xref: /sqlite-3.40.0/doc/lemon.html (revision 9a243e69)
175897234Sdrh<html>
275897234Sdrh<head>
375897234Sdrh<title>The Lemon Parser Generator</title>
475897234Sdrh</head>
5*9a243e69Sdrh<body bgcolor='white'>
6*9a243e69Sdrh<h1 align='center'>The Lemon Parser Generator</h1>
775897234Sdrh
89bccde3dSdrh<p>Lemon is an LALR(1) parser generator for C.
99bccde3dSdrhIt does the same job as "bison" and "yacc".
10*9a243e69SdrhBut Lemon is not a bison or yacc clone.  Lemon
1175897234Sdrhuses a different grammar syntax which is designed to
129bccde3dSdrhreduce the number of coding errors.  Lemon also uses a
139bccde3dSdrhparsing engine that is faster than yacc and
149bccde3dSdrhbison and which is both reentrant and threadsafe.
159bccde3dSdrh(Update: Since the previous sentence was written, bison
169bccde3dSdrhhas also been updated so that it too can generate a
179bccde3dSdrhreentrant and threadsafe parser.)
189bccde3dSdrhLemon also implements features that can be used
19*9a243e69Sdrhto eliminate resource leaks, making it suitable for use
2075897234Sdrhin long-running programs such as graphical user interfaces
2175897234Sdrhor embedded controllers.</p>
2275897234Sdrh
2375897234Sdrh<p>This document is an introduction to the Lemon
2475897234Sdrhparser generator.</p>
2575897234Sdrh
26c5e56b34Sdrh<h2>Security Note</h2>
27c5e56b34Sdrh
28c5e56b34Sdrh<p>The language parser code created by Lemon is very robust and
29c5e56b34Sdrhis well-suited for use in internet-facing applications that need to
30c5e56b34Sdrhsafely process maliciously crafted inputs.
31c5e56b34Sdrh
32c5e56b34Sdrh<p>The "lemon.exe" command-line tool itself works great when given a valid
33c5e56b34Sdrhinput grammar file and almost always gives helpful
34c5e56b34Sdrherror messages for malformed inputs.  However,  it is possible for
35c5e56b34Sdrha malicious user to craft a grammar file that will cause
36c5e56b34Sdrhlemon.exe to crash.
37c5e56b34SdrhWe do not see this as a problem, as lemon.exe is not intended to be used
38c5e56b34Sdrhwith hostile inputs.
39c5e56b34SdrhTo summarize:</p>
40c5e56b34Sdrh
41c5e56b34Sdrh<ul>
42c5e56b34Sdrh<li>Parser code generated by lemon &rarr; Robust and secure
43c5e56b34Sdrh<li>The "lemon.exe" command line tool itself &rarr; Not so much
44c5e56b34Sdrh</ul>
45c5e56b34Sdrh
4675897234Sdrh<h2>Theory of Operation</h2>
4775897234Sdrh
4875897234Sdrh<p>The main goal of Lemon is to translate a context free grammar (CFG)
4975897234Sdrhfor a particular language into C code that implements a parser for
5075897234Sdrhthat language.
5175897234SdrhThe program has two inputs:
5275897234Sdrh<ul>
5375897234Sdrh<li>The grammar specification.
5475897234Sdrh<li>A parser template file.
5575897234Sdrh</ul>
5675897234SdrhTypically, only the grammar specification is supplied by the programmer.
5775897234SdrhLemon comes with a default parser template which works fine for most
5875897234Sdrhapplications.  But the user is free to substitute a different parser
5975897234Sdrhtemplate if desired.</p>
6075897234Sdrh
61*9a243e69Sdrh<p>Depending on command-line options, Lemon will generate up to
62*9a243e69Sdrhthree output files.
6375897234Sdrh<ul>
6475897234Sdrh<li>C code to implement the parser.
6575897234Sdrh<li>A header file defining an integer ID for each terminal symbol.
6675897234Sdrh<li>An information file that describes the states of the generated parser
6775897234Sdrh    automaton.
6875897234Sdrh</ul>
6975897234SdrhBy default, all three of these output files are generated.
709bccde3dSdrhThe header file is suppressed if the "-m" command-line option is
719bccde3dSdrhused and the report file is omitted when "-q" is selected.</p>
7275897234Sdrh
739bccde3dSdrh<p>The grammar specification file uses a ".y" suffix, by convention.
7475897234SdrhIn the examples used in this document, we'll assume the name of the
759bccde3dSdrhgrammar file is "gram.y".  A typical use of Lemon would be the
7675897234Sdrhfollowing command:
7775897234Sdrh<pre>
7875897234Sdrh   lemon gram.y
7975897234Sdrh</pre>
809bccde3dSdrhThis command will generate three output files named "gram.c",
819bccde3dSdrh"gram.h" and "gram.out".
8275897234SdrhThe first is C code to implement the parser.  The second
8375897234Sdrhis the header file that defines numerical values for all
8475897234Sdrhterminal symbols, and the last is the report that explains
8575897234Sdrhthe states used by the parser automaton.</p>
8675897234Sdrh
8775897234Sdrh<h3>Command Line Options</h3>
8875897234Sdrh
8975897234Sdrh<p>The behavior of Lemon can be modified using command-line options.
9075897234SdrhYou can obtain a list of the available command-line options together
9175897234Sdrhwith a brief explanation of what each does by typing
9275897234Sdrh<pre>
93*9a243e69Sdrh   lemon "-?"
9475897234Sdrh</pre>
9575897234SdrhAs of this writing, the following command-line options are supported:
9675897234Sdrh<ul>
979bccde3dSdrh<li><b>-b</b>
989bccde3dSdrhShow only the basis for each parser state in the report file.
999bccde3dSdrh<li><b>-c</b>
100*9a243e69SdrhDo not compress the generated action tables.  The parser will be a
101*9a243e69Sdrhlittle larger and slower, but it will detect syntax errors sooner.
1029bccde3dSdrh<li><b>-D<i>name</i></b>
103*9a243e69SdrhDefine C preprocessor macro <i>name</i>.  This macro is usable by
104*9a243e69Sdrh"<tt><a href='#pifdef'>%ifdef</a></tt>" and
105*9a243e69Sdrh"<tt><a href='#pifdef'>%ifndef</a></tt>" lines
106*9a243e69Sdrhin the grammar file.
1079bccde3dSdrh<li><b>-g</b>
1089bccde3dSdrhDo not generate a parser.  Instead write the input grammar to standard
1099bccde3dSdrhoutput with all comments, actions, and other extraneous text removed.
1109bccde3dSdrh<li><b>-l</b>
111dfe4e6bbSdrhOmit "#line" directives in the generated parser C code.
1129bccde3dSdrh<li><b>-m</b>
1139bccde3dSdrhCause the output C source code to be compatible with the "makeheaders"
1149bccde3dSdrhprogram.
1159bccde3dSdrh<li><b>-p</b>
1169bccde3dSdrhDisplay all conflicts that are resolved by
1179bccde3dSdrh<a href='#precrules'>precedence rules</a>.
1189bccde3dSdrh<li><b>-q</b>
1199bccde3dSdrhSuppress generation of the report file.
1209bccde3dSdrh<li><b>-r</b>
1219bccde3dSdrhDo not sort or renumber the parser states as part of optimization.
1229bccde3dSdrh<li><b>-s</b>
1239bccde3dSdrhShow parser statistics before existing.
1249bccde3dSdrh<li><b>-T<i>file</i></b>
1259bccde3dSdrhUse <i>file</i> as the template for the generated C-code parser implementation.
1269bccde3dSdrh<li><b>-x</b>
1279bccde3dSdrhPrint the Lemon version number.
12875897234Sdrh</ul>
12975897234Sdrh
13075897234Sdrh<h3>The Parser Interface</h3>
13175897234Sdrh
13275897234Sdrh<p>Lemon doesn't generate a complete, working program.  It only generates
13375897234Sdrha few subroutines that implement a parser.  This section describes
13475897234Sdrhthe interface to those subroutines.  It is up to the programmer to
13575897234Sdrhcall these subroutines in an appropriate way in order to produce a
13675897234Sdrhcomplete system.</p>
13775897234Sdrh
13875897234Sdrh<p>Before a program begins using a Lemon-generated parser, the program
13975897234Sdrhmust first create the parser.
