntactic construct. Bison provides a mechanism for handling these locations.
Each token has a semantic value. In a similar fashion, each token has an associated location, but the type of locations is the same for all tokens and groupings. Moreover, the output parser is equipped with a default data structure for storing locations (see Locations, for more details).
Like semantic values, locations can be reached in actions using a dedicated
set of constructs. In the example above, the location of the whole grouping
is @$, while the locations of the subexpressions are @1 and
@3.
When a rule is matched, a default action is used to compute the semantic value
of its left hand side (see Actions). In the same way, another default
action is used for locations. However, the action for locations is general
enough for most cases, meaning there is usually no need to describe for each
rule how @$ should be formed. When building a new location for a given
grouping, the default behavior of the output parser is to take the beginning
of the first symbol, and the end of the last symbol.
When you run Bison, you give it a Bison grammar file as input. The output is a C source file that parses the language described by the grammar. This file is called a Bison parser. Keep in mind that the Bison utility and the Bison parser are two distinct programs: the Bison utility is a program whose output is the Bison parser that becomes part of your program.
The job of the Bison parser is to group tokens into groupings according to the grammar rules--for example, to build identifiers and operators into expressions. As it does this, it runs the actions for the grammar rules it uses.
The tokens come from a function called the lexical analyzer that you
must supply in some fashion (such as by writing it in C). The Bison parser
calls the lexical analyzer each time it wants a new token. It doesn't know
what is "inside" the tokens (though their semantic values may reflect
this). Typically the lexical analyzer makes the tokens by parsing
characters of text, but Bison does not depend on this. See The Lexical Analyzer Function yylex.
The Bison parser file is C code which defines a function named
yyparse which implements that grammar. This function does not make
a complete C program: you must supply some additional functions. One is
the lexical analyzer. Another is an error-reporting function which the
parser calls to report an error. In addition, a complete C program must
start with a function called main; you have to provide this, and
arrange for it to call yyparse or the parser will never run.
See Parser C-Language Interface.
Aside from the token type names and the symbols in the actions you
write, all symbols defined in the Bison parser file itself
begin with yy or YY. This includes interface functions
such as the lexical analyzer function yylex, the error reporting
function yyerror and the parser function yyparse itself.
This also includes numerous identifiers used for internal purposes.
Therefore, you should avoid using C identifiers starting with yy
or YY in the Bison grammar file except for the ones defined in
this manual.
In some cases the Bison parser file includes system headers, and in
those cases your code should respect the identifiers reserved by those
headers. On some non-GNU hosts, <alloca.h>,
<stddef.h>, and <stdlib.h> are included as needed to
declare memory allocators and related types.
Other system headers may be included if you define YYDEBUG to a
nonzero value (see Debugging Your Parser).
The actual language-design process using Bison, from grammar specification to a working compiler or interpreter, has these parts:
yylex). It could also be produced using Lex, but the use
of Lex is not discussed in this manual.
To turn this source code as written into a runnable program, you must follow these steps:
The input file for the Bison utility is a Bison grammar file. The
general form of a Bison grammar file is as follows:
%{
C declarations
%}
Bison declarations
%%
Grammar rules
%%
Additional C code
The %%, %{ and %} are punctuation that appears
in every Bison grammar file to separate the sections.
The C declarations may define types and variables used in the actions.
You can also use preprocessor commands to define macros used there, and use
#include to include header files that do any of these things.
The Bison declarations declare the names of the terminal and nonterminal symbols, and may also describe operator precedence and the data types of semantic values of various symbols.
The grammar rules define how to construct each nonterminal symbol from its parts.
The additional C code can contain any C code you want to use. Often the
definition of the lexical analyzer yylex goes here, plus subroutines
called by the actions in the grammar rules. In a simple program, all the
rest of the program can go here.
Now we show and explain three sample programs written using Bison: a reverse polish notation calculator, an algebraic (infix) notation calculator, and a multi-function calculator. All three have been tested under BSD Unix 4.3; each produces a usable, though limited, interactive desk-top calculator.
These examples are simple, but Bison grammars for real programming languages are written the same way.
The first example is that of a simple double-precision reverse polish notation calculator (a calculator using postfix operators). This example provides a good starting point, since operator precedence is not an issue. The second example will illustrate how operator precedence is handled.
The source code for this calculator is named rpcalc.y. The
.y extension is a convention used for Bison input files.
rpcalcHere are the C and Bison declarations for the reverse polish notation
calculator. As in C, comments are placed between /*...*/.
/* Reverse polish notation calculator. */
%{
#define YYSTYPE double
#include <math.h>
%}
%token NUM
%% /* Grammar rules and actions follow */
The C declarations section (see The C Declarations Section) contains two preprocessor directives.
The #define directive defines the macro YYSTYPE, thus
specifying the C data type for semantic values of both tokens and
groupings (see Data Types of Semantic Values). The
Bison parser will use whatever type YYSTYPE is defined as; if you
don't define it, int is the default. Because we specify
double, each token and each expression has an associated value,
which is a floating point number.
The #include directive is used to declare the exponentiation
function pow.