14075897234SdrhA new parser is created as follows:
14175897234Sdrh<pre>
14275897234Sdrh   void *pParser = ParseAlloc( malloc );
14375897234Sdrh</pre>
14475897234SdrhThe ParseAlloc() routine allocates and initializes a new parser and
14575897234Sdrhreturns a pointer to it.
1469bccde3dSdrhThe actual data structure used to represent a parser is opaque &mdash;
14775897234Sdrhits internal structure is not visible or usable by the calling routine.
14875897234SdrhFor this reason, the ParseAlloc() routine returns a pointer to void
14975897234Sdrhrather than a pointer to some particular structure.
15075897234SdrhThe sole argument to the ParseAlloc() routine is a pointer to the
1519bccde3dSdrhsubroutine used to allocate memory.  Typically this means malloc().</p>
15275897234Sdrh
15375897234Sdrh<p>After a program is finished using a parser, it can reclaim all
15475897234Sdrhmemory allocated by that parser by calling
15575897234Sdrh<pre>
15675897234Sdrh   ParseFree(pParser, free);
15775897234Sdrh</pre>
15875897234SdrhThe first argument is the same pointer returned by ParseAlloc().  The
15975897234Sdrhsecond argument is a pointer to the function used to release bulk
16075897234Sdrhmemory back to the system.</p>
16175897234Sdrh
16275897234Sdrh<p>After a parser has been allocated using ParseAlloc(), the programmer
16375897234Sdrhmust supply the parser with a sequence of tokens (terminal symbols) to
16475897234Sdrhbe parsed.  This is accomplished by calling the following function
16575897234Sdrhonce for each token:
16675897234Sdrh<pre>
16775897234Sdrh   Parse(pParser, hTokenID, sTokenData, pArg);
16875897234Sdrh</pre>
16975897234SdrhThe first argument to the Parse() routine is the pointer returned by
17075897234SdrhParseAlloc().
171*9a243e69SdrhThe second argument is a small positive integer that tells the parser the
17275897234Sdrhtype of the next token in the data stream.
17375897234SdrhThere is one token type for each terminal symbol in the grammar.
17475897234SdrhThe gram.h file generated by Lemon contains #define statements that
17575897234Sdrhmap symbolic terminal symbol names into appropriate integer values.
1769bccde3dSdrhA value of 0 for the second argument is a special flag to the
1779bccde3dSdrhparser to indicate that the end of input has been reached.
17875897234SdrhThe third argument is the value of the given token.  By default,
179*9a243e69Sdrhthe type of the third argument is "void*", but the grammar will
18075897234Sdrhusually redefine this type to be some kind of structure.
18175897234SdrhTypically the second argument will be a broad category of tokens
1829bccde3dSdrhsuch as "identifier" or "number" and the third argument will
18375897234Sdrhbe the name of the identifier or the value of the number.</p>
18475897234Sdrh
18575897234Sdrh<p>The Parse() function may have either three or four arguments,
18645f31be8Sdrhdepending on the grammar.  If the grammar specification file requests
187*9a243e69Sdrhit (via the <tt><a href='#extraarg'>%extra_argument</a></tt> directive),
18845f31be8Sdrhthe Parse() function will have a fourth parameter that can be
18975897234Sdrhof any type chosen by the programmer.  The parser doesn't do anything
19075897234Sdrhwith this argument except to pass it through to action routines.
19175897234SdrhThis is a convenient mechanism for passing state information down
19275897234Sdrhto the action routines without having to use global variables.</p>
19375897234Sdrh
19475897234Sdrh<p>A typical use of a Lemon parser might look something like the
19575897234Sdrhfollowing:
19675897234Sdrh<pre>
197*9a243e69Sdrh    1 ParseTree *ParseFile(const char *zFilename){
198*9a243e69Sdrh    2    Tokenizer *pTokenizer;
199*9a243e69Sdrh    3    void *pParser;
200*9a243e69Sdrh    4    Token sToken;
201*9a243e69Sdrh    5    int hTokenId;
202*9a243e69Sdrh    6    ParserState sState;
203*9a243e69Sdrh    7
204*9a243e69Sdrh    8    pTokenizer = TokenizerCreate(zFilename);
205*9a243e69Sdrh    9    pParser = ParseAlloc( malloc );
206*9a243e69Sdrh   10    InitParserState(&amp;sState);
207*9a243e69Sdrh   11    while( GetNextToken(pTokenizer, &amp;hTokenId, &amp;sToken) ){
208*9a243e69Sdrh   12       Parse(pParser, hTokenId, sToken, &amp;sState);
20975897234Sdrh   13    }
210*9a243e69Sdrh   14    Parse(pParser, 0, sToken, &amp;sState);
21175897234Sdrh   15    ParseFree(pParser, free );
21275897234Sdrh   16    TokenizerFree(pTokenizer);
21375897234Sdrh   17    return sState.treeRoot;
21475897234Sdrh   18 }
21575897234Sdrh</pre>
21675897234SdrhThis example shows a user-written routine that parses a file of
21775897234Sdrhtext and returns a pointer to the parse tree.
2189bccde3dSdrh(All error-handling code is omitted from this example to keep it
21975897234Sdrhsimple.)
22075897234SdrhWe assume the existence of some kind of tokenizer which is created
22175897234Sdrhusing TokenizerCreate() on line 8 and deleted by TokenizerFree()
22275897234Sdrhon line 16.  The GetNextToken() function on line 11 retrieves the
22375897234Sdrhnext token from the input file and puts its type in the
22475897234Sdrhinteger variable hTokenId.  The sToken variable is assumed to be
22575897234Sdrhsome kind of structure that contains details about each token,
22675897234Sdrhsuch as its complete text, what line it occurs on, etc.</p>
22775897234Sdrh
22875897234Sdrh<p>This example also assumes the existence of structure of type
22975897234SdrhParserState that holds state information about a particular parse.
23075897234SdrhAn instance of such a structure is created on line 6 and initialized
23175897234Sdrhon line 10.  A pointer to this structure is passed into the Parse()
23275897234Sdrhroutine as the optional 4th argument.
23375897234SdrhThe action routine specified by the grammar for the parser can use
23475897234Sdrhthe ParserState structure to hold whatever information is useful and
23575897234Sdrhappropriate.  In the example, we note that the treeRoot field of
23675897234Sdrhthe ParserState structure is left pointing to the root of the parse
23775897234Sdrhtree.</p>
23875897234Sdrh
23975897234Sdrh<p>The core of this example as it relates to Lemon is as follows:
24075897234Sdrh<pre>
24175897234Sdrh   ParseFile(){
24275897234Sdrh      pParser = ParseAlloc( malloc );
243*9a243e69Sdrh      while( GetNextToken(pTokenizer,&amp;hTokenId, &amp;sToken) ){
24475897234Sdrh         Parse(pParser, hTokenId, sToken);
24575897234Sdrh      }
24675897234Sdrh      Parse(pParser, 0, sToken);
24775897234Sdrh      ParseFree(pParser, free );
24875897234Sdrh   }
24975897234Sdrh</pre>
25075897234SdrhBasically, what a program has to do to use a Lemon-generated parser
25175897234Sdrhis first create the parser, then send it lots of tokens obtained by
25275897234Sdrhtokenizing an input source.  When the end of input is reached, the
25375897234SdrhParse() routine should be called one last time with a token type
25475897234Sdrhof 0.  This step is necessary to inform the parser that the end of
25575897234Sdrhinput has been reached.  Finally, we reclaim memory used by the
25675897234Sdrhparser by calling ParseFree().</p>
25775897234Sdrh
25875897234Sdrh<p>There is one other interface routine that should be mentioned
25975897234Sdrhbefore we move on.