The second section, Bison declarations, provides information to Bison about
the token types (see The Bison Declarations Section). Each terminal symbol that is
not a single-character literal must be declared here. (Single-character
literals normally don't need to be declared.) In this example, all the
arithmetic operators are designated by single-character literals, so the
only terminal symbol that needs to be declared is NUM, the token
type for numeric constants.
rpcalcHere are the grammar rules for the reverse polish notation calculator.
input: /* empty */
| input line
;
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
exp: NUM { $$ = $1; }
| exp exp '+' { $$ = $1 + $2; }
| exp exp '-' { $$ = $1 - $2; }
| exp exp '*' { $$ = $1 * $2; }
| exp exp '/' { $$ = $1 / $2; }
/* Exponentiation */
| exp exp '^' { $$ = pow ($1, $2); }
/* Unary minus */
| exp 'n' { $$ = -$1; }
;
%%
The groupings of the rpcalc "language" defined here are the expression
(given the name exp), the line of input (line), and the
complete input transcript (input). Each of these nonterminal
symbols has several alternate rules, joined by the | punctuator
which is read as "or". The following sections explain what these rules
mean.
The semantics of the language is determined by the actions taken when a grouping is recognized. The actions are the C code that appears inside braces. See Actions.
You must specify these actions in C, but Bison provides the means for
passing semantic values between the rules. In each action, the
pseudo-variable $$ stands for the semantic value for the grouping
that the rule is going to construct. Assigning a value to $$ is the
main job of most actions. The semantic values of the components of the
rule are referred to as $1, $2, and so on.
inputConsider the definition of input:
input: /* empty */
| input line
;
This definition reads as follows: "A complete input is either an empty
string, or a complete input followed by an input line". Notice that
"complete input" is defined in terms of itself. This definition is said
to be left recursive since input appears always as the
leftmost symbol in the sequence. See Recursive Rules.
The first alternative is empty because there are no symbols between the
colon and the first |; this means that input can match an
empty string of input (no tokens). We write the rules this way because it
is legitimate to type Ctrl-d right after you start the calculator.
It's conventional to put an empty alternative first and write the comment
/* empty */ in it.
The second alternate rule (input line) handles all nontrivial input.
It means, "After reading any number of lines, read one more line if
possible." The left recursion makes this rule into a loop. Since the
first alternative matches empty input, the loop can be executed zero or
more times.
The parser function yyparse continues to process input until a
grammatical error is seen or the lexical analyzer says there are no more
input tokens; we will arrange for the latter to happen at end of file.
lineNow consider the definition of line:
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
The first alternative is a token which is a newline character; this means
that rpcalc accepts a blank line (and ignores it, since there is no
action). The second alternative is an expression followed by a newline.
This is the alternative that makes rpcalc useful. The semantic value of
the exp grouping is the value of $1 because the exp in
question is the first symbol in the alternative. The action prints this
value, which is the result of the computation the user asked for.
This action is unusual because it does not assign a value to $$. As
a consequence, the semantic value associated with the line is
uninitialized (its value will be unpredictable). This would be a bug if
that value were ever used, but we don't use it: once rpcalc has printed the
value of the user's input line, that value is no longer needed.
exprThe exp grouping has several rules, one for each kind of expression.
The first rule handles the simplest expressions: those that are just numbers.
The second handles an addition-expression, which looks like two expressions
followed by a plus-sign. The third handles subtraction, and so on.
exp: NUM
| exp exp '+' { $$ = $1 + $2; }
| exp exp '-' { $$ = $1 - $2; }
...
;
We have used | to join all the rules for exp, but we could
equally well have written them separately:
exp: NUM ;
exp: exp exp '+' { $$ = $1 + $2; } ;
exp: exp exp '-' { $$ = $1 - $2; } ;
...
Most of the rules have actions that compute the value of the expression in
terms of the value of its parts. For example, in the rule for addition,
$1 refers to the first component exp and $2 refers to
the second one. The third component, '+', has no meaningful
associated semantic value, but if it had one you could refer to it as
$3. When yyparse recognizes a sum expression using this
rule, the sum of the two subexpressions' values is produced as the value of
the entire expression. See Actions.
You don't have to give an action for every rule. When a rule has no
action, Bison by default copies the value of $1 into $$.
This is what happens in the first rule (the one that uses NUM).
The formatting shown here is the recommended convention, but Bison does
not require it. You can add or change whitespace as much as you wish.
For example, this:
exp : NUM | exp exp '+' {$$ = $1 + $2; } | ...
means the same thing as this:
exp: NUM
| exp exp '+' { $$ = $1 + $2; }
| ...
The latter, however, is much more readable.
rpcalc Lexical AnalyzerThe lexical analyzer's job is low-level parsing: converting characters or
sequences of characters into tokens. The Bison parser gets its tokens by
calling the lexical analyzer. See The Lexical Analyzer Function yylex.
Only a simple lexical analyzer is needed for the RPN calculator. This
lexical analyzer skips blanks and tabs, then reads in numbers as
double and returns them as NUM tokens. Any other character
that isn't part of a number is a separate token. Note that the token-code
for such a single-character token is the character itself.
The return value of the lexical analyzer function is a numeric code which
represents a token type. The same text used in Bison rules to stand for
this token type is also a C expression for the numeric code for the type.
This works in two ways. If the token type is a character literal, then its
numeric code is the ASCII code for that character; you can use the same
character literal in the lexical analyzer to express the number. If the
token type is an identifier, that identifier is defined by Bison as a C
macro whose definition is the appropriate number. In this example,
therefore, NUM becomes a macro for yylex to use.