26075897234SdrhThe ParseTrace() function can be used to generate debugging output
26175897234Sdrhfrom the parser.  A prototype for this routine is as follows:
26275897234Sdrh<pre>
26375897234Sdrh   ParseTrace(FILE *stream, char *zPrefix);
26475897234Sdrh</pre>
26575897234SdrhAfter this routine is called, a short (one-line) message is written
26675897234Sdrhto the designated output stream every time the parser changes states
26775897234Sdrhor calls an action routine.  Each such message is prefaced using
26875897234Sdrhthe text given by zPrefix.  This debugging output can be turned off
26975897234Sdrhby calling ParseTrace() again with a first argument of NULL (0).</p>
27075897234Sdrh
27175897234Sdrh<h3>Differences With YACC and BISON</h3>
27275897234Sdrh
27375897234Sdrh<p>Programmers who have previously used the yacc or bison parser
27475897234Sdrhgenerator will notice several important differences between yacc and/or
27575897234Sdrhbison and Lemon.
27675897234Sdrh<ul>
27775897234Sdrh<li>In yacc and bison, the parser calls the tokenizer.  In Lemon,
27875897234Sdrh    the tokenizer calls the parser.
27975897234Sdrh<li>Lemon uses no global variables.  Yacc and bison use global variables
28075897234Sdrh    to pass information between the tokenizer and parser.
28175897234Sdrh<li>Lemon allows multiple parsers to be running simultaneously.  Yacc
28275897234Sdrh    and bison do not.
28375897234Sdrh</ul>
28475897234SdrhThese differences may cause some initial confusion for programmers
28575897234Sdrhwith prior yacc and bison experience.
28675897234SdrhBut after years of experience using Lemon, I firmly
28775897234Sdrhbelieve that the Lemon way of doing things is better.</p>
28875897234Sdrh
28945f31be8Sdrh<p><i>Updated as of 2016-02-16:</i>
29045f31be8SdrhThe text above was written in the 1990s.
29145f31be8SdrhWe are told that Bison has lately been enhanced to support the
29245f31be8Sdrhtokenizer-calls-parser paradigm used by Lemon, and to obviate the
29345f31be8Sdrhneed for global variables.</p>
29445f31be8Sdrh
29575897234Sdrh<h2>Input File Syntax</h2>
29675897234Sdrh
29775897234Sdrh<p>The main purpose of the grammar specification file for Lemon is
29875897234Sdrhto define the grammar for the parser.  But the input file also
29975897234Sdrhspecifies additional information Lemon requires to do its job.
30075897234SdrhMost of the work in using Lemon is in writing an appropriate
30175897234Sdrhgrammar file.</p>
30275897234Sdrh
303*9a243e69Sdrh<p>The grammar file for Lemon is, for the most part, free format.
30475897234SdrhIt does not have sections or divisions like yacc or bison.  Any
30575897234Sdrhdeclaration can occur at any point in the file.
30675897234SdrhLemon ignores whitespace (except where it is needed to separate
307*9a243e69Sdrhtokens), and it honors the same commenting conventions as C and C++.</p>
30875897234Sdrh
30975897234Sdrh<h3>Terminals and Nonterminals</h3>
31075897234Sdrh
31175897234Sdrh<p>A terminal symbol (token) is any string of alphanumeric
3129bccde3dSdrhand/or underscore characters
31375897234Sdrhthat begins with an uppercase letter.
314c8eee5e5SdrhA terminal can contain lowercase letters after the first character,
31575897234Sdrhbut the usual convention is to make terminals all uppercase.
31675897234SdrhA nonterminal, on the other hand, is any string of alphanumeric
31775897234Sdrhand underscore characters than begins with a lowercase letter.
318*9a243e69SdrhAgain, the usual convention is to make nonterminals use all lowercase
319*9a243e69Sdrhletters.</p>
32075897234Sdrh
32175897234Sdrh<p>In Lemon, terminal and nonterminal symbols do not need to
32275897234Sdrhbe declared or identified in a separate section of the grammar file.
32375897234SdrhLemon is able to generate a list of all terminals and nonterminals
32475897234Sdrhby examining the grammar rules, and it can always distinguish a
32575897234Sdrhterminal from a nonterminal by checking the case of the first
32675897234Sdrhcharacter of the name.</p>
32775897234Sdrh
32875897234Sdrh<p>Yacc and bison allow terminal symbols to have either alphanumeric
32975897234Sdrhnames or to be individual characters included in single quotes, like
33075897234Sdrhthis: ')' or '$'.  Lemon does not allow this alternative form for
33175897234Sdrhterminal symbols.  With Lemon, all symbols, terminals and nonterminals,
33275897234Sdrhmust have alphanumeric names.</p>
33375897234Sdrh
33475897234Sdrh<h3>Grammar Rules</h3>
33575897234Sdrh
33675897234Sdrh<p>The main component of a Lemon grammar file is a sequence of grammar
33775897234Sdrhrules.
33875897234SdrhEach grammar rule consists of a nonterminal symbol followed by
3399bccde3dSdrhthe special symbol "::=" and then a list of terminals and/or nonterminals.
34075897234SdrhThe rule is terminated by a period.
34175897234SdrhThe list of terminals and nonterminals on the right-hand side of the
34275897234Sdrhrule can be empty.
34375897234SdrhRules can occur in any order, except that the left-hand side of the
34475897234Sdrhfirst rule is assumed to be the start symbol for the grammar (unless
345*9a243e69Sdrhspecified otherwise using the <tt><a href='#start_symbol'>%start_symbol</a></tt>
346*9a243e69Sdrhdirective described below.)
34775897234SdrhA typical sequence of grammar rules might look something like this:
34875897234Sdrh<pre>
34975897234Sdrh  expr ::= expr PLUS expr.
35075897234Sdrh  expr ::= expr TIMES expr.
35175897234Sdrh  expr ::= LPAREN expr RPAREN.
35275897234Sdrh  expr ::= VALUE.
35375897234Sdrh</pre>
35475897234Sdrh</p>
35575897234Sdrh
3569bccde3dSdrh<p>There is one non-terminal in this example, "expr", and five
3579bccde3dSdrhterminal symbols or tokens: "PLUS", "TIMES", "LPAREN",
3589bccde3dSdrh"RPAREN" and "VALUE".</p>
35975897234Sdrh
36075897234Sdrh<p>Like yacc and bison, Lemon allows the grammar to specify a block
36175897234Sdrhof C code that will be executed whenever a grammar rule is reduced
36275897234Sdrhby the parser.
36375897234SdrhIn Lemon, this action is specified by putting the C code (contained
36475897234Sdrhwithin curly braces <tt>{...}</tt>) immediately after the
36575897234Sdrhperiod that closes the rule.
36675897234SdrhFor example:
36775897234Sdrh<pre>
36875897234Sdrh  expr ::= expr PLUS expr.   { printf("Doing an addition...\n"); }
36975897234Sdrh</pre>
37075897234Sdrh</p>
37175897234Sdrh
37275897234Sdrh<p>In order to be useful, grammar actions must normally be linked to
37375897234Sdrhtheir associated grammar rules.
3749bccde3dSdrhIn yacc and bison, this is accomplished by embedding a "$$" in the
37575897234Sdrhaction to stand for the value of the left-hand side of the rule and
3769bccde3dSdrhsymbols "$1", "$2", and so forth to stand for the value of
37775897234Sdrhthe terminal or nonterminal at position 1, 2 and so forth on the
37875897234Sdrhright-hand side of the rule.
37975897234SdrhThis idea is very powerful, but it is also very error-prone.  The
38075897234Sdrhsingle most common source of errors in a yacc or bison grammar is
38175897234Sdrhto miscount the number of symbols on the right-hand side of a grammar
3829bccde3dSdrhrule and say "$7" when you really mean "$8".</p>
38375897234Sdrh
38475897234Sdrh<p>Lemon avoids the need to count grammar symbols by assigning symbolic
38575897234Sdrhnames to each symbol in a grammar rule and then using those symbolic
38675897234Sdrhnames in the action.
38775897234SdrhIn yacc or bison, one would write this:
38875897234Sdrh<pre>
389*9a243e69Sdrh  expr -&gt; expr PLUS expr  { $$ = $1 + $3; };
39075897234Sdrh</pre>
39175897234SdrhBut in Lemon, the same rule becomes the following:
39275897234Sdrh<pre>
39375897234Sdrh  expr(A) ::= expr(B) PLUS expr(C).  { A = B+C; }
39475897234Sdrh</pre>
39575897234SdrhIn the Lemon rule, any symbol in parentheses after a grammar rule
39675897234Sdrhsymbol becomes a place holder for that symbol in the grammar rule.