The semantic value of the token (if it has one) is stored into the
global variable yylval, which is where the Bison parser will look
for it. (The C data type of yylval is YYSTYPE, which was
defined at the beginning of the grammar; see Declarations for rpcalc.)
A token type code of zero is returned if the end-of-file is encountered. (Bison recognizes any nonpositive value as indicating the end of the input.)
Here is the code for the lexical analyzer:
/* Lexical analyzer returns a double floating point
number on the stack and the token NUM, or the ASCII
character read if not a number. Skips all blanks
and tabs, returns 0 for EOF. */
#include <ctype.h>
int
yylex (void)
{
int c;
/* skip white space */
while ((c = getchar ()) == ' ' || c == '\t')
;
/* process numbers */
if (c == '.' || isdigit (c))
{
ungetc (c, stdin);
scanf ("%lf", &yylval);
return NUM;
}
/* return end-of-file */
if (c == EOF)
return 0;
/* return single chars */
return c;
}
In keeping with the spirit of this example, the controlling function is
kept to the bare minimum. The only requirement is that it call
yyparse to start the process of parsing.
int
main (void)
{
return yyparse ();
}
When yyparse detects a syntax error, it calls the error reporting
function yyerror to print an error message (usually but not
always "parse error"). It is up to the programmer to supply
yyerror (see Parser C-Language Interface), so
here is the definition we will use:
#include <stdio.h>
void
yyerror (const char *s) /* Called by yyparse on error */
{
printf ("%s\n", s);
}
After yyerror returns, the Bison parser may recover from the error
and continue parsing if the grammar contains a suitable error rule
(see Error Recovery). Otherwise, yyparse returns nonzero. We
have not written any error rules in this example, so any invalid input will
cause the calculator program to exit. This is not clean behavior for a
real calculator, but it is adequate for the first example.
Before running Bison to produce a parser, we need to decide how to
arrange all the source code in one or more source files. For such a
simple example, the easiest thing is to put everything in one file. The
definitions of yylex, yyerror and main go at the
end, in the "additional C code" section of the file (see The Overall Layout of a Bison Grammar).
For a large project, you would probably have several source files, and use
make to arrange to recompile them.
With all the source in a single file, you use the following command to
convert it into a parser file:
bison file_name.y
In this example the file was called rpcalc.y (for "Reverse Polish
CALCulator"). Bison produces a file named file_name.tab.c,
removing the .y from the original file name. The file output by
Bison contains the source code for yyparse. The additional
functions in the input file (yylex, yyerror and main)
are copied verbatim to the output.
Here is how to compile and run the parser file:
# List files in current directory. $ ls rpcalc.tab.c rpcalc.y # Compile the Bison parser. #-lmtells compiler to search math library forpow. $ cc rpcalc.tab.c -lm -o rpcalc # List files again. $ ls rpcalc rpcalc.tab.c rpcalc.y
The file rpcalc now contains the executable code. Here is an
example session using rpcalc.
$ rpcalc
4 9 +
13
3 7 + 3 4 5 *+-
-13
3 7 + 3 4 5 * + - n Note the unary minus, n
13
5 6 / 4 n +
-3.166666667
3 4 ^ Exponentiation
81
^D End-of-file indicator
$
calcWe now modify rpcalc to handle infix operators instead of postfix. Infix
notation involves the concept of operator precedence and the need for
parentheses nested to arbitrary depth. Here is the Bison code for
calc.y, an infix desk-top calculator.
/* Infix notation calculator--calc */
%{
#define YYSTYPE double
#include <math.h>
%}
/* BISON Declarations */
%token NUM
%left '-' '+'
%left '*' '/'
%left NEG /* negation--unary minus */
%right '^' /* exponentiation */
/* Grammar follows */
%%
input: /* empty string */
| input line
;
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
;
exp: NUM { $$ = $1; }
| exp '+' exp { $$ = $1 + $3; }
| exp '-' exp { $$ = $1 - $3; }
| exp '*' exp { $$ = $1 * $3; }
| exp '/' exp { $$ = $1 / $3; }
| '-' exp %prec NEG { $$ = -$2; }
| exp '^' exp { $$ = pow ($1, $3); }
| '(' exp ')' { $$ = $2; }
;
%%
The functions yylex, yyerror and main can be the
same as before.
There are two important new features shown in this code.
In the second section (Bison declarations), %left declares token
types and says they are left-associative operators. The declarations
%left and %right (right associativity) take the place of
%token which is used to declare a token type name without
associativity. (These tokens are single-character literals, which
ordinarily don't need to be declared. We declare them here to specify
the associativity.)
Operator precedence is determined by the line ordering of the
declarations; the higher the line number of the declaration (lower on
the page or screen), the higher the precedence. Hence, exponentiation
has the highest precedence, unary minus (NEG) is next, followed
by * and /, and so on. See Operator Precedence.
The other important new feature is the %prec in the grammar section
for the unary minus operator. The %prec simply instructs Bison that
the rule | '-' exp has the same precedence as NEG--in this
case the next-to-highest. See Context-Dependent Precedence.
Here is a sample run of calc.y:
$ calc 4 + 4.5 - (34/(8*3+-3)) 6.880952381 -56 + 2 -54 3 ^ 2 9
Up to this point, this manual has not addressed the issue of error
recovery--how to continue parsing after the parser detects a syntax
error. All we have handled is error reporting with yyerror.
Recall that by default yyparse returns after calling
yyerror. This means that an erroneous input line causes the
calculator program to exit. Now we show how to rectify this deficiency.
The Bison language itself includes the reserved word error, which
may be included in the grammar rules. In the example below it has
been added to one of the alternatives for line:
line: '\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
| error '\n' { yyerrok; }
;
This addition to the grammar allows for simple error recovery in the
event of a parse error. If an expression that cannot be evaluated is
read, the error will be recognized by the third rule for line,
and parsing will continue. (The yyerror function is still called
upon to print its message as well.) The action executes the statement
yyerrok, a macro defined automatically by Bison; its meaning is
that error recovery is complete (see Error Recovery). Note the
difference between yyerrok and yyerror; neither one is a
misprint.
This form of error recovery deals with syntax errors. There are other
kinds of errors; for example, division by zero, which raises an exception
signal that is normally fatal. A real calculator program must handle this
signal and use longjmp to return to main and resume parsing
input lines; it would also have to discard the rest of the current line of
input. We won't discuss this issue further because it is not specific to
Bison programs.
ltcalcThis example extends the infix notation calculator with location tracking. This feature will be used to improve the error messages. For the sake of clarity, this example is a simple integer calculator, since most of the work needed to use locations will be done in the lexical analyser.
ltcalcThe C and Bison declarations for the location tracking calculator are
the same as the declarations for the infix notation calculator.
/* Location tracking calculator. */
%{
#define YYSTYPE int
#include <math.h>
%}
/* Bison declarations. */
%token NUM
%left '-' '+'
%left '*' '/'
%left NEG
%right '^'
%% /* Grammar follows */
Note there are no declarations specific to locations. Defining a data
type for storing locations is not needed: we will use the type provided
by default (see Data Types of Locations), which is a
four member structure with the following integer fields:
first_line, first_column, last_line and
last_column.
ltcalcWhether handling locations or not has no effect on the syntax of your language. Therefore, grammar rules for this example will be very close to those of the previous example: we will only modify them to benefit from the new information.
Here, we will use locations to report divisions by zero, and locate the
wrong expressions or subexpressions.
input : /* empty */
| input line
;
line : '\n'
| exp '\n' { printf ("%d\n", $1); }
;
exp : NUM { $$ = $1; }
| exp '+' exp { $$ = $1 + $3; }
| exp '-' exp { $$ = $1 - $3; }
| exp '*' exp { $$ = $1 * $3; }
| exp '/' exp
{
if ($3)
$$ = $1 / $3;
else
{
$$ = 1;
fprintf (stderr, "%d.%d-%d.%d: division by zero",
@3.first_line, @3.first_column,
@3.last_line, @3.last_column);
}
}
| '-' exp %preg NEG { $$ = -$2; }
| exp '^' exp { $$ = pow ($1, $3); }
| '(' exp ')' { $$ = $2; }
This code shows how to reach locations inside of semantic actions, by
using the pseudo-variables @n for rule components, and the
pseudo-variable @$ for groupings.
We don't need to assign a value to @$: the output parser does it
automatically. By default, before executing the C code of each action,
@$ is set to range from the beginning of @1 to the end
of @n, for a rule with n components. This behavior
can be redefined (see Default Action for Locations), and for very specific rules, @$ can be computed by
hand.
ltcalc Lexical Analyzer.Until now, we relied on Bison's defaults to enable location tracking. The next step is to rewrite the lexical analyser, and make it able to feed the parser with the token locations, as it already does for semantic values.
To this end, we must take into account every single character of the
input text, to avoid the computed locations of being fuzzy or wrong:
int
yylex (void)
{
int c;
/* skip white space */
while ((c = getchar ()) == ' ' || c == '\t')
++yylloc.last_column;
/* step */
yylloc.first_line = yylloc.last_line;
yylloc.first_column = yylloc.last_column;
/* process numbers */
if (isdigit (c))
{
yylval = c - '0';
++yylloc.last_column;
while (isdigit (c = getchar ()))
{
++yylloc.last_column;
yylval = yylval * 10 + c - '0';
}
ungetc (c, stdin);
return NUM;
}
/* return end-of-file */
if (c == EOF)
return 0;
/* return single chars and update location */
if (c == '\n')
{
++yylloc.last_line;
yylloc.last_column = 0;
}
else
++yylloc.last_column;
return c;
}
Basically, the lexical analyzer performs the same processing as before:
it skips blanks and tabs, and reads numbers or single-character tokens.
In addition, it updates yylloc, the global variable (of type
YYLTYPE) containing the token's location.
Now, each time this function returns a token, the parser has its number
as well as its semantic value, and its location in the text. The last
needed change is to initialize yylloc, for example in the
controlling function:
int
main (void)
{
yylloc.first_line = yylloc.last_line = 1;
yylloc.first_column = yylloc.last_column = 0;
return yyparse ();
}
Remember that computing locations is not a matter of syntax. Every character must be associated to a location update, whether it is in valid input, in comments, in literal strings, and so on.
mfcalcNow that the basics of Bison have been discussed, it is time to move on to
a more advanced problem. The above calculators provided only five
functions, +, -, *, / and ^. It would
be nice to have a calculator that provides other mathematical functions such
as sin, cos, etc.