39775897234SdrhThis place holder can then be used in the associated C action to
39875897234Sdrhstand for the value of that symbol.<p>
39975897234Sdrh
40075897234Sdrh<p>The Lemon notation for linking a grammar rule with its reduce
40175897234Sdrhaction is superior to yacc/bison on several counts.
40275897234SdrhFirst, as mentioned above, the Lemon method avoids the need to
40375897234Sdrhcount grammar symbols.
40475897234SdrhSecondly, if a terminal or nonterminal in a Lemon grammar rule
40575897234Sdrhincludes a linking symbol in parentheses but that linking symbol
40675897234Sdrhis not actually used in the reduce action, then an error message
40775897234Sdrhis generated.
40875897234SdrhFor example, the rule
40975897234Sdrh<pre>
41075897234Sdrh  expr(A) ::= expr(B) PLUS expr(C).  { A = B; }
41175897234Sdrh</pre>
4129bccde3dSdrhwill generate an error because the linking symbol "C" is used
41375897234Sdrhin the grammar rule but not in the reduce action.</p>
41475897234Sdrh
41575897234Sdrh<p>The Lemon notation for linking grammar rules to reduce actions
41675897234Sdrhalso facilitates the use of destructors for reclaiming memory
41775897234Sdrhallocated by the values of terminals and nonterminals on the
41875897234Sdrhright-hand side of a rule.</p>
41975897234Sdrh
4209bccde3dSdrh<a name='precrules'></a>
42175897234Sdrh<h3>Precedence Rules</h3>
42275897234Sdrh
42375897234Sdrh<p>Lemon resolves parsing ambiguities in exactly the same way as
42475897234Sdrhyacc and bison.  A shift-reduce conflict is resolved in favor
42575897234Sdrhof the shift, and a reduce-reduce conflict is resolved by reducing
42675897234Sdrhwhichever rule comes first in the grammar file.</p>
42775897234Sdrh
42875897234Sdrh<p>Just like in
42975897234Sdrhyacc and bison, Lemon allows a measure of control
430*9a243e69Sdrhover the resolution of parsing conflicts using precedence rules.
43175897234SdrhA precedence value can be assigned to any terminal symbol
4329bccde3dSdrhusing the
433*9a243e69Sdrh<tt><a href='#pleft'>%left</a></tt>,
434*9a243e69Sdrh<tt><a href='#pright'>%right</a></tt> or
435*9a243e69Sdrh<tt><a href='#pnonassoc'>%nonassoc</a></tt> directives.  Terminal symbols
436*9a243e69Sdrhmentioned in earlier directives have a lower precedence than
43775897234Sdrhterminal symbols mentioned in later directives.  For example:</p>
43875897234Sdrh
43975897234Sdrh<p><pre>
44075897234Sdrh   %left AND.
44175897234Sdrh   %left OR.
44275897234Sdrh   %nonassoc EQ NE GT GE LT LE.
44375897234Sdrh   %left PLUS MINUS.
44475897234Sdrh   %left TIMES DIVIDE MOD.
44575897234Sdrh   %right EXP NOT.
44675897234Sdrh</pre></p>
44775897234Sdrh
44875897234Sdrh<p>In the preceding sequence of directives, the AND operator is
44975897234Sdrhdefined to have the lowest precedence.  The OR operator is one
45075897234Sdrhprecedence level higher.  And so forth.  Hence, the grammar would
45175897234Sdrhattempt to group the ambiguous expression
45275897234Sdrh<pre>
45375897234Sdrh     a AND b OR c
45475897234Sdrh</pre>
45575897234Sdrhlike this
45675897234Sdrh<pre>
45775897234Sdrh     a AND (b OR c).
45875897234Sdrh</pre>
45975897234SdrhThe associativity (left, right or nonassoc) is used to determine
46075897234Sdrhthe grouping when the precedence is the same.  AND is left-associative
46175897234Sdrhin our example, so
46275897234Sdrh<pre>
46375897234Sdrh     a AND b AND c
46475897234Sdrh</pre>
46575897234Sdrhis parsed like this
46675897234Sdrh<pre>
46775897234Sdrh     (a AND b) AND c.
46875897234Sdrh</pre>
46975897234SdrhThe EXP operator is right-associative, though, so
47075897234Sdrh<pre>
47175897234Sdrh     a EXP b EXP c
47275897234Sdrh</pre>
47375897234Sdrhis parsed like this
47475897234Sdrh<pre>
47575897234Sdrh     a EXP (b EXP c).
47675897234Sdrh</pre>
47775897234SdrhThe nonassoc precedence is used for non-associative operators.
47875897234SdrhSo
47975897234Sdrh<pre>
48075897234Sdrh     a EQ b EQ c
48175897234Sdrh</pre>
48275897234Sdrhis an error.</p>
48375897234Sdrh
48475897234Sdrh<p>The precedence of non-terminals is transferred to rules as follows:
48575897234SdrhThe precedence of a grammar rule is equal to the precedence of the
48675897234Sdrhleft-most terminal symbol in the rule for which a precedence is
48775897234Sdrhdefined.  This is normally what you want, but in those cases where
48875897234Sdrhyou want to precedence of a grammar rule to be something different,
48975897234Sdrhyou can specify an alternative precedence symbol by putting the
49075897234Sdrhsymbol in square braces after the period at the end of the rule and
49175897234Sdrhbefore any C-code.  For example:</p>
49275897234Sdrh
49375897234Sdrh<p><pre>
49475897234Sdrh   expr = MINUS expr.  [NOT]
49575897234Sdrh</pre></p>
49675897234Sdrh
49775897234Sdrh<p>This rule has a precedence equal to that of the NOT symbol, not the
49875897234SdrhMINUS symbol as would have been the case by default.</p>
49975897234Sdrh
50075897234Sdrh<p>With the knowledge of how precedence is assigned to terminal
50175897234Sdrhsymbols and individual
50275897234Sdrhgrammar rules, we can now explain precisely how parsing conflicts
50375897234Sdrhare resolved in Lemon.  Shift-reduce conflicts are resolved
50475897234Sdrhas follows:
50575897234Sdrh<ul>
50675897234Sdrh<li> If either the token to be shifted or the rule to be reduced
50775897234Sdrh     lacks precedence information, then resolve in favor of the
50875897234Sdrh     shift, but report a parsing conflict.
50975897234Sdrh<li> If the precedence of the token to be shifted is greater than
51075897234Sdrh     the precedence of the rule to reduce, then resolve in favor
51175897234Sdrh     of the shift.  No parsing conflict is reported.
512*9a243e69Sdrh<li> If the precedence of the token to be shifted is less than the
51375897234Sdrh     precedence of the rule to reduce, then resolve in favor of the
51475897234Sdrh     reduce action.  No parsing conflict is reported.
51575897234Sdrh<li> If the precedences are the same and the shift token is
51675897234Sdrh     right-associative, then resolve in favor of the shift.
51775897234Sdrh     No parsing conflict is reported.
518*9a243e69Sdrh<li> If the precedences are the same and the shift token is
51975897234Sdrh     left-associative, then resolve in favor of the reduce.
52075897234Sdrh     No parsing conflict is reported.
521*9a243e69Sdrh<li> Otherwise, resolve the conflict by doing the shift, and
522*9a243e69Sdrh     report a parsing conflict.
52375897234Sdrh</ul>
52475897234SdrhReduce-reduce conflicts are resolved this way:
52575897234Sdrh<ul>
52675897234Sdrh<li> If either reduce rule
52775897234Sdrh     lacks precedence information, then resolve in favor of the
528*9a243e69Sdrh     rule that appears first in the grammar, and report a parsing
52975897234Sdrh     conflict.
530*9a243e69Sdrh<li> If both rules have precedence and the precedence is different,
53175897234Sdrh     then resolve the dispute in favor of the rule with the highest
532*9a243e69Sdrh     precedence, and do not report a conflict.