It is easy to add new operators to the infix calculator as long as they are
only single-character literals. The lexical analyzer yylex passes
back all nonnumber characters as tokens, so new grammar rules suffice for
adding a new operator. But we want something more flexible: built-in
functions whose syntax has this form:
function_name (argument)
At the same time, we will add memory to the calculator, by allowing you
to create named variables, store values in them, and use them later.
Here is a sample session with the multi-function calculator:
$ mfcalc pi = 3.141592653589 3.1415926536 sin(pi) 0.0000000000 alpha = beta1 = 2.3 2.3000000000 alpha 2.3000000000 ln(alpha) 0.8329091229 exp(ln(beta1)) 2.3000000000 $
Note that multiple assignment and nested function calls are permitted.
mfcalcHere are the C and Bison declarations for the multi-function calculator.
%{
#include <math.h> /* For math functions, cos(), sin(), etc. */
#include "calc.h" /* Contains definition of `symrec' */
%}
%union {
double val; /* For returning numbers. */
symrec *tptr; /* For returning symbol-table pointers */
}
%token <val> NUM /* Simple double precision number */
%token <tptr> VAR FNCT /* Variable and Function */
%type <val> exp
%right '='
%left '-' '+'
%left '*' '/'
%left NEG /* Negation--unary minus */
%right '^' /* Exponentiation */
/* Grammar follows */
%%
The above grammar introduces only two new features of the Bison language. These features allow semantic values to have various data types (see More Than One Value Type).
The %union declaration specifies the entire list of possible types;
this is instead of defining YYSTYPE. The allowable types are now
double-floats (for exp and NUM) and pointers to entries in
the symbol table. See The Collection of Value Types.
Since values can now have various types, it is necessary to associate a
type with each grammar symbol whose semantic value is used. These symbols
are NUM, VAR, FNCT, and exp. Their
declarations are augmented with information about their data type (placed
between angle brackets).
The Bison construct %type is used for declaring nonterminal symbols,
just as %token is used for declaring token types. We have not used
%type before because nonterminal symbols are normally declared
implicitly by the rules that define them. But exp must be declared
explicitly so we can specify its value type. See Nonterminal Symbols.
mfcalcHere are the grammar rules for the multi-function calculator.
Most of them are copied directly from calc; three rules,
those which mention VAR or FNCT, are new.
input: /* empty */
| input line
;
line:
'\n'
| exp '\n' { printf ("\t%.10g\n", $1); }
| error '\n' { yyerrok; }
;
exp: NUM { $$ = $1; }
| VAR { $$ = $1->value.var; }
| VAR '=' exp { $$ = $3; $1->value.var = $3; }
| FNCT '(' exp ')' { $$ = (*($1->value.fnctptr))($3); }
| exp '+' exp { $$ = $1 + $3; }
| exp '-' exp { $$ = $1 - $3; }
| exp '*' exp { $$ = $1 * $3; }
| exp '/' exp { $$ = $1 / $3; }
| '-' exp %prec NEG { $$ = -$2; }
| exp '^' exp { $$ = pow ($1, $3); }
| '(' exp ')' { $$ = $2; }
;
/* End of grammar */
%%
mfcalc Symbol TableThe multi-function calculator requires a symbol table to keep track of the names and meanings of variables and functions. This doesn't affect the grammar rules (except for the actions) or the Bison declarations, but it requires some additional C functions for support.
The symbol table itself consists of a linked list of records. Its
definition, which is kept in the header calc.h, is as follows. It
provides for either functions or variables to be placed in the table.
/* Fonctions type. */
typedef double (*func_t) (double);
/* Data type for links in the chain of symbols. */
struct symrec
{
char *name; /* name of symbol */
int type; /* type of symbol: either VAR or FNCT */
union
{
double var; /* value of a VAR */
func_t fnctptr; /* value of a FNCT */
} value;
struct symrec *next; /* link field */
};
typedef struct symrec symrec;
/* The symbol table: a chain of `struct symrec'. */
extern symrec *sym_table;
symrec *putsym (const char *, func_t);
symrec *getsym (const char *);
The new version of main includes a call to init_table, a
function that initializes the symbol table. Here it is, and
init_table as well:
#include <stdio.h>
int
main (void)
{
init_table ();
return yyparse ();
}
void
yyerror (const char *s) /* Called by yyparse on error */
{
printf ("%s\n", s);
}
struct init
{
char *fname;
double (*fnct)(double);
};
struct init arith_fncts[] =
{
"sin", sin,
"cos", cos,
"atan", atan,
"ln", log,
"exp", exp,
"sqrt", sqrt,
0, 0
};
/* The symbol table: a chain of `struct symrec'. */
symrec *sym_table = (symrec *) 0;
/* Put arithmetic functions in table. */
void
init_table (void)
{
int i;
symrec *ptr;
for (i = 0; arith_fncts[i].fname != 0; i++)
{
ptr = putsym (arith_fncts[i].fname, FNCT);
ptr->value.fnctptr = arith_fncts[i].fnct;
}
}
By simply editing the initialization list and adding the necessary include files, you can add additional functions to the calculator.