53375897234Sdrh<li> Otherwise, resolve the conflict by reducing by the rule that
534*9a243e69Sdrh     appears first in the grammar, and report a parsing conflict.
53575897234Sdrh</ul>
53675897234Sdrh
53775897234Sdrh<h3>Special Directives</h3>
53875897234Sdrh
53975897234Sdrh<p>The input grammar to Lemon consists of grammar rules and special
54075897234Sdrhdirectives.  We've described all the grammar rules, so now we'll
54175897234Sdrhtalk about the special directives.</p>
54275897234Sdrh
543*9a243e69Sdrh<p>Directives in Lemon can occur in any order.  You can put them before
544*9a243e69Sdrhthe grammar rules, or after the grammar rules, or in the midst of the
54575897234Sdrhgrammar rules.  It doesn't matter.  The relative order of
54675897234Sdrhdirectives used to assign precedence to terminals is important, but
54775897234Sdrhother than that, the order of directives in Lemon is arbitrary.</p>
54875897234Sdrh
54975897234Sdrh<p>Lemon supports the following special directives:
55075897234Sdrh<ul>
551*9a243e69Sdrh<li><tt><a href='#pcode'>%code</a></tt>
552*9a243e69Sdrh<li><tt><a href='#default_destructor'>%default_destructor</a></tt>
553*9a243e69Sdrh<li><tt><a href='#default_type'>%default_type</a></tt>
554*9a243e69Sdrh<li><tt><a href='#destructor'>%destructor</a></tt>
555*9a243e69Sdrh<li><tt><a href='#pifdef'>%endif</a></tt>
556*9a243e69Sdrh<li><tt><a href='#extraarg'>%extra_argument</a></tt>
557*9a243e69Sdrh<li><tt><a href='#pfallback'>%fallback</a></tt>
558*9a243e69Sdrh<li><tt><a href='#pifdef'>%ifdef</a></tt>
559*9a243e69Sdrh<li><tt><a href='#pifdef'>%ifndef</a></tt>
560*9a243e69Sdrh<li><tt><a href='#pinclude'>%include</a></tt>
561*9a243e69Sdrh<li><tt><a href='#pleft'>%left</a></tt>
562*9a243e69Sdrh<li><tt><a href='#pname'>%name</a></tt>
563*9a243e69Sdrh<li><tt><a href='#pnonassoc'>%nonassoc</a></tt>
564*9a243e69Sdrh<li><tt><a href='#parse_accept'>%parse_accept</a></tt>
565*9a243e69Sdrh<li><tt><a href='#parse_failure'>%parse_failure</a></tt>
566*9a243e69Sdrh<li><tt><a href='#pright'>%right</a></tt>
567*9a243e69Sdrh<li><tt><a href='#stack_overflow'>%stack_overflow</a></tt>
568*9a243e69Sdrh<li><tt><a href='#stack_size'>%stack_size</a></tt>
569*9a243e69Sdrh<li><tt><a href='#start_symbol'>%start_symbol</a></tt>
570*9a243e69Sdrh<li><tt><a href='#syntax_error'>%syntax_error</a></tt>
571*9a243e69Sdrh<li><tt><a href='#token_class'>%token_class</a></tt>
572*9a243e69Sdrh<li><tt><a href='#token_destructor'>%token_destructor</a></tt>
573*9a243e69Sdrh<li><tt><a href='#token_prefix'>%token_prefix</a></tt>
574*9a243e69Sdrh<li><tt><a href='#token_type'>%token_type</a></tt>
575*9a243e69Sdrh<li><tt><a href='#ptype'>%type</a></tt>
576*9a243e69Sdrh<li><tt><a href='#pwildcard'>%wildcard</a></tt>
57775897234Sdrh</ul>
57875897234SdrhEach of these directives will be described separately in the
57975897234Sdrhfollowing sections:</p>
58075897234Sdrh
5819bccde3dSdrh<a name='pcode'></a>
582f2340fc7Sdrh<h4>The <tt>%code</tt> directive</h4>
583f2340fc7Sdrh
584*9a243e69Sdrh<p>The <tt>%code</tt> directive is used to specify additional C code that
585f2340fc7Sdrhis added to the end of the main output file.  This is similar to
586*9a243e69Sdrhthe <tt><a href='#pinclude'>%include</a></tt> directive except that
587*9a243e69Sdrh<tt>%include</tt> is inserted at the beginning of the main output file.</p>
588f2340fc7Sdrh
589*9a243e69Sdrh<p><tt>%code</tt> is typically used to include some action routines or perhaps
5909bccde3dSdrha tokenizer or even the "main()" function
5919bccde3dSdrhas part of the output file.</p>
592f2340fc7Sdrh
5939bccde3dSdrh<a name='default_destructor'></a>
594f2340fc7Sdrh<h4>The <tt>%default_destructor</tt> directive</h4>
595f2340fc7Sdrh
596*9a243e69Sdrh<p>The <tt>%default_destructor</tt> directive specifies a destructor to
597f2340fc7Sdrhuse for non-terminals that do not have their own destructor
598*9a243e69Sdrhspecified by a separate <tt>%destructor</tt> directive.  See the documentation
599*9a243e69Sdrhon the <tt><a name='#destructor'>%destructor</a></tt> directive below for
6009bccde3dSdrhadditional information.</p>
601f2340fc7Sdrh
602*9a243e69Sdrh<p>In some grammars, many different non-terminal symbols have the
603f2340fc7Sdrhsame data type and hence the same destructor.  This directive is
604*9a243e69Sdrha convenient way to specify the same destructor for all those
605f2340fc7Sdrhnon-terminals using a single statement.</p>
606f2340fc7Sdrh
6079bccde3dSdrh<a name='default_type'></a>
608f2340fc7Sdrh<h4>The <tt>%default_type</tt> directive</h4>
609f2340fc7Sdrh
610*9a243e69Sdrh<p>The <tt>%default_type</tt> directive specifies the data type of non-terminal
611*9a243e69Sdrhsymbols that do not have their own data type defined using a separate
612*9a243e69Sdrh<tt><a href='#ptype'>%type</a></tt> directive.</p>
613f2340fc7Sdrh
6149bccde3dSdrh<a name='destructor'></a>
61575897234Sdrh<h4>The <tt>%destructor</tt> directive</h4>
61675897234Sdrh
617*9a243e69Sdrh<p>The <tt>%destructor</tt> directive is used to specify a destructor for
61875897234Sdrha non-terminal symbol.
619*9a243e69Sdrh(See also the <tt><a href='#token_destructor'>%token_destructor</a></tt>
6209bccde3dSdrhdirective which is used to specify a destructor for terminal symbols.)</p>
62175897234Sdrh
62275897234Sdrh<p>A non-terminal's destructor is called to dispose of the
62375897234Sdrhnon-terminal's value whenever the non-terminal is popped from
62475897234Sdrhthe stack.  This includes all of the following circumstances:
62575897234Sdrh<ul>
62675897234Sdrh<li> When a rule reduces and the value of a non-terminal on
62775897234Sdrh     the right-hand side is not linked to C code.
62875897234Sdrh<li> When the stack is popped during error processing.
62975897234Sdrh<li> When the ParseFree() function runs.
63075897234Sdrh</ul>
63175897234SdrhThe destructor can do whatever it wants with the value of
63275897234Sdrhthe non-terminal, but its design is to deallocate memory
63375897234Sdrhor other resources held by that non-terminal.</p>
63475897234Sdrh
63575897234Sdrh<p>Consider an example:
63675897234Sdrh<pre>
63775897234Sdrh   %type nt {void*}
63875897234Sdrh   %destructor nt { free($$); }
63975897234Sdrh   nt(A) ::= ID NUM.   { A = malloc( 100 ); }
64075897234Sdrh</pre>
641*9a243e69SdrhThis example is a bit contrived, but it serves to illustrate how
64275897234Sdrhdestructors work.  The example shows a non-terminal named
6439bccde3dSdrh"nt" that holds values of type "void*".  When the rule for
6449bccde3dSdrhan "nt" reduces, it sets the value of the non-terminal to
64575897234Sdrhspace obtained from malloc().  Later, when the nt non-terminal
64675897234Sdrhis popped from the stack, the destructor will fire and call
64775897234Sdrhfree() on this malloced space, thus avoiding a memory leak.
6489bccde3dSdrh(Note that the symbol "$$" in the destructor code is replaced
64975897234Sdrhby the value of the non-terminal.)</p>
65075897234Sdrh
65175897234Sdrh<p>It is important to note that the value of a non-terminal is passed
65275897234Sdrhto the destructor whenever the non-terminal is removed from the
65375897234Sdrhstack, unless the non-terminal is used in a C-code action.  If
65475897234Sdrhthe non-terminal is used by C-code, then it is assumed that the
6559bccde3dSdrhC-code will take care of destroying it.
6569bccde3dSdrhMore commonly, the value is used to build some
657*9a243e69Sdrhlarger structure, and we don't want to destroy it, which is why
65875897234Sdrhthe destructor is not called in this circumstance.</p>
65975897234Sdrh
6609bccde3dSdrh<p>Destructors help avoid memory leaks by automatically freeing
6619bccde3dSdrhallocated objects when they go out of scope.
66275897234SdrhTo do the same using yacc or bison is much more difficult.</p>
66375897234Sdrh
664*9a243e69Sdrh<a name='extraarg'></a>
66575897234Sdrh<h4>The <tt>%extra_argument</tt> directive</h4>
66675897234Sdrh
667*9a243e69SdrhThe <tt>%extra_argument</tt> directive instructs Lemon to add a 4th parameter
66875897234Sdrhto the parameter list of the Parse() function it generates.  Lemon
66975897234Sdrhdoesn't do anything itself with this extra argument, but it does
67075897234Sdrhmake the argument available to C-code action routines, destructors,
67175897234Sdrhand so forth.  For example, if the grammar file contains:</p>
67275897234Sdrh
67375897234Sdrh<p><pre>
67475897234Sdrh    %extra_argument { MyStruct *pAbc }
67575897234Sdrh</pre></p>
67675897234Sdrh
67775897234Sdrh<p>Then the Parse() function generated will have an 4th parameter
6789bccde3dSdrhof type "MyStruct*" and all action routines will have access to
6799bccde3dSdrha variable named "pAbc" that is the value of the 4th parameter
68075897234Sdrhin the most recent call to Parse().</p>
68175897234Sdrh
6829bccde3dSdrh<a name='pfallback'></a>
6839bccde3dSdrh<h4>The <tt>%fallback</tt> directive</h4>
6849bccde3dSdrh
685*9a243e69Sdrh<p>The <tt>%fallback</tt> directive specifies an alternative meaning for one
6869bccde3dSdrhor more tokens.  The alternative meaning is tried if the original token
687*9a243e69Sdrhwould have generated a syntax error.</p>
6889bccde3dSdrh
689*9a243e69Sdrh<p>The <tt>%fallback</tt> directive was added to support robust parsing of SQL
690*9a243e69Sdrhsyntax in <a href='https://www.sqlite.org/'>SQLite</a>.
6919bccde3dSdrhThe SQL language contains a large assortment of keywords, each of which
6929bccde3dSdrhappears as a different token to the language parser.  SQL contains so
693*9a243e69Sdrhmany keywords that it can be difficult for programmers to keep up with
6949bccde3dSdrhthem all.  Programmers will, therefore, sometimes mistakenly use an
695*9a243e69Sdrhobscure language keyword for an identifier.  The <tt>%fallback</tt> directive
6969bccde3dSdrhprovides a mechanism to tell the parser:  "If you are unable to parse
697*9a243e69Sdrhthis keyword, try treating it as an identifier instead."</p>
6989bccde3dSdrh
699*9a243e69Sdrh<p>The syntax of <tt>%fallback</tt> is as follows:
7009bccde3dSdrh
7019bccde3dSdrh<blockquote>
7029bccde3dSdrh<tt>%fallback</tt> <i>ID</i> <i>TOKEN...</i> <b>.</b>
703*9a243e69Sdrh</blockquote></p>
7049bccde3dSdrh
705*9a243e69Sdrh<p>In words, the <tt>%fallback</tt> directive is followed by a list of token
706*9a243e69Sdrhnames terminated by a period.
707*9a243e69SdrhThe first token name is the fallback token &mdash; the
7089bccde3dSdrhtoken to which all the other tokens fall back to.  The second and subsequent
7099bccde3dSdrharguments are tokens which fall back to the token identified by the first
710*9a243e69Sdrhargument.</p>
7119bccde3dSdrh
7129bccde3dSdrh<a name='pifdef'></a>
713*9a243e69Sdrh<h4>The <tt>%ifdef</tt>, <tt>%ifndef</tt>, and <tt>%endif</tt> directives</h4>
7149bccde3dSdrh
715*9a243e69Sdrh<p>The <tt>%ifdef</tt>, <tt>%ifndef</tt>, and <tt>%endif</tt> directives
716*9a243e69Sdrhare similar to #ifdef, #ifndef, and #endif in the C-preprocessor,
717*9a243e69Sdrhjust not as general.
7189bccde3dSdrhEach of these directives must begin at the left margin.  No whitespace
719*9a243e69Sdrhis allowed between the "%" and the directive name.</p>
7209bccde3dSdrh
721*9a243e69Sdrh<p>Grammar text in between "<tt>%ifdef MACRO</tt>" and the next nested
722*9a243e69Sdrh"<tt>%endif</tt>" is
7239bccde3dSdrhignored unless the "-DMACRO" command-line option is used.  Grammar text
724*9a243e69Sdrhbetwen "<tt>%ifndef MACRO</tt>" and the next nested "<tt>%endif</tt>" is
725*9a243e69Sdrhincluded except when the "-DMACRO" command-line option is used.</p>
7269bccde3dSdrh
727*9a243e69Sdrh<p>Note that the argument to <tt>%ifdef</tt> and <tt>%ifndef</tt> must
728*9a243e69Sdrhbe a single preprocessor symbol name, not a general expression.
729*9a243e69SdrhThere is no "<tt>%else</tt>" directive.</p>
7309bccde3dSdrh
7319bccde3dSdrh
7329bccde3dSdrh<a name='pinclude'></a>
73375897234Sdrh<h4>The <tt>%include</tt> directive</h4>
73475897234Sdrh
735*9a243e69Sdrh<p>The <tt>%include</tt> directive specifies C code that is included at the
736*9a243e69Sdrhtop of the generated parser.  You can include any text you want &mdash;
737f2340fc7Sdrhthe Lemon parser generator copies it blindly.  If you have multiple
738*9a243e69Sdrh<tt>%include</tt> directives in your grammar file, their values are concatenated
739*9a243e69Sdrhso that all <tt>%include</tt> code ultimately appears near the top of the
740*9a243e69Sdrhgenerated parser, in the same order as it appeared in the grammar.</p>
74175897234Sdrh
742*9a243e69Sdrh<p>The <tt>%include</tt> directive is very handy for getting some extra #include
74375897234Sdrhpreprocessor statements at the beginning of the generated parser.
74475897234SdrhFor example:</p>
74575897234Sdrh
74675897234Sdrh<p><pre>
74775897234Sdrh   %include {#include &lt;unistd.h&gt;}
74875897234Sdrh</pre></p>
74975897234Sdrh
75075897234Sdrh<p>This might be needed, for example, if some of the C actions in the
751*9a243e69Sdrhgrammar call functions that are prototyped in unistd.h.</p>
75275897234Sdrh
7539bccde3dSdrh<a name='pleft'></a>
75475897234Sdrh<h4>The <tt>%left</tt> directive</h4>
75575897234Sdrh
756*9a243e69SdrhThe <tt>%left</tt> directive is used (along with the
757*9a243e69Sdrh<tt><a href='#pright'>%right</a></tt> and
758*9a243e69Sdrh<tt><a href='#pnonassoc'>%nonassoc</a></tt> directives) to declare
759*9a243e69Sdrhprecedences of terminal symbols.