Two important functions allow look-up and installation of symbols in the
symbol table. The function putsym is passed a name and the type
(VAR or FNCT) of the object to be installed. The object is
linked to the front of the list, and a pointer to the object is returned.
The function getsym is passed the name of the symbol to look up. If
found, a pointer to that symbol is returned; otherwise zero is returned.
symrec *
putsym (char *sym_name, int sym_type)
{
symrec *ptr;
ptr = (symrec *) malloc (sizeof (symrec));
ptr->name = (char *) malloc (strlen (sym_name) + 1);
strcpy (ptr->name,sym_name);
ptr->type = sym_type;
ptr->value.var = 0; /* set value to 0 even if fctn. */
ptr->next = (struct symrec *)sym_table;
sym_table = ptr;
return ptr;
}
symrec *
getsym (const char *sym_name)
{
symrec *ptr;
for (ptr = sym_table; ptr != (symrec *) 0;
ptr = (symrec *)ptr->next)
if (strcmp (ptr->name,sym_name) == 0)
return ptr;
return 0;
}
The function yylex must now recognize variables, numeric values, and
the single-character arithmetic operators. Strings of alphanumeric
characters with a leading non-digit are recognized as either variables or
functions depending on what the symbol table says about them.
The string is passed to getsym for look up in the symbol table. If
the name appears in the table, a pointer to its location and its type
(VAR or FNCT) is returned to yyparse. If it is not
already in the table, then it is installed as a VAR using
putsym. Again, a pointer and its type (which must be VAR) is
returned to yyparse.
No change is needed in the handling of numeric values and arithmetic
operators in yylex.
#include <ctype.h>
int
yylex (void)
{
int c;
/* Ignore whitespace, get first nonwhite character. */
while ((c = getchar ()) == ' ' || c == '\t');
if (c == EOF)
return 0;
/* Char starts a number => parse the number. */
if (c == '.' || isdigit (c))
{
ungetc (c, stdin);
scanf ("%lf", &yylval.val);
return NUM;
}
/* Char starts an identifier => read the name. */
if (isalpha (c))
{
symrec *s;
static char *symbuf = 0;
static int length = 0;
int i;
/* Initially make the buffer long enough
for a 40-character symbol name. */
if (length == 0)
length = 40, symbuf = (char *)malloc (length + 1);
i = 0;
do
{
/* If buffer is full, make it bigger. */
if (i == length)
{
length *= 2;
symbuf = (char *)realloc (symbuf, length + 1);
}
/* Add this character to the buffer. */
symbuf[i++] = c;
/* Get another character. */
c = getchar ();
}
while (c != EOF && isalnum (c));
ungetc (c, stdin);
symbuf[i] = '\0';
s = getsym (symbuf);
if (s == 0)
s = putsym (symbuf, VAR);
yylval.tptr = s;
return s->type;
}
/* Any other character is a token by itself. */
return c;
}
This program is both powerful and flexible. You may easily add new
functions, and it is a simple job to modify this code to install predefined
variables such as pi or e as well.
math.h to the initialization list.
init_table to add these constants to the symbol table.
It will be easiest to give the constants type VAR.
Bison takes as input a context-free grammar specification and produces a C-language function that recognizes correct instances of the grammar.
The Bison grammar input file conventionally has a name ending in .y.
See Invoking Bison.
A Bison grammar file has four main sections, shown here with the
appropriate delimiters:
%{
C declarations
%}
Bison declarations
%%
Grammar rules
%%
Additional C code
Comments enclosed in /* ... */ may appear in any of the sections.
The C declarations section contains macro definitions and
declarations of functions and variables that are used in the actions in the
grammar rules. These are copied to the beginning of the parser file so
that they precede the definition of yyparse. You can use
#include to get the declarations from a header file. If you don't
need any C declarations, you may omit the %{ and %}
delimiters that bracket this section.
The Bison declarations section contains declarations that define terminal and nonterminal symbols, specify precedence, and so on. In some simple grammars you may not need any declarations. See Bison Declarations.
The grammar rules section contains one or more Bison grammar rules, and nothing else. See Syntax of Grammar Rules.
There must always be at least one grammar rule, and the first
%% (which precedes the grammar rules) may never be omitted even
if it is the first thing in the file.
The additional C code section is copied verbatim to the end of the
parser file, just as the C declarations section is copied to the
beginning. This is the most convenient place to put anything that you
want to have in the parser file but which need not come before the
definition of yyparse. For example, the definitions of
yylex and yyerror often go here. See Parser C-Language Interface.
If the last section is empty, you may omit the %% that separates it
from the grammar rules.
The Bison parser itself contains many static variables whose names start
with yy and many macros whose names start with YY. It is a
good idea to avoid using any such names (except those documented in this
manual) in the additional C code section of the grammar file.
Symbols in Bison grammars represent the grammatical classifications of the language.
A terminal symbol (also known as a token type) represents a
class of syntactically equivalent tokens. You use the symbol in grammar
rules to mean that a token in that class is allowed. The symbol is
represented in the Bison parser by a numeric code, and the yylex
function returns a token type code to indicate what kind of token has been
read. You don't need to know what the code value is; you can use the
symbol to stand for it.
A nonterminal symbol stands for a class of syntactically equivalent groupings. The symbol name is used in writing grammar rules. By convention, it should be all lower case.