760*9a243e69SdrhEvery terminal symbol whose name appears after
761*9a243e69Sdrha <tt>%left</tt> directive but before the next period (".") is
76275897234Sdrhgiven the same left-associative precedence value.  Subsequent
763*9a243e69Sdrh<tt>%left</tt> directives have higher precedence.  For example:</p>
76475897234Sdrh
76575897234Sdrh<p><pre>
76675897234Sdrh   %left AND.
76775897234Sdrh   %left OR.
76875897234Sdrh   %nonassoc EQ NE GT GE LT LE.
76975897234Sdrh   %left PLUS MINUS.
77075897234Sdrh   %left TIMES DIVIDE MOD.
77175897234Sdrh   %right EXP NOT.
77275897234Sdrh</pre></p>
77375897234Sdrh
774*9a243e69Sdrh<p>Note the period that terminates each <tt>%left</tt>,
775*9a243e69Sdrh<tt>%right</tt> or <tt>%nonassoc</tt>
77675897234Sdrhdirective.</p>
77775897234Sdrh
77875897234Sdrh<p>LALR(1) grammars can get into a situation where they require
77975897234Sdrha large amount of stack space if you make heavy use or right-associative
780*9a243e69Sdrhoperators.  For this reason, it is recommended that you use <tt>%left</tt>
781*9a243e69Sdrhrather than <tt>%right</tt> whenever possible.</p>
78275897234Sdrh
7839bccde3dSdrh<a name='pname'></a>
78475897234Sdrh<h4>The <tt>%name</tt> directive</h4>
78575897234Sdrh
78675897234Sdrh<p>By default, the functions generated by Lemon all begin with the
7879bccde3dSdrhfive-character string "Parse".  You can change this string to something
788*9a243e69Sdrhdifferent using the <tt>%name</tt> directive.  For instance:</p>
78975897234Sdrh
79075897234Sdrh<p><pre>
79175897234Sdrh   %name Abcde
79275897234Sdrh</pre></p>
79375897234Sdrh
79475897234Sdrh<p>Putting this directive in the grammar file will cause Lemon to generate
79575897234Sdrhfunctions named
79675897234Sdrh<ul>
79775897234Sdrh<li> AbcdeAlloc(),
79875897234Sdrh<li> AbcdeFree(),
79975897234Sdrh<li> AbcdeTrace(), and
80075897234Sdrh<li> Abcde().
80175897234Sdrh</ul>
802*9a243e69SdrhThe <tt>%name</tt> directive allows you to generate two or more different
803*9a243e69Sdrhparsers and link them all into the same executable.</p>
80475897234Sdrh
8059bccde3dSdrh<a name='pnonassoc'></a>
80675897234Sdrh<h4>The <tt>%nonassoc</tt> directive</h4>
80775897234Sdrh
80875897234Sdrh<p>This directive is used to assign non-associative precedence to
8099bccde3dSdrhone or more terminal symbols.  See the section on
8109bccde3dSdrh<a href='#precrules'>precedence rules</a>
811*9a243e69Sdrhor on the <tt><a href='#pleft'>%left</a></tt> directive
812*9a243e69Sdrhfor additional information.</p>
81375897234Sdrh
8149bccde3dSdrh<a name='parse_accept'></a>
81575897234Sdrh<h4>The <tt>%parse_accept</tt> directive</h4>
81675897234Sdrh
817*9a243e69Sdrh<p>The <tt>%parse_accept</tt> directive specifies a block of C code that is
8189bccde3dSdrhexecuted whenever the parser accepts its input string.  To "accept"
81975897234Sdrhan input string means that the parser was able to process all tokens
82075897234Sdrhwithout error.</p>
82175897234Sdrh
82275897234Sdrh<p>For example:</p>
82375897234Sdrh
82475897234Sdrh<p><pre>
82575897234Sdrh   %parse_accept {
82675897234Sdrh      printf("parsing complete!\n");
82775897234Sdrh   }
82875897234Sdrh</pre></p>
82975897234Sdrh
8309bccde3dSdrh<a name='parse_failure'></a>
83175897234Sdrh<h4>The <tt>%parse_failure</tt> directive</h4>
83275897234Sdrh
833*9a243e69Sdrh<p>The <tt>%parse_failure</tt> directive specifies a block of C code that
83475897234Sdrhis executed whenever the parser fails complete.  This code is not
83575897234Sdrhexecuted until the parser has tried and failed to resolve an input
83675897234Sdrherror using is usual error recovery strategy.  The routine is
83775897234Sdrhonly invoked when parsing is unable to continue.</p>
83875897234Sdrh
83975897234Sdrh<p><pre>
84075897234Sdrh   %parse_failure {
84175897234Sdrh     fprintf(stderr,"Giving up.  Parser is hopelessly lost...\n");
84275897234Sdrh   }
84375897234Sdrh</pre></p>
84475897234Sdrh
8459bccde3dSdrh<a name='pright'></a>
84675897234Sdrh<h4>The <tt>%right</tt> directive</h4>
84775897234Sdrh
84875897234Sdrh<p>This directive is used to assign right-associative precedence to
8499bccde3dSdrhone or more terminal symbols.  See the section on
8509bccde3dSdrh<a href='#precrules'>precedence rules</a>
8519bccde3dSdrhor on the <a href='#pleft'>%left</a> directive for additional information.</p>
85275897234Sdrh
8539bccde3dSdrh<a name='stack_overflow'></a>
85475897234Sdrh<h4>The <tt>%stack_overflow</tt> directive</h4>
85575897234Sdrh
856*9a243e69Sdrh<p>The <tt>%stack_overflow</tt> directive specifies a block of C code that
85775897234Sdrhis executed if the parser's internal stack ever overflows.  Typically
85875897234Sdrhthis just prints an error message.  After a stack overflow, the parser
85975897234Sdrhwill be unable to continue and must be reset.</p>
86075897234Sdrh
86175897234Sdrh<p><pre>
86275897234Sdrh   %stack_overflow {
86375897234Sdrh     fprintf(stderr,"Giving up.  Parser stack overflow\n");
86475897234Sdrh   }
86575897234Sdrh</pre></p>
86675897234Sdrh
86775897234Sdrh<p>You can help prevent parser stack overflows by avoiding the use
86875897234Sdrhof right recursion and right-precedence operators in your grammar.
869*9a243e69SdrhUse left recursion and and left-precedence operators instead to
87075897234Sdrhencourage rules to reduce sooner and keep the stack size down.
87175897234SdrhFor example, do rules like this:
87275897234Sdrh<pre>
87375897234Sdrh   list ::= list element.      // left-recursion.  Good!
87475897234Sdrh   list ::= .
87575897234Sdrh</pre>
87675897234SdrhNot like this:
87775897234Sdrh<pre>
87875897234Sdrh   list ::= element list.      // right-recursion.  Bad!
87975897234Sdrh   list ::= .
880*9a243e69Sdrh</pre></p>
88175897234Sdrh
8829bccde3dSdrh<a name='stack_size'></a>
88375897234Sdrh<h4>The <tt>%stack_size</tt> directive</h4>
88475897234Sdrh
88575897234Sdrh<p>If stack overflow is a problem and you can't resolve the trouble
88675897234Sdrhby using left-recursion, then you might want to increase the size
88775897234Sdrhof the parser's stack using this directive.  Put an positive integer
888*9a243e69Sdrhafter the <tt>%stack_size</tt> directive and Lemon will generate a parse
88975897234Sdrhwith a stack of the requested size.  The default value is 100.</p>
89075897234Sdrh
89175897234Sdrh<p><pre>
89275897234Sdrh   %stack_size 2000
89375897234Sdrh</pre></p>
89475897234Sdrh
8959bccde3dSdrh<a name='start_symbol'></a>
89675897234Sdrh<h4>The <tt>%start_symbol</tt> directive</h4>
89775897234Sdrh
898*9a243e69Sdrh<p>By default, the start symbol for the grammar that Lemon generates
89975897234Sdrhis the first non-terminal that appears in the grammar file.  But you
900*9a243e69Sdrhcan choose a different start symbol using the
901*9a243e69Sdrh<tt>%start_symbol</tt> directive.</p>
90275897234Sdrh
90375897234Sdrh<p><pre>
90475897234Sdrh   %start_symbol  prog
90575897234Sdrh</pre></p>
90675897234Sdrh
907*9a243e69Sdrh<a name='syntax_error'></a>
908*9a243e69Sdrh<h4>The <tt>%syntax_error</tt> directive</h4>
909*9a243e69Sdrh
910*9a243e69Sdrh<p>See <a href='#error_processing'>Error Processing</a>.</p>
911*9a243e69Sdrh
912*9a243e69Sdrh<a name='token_class'></a>
913*9a243e69Sdrh<h4>The <tt>%token_class</tt> directive</h4>
914*9a243e69Sdrh
915*9a243e69Sdrh<p>Undocumented.  Appears to be related to the MULTITERMINAL concept.