Symbol names can contain letters, digits (not at the beginning), underscores and periods. Periods make sense only in nonterminals.
There are three ways of writing terminal symbols in the grammar:
%token. See Token Type Names.
'+' is a character token type. A
character token type doesn't need to be declared unless you need to
specify its semantic value data type (see Data Types of Semantic Values), associativity, or precedence (see Operator Precedence).
By convention, a character token type is used only to represent a
token that consists of that particular character. Thus, the token
type '+' is used to represent the character + as a
token. Nothing enforces this convention, but if you depart from it,
your program will confuse other readers.
All the usual escape sequences used in character literals in C can be
used in Bison as well, but you must not use the null character as a
character literal because its ASCII code, zero, is the code yylex
returns for end-of-input (see Calling Convention for yylex).
"<=" is a literal string token. A literal string token
doesn't need to be declared unless you need to specify its semantic
value data type (see Value Type), associativity, or precedence
(see Precedence).
You can associate the literal string token with a symbolic name as an
alias, using the %token declaration (see Token Declarations). If you don't do that, the lexical analyzer has to
retrieve the token number for the literal string token from the
yytname table (see Calling Convention).
WARNING: literal string tokens do not work in Yacc.
By convention, a literal string token is used only to represent a token
that consists of that particular string. Thus, you should use the token
type "<=" to represent the string <= as a token. Bison
does not enforce this convention, but if you depart from it, people who
read your program will be confused.
All the escape sequences used in string literals in C can be used in Bison as well. A literal string token must contain two or more characters; for a token containing just one character, use a character token (see above).
How you choose to write a terminal symbol has no effect on its grammatical meaning. That depends only on where it appears in rules and on when the parser function returns that symbol.
The value returned by yylex is always one of the terminal symbols
(or 0 for end-of-input). Whichever way you write the token type in the
grammar rules, you write it the same way in the definition of yylex.
The numeric code for a character token type is simply the ASCII code for
the character, so yylex can use the identical character constant to
generate the requisite code. Each named token type becomes a C macro in
the parser file, so yylex can use the name to stand for the code.
(This is why periods don't make sense in terminal symbols.)
See Calling Convention for yylex.
If yylex is defined in a separate file, you need to arrange for the
token-type macro definitions to be available there. Use the -d
option when you run Bison, so that it will write these macro definitions
into a separate header file name.tab.h which you can include
in the other source files that need it. See Invoking Bison.
The symbol error is a terminal symbol reserved for error recovery
(see Error Recovery); you shouldn't use it for any other purpose.
In particular, yylex should never return this value.
A Bison grammar rule has the following general form:
result: components...
;
where result is the nonterminal symbol that this rule describes, and components are various terminal and nonterminal symbols that are put together by this rule (see Symbols).
For example,
exp: exp '+' exp
;
says that two groupings of type exp, with a + token in between,
can be combined into a larger grouping of type exp.
Whitespace in rules is significant only to separate symbols. You can add extra whitespace as you wish.
Scattered among the components can be actions that determine
the semantics of the rule. An action looks like this:
{C statements}
Usually there is only one action and it follows the components. See Actions.
Multiple rules for the same result can be written separately or can
be joined with the vertical-bar character | as follows:
They are still considered distinct rules even when joined in this way.
If components in a rule is empty, it means that result can
match the empty string. For example, here is how to define a
comma-separated sequence of zero or more exp groupings:
expseq: /* empty */
| expseq1
;
expseq1: exp
| expseq1 ',' exp
;
It is customary to write a comment /* empty */ in each rule
with no components.
A rule is called recursive when its result nonterminal appears
also on its right hand side. Nearly all Bison grammars need to use
recursion, because that is the only way to define a sequence of any number
of a particular thing. Consider this recursive definition of a
comma-separated sequence of one or more expressions:
expseq1: exp
| expseq1 ',' exp
;
Since the recursive use of expseq1 is the leftmost symbol in the
right hand side, we call this left recursion. By contrast, here
the same construct is defined using right recursion:
expseq1: exp
| exp ',' expseq1
;
Any kind of sequence can be defined using either left recursion or right recursion, but you should always use left recursion, because it can parse a sequence of any number of elements with bounded stack space. Right recursion uses up space on the Bison stack in proportion to the number of elements in the sequence, because all the elements must be shifted onto the stack before the rule can be applied even once. See The Bison Parser Algorithm, for further explanation of this.
Indirect or mutual recursion occurs when the result of the rule does not appear directly on its right hand side, but does appear in rules for other nonterminals which do appear on its right hand side.
For example:
expr: primary
| primary '+' primary
;
primary: constant
| '(' expr ')'
;
defines two mutually-recursive nonterminals, since each refers to the other.
The grammar rules for a language determine only the syntax. The semantics are determined by the semantic values associated with various tokens and groupings, and by the actions taken when various groupings are recognized.
For example, the calculator calculates properly because the value
associated with each expression is the proper number; it adds properly
because the action for the grouping x + y is to add
the numbers associated with x and y.
In a simple program it may be sufficient to use the same data type for the semantic values of all language constructs. This was true in the RPN and infix calculator examples (see Reverse Polish Notation Calculator).
Bison's default is to use type int for all semantic values. To
specify some other type, define YYSTYPE as a macro, like this:
#define YYSTYPE double
This macro definition must go in the C declarations section of the grammar file (see Outline of a Bison Grammar).
In most programs, you will need different data types for different kinds
of tokens and groupings. For example, a numeric constant may need type
int or long, while a string constant needs type char *,
and an identifier might need a pointer to an entry in the symbol table.
To use more than one data type for semantic values in one parser, Bison requires you to do two things:
%union Bison declaration (see The Collection of Value Types).
%token Bison declaration (see Token Type Names)
and for groupings with the %type Bison declaration (see Nonterminal Symbols).
An action accompanies a syntactic rule and contains C code to be executed each time an instance of that rule is recognized. The task of most actions is to compute a semantic value for the grouping built by the rule from the semantic values associated with tokens or smaller groupings.
An action consists of C statements surrounded by braces, much like a compound statement in C. It can be placed at any position in the rule; it is executed at that position. Most rules have just one action at the end of the rule, following all the components. Actions in the middle of a rule are tricky and used only for special purposes (see Actions in Mid-Rule).
The C code in an action can refer to the semantic values of the components
matched by the rule with the construct $n, which stands for
the value of the nth component. The semantic value for the grouping
being constructed is $$. (Bison translates both of these constructs
into array element references when it copies the actions into the parser
file.)
Here is a typical example:
exp: ...
| exp '+' exp
{ $$ = $1 + $3; }
This rule constructs an exp from two smaller exp groupings
connected by a plus-sign token. In the action, $1 and $3
refer to the semantic values of the two component exp groupings,
which are the first and third symbols on the right hand side of the rule.
The sum is stored into $$ so that it becomes the semantic value of
the addition-expression just recognized by the rule. If there were a
useful semantic value associated with the + token, it could be
referred to as $2.
If you don't specify an action for a rule, Bison supplies a default:
$$ = $1. Thus, the value of the first symbol in the rule becomes
the value of the whole rule. Of course, the default rule is valid only
if the two data types match. There is no meaningful default action for
an empty rule; every empty rule must have an explicit action unless the
rule's value does not matter.
$n with n zero or negative is allowed for reference
to tokens and groupings on the stack before those that match the
current rule. This is a very risky practice, and to use it reliably
you must be certain of the context in which the rule is applied. Here
is a case in which you can use this reliably:
foo: expr bar '+' expr { ... }
| expr bar '-' expr { ... }
;
bar: /* empty */
{ previous_expr = $0; }
;
As long as bar is used only in the fashion shown here, $0
always refers to the expr which precedes bar in the
definition of foo.
If you have chosen a single data type for semantic values, the $$
and $n constructs always have that data type.
If you have used %union to specify a variety of data types, then you
must declare a choice among these types for each terminal or nonterminal
symbol that can have a semantic value. Then each time you use $$ or
$n, its data type is determined by which symbol it refers to
in the rule. In this example,
exp: ...
| exp '+' exp
{ $$ = $1 + $3; }
$1 and $3 refer to instances of exp, so they all
have the data type declared for the nonterminal symbol exp. If
$2 were used, it would have the data type declared for the
terminal symbol '+', whatever that might be.
Alternatively, you can specify the data type when you refer to the value,
by inserting <type> after the $ at the beginning of the
reference. For example, if you have defined types as shown here:
%union {
int itype;
double dtype;
}
then you can write $<itype>1 to refer to the first subunit of the
rule as an integer, or $<dtype>1 to refer to it as a double.
Occasionally it is useful to put an action in the middle of a rule. These actions are written just like usual end-of-rule actions, but they are executed before the parser even recognizes the following components.
A mid-rule action may refer to the components preceding it using
$n, but it may not refer to subsequent components because
it is run before they are parsed.
The mid-rule action itself counts as one of the components of the rule.
This makes a difference when there is another action later in the same rule
(and usually there is another at the end): you have to count the actions
along with the symbols when working out which number n to use in
$n.
The mid-rule action can also have a semantic value. The action can set
its value with an assignment to $$, and actions later in the rule
can refer to the value using $n. Since there is no symbol
to name the action, there is no way to declare a data type for the value
in advance, so you must use the $<...>n construct to
specify a data type each time you refer to this value.
There is no way to set the value of the entire rule with a mid-rule
action, because assignments to $$ do not have that effect. The
only way to set the value for the entire rule is with an ordinary action
at the end of the rule.
Here is an example from a hypothetical compiler, handling a let
statement that looks like let (variable) statement and
serves to create a variable named variable temporarily for the
duration of statement. To parse this construct, we must put
variable into the symbol table while statement is parsed, then
remove it afterward. Here is how it is done:
stmt: LET '(' var ')'
{ $<context>$ = push_context ();
declare_variable ($3); }
stmt { $$ = $6;
pop_context ($<context>5); }
As soon as let (variable) has been recognized, the first
action is run. It saves a copy of the current semantic context (the
list of accessible variables) as its semantic value, using alternative
context in the data-type union. Then it calls
declare_variable to add the new variable to that list. Once the
first action is finished, the embedded statement stmt can be
parsed. Note that the mid-rule action is component number 5, so the
stmt is component number 6.
After the embedded statement is parsed, its semantic value becomes the
value of the entire let-statement. Then the semantic value