916*9a243e69Sdrh<a href='http://sqlite.org/src/fdiff?v1=796930d5fc2036c7&v2=624b24c5dc048e09&sbs=0'>Implementation</a>.</p>
917*9a243e69Sdrh
9189bccde3dSdrh<a name='token_destructor'></a>
91975897234Sdrh<h4>The <tt>%token_destructor</tt> directive</h4>
92075897234Sdrh
921*9a243e69Sdrh<p>The <tt>%destructor</tt> directive assigns a destructor to a non-terminal
922*9a243e69Sdrhsymbol.  (See the description of the
923*9a243e69Sdrh<tt><a href='%destructor'>%destructor</a></tt> directive above.)
924*9a243e69SdrhThe <tt>%token_destructor</tt> directive does the same thing
925*9a243e69Sdrhfor all terminal symbols.</p>
92675897234Sdrh
92775897234Sdrh<p>Unlike non-terminal symbols which may each have a different data type
92875897234Sdrhfor their values, terminals all use the same data type (defined by
929*9a243e69Sdrhthe <tt><a href='#token_type'>%token_type</a></tt> directive)
930*9a243e69Sdrhand so they use a common destructor.
931*9a243e69SdrhOther than that, the token destructor works just like the non-terminal
93275897234Sdrhdestructors.</p>
93375897234Sdrh
9349bccde3dSdrh<a name='token_prefix'></a>
93575897234Sdrh<h4>The <tt>%token_prefix</tt> directive</h4>
93675897234Sdrh
93775897234Sdrh<p>Lemon generates #defines that assign small integer constants
93875897234Sdrhto each terminal symbol in the grammar.  If desired, Lemon will
93975897234Sdrhadd a prefix specified by this directive
940*9a243e69Sdrhto each of the #defines it generates.</p>
941*9a243e69Sdrh
942*9a243e69Sdrh<p>So if the default output of Lemon looked like this:
94375897234Sdrh<pre>
94475897234Sdrh    #define AND              1
94575897234Sdrh    #define MINUS            2
94675897234Sdrh    #define OR               3
94775897234Sdrh    #define PLUS             4
94875897234Sdrh</pre>
94975897234SdrhYou can insert a statement into the grammar like this:
95075897234Sdrh<pre>
95175897234Sdrh    %token_prefix    TOKEN_
95275897234Sdrh</pre>
95375897234Sdrhto cause Lemon to produce these symbols instead:
95475897234Sdrh<pre>
95575897234Sdrh    #define TOKEN_AND        1
95675897234Sdrh    #define TOKEN_MINUS      2
95775897234Sdrh    #define TOKEN_OR         3
95875897234Sdrh    #define TOKEN_PLUS       4
959*9a243e69Sdrh</pre></p>
96075897234Sdrh
9619bccde3dSdrh<a name='token_type'></a><a name='ptype'></a>
96275897234Sdrh<h4>The <tt>%token_type</tt> and <tt>%type</tt> directives</h4>
96375897234Sdrh
96475897234Sdrh<p>These directives are used to specify the data types for values
96575897234Sdrhon the parser's stack associated with terminal and non-terminal
96675897234Sdrhsymbols.  The values of all terminal symbols must be of the same
96775897234Sdrhtype.  This turns out to be the same data type as the 3rd parameter
96875897234Sdrhto the Parse() function generated by Lemon.  Typically, you will
96975897234Sdrhmake the value of a terminal symbol by a pointer to some kind of
97075897234Sdrhtoken structure.  Like this:</p>
97175897234Sdrh
97275897234Sdrh<p><pre>
97375897234Sdrh   %token_type    {Token*}
97475897234Sdrh</pre></p>
97575897234Sdrh
97675897234Sdrh<p>If the data type of terminals is not specified, the default value
977dfe4e6bbSdrhis "void*".</p>
97875897234Sdrh
97975897234Sdrh<p>Non-terminal symbols can each have their own data types.  Typically
980*9a243e69Sdrhthe data type of a non-terminal is a pointer to the root of a parse tree
98175897234Sdrhstructure that contains all information about that non-terminal.
98275897234SdrhFor example:</p>
98375897234Sdrh
98475897234Sdrh<p><pre>
98575897234Sdrh   %type   expr  {Expr*}
98675897234Sdrh</pre></p>
98775897234Sdrh
98875897234Sdrh<p>Each entry on the parser's stack is actually a union containing
98975897234Sdrhinstances of all data types for every non-terminal and terminal symbol.
99075897234SdrhLemon will automatically use the correct element of this union depending
99175897234Sdrhon what the corresponding non-terminal or terminal symbol is.  But
99275897234Sdrhthe grammar designer should keep in mind that the size of the union
99375897234Sdrhwill be the size of its largest element.  So if you have a single
99475897234Sdrhnon-terminal whose data type requires 1K of storage, then your 100
99575897234Sdrhentry parser stack will require 100K of heap space.  If you are willing
99675897234Sdrhand able to pay that price, fine.  You just need to know.</p>
99775897234Sdrh
9989bccde3dSdrh<a name='pwildcard'></a>
9999bccde3dSdrh<h4>The <tt>%wildcard</tt> directive</h4>
10009bccde3dSdrh
1001*9a243e69Sdrh<p>The <tt>%wildcard</tt> directive is followed by a single token name and a
10029bccde3dSdrhperiod.  This directive specifies that the identified token should
1003*9a243e69Sdrhmatch any input token.</p>
10049bccde3dSdrh
10059bccde3dSdrh<p>When the generated parser has the choice of matching an input against
10069bccde3dSdrhthe wildcard token and some other token, the other token is always used.
1007*9a243e69SdrhThe wildcard token is only matched if there are no alternatives.</p>
10089bccde3dSdrh
1009*9a243e69Sdrh<a name='error_processing'></a>
101075897234Sdrh<h3>Error Processing</h3>
101175897234Sdrh
101275897234Sdrh<p>After extensive experimentation over several years, it has been
101375897234Sdrhdiscovered that the error recovery strategy used by yacc is about
101475897234Sdrhas good as it gets.  And so that is what Lemon uses.</p>
101575897234Sdrh
101675897234Sdrh<p>When a Lemon-generated parser encounters a syntax error, it
1017*9a243e69Sdrhfirst invokes the code specified by the <tt>%syntax_error</tt> directive, if
101875897234Sdrhany.  It then enters its error recovery strategy.  The error recovery
101975897234Sdrhstrategy is to begin popping the parsers stack until it enters a
102075897234Sdrhstate where it is permitted to shift a special non-terminal symbol
10219bccde3dSdrhnamed "error".  It then shifts this non-terminal and continues
1022*9a243e69Sdrhparsing.  The <tt>%syntax_error</tt> routine will not be called again
102375897234Sdrhuntil at least three new tokens have been successfully shifted.</p>
102475897234Sdrh
102575897234Sdrh<p>If the parser pops its stack until the stack is empty, and it still
1026*9a243e69Sdrhis unable to shift the error symbol, then the
1027*9a243e69Sdrh<tt><a href='#parse_failure'>%parse_failure</a></tt> routine
102875897234Sdrhis invoked and the parser resets itself to its start state, ready
102975897234Sdrhto begin parsing a new file.  This is what will happen at the very
103075897234Sdrhfirst syntax error, of course, if there are no instances of the
10319bccde3dSdrh"error" non-terminal in your grammar.</p>
103275897234Sdrh
103375897234Sdrh</body>
103475897234Sdrh</html>
1